METEOROLOGICAL MONOGRAPHS

EDITOR SENIOR ASSOCIATE EDITOR TECHNICAL EDITOR

JoHN A. DuTTON RONALD C. TAYLOR JOHN R. GERHARDT
Pennsylvania State Unh•ersity National Science Foundation American Meteorological Society

ASSOCIATE EDITORS
ROBERT C. BEARDSLEY JAMES R. HoLTON
Massachusetts Institute of Technology University of Washington

RICHARD E. ORVILLE JESSE J. STEPHENS

State University of New York at Albany Florida State University

•
METEOROLOGICAL MONOGRAPHS, a serial publication of the American Meteorological Society, serves as a medium for original
papers, survey articles, and other material in meteorology and closely related fields; it is intended for material which is better suited in
length or nature for publication in monograph form than for publication in the JouRNAL OF THE ATMOSPHERIC SciENCES, the JouRNAL
OF APPLIED METEOROLOGY, the JOURNAL OF PHYSICAL OCEANOGRAPHY, the MONTHLY WEATHER REVIEW and the BULLETIN OF THE
AMERICAN METEOROLOGICAL SociETY. A METEOROLOGICAL MONOGRAPH may consist of a single paper or of a group of papers
concerned with a single general topic.

•
INFORMATION FOR CONTRIBUTORS
Manuscripts for the METEOROLOGICAL MONOGRAPHS
should be sent directly to Dr. John Dutton, Editor, Department
of Meteorology, 503 Walker Building, The Pennsylvania State
University, University Park, PA 16802. Manuscripts may be
submitted by persons of any nationality who are members or
nonmembers of the Society, but only manuscripts in the
English language can be accepted. Every manuscript submitted
is reviewed and in no case does the editor advise the author as
to acceptability until at least one review has been obtained.
Authors will receive galley proof but not page proof.
tion headings, if needed, should be located at the beginning of
certain paragraphs and underlined.
7. References. References should be arranged alphabetically
without numbering. The text citation should consist of the name
of the author and the year of publication. Thus, "according to
Halley (1686)," or "as shown by an earlier study (Halley, 1686)."
When there are two or more papers by the same author published
in the same year, the distinguishing letters, a, b, etc., should
be added to the year.

Manuscripts. The manuscript must be complete and in final
form when submitted. It must be original typewritten copy on
one side only of white paper sheets 8\-2 x II inches, consecu-
tively numbered; double spacing and wide margins are essential.
Carbon copy and single spacing are not acceptable.
Each manuscript may include the following components,
which should be presented in the order listed. Of these, the
table of contents: title, author's name and affiliation; abstract;
text; references; and l~gends are obligatory.
I. Title page. This will be prepared by the editor if the
manuscript is accepted for publication.
2. Preface or foreword. A preface may be contributed by
the sponsors of the investigation, or by some other interested
group or individual. The preface should indicate the original
of the study and should present other facts of general interest
which emphasize its importance and significance.
3. Table of contents. Chapter, section, and subsection head-
ing should all be listed in the table of contents.
4. Title, author's name and affiliation. The affiliation should
be stated as concisely as possible and should not constitute
a complete address. The date of receipt of the manuscript is
supplied by the editor.
5. Abstract. This should summarize the principal hypoth-
eses, methods, and conclusions of the investigation. It should
not include mathematical symbols or references to equation
numbers, since the abstract is sometimes quoted verbatim in
abstracting or reviewing journals.
6. Text. For one of a group of papers which together consti-
tute a MONOGRAPH, it is sufficient to divide the text into
sections, each with a separate heading, numbered consecu-
tively. The section heading should be placed on a separate line,
flush with the margin, and should not be underlined. Subsec-
In the listing of references, each reference must be complete
and in the following form. For an article: author(s), year, title
of article, title of journal (abbreviated and underlined), volume
number, pages. For a book: author(s), year, title of book (under-
lined), city of publication, publisher, pages. Abbreviations for
journal titles should in general confirm to the·· List of Periodicals''
published by CHEMICAL ABSTRACTS.
8. Appendix. Essential material which is of interest to a
limited group of readers should not be included in the main body
ofthe text but should be presented in an appendix. It is sufficient
to outline in the text the ideas, procedures, assumptions, etc.,
involved, and to refer the reader to the appendix for fuller de-
tails. For example, lengthy and involved mathematical analyses
are better placed in an appendix than in the main text.
Illustrations. The illustrations should accompany the manu-
script and be in final form. Each figure should be mentioned
specifically in the text. Figure number and legend will be set
in type and must not be part of the drawing. A separate list of
legends should be provided. The following details should be
provided:
I. Drawings should be done entirely in black India ink. It
is often desirable to submit photographic copies of original
drawings, retaining the originals until the manuscript has been
accepted and is ready to go to the printer. If the drawings are
large, photographic copies should be no larger than 8\-2 x 11
inches to facilitate reviewing and editing.
2. The width of a figure as printed is 3ll.! inches or, less fre-
quently,~ inches. Original drawings are preferably about twice
final size.
3. Lettering must be large enough to remain clearly legible
when reduced; after reduction the smallest letters should not be
less than 1!t6 inch or I mm high.
Abbreviations and mathematical symbols. See inside covers
of the JOURNAL OF THE ATMOSPHERIC SCIENCES.
METROMEX: A REVIEW AND SUMMARY
 METEOROLOGICAL MONOGRAPHS

VOLUME 18 OCTOBER 1981 NUMBER 40

METROMEX: A REVIEW AND SUMMARY

Stanley A. Changnon, Jr.
Editor

Stanley A. Changnon, Jr., Richard G. Semonin,

August H. Auer, Roscoe R. Braham, Jr., and
Jeremy M. Hales
Contributing Authors

Published by the American Meteorological Society

45 Beacon St., Boston, MA 02108
© Copyright 1981, American Meteorological Society. Permis-
sion to use figures, tables and brief excerpts from this mono-
graph in scientific and educational works is hereby granted pro-
vided the source is acknowledged. All rights reserved. No part
of this publication may be reproduced, stored in a retrieval
system, or transmitted, in any form or by any means, elec-
tronic, mechanical, photocopying, recording, or otherwise,
without the prior written permission of the publisher.

ISSN 0065-9401
ISBN 978-1-935704-29-4 (eBook)
DOI 10.1007/978-1-935704-29-4

American Meteorological Society

45 Beacon Street, Boston, MA 02108
 TABLE OF CONTENTS

CHAPTER 1: Introduction
-STANLEY A. CHANGNON, JR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
CHAPTER 2: Surface Weather Conditions
-RICHARD G. SEMONIN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

CHAPTER 3: Urban Boundary Layer

-AUGUST H. AUER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

CHAPTER 4: Cloud Characteristics

-RICHARD G. SEMONIN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
CHAPTER 5: Urban Precipitation Processes
-ROSCOE R. BRAHAM, JR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

CHAPTER 6: Atmospheric Chemistry and Source-Receptor Relationships

-JEREMY M. HALES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

CHAPTER 7: Summary of Urban Effects on Cloud and Rain

-ROSCOE R. BRAHAM, JR., RICHARD G. SEMONIN,
AUGUST H. AUER, STANLEY A. CHANGNON, JR.,
and JEREMY M. HALES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

CHAPTER 8: Impacts of Urban Modified Precipitation Conditions

-STANLEY A. CHANGNON, JR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
APPENDIX: Management and Organization .................................... 179
 PREFACE

This report is the result of the interest and dedication of certain principal sci-
entists involved in a major national atmospheric research program. This program,
the Metropolitan Meteorological Experiment (METROMEX), was conceived and
planned in 1970 by the authors and others. Subsequently, the experiment was
conducted by the authors and scientists from other participating institutions.
In 1977, the concept and plan of this summary document for METROMEX
was evolved. The authors' pioneering and leadership roles in METROMEX pro-
vided the impetus and commitment for such an undertaking. Furthermore, the
many interesting findings emanating from the diverse METROMEX research ef-
forts were being published in a wide variety of journal articles and organizational
reports that defied practical overall assessment and integration of the findings.
Thus, there was a clear need and desire to review these findings and to assemble
a review of them in a single condensed summary document.
After I contacted the potential authors and got their agreement to pursue such
an endeavor, the Water Survey sought funding from the National Science Foun-
dation to cover costs of meetings of the authors, the typing of the final manuscript
and preparation of illustrations, and the publication charges. In the following two
years, 1977-79, the five authors of the chapters of this document have met, written
at least three drafts of all texts, reviewed the texts of the others, revised texts
accordingly, communicated frequently, and cooperated extensively. My role as
editor has been eased by the friendly interactions among the authors and their
great scientific dedication to this document. The support of our agencies and
universities in allowing these research and writing efforts has been extensive and
is greatly appreciated, as is the funding for the document from NSF under Grant
ENV77-15375.
The strong encouragement of Dr. William C. Ackermann, Chief of the Illinois
State Water Survey, has been a vital factor in the Water Survey's major involve-
ment in this effort. Dr. Donald L. Veal, Head of the Department of Atmospheric
Science at the University of Wyoming, also provided strong support for the pro-
grams.
Several of our staff members and associates have been helpful in the review
of the draft texts, including Bernice Ackerman, Donald F. Gatz, J. Loreena Ivens,
David Johnson and Harry T. Ochs III ofthe Water Survey. University of Chicago
staff contributors included Robert Kelly, Haya Berman and Julia Jacobs. A staff
contributor at the University of Wyoming included John M. White.
Secretaries of the Water Survey staff typed the manuscripts and the efforts of
Debra K. Hayn, Julia K. Lewis and Rebecca A. Runge are appreciated. The
drafting of more than 100 illustrations was done by John Brother of the Water
Survey and their high quality attests his skill.
Most importantly, the understanding and support of our families, in relation to
this time-consuming effort, often conducted in long periods at home, are gratefully
acknowledged. We dedicate this document to them.

STANLEY A. CHANGNON, JR.

December 1979
 CHAPTER 1

Introduction
STANLEY A. CHANGNON' JR.

CONTENTS

ABSTRACT 1
1.1 Overview 2
1.2 Background .................................................... ................... . 3
1.2.1 GENERAL EFFECTS OF URBAN AREAS ON WEATHER AND CLIMATE ................. . 3
1.2.2 URBAN-RELATED PRECIPITATION CHANGES ..................................... . 4
1.3 METROMEX evolves .................................................... ........... . 5
1.3.1 PLANS, PHASE 1 .................................................... ......... . 5
Climatological Studies of Urban Influences at St. Louis . . . . . . . . . . . . . . . . . . . . 6
1.3.2 PLANS, PHASE 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3.3 STUDY AREA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3.4 METROMEX GROUPS .................................................... ..... 10
1.3.5 INSTRUMENTATION AND OPERATIONS ........................................... 11
Surface Networks and Instruments ....................................... 11
Vertical Measurement Systems ........................................... 12
Meteorological aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
REFERENCES .................................................... ................... 15

ABSTRACT

This is a book about how a large metropolitan area in the humid continental climate zone of the central
United States affects the summer atmosphere and how these alterations change the weather and influence man.
A sizable multi-group field research effort was pursued with a wide variety of instruments over a 6-year period
to gather data, and this book assembles and reviews the key findings of METROMEX (Metropolitan Meteor-
ological Experiment).
The findings are organized in a deductive manner, beginning in this chapter with the climatological basis for
the choice of St. Louis as the study area for METROMEX. The apparent urban-induced changes in climate at
St. Louis are set in the general context of what has been learned elsewhere, largely from analyses of historical
data, about how cities influence weather and change the climate. This chapter also describes the research
strategy used for METROMEX including the organizational approach, the instrumentation and studies used to
address the issue.
This chapter thus sets the background and basis for understanding METROMEX and its findings. The ensuing
five chapters treat, in a logical sequence of effects, the findings about surface weather conditions (Chapter 2),
the urban boundary layer (Chapter 3), cloud characteristics (Chapter 4), precipitation processes (Chapter 5),
and atmospheric chemistry (Chapter 6). Chapter 7 assembles relevant findings into a summary of urban effects
on clouds and rainfall. Chapter 8 describes the impacts of the altered precipitation on man and the biosphere.
Each chapter is preceded by an abstract that summarizes its major findings.

1
2 METEOROLOGICAL MONOGRAPHS
1.1 Overview
METROMEX (Metropolitan Meteorological Exper-
iment) was a major national program designed and
conducted by several research groups to study inad-
vertent weather and climate modification by urban-in-
dustrial effects, and, in particular, man-made changes
of precipitation. Urban areas affect most forms of
weather and some, such as winds, temperature and
visibility, are obvious and their changes easily mea-
sured. Inadvertent precipitation changes are harder to
measure and explain, and prior to METROMEX, ur-
ban-related rain changes had had only limited study.
The general program goals were 1) to study the ef-
fects of urban environments upon the frequency,
amount, intensity and duration of precipitation and
related severe weather; 2) to identify the physical pro-
cesses of the atmosphere which are responsible for
producing the observed urban weather effects; 3) to
isolate the factors of the city complex which are the
causative agents of the observed effects; and 4) to as-
sess the impact of urban-induced inadvertent weather
changes upon the wider issues of society.
Study of climatic data of St. Louis by Huff and
Changnon (1972) suggested summer increases in the
immediate downwind area of 1) rainfall (10 to 20%);
2) moderate rain days (10 to 25%); 3) heavy rainstorms
(80%); 4) thunderstorms (20%); and 5) hailstorms
(30%). These increases were in relation to values in
areas west of St. Louis and changes with time. For
this and other reasons, St. Louis was chosen as the
site for the field study portion of METROMEX.
A host of instruments was used to study the summer
weather including 250 recording raingages and hail-
pads, 5 weather radars, 70 rainwater collectors, 15
pibal stations, 5 meteorological aircraft, 5 radiosonde
stations, unique atmospheric tracers, and a wide va-
riety of other standard and unusual meteorological
equipment. These measurement tools were used to
provide information on 1) the processes of cloud and
precipitation formation, 2) the chemistry of aerosols
and rainwater, 3) the urban heat budget, 4) the three-
dimensional portrayals of precipitation elements, and
5) the airflow and cloud development for numerical
models. After eight years of data collection, analysis
and study, a substantial amount of the information
sought about urban weather and climate has been ob-
tained and is summarized herein.
The identification of urban weather effects and sort-
ing out their causes is difficult. Local physiographic
features such as hills and wetlands, as well as features
of urban-industrial areas such as the urban heat island,
surface roughness, and gaseous and particulate emis-
sions, must be considered as potential causal agents
for modifying clouds and rainfall. The problem is to
sort out the relative contributions of these potential
causal agents in the face of the natural variability of
weather systems. Success in establishing cause and
effect relationships is directly related to the frequency
and universality with which the effect occurs, the ex-
tent to which the effect can be linked to the cause
through physical-mathematical models, and our ability
to identify and measure the key variables involved.
In METROMEX, urban effects on weather and the
ensuing climate changes having all degrees of com-
plexity were studied. Although the urban heat island
was well documented before METROMEX, it was
thoroughly studied at St. Louis to provide input to
studies of more complex phenomena. Urban effects
upon the boundary layer winds, cloud microstructure,
and rain initiation were occasionally discussed before
METROMEX; but now they have been documented
to the point that they can be considered as established
fact. Urban effects upon rainfall and severe storms are
the most complex of all urban weather effects, and, at
present, the causes for them can only be partially
understood.
One of the most difficult issues is whether "urban"
effects on rainfall could have arisen purely from
chance placement of events (storms) over the raingage
network in a 5-year period. This leads to questions
concerning sample sizes and statistical independence
of data, which can be very troublesome questions in
projects such as METROMEX. In studies where we
have several measurements of a parameter on each of
several days, such as radar first echoes and raincells,
we have tested each against the null hypothesis of no
effects across days, assuming independence within
days-a questionable assumption. In our view, how-
ever, the strongest evidence for urban effects upon
rainfall and severe storms lies not in the strict statis-
tical evidence, but in the cohesion of many different
kinds of evidence. We hope that ultimately we will
have soundly based cause-effect models linking ob-
served initial conditions to observed final conditions
through a well-documented chain of physical argu-
ments.
As METROMEX findings are woven into the fabric
of atmospheric science, and the results from St. Louis
are compared with those from future studies at other
cities, the present hypotheses will be sharpened and
made more certain. We can also anticipate further
contributions from METROMEX, for although this
report represents the formal conclusion of the project,
the data bank will continue to provide data for study
for many years to come.
Initially, scientists from the Argonne National Lab-
oratory, University of Chicago, Illinois State Water
Survey, and University of Wyoming formed the nu-
cleus of the 8-year cooperative scientific program to
METROMEX: A REVIEW AND SUMMARY
study the inadvertent modification of weather by an
urban-industrial complex. Emphasis of the study was
on urban-related alterations in precipitation processes,
quantitative changes in surface precipitation and allied
weather conditions, and on the resulting impacts.
Consideration of the available climatological results,
available resources, reasonable access to the urban
atmosphere by meteorological research aircraft, and
the location of the nucleus research groups resulted in
the choice of the St. Louis area as the site of the field
project of METROMEX. The field project was the
major effort of METROMEX, but the program also
involved laboratory and atmospheric modeling proj-
ects and the all-encompassing research. Program plan-
ning began in 1969; most field operations and data col-
lection occurred in the summers (June-August) from
1971 through 1976; and final analyses have been in
progress since 1975.
METROMEX was designed and managed by the
participating scientists. Each sought funding from
state and federal agencies and the project was open-
ended-anyone interested in the subject was wel-
comed to focus his research at St. Louis. As a result,
scientists from 12 groups participated in METRO-
MEX.
1.2 Background
Man has suffered the excesses and enjoyed the plea-
sures of weather throughout his existence on Earth,
but it has only been in the last 150 years that he has
begun to affect the weather significantly and persis-
tently. Although man for many centuries has specu-
lated on his effect on weather through such activities
as major battles, incantations, large fires, and atomic
explosions, the major means whereby he has affected
weather has been accidental and through his urban-
industrial developments. More than 700 years ago the
London urban complex had achieved a size such that
it produced a recognizable effect on its local climate,
at least in terms of reduced visibility and increased
temperature (Chandler, 1970). Serious scientific at-
tempts by man to modify weather have been confined
to the last 30 years, and have been limited to a few
localized projects indicating precipitation increases
(and decreases) and fog dissipation. His intentional
efforts have yet to produce major, widespread changes
in rain.
The increase in urbanization that began 200 years
ago with the industrial revolution has led to microscale
and mesoscale changes in the weather and climate in
and near urban areas in the midlatitudes (Peterson,
1969; Landsberg, 1970). Considerable scientific atten-
tion has been directed to this urban problem in the
European area during the last 100 years (Kratzer,
3
1956). As major urban-industrial complexes became
prevalent in many countries, worldwide attention to
this problem developed rapidly in the last 20 years.
Furthermore, the development of megalopolises in the
United States during the past 10 to 30 years has
brought with it increasing public and scientific aware-
ness of the existence and, in some cases, the serious-
ness of urban effects on weather leading to demon-
strable local changes in climate (Weather Modification
Advisory Board, 1978).
1.2.1 GENERAL EFFECTS OF URBAN AREAS ON
WEATHER AND CLIMATE
The weather changes wrought by urbanization in-
clude most observable weather conditions. They may
be categorized according to two classes: the obvious,
or easy to measure, and the subtle, or hard to measure.
The more obvious effects of urban areas on weather
include such things as decreased visibility resulting
from contaminants in the air, reduced winds and in-
creased air temperature.
Certainly, many are aware that in a major complex
the visibility is more frequently restricted than in rural
areas. Smog is now a household word. Most Ameri-
cans also are aware that the temperature in small cities
as well as large urban areas is generally higher at any
given time than it is in nearby rural areas. The tem-
perature effect has been well recognized and well mea-
sured for many years because its direct measurement,
at least at the surface, is easily accomplished. Urban
areas act as an obstacle to decrease winds near the
surface, to increase turbulence and vertical motions
in the atmosphere above cities, and to create occa-
sionally a localized rural-urban circulation pattern
(Landsberg, 1970).
The more subtle, less recognized weather changes
wrought by urban areas include alterations of foggi-
ness (as opposed to smog), cloudiness, rainfall, snow-
fall, solar radiation, humidity, atmospheric electricity,
severe weather events (such as thunderstorms, hail or
severe rainstorms), and certain mesoscale synoptic
weather features. There have been sufficient descrip-
tive studies to reveal that these more difficult-to-mea-
sure weather conditions are being changed, some dra-
matically, by urban complexes.
The degree of change in any of these weather ele-
ments at any time is dependent upon several factors.
Some of these are the areal extent of the urban com-
plex, the components of the industrial complex, its
juxtaposition to major water bodies and other topo-
graphic features, time of day, season of year, existing
weather conditions and climate. The degrees of cli-
mate change reported at various American and Eu-
ropean cities have been compiled and used to develop
4 METEOROLOGICAL MONOGRAPHS

tion downwind and 50 km east of the Chicago-Gary

TABLE l. Northern Hemisphere results showing average urban
climate changes expressed as percent* of rural conditions (from
area at La Porte, Indiana (Stout, 1962).
Changnon, 1976).
All of the precipitation-related conditions at La
Annual Cold season Warm season Porte underwent thorough study in the 1965--67 period
(Changnon, 1968). This study found rather significant
Contaminants +1000 +2000 +500
Solar radiation -22 -34 -20 warm season increases in total precipitation and in the
Temperature ("C) +2 +3 +I frequency of rain days, thunderstorm days and hail
Humidity -6 -2 -8
Visibility -26 -34 -17 days.
Fog +60 +100 +30 The La Porte results met with considerable national
Wind speed -25 -20 -30 interest, both in the scientific community and by the
Cloudiness +8 +5 +10
Rainfall +14 + 13 +15 general public. Scientific interest revolved around the
Snowfall ±10 ±10 fact that the inadvertent precipitation increases had
Thunderstorms +15 +5 +30
particular bearing upon the general areas of weather
*Except for temperature. modification and air pollution (probably the two great-
est current interest areas in meteorology).
Certain meaningful reactions and observations re-
sulted from the presentation of the La Porte results.
the average percentage changes (urban values versus
First, there was the previously mentioned widespread
rural values) shown in Table 1 (Changnon, 1976).
recognition and interest. Although there was wide-
Naturally, large deviations from these averages can
spread acceptance of the validity of the results, there
occur for a given month, day or hour. For instance,
also was some questioning of the findings, a condition
solar radiation received in an urban area is decreased
not unexpected with the subject of weather modifi-
much more in winter than in summer. The diurnal var-
cation. These reactions and counter-reactions (Ogden,
iation of the urban-rural temperature difference is par-
1969; Holzman and Thorn, 1970; Changnon, 1970;
ticularly pronounced, and may exceed soc in the early
Changnon, 1971) emphasized the need to evaluate in-
morning at major cities.
advertent modification of the weather in the assess-
ment of the environment. Other data such as runoff,
1.2.2 URBAN-RELATED PRECIPITATION CHANGES
crop yields and tree rings were studied and the results
Several European climatic studies of the 20th Cen- generally supported the reality of the precipitation
tury were concerned with added cloudiness, precipi- anomaly at La Porte throughout the 1960's (Changnon
tation and severe weather within and downwind of and Huff, 1977).
certain cities (Ashworth, 1929; Kratzer, 1956). How- Climatological-statistical investigations similar to
ever, modification of precipitation conditions in the that at La Porte were subsequently pursued for other
United States through urban-related influences was major urban-industrial complexes to determine whether
not noted until the 1950's when Landsberg (1956) pre- urban effects on precipitation existed elsewhere. Re-
sented indications of urban-produced rain increases at sults from studies of nine cities are presented in Table
Tulsa. It should be noted that North American weather 2. Seven cities showed increases in summer rainfall,
observers had reported clouds developing from large thunderstorms and hailstorms (Changnon, 1976). These
forest, industrial and prairie fires as early as 1900 cities reflect influences in different climates, city sizes,
(Broeck, 1900; Mitchell, 1901). basic industrial types and urban growth rates. Expla-
Some of the primary reasons for the lack of aware- nations offered for these increases included 1) urban-
ness and study of urban influences on precipitation
were a general lack of meteorological expertise in the
United States, and, more importantly, the lack of TABLE 2. Maximum urban-rural differences in summer rainfall
weather station networks adequate for measuring and and severe weather events, expressed as percent of rural values,
based on available historical data (from Changnon, 1976).
small-scale precipitation changes and defining other
mesosynoptic weather conditions. Thus, the rather lo- City Rainfall Thunderstorms Hailstorms
calized nature of the precipitation changes went un-
detected. St. Louis + 15 +25 +276
Chicago +17 +42 +246
The Illinois State Water Survey initiated studies of Cleveland +27 +38 +90
possible urban-industrial effects on precipitation pro- Indianapolis 0 0 0
Washington, DC +9 +36 +67
cesses in 1958 with the study of two middle-sized ur- Houston +9 +10 +430
ban areas in Central Illinois (Pearson, 1958; Chang- New Orleans +10 +27 +350
non, 1962). Further climatological investigations in the Tulsa
Detroit +25
0 0
No data
0
No data
early 1960's revealed a startling increase in precipita-
METROMEX: A REVIEW AND SUMMARY 5

related increases in active cloud condensation nuclei urban weather modification was obviously too large
and ice nuclei, both important in the cloud and rain- and complex to be handled effectively by any one of
drop formation; 2) roughness of the city resulting in the four groups. Moreover, the complementary nature
low-level convergence and increased upward vertical of the strengths of these groups made it apparent that
velocities; 3) urban heating (due to altered radiation much could be gained by pooling efforts toward a few
and combustion processes) which would also lead to common objectives.
upward motions of the air to initiate or enhance con- The investigative conglomerate of METROMEX
vection; and 4) additional moisture from industrial was unique in that it included groups from state uni-
processes (cooling towers). versities, a state research institution, a private uni-
versity, and federal laboratories, all acting coopera-
1.3 METROMEX evolves tively to establish their own measurement and study
areas within the total plan. The management and or-
The urban-rain climatological studies and ensuing
ganizational aspects of METROMEX and the groups
scientific questions made it clear that measurements
that participated are described in the Appendix.
of precipitation processes in urban areas were essen-
A flow diagram designed in 1970 for the approach
tial to evaluate, substantiate and describe the potential
used in METROMEX is shown in Fig. 1. The project
causes of urban-induced precipitation. The formation
adhered closely to this approach. The climatological
of precipitation involves a complicated system of in-
studies at St. Louis carne first, as part of the choice
teractive physico-chemical and fluid-dynamical pro-
of the field site. The first reported urban weather effect
cesses, and depends critically upon values of many
at St. Louis was in relation to the urban heat island
meteorological variables even prior to the formation
during 1891-95 (Hammon and Duenckel, 1902). The
of clouds. Hence, only definitive field studies involv-
temperature difference between the city center and the
ing a complex set of measurements, both at the surface
nearby rural Forest Park (located 6 krn west of down-
and in the low-to-middle atmosphere, can furnish the
town St. Louis) in summer was 0. 7°C for the maximum
information needed to establish the connection be-
temperature and 1.2°C for the minimum temperature.
tween urban effects and precipitation increases, and
By the way, this compares favorably with the MET-
to ascertain the cause or causes for these increases.
Thus, any substantial increase in knowledge about ur-
ban weather effects required two things: 1) much more
complete surface weather data; and 2) physical mea-
surements and experiments in and around clouds af-
fected by the urban-industrial complex. Therefore, an
urban-weather research program focusing on a field
project adequate to make the necessary measurements
was planned and organized by the four atmospheric
research groups. The site to be chosen for the field
project had to be where urban climatic studies had
identified the existence of an apparent urban effect on
precipitation.

1.3.1 PLANS, PHASE 1

Scientists from the Argonne National Laboratory,
University of Chicago, and the University ofWyorning
began discussions with scientists from the Illinois
State Water Survey during the summer of 1969 to de-
velop specific scientific plans for an urban-rain re-
search program. Consideration of the available cli-
matological results, available resources and the
locations of the four research groups resulted in the
TRANSLATE
choice of the St. Louis area as the site of the field RESULTS
TO OTHER
project, the major effort of the program. Through a CIT! ES
series of meetings which ended in March 1970, a com-
prehensive project plan was evolved. This plan iden-
tified all measurements considered essential to the FIG. I. Outline ofMETROMEX, as planned in 1970. After Chang-
study of urban effects on precipitation. The subject of non et al. (1971).
6 METEOROLOGICAL MONOGRAPHS

CONTROL .,\U I
EFFECT) I
I
• I
I

0 • STATIO'!US[D TO CAl.Ctl.ATEURSA:.Vltt.U[

;.~Al..i.
OF r:.l t.at.r~Ut.~
E: X P l ;. 'l A T ! 0 1l
FIG . 4. Pattern of ruraVurban ratios for number of days with rain-
0 8 1(, 24 J2 40
Q• CPITRAL UR~A!iSTATIIJ,IS USED TO DHH OP URJA'I 'iAlUE fall ;;.6 em during 1949--68, at St . Louis . After Huff and Changnon
e--3 ---- ....-.-.
• : STAT!Oi US(O Iii l>l\'( LO?I;lG SU:cao: ~ AVCR.AGtS {loilble 3) (1972).
FIG . 2. Pattern of average ruraVurban ratios(%) of summer rain-
fall in the St. Louis area, 1949--68, with urban based on two stations.
After Huff and Changnon (1972).

0 ·~~L
I I I I I
4 1----:-:19"='4s=-----:-:
19":-49::-----:-:19':csJ:-----:-:
19':cs7:-----:-:
19':-61,.------::-::
l9~65
START OF 5-YEAR PERIOO

FIG. 3. Five-year moving averages of ratio of Centreville (rural

area east of St. Louis) to St. Louis summer rainfall, 1941--69. After FIG . .5. Frequencies of thunder days when isolated thunderstorms
Huff and Changnon (1972). occurred only over the St. Louis area during 1950-58.

ROMEX 1971-75 differences between the city center Figs. 2-6 and Table 3 (Huff and Changnon, 1972). Es-
and the western rural areas though those were now sentially, the St. Louis data show a downwind in-
located 30 km or more west of St. Louis (Hilberg , crease in warm season rainfall, frequency of rain days
1978). and heavy rainstorms, thunderstorm occurrences, and
hail-day frequencies, all probably urban produced.
Climatological Studies of Urban Influences at St. However, little climatic evidence was found of urban
effects on cold season precipitation in the St. Louis
Louis
area.
Selected results from recent climatological studies Fig. 2 shows the average pattern of summer rainfall
of precipitation changes at St. Louis are illustrated in (June-August) in the St. Louis areas for the 1949-68
METROMEX: A REVIEW AND SUMMARY 7
15.------------.------------.------------.
Louis urban-industrial complex (Fig. 2). This down-
wind high began to appear during the 1945-55 period
as illustrated in Fig. 3, in which the rural/urban ratios
are shown for 5-year moving averages at Centreville
(see Fig. 4) near the center of the rainfall high. In a
stochastic modeling analysis of changes in the mean
and trends of the annual precipitation at six climatic
stations in St. Louis area with long records (1916-70),
Rao ( 1979) concluded that the Edwardsville precipi-
tation values show an increasing trend that is signifi-
cant at the 95% level. The other five stations in and
west of St. Louis did not. Rao concludes that the Ed-
wardsville results indicate that the change in precipi-
tation occurred very gradually and in the period after
1940.
Table 3 summarizes the urban effect upon warm
1944 1949 1954 1959
START OF 5-YEAR PERIOD season rain-day frequencies in the St. Louis region.
FIG. 6. Five-year average frequency of hail days per year in areas The rain-day increases for the 2.5 mm and higher
upwind (west) and downwind Gust east) of St. Louis (based on three classes for the Major Effect Area are similar to those
stations in each area) in summer. After Huff and Changnon (1972). found earlier at La Porte (Huff and Changnon, 1970).
The warm season increases were most notable when
TABLE 3. Point average rain-day frequencies in St. Louis area, rain days were measured at the 6.4 mm level and
1951 through 1968 1 (from Huff and Changnon, 1972). higher, indicating that the St. Louis urban effect(s) is
likely one of rain enhancement in existing, moderately
Warm season (April-October) frequencies of
days in each area strong convection systems. This suggestion is sup-
ported further by the fact that sizable local increases
Daily rainfall uc• MP MN De• in heavy rainfall occurrences, thunderstorm days and
;;.0.25 mm 1030 1060 1035 966 hail days were found. Dettwiller and Changnon ( 1976)
;;;.2.5 mm 731 716 730 713 found that the maximum daily warm season rainfall
;;;.6.4 mm 469 492 506 480
;;.!2.8 mm 272 277 294 272 values in St. Louis had increased by 37% from 1872
to 1954 as the city population grew from 300 000 to 1.3
1Point average values based on four stations in each area, as des-
million. After the station was relocated on the north-
ignated in Fig. 2.
2 UC = Upwind control area; MI =combined minor effect areas west edge of St. Louis (Lambert Field) in 1955, an
(see Fig. 2); MA =major effect area; DC = downwind control area. increase could no longer be detected.
Urban effects on severe weather events were ap-
period, expressed in terms of percentage of the urban parent in the climatic data of the St. Louis region. The
summer total. That is, the ratio of the average summer ratio map of Fig. 4 shows that the frequency of heavy
rainfall at each station to that at the two most central rainstorms (days with ;;;;.6 em) is much greater in the
(city and St. Louis University) urban stations (rural/ Major Effect Area where 80% more occur, on the av-
urban ratios) was calculated and mapped. In the back- erage, than at the two urban stations used to calculate
ground, the pie-shaped areas extending 25 miles east- the urban frequency. A comparison between the fre-
ward from the city represent the area in which we quency of thunderstorms (Fig. 5) occurring only in the
thought an urban effect would most likely occur (la- urban area (city station in downtown St. Louis) and
beled the Major Effect Area) based upon climatolog- those only in the upwind control (rural) area northwest
ical considerations of storm motions and durations of the city (Lambert Field) indicates the urban envi-
(Huff and Changnon, 1972). The city area, plus those ronment is instrumental in the development of summer
extending north-northeast and south-southeast from thunderstorms. Fig. 6 allows comparison of summer
the city, comprise a Minor Effect Area where urban hail-day occurrences between the Major Effect Area
effects were considered to be less pronounced. Two and the Upwind Control Area (based on 3 stations in
no-effect areas, the Upwind Control Area extending each area). Enhancement of hail occurrences east of
westward from the city and the Downwind Control St. Louis is apparent beginning in the late 1940's.
Area extending eastward from the Major Effect Area, The historical data on rainfall, heavy rain events and
were identified for comparative analyses. hail collectively reveal a gradual temporal increase in
Of particular interest is the 10 to 20% higher summer convective precipitation in and east of St. Louis (rel-
rainfall that exists in an area 5 to 25 km east of the St. ative to that elsewhere) for many years. Taken alone,
8 METEOROLOGICAL MONOGRAPHS
these findings are not conclusive evidence of the urban
effects on summer precipitation. However, all avail-
able climatic evidence, plus that found on local in-
creases in streamflow and crop yields (see Chapter 8)
suggest that local influences on summer rainfall were
sufficient to become detectable during the 1940-55 pe-
riod. The results of the METROMEX studies gener-
ally show that urban effects on cloud and precipitation
processes are related to city size, and as St. Louis
grew, the amount of urban influence on the atmo-
sphere and rain processes grew from no effect to a
detectable one. The detection of an urban influence
depends so much on the very limited historical data
that an exact place and time of detection cannot be
realistically expected. However, the climate changes
detected in the 1940-55 period were at the time when
the contiguous metropolitan region exceeded 100
square miles (260 km 2 ) in area and one million in pop-
ulation, both of which occurred in the 1940's. These
temporal changes in rain, hail, streamflow, crop yields
and city size tend to counter arguments that localized
rain changes are due to topographic influences. Top-
ographic and marine influences on local weather can
produce weather changes as sizable as those produced
by a major city. Hence it is difficult to isolate urban
effects at many cities along the coast of major water
bodies or those in hilly or mountainous terrain. St.
Louis represents one of the few major U.S. metro-
politan areas with only minimal topographic variations
and no marine influences. However, only moderate
terrain changes in certain flatland sites can induce con-
siderable atmospheric effects and this possibility had
to be reckoned with in the St. Louis studies.
1.3.2
PLANS, PHASE 2
Since the St. Louis climatic studies ongoing in
1969-70 answered the first question in the flow dia-
gram of Fig. I with a "yes," the program planning
proceeded to the field project. Concern over potential
impacts of urban related rain anomalies led to various
studies in the social and other physical sciences. This
was noted by the inclusion of allied interdisciplinary
research in the program (see Appendix). Importantly,
the research groups also recognized the need to pro-
vide results from the field project that would be trans-
latable to other cities through modeling research dur-
ing METROMEX.
Climatological research results indicated that a 5-
year study period would be minimal to collect data
adequate to define the phenomena, and the plan called
for the initiation of the METROMEX field project in
1971. The primary operational period in summer was
dictated by the climatological results which indicated
local rain changes during summer. Some data were
collected in the other seasons, both primarily from the
raingage network operated continuously and from se-
lected field sorties by project aircraft in winter.
1.3.3 STUDY AREA
There were several reasons for selecting the St.
Louis metropolitan area as the site of METROMEX.
It was large enough to measurably affect the atmo-
sphere, but small enough to be encompassed by a proj-
ect of practical size. It was relatively free of topo-
graphic influences and the effects of nearby large
bodies of water, and it had been the subject of prior
meteorological studies and possessed a documented
meteorological history (its climate is described in
Chapter 2). It had a large and diverse industrial base,
typical of those found in many North American urban
centers.
St. Louis is located a few kilometers south of the
confluence of the Missouri and Mississippi rivers (Fig.
7). Its metropolitan area is characterized by relatively
flat terrain at an elevation of 150 m (500 ft) msl, sur-
rounded by predominantly agricultural, rolling terrain.
The topography is shown by 200ft (61 m) intervals of
altitude above mean sea level in Fig. 7. With the ex-
ception of the extreme west-southwest edge of the re-
search circle, very little change in the terrain occurs.
There are sharp low bluffs along the Mississippi River
north-northwest of St. Louis on the Illinois side and
south of St. Louis on the Missouri and Illinois sides.
Elevations between downtown St. Louis and the sub-
urban areas differ by no more than 160 ft (50 m); the
highest terrain in the research area is located 40 km
to the southwest of the city, where the land rises to
about 1100 ft (340m) msl into the low Ozark foothills.
About 35-40 km northwest of St. Louis is a large flat
floodplain between the Missouri and Mississippi riv-
ers. A more detailed topographic map is available in
Wong and Dirks (1978).
The five major land use characteristics of the study
area are shown in Fig. 7. Most of the land in the MET-
ROMEX circular study area is devoted to agriculture
(row crops). There are essentially four cultural-phys-
iographic areas around the metropolitan area. South-
west of St. Louis in the hills are widespread wooded
areas. To the northwest is a large river bottomland
containing farms and small wooded tracts. Along the
river east of St. Louis is a large floodplain with many
industries, suburbs, open swampy spaces and farms.
Names of the principal cities are shown in Fig. 8. East,
northeast, and southeast of the floodplain is low rolling
farmland. Vegetative cover for downtown St. Louis,
and the areas of compact single-space and multi-family
dwellings located within 10 km northwest and south-
west of the downtown area is only 25% (Auer, 1978).
METROMEX: A REVIEW AND SUMMARY 9

TOPOGRAPHIC CONTOUR
INTERVAL IN FEE T
(200 ft • 61 m)
LAND USE
AGRICULTURAL

>S~ INOUSTRIAL/sq km

Q >5~ RES I DENTIAL/sq l<m

ffi >SOt COI'I1ERI CAL & URB AN

RES I DENT! AL/ sq km
SCALE Of KI LQ,.,..ETt;R,!;i
8 lG 24 J2

FIG. 7. Topographic features and major land use types within the METROMEX research area.

Much of the remaining metropolitan area consists of north at Alton-Wood River (Fig. 8). However, there
large residential areas with more than 50% vegetative are many light industries in St. Louis, to the west of
cover. Although St. Louis and the circular area shown the city, and near St. Charles. The major power plants
on Fig. 7 formed the primary study area, the natural are located 5 to 25 km away and around the city to the
west-to-east movement of weather made it necessary west, north and south.
to study precipitation to the east and northeast of the Because the metropolitan St. Louis area is so iso-
circle. A more detailed land use map of the metropol-
itan area is shown in Auer (1978).
According to 1970 U.S. Census Bureau statistics,
the city af St. Louis had a population of 622 236 (ranks
18th nationally) with a population density of 3938 peo-
ple per square kilometer (ranks 7th). U.S. Census Bu-
reau figures showed a population of2 363 017 for met-
ropolitan St. Louis in 1970. Estimates of the 1975
urban population by the Census Bureau place the met-
ropolitan total at 2.4 million with the city at 532 000.
Industrial areas are located primarily along the Mis-
sissippi River (Fig. 7). St. Louis is a typical major
American industrial city. It has more than 1000 indus-
tries with great diversification (Koepke, 1970). There
are major steel mills, a large complex of refineries,
metal processing plants, several sizable power plants,
and a wide assortment of manufacturing plants. St.
Louis is also a major transportation center with 18
railroads, ranking second only to Chicago in surface
traffic flow in the United States. The heavy industry Kll..oMt..'TER5

is concentrated along, and largely east, of the Missis- 0 0

r::::--=r----=::>
0 s
lf.
10
sippi River, both to the east of St. Louis and to the M!l.l·:.'?
:>1'A'JVTE

east of the other, separate industrial center 15 km FIG . 8. Principal cultural feature s in the St. Louis a rea.
10 METEOROLOGICAL MONOGRAPHS

Pollutants at Granite City (the heart of the heavy in-

• Strong condensation
nucl ei
~ Zinc
<) Lead
0 !~anganese
FIG. 9. Principal sources of Zn, Mn, non-vehicular Pb, and con-
densation nuclei.
lated (the nearest neighboring city of 50 000 or more
population being at a distance of 135 km), its impact
on regional air quality is relatively easy to identify.
METROMEX research aided the EPA in identifying
principal sources of several pollutants and condensa-
tion nuclei (Fig. 9). These reveal the industrial con-
centration along the river.
Monitoring of the nation's most common five air
pollutants in the St. Louis area during the main MET-
ROMEX field period (1971-75) by local air pollution
agencies provided information on the temporal and
spatial variations of local air quality (Southwestern
Illinois Metropolitan and Regional Planning Commis-
sion, 1976). The annual and summer mean values of
atmospheric particulates, S02 , N0 2 , and CO at 3 cities
(see their locales on Fig. 8) are shown in Table 4.
dustry sector with a major steel mill) are the area's
highest except for the S02 values for which the Wood
River mean value is the greatest, likely as a result of
the nearby refinery complex and power plants. Pol-
lutant values at suburban Belleville are sizably lower.
The national air quality standard for suspended par-
ticulates is 75 p.,g m-3 , and the 1971 through 1975 Gran-
ite City and Wood River values exceeded the stan-
dard. No other pollutant means exceeded standards.
The particulates, S02 and ozone all showed their max-
imums in summer, with cold season maximums for
N02 and CO. Motor vehicles provide -75% of the St.
Louis CO, and they, plus power plants, are the pri-
mary sources of the N02 • Four of the five air pollu-
tants (S0 2 , N02 , CO and ozone) increased during the
1971-75 period, with increases ranging from 40% in
0
~t£T£RS
KH..O
B 16
S02 , to 60% in N02 and CO. Particulates decreased
0 5 10
25% between 1971 and 1975.
S1'ATUT£ MILES
1.3.4 METROMEX GROUPS
The management and funding of METROMEX were
unique in that each group secured its own funding from
whatever private, state or federal sources that wished
to furnish support. This meant that each group had to
have one or more research projects that were viable
scientific experiments even if all other projects failed
to be funded or accomplished. However, the conglom-
erate believed that the research results would be more
than the sum of the parts because of the synergism of
the combined projects.
Each of the four original planning groups identified
specific portions of five main measurement areas of
the field project shown in Fig. 1. The major research
areas and efforts of each group are shown in Table 5
along with those of eight other groups which became
involved. More information on the organizations who
participated in METROMEX and their programs is
presented in the Appendix.

TABLE 4. Major air pollutants at various locales in 1975.

Mean suspended Mean sulfur Mean nitrogen Mean carbon

particulates dioxide dioxide monoxide
(J.tg m-3) (ppm) (ppm) (ppm)
Locales 1971 Annual Summer Annual Summer Annual Summer Annual Summer

Granite City (heavy 170 175 0.012 0.012 0.028 0.028 1.1 1.1
industry)
Wood River (refinery 85 112 0 .023 0.024 0.021 0.020 0.9 0.4
area)
Belleville (downwind 65 74 0.005 M 0.007 M 0.4 M
suburb)
M = Missing.
METROMEX: A REVIEW AND SUMMARY 11

TABLE 5. Research areas of METROMEX and principal groups
involved in each area.
Participating
Research area groups*
Assessment of weather elements ISWS
by surface networks
Assessment by radars ISWS, UC, NOAA
Statistical evaluation of ISWS, UC
precipitation
Cloud and precipitation UC, UW, ISWS, UMR, DRI
formation from nuclei
measurements
Cloud and precipitation UW, UC, ISWS
characteristics
Modeling ISWS, UW, UC
Heat budget UW, UC, SRI, ISWS
Airflow measurements ANL, UW, ISWS, NOAA
Atmospheric chemistry ANL, ISWS, PNL
Impact research ISWS, UW, IBS, UA, UCSD
*ISWS = Illinois State Water Survey; UC = University of Chi-
cago; ANL = Argonne National Laboratory (AEC); UW = Uni-
versity of Wyoming; IBS = Institute of Behavioral Sciences; UA
= University of Arizona; PNL = Battelle Pacific Northwest Lab-
oratory; UMR = University of Missouri at Rolla; SRI = Stanford
Research Institute; NOAA = National Oceanic and Atmospheric
Administration; UCSD = University of California at San Diego;
DRI = Denver Research Institute.
1.3.5 INSTRUMENTATION AND OPERATIONS
Detailed descriptions of the field operations and in-
strumentation appear in a series of five annual project
reports. 1 An important facet of the 5-year field pro-
gram, unique to most METROMEX participants, was
rapid data processing and a quick initial analysis to
inspect the results of each summer. A primary reason
for this approach was to continually refocus the field
operations and analyses as new findings of interest
developed. In other words, the experiment over five
years was kept as flexible as possible to adapt data
collection to new findings. For the ISWS, UC and
UW, several major shifts in research and operational
emphasis resulted.
Surface Networks and Instruments
The scope and size of the METROMEX effort to
meet the goals of the 5-year program were sizable, as
reflected in all facets of instrumentation utilized by the
program. The data compiled as input for METRO-
MEX investigations were gathered from more than 700
instruments ranging in sophistication from simple hail
sensors (hailpads) to instrumented aircraft.
Central to the METROMEX instrumentation array
was the raingage and hailpad network operated by the
ISWS. Their previous consideration of results from
'METROMEX Annual Operational Reports for 1971, 1972, 1973,
1974 and 1975; copies available at the ISWS, Box 232, Urbana,
Illinois.
studies of radar echoes and weather modification led
to a design having a circular study area, one that was
at least the length of a thunderstorm's typical duration-
length of 50 km, and a network centered on the forcing
function, the city. Coupling of these size and shape
designs with the desired gage density and the available
resources resulted in a network with 225 recording
raingages and hailpads (Fig. 10) in an 80 km diameter
circle encompassing 5540 km 2 •
The raingage and hailpad network of 1971 was ex-
tended eastward in 1972 (see Fig. 10) to include an
additional 4403 km2 with one raingage per 210 km 2 •
Gage spacing in the extension was 14 km compared to
the 4.8 km in the research circle. This allowed a better
definition of the rainfall distribution at greater down-
wind distances which 1971 rainfall and echo results
indicated were also influenced. In 1974, 21 more re-
cording raingages (and hailpads) were added northeast
of the Alton area (Fig. 10) because 1971-73 rainfall
and echo results indicated influences from the Alton
area with extensions to the northeast of Alton. In the
nine non-summer months, the raingage network was
kept operating but with only half (every other gage)
open in the 1971-75 period. Recording raindrop spec-
trometers of the ISWS were installed at 8 sites (3 in
1971-72, and 8 in 1973-75) and operated in the sum-
mers. Dropsize data from the 8 sites (Fig. 10) were
used to study regional differences. Six devices to re-
cord lightning and to record thunder were built by
ISWS engineers and installed, as shown in Fig. 10, to
measure upwind, city and downwind activity. These
data plus the standard thunderstorm observations at
six weather stations (Fig. 10) provided the basis for
study of urban influences on thunderstorm activity.
A series of 7 surface weather stations that recorded
temperature, humidity and wind was established in
1971 by the ISWS. They were installed to supplement
the existing network of 9 stations employed by the
Illinois EPA and the St. Louis city-county air pollution
network. However, the network of 1971 was found to
be inadequate for mesoscale thermodynamic analyses,
and the network was expanded to 32 temperature-hu-
midity recording stations for the 1972-75 seasons (Fig.
11). A wind recording network comprising 14 stations
was installed and operated by the ISWS during the
summers of 1972-75 (Fig. 11).
Within the study area, an array of instruments mea-
sured aerosols and pollutants in the rainfall (Fig. 12).
The ISWS maintained a network of about 80 rainwater
collectors in two networks, one for 1971-74, and an-
other in 1975 when the emphasis shifted to the study
of the Alton area. The PNL put out lines of rain sam-
plers in 1972-74. The ISWS also typically had 7 wet-
dry samplers (atmospheric deposition), 15 aerosol
12 METEOROLOGICAL MONOGRAPHS

EXPLANATION

~.
R
0 . .()
·y
• \1 971-1975 Beaver Cr'cek

St:ot l. AF8
.I
* -1
I
·;
Okawv i lle

R
·;
/ _/
~ •
* ·.ta~.erloo. ./
./ ---- _ --
----
SCAI.E Oi,. I(.ILOMI':Ti:::RS
--- -----~---- 15 30

FIG . 10. The METROMEX raingage- hailpad network and sites of other surface data
measurements.

samplers, 4 high volume air samplers, and 3 sites with
sequential rain samplers and Andersen impactors.
These were arrayed to obtain upwind, urban, indus-
trial and downwind values.
Vertical Measurement Systems
Throughout the extensive summer study periods of
1971-75, the ISWS maintained two all-sky cameras,
and a third was added in 1975 (Fig. 13). They photo-
graphed cloud developments in the northwest rural
areas and the two metropolitan areas.
Radiosondes and pibal data were effectively used to
define urban-related effects on the thermal, humidity
and airflow structure of the low-level atmosphere.
These upper air observations were most commonly
acquired from fixed networks supervised primarily by
the ISWS and UW.
0

•
Total pr-ecipitation samplers irl
networks ( ISWS)
lines of total pretipitation KI LO.H.E.."TERS

e fUWERAT\IRt ANO
+ S<llllplers (PNL)
Wet-dry sarnple-r
~6
0 5 10

+
STATln'E Mll.SS
HUHII.liTY J;.I~I:!AS lligh volume air samp l er
=
0 8 16
J[KPUIATUR[. ftUHIO[Tl Aerosol (nuclcpore) sampl ers
ANDVIH:I) 0 5 10 0
.!U
5'T.\TUT& ~ (1973-75)
• Sequential rain samp l er and
Anderson Impactor
FIG. II. Stations measuring surface temperature, humidity and
winds during most METROMEX summers. FIG. 12. Instrumentation sites for atmospheric chemistry studies.
METROMEX: A REVIEW AND SUMMARY 13

EXP LA N A TIO N
-1 Dopp l er rada r

a Radlosond~
(yea" Indicated )
X Pfba l s i te, 1973-74
t:c> Profil er

.E-
"
Grccnvi I tc \

\
\
\
\
\
I
I
I
I
I
I
/

-- -- -__/
~.;&TEFIS
SCA.LE OF Kll.O
1$ JO 4S

F IG. 13. Instruments used to make vertical measurements of the atmosphere .

The Argonne National Laboratory (in 1971) and the
Illinois State Water Survey (in 1972) utilized a variable
array of nine pibal stations chosen from 20 sites, de-
pending on the type of day. These double theodolite
pibal stations were manned by Air Weather Service
personnel. The University of Wyoming also utilized
four such Air Force double theodolite teams in 1971.
In 1973-75, the ISWS trained observers and expanded
its pibal network to eleven single theodolite stations ,
as shown in Fig. 13, with the 1975 effort being devoted
to a detailed study of the airflow in the Alton metro-
politan area. The University of Wyoming maintained
a mobile single theodolite pibal capability with two
meteorologically equipped vans in 1971-72, and added
a third unit for the 1973-76 period.
Networks of radiosonde sites were also a major part
of the observational effort of METROMEX. Typi-
cally, METROMEX had four sites operating in July-
August of each year, and the various sites and their
years of operation are shown in Fig. 13. The UW op-
erated a sonde approximately 50 km WSW of St. Louis
during 1971-74 to sample environmental conditions
upwind from St. Louis. The ISWS operated 3 sondes
each summer (1971-75) with them being aligned west-
east (1971-72), northwest-southeast (1973- 74), and
southwest-northeast (1975) , and located to sample up-
wind, urban and downwind conditions. The PNL op-
erated a sonde in 1972-73 at Lambert Field. For its
1973-76 studies, the UW equipped a portable radio-
sonde unit to acquire upper air observations at stra-
tegic times and locations not afforded by the fixed net-
work. During the 1971-73 period, local NOAA officials
made available to METROMEX personnel the EMSU
radiosonde data taken at the Gateway Arch in the cen-
ter of St. Louis (see Fig. 13), and they also obtained
special radiosonde data to supplement the METRO-
MEX effort. In all, the METROMEX complement of
upper air observations included approximately 6150
pilot balloon observations and over 1000 radiosonde
releases.
In 1975, the ISWS utilized two NCAR-owned teth-
ered balloon-borne profilers for acquiring standard
weather data for boundary layer studies . These pro-
filers obtained measurements of dry and wet bulb tem-
peratures , pressure, and wind speed and direction
throughout the first 610 m above the ground.
The effects ofurban environments on the frequency,
amount, intensity and duration of precipitation and
related severe weather were monitored in METRO-
MEX through the use of three ground-based weather
radar systems. Their scanning areas are shown in Fig.
14. Two of these radar systems were located at the
Pere Marquette field headquarters 45 km northwest of
St. Louis and operated by the ISWS (Fig. 14). A FPS-
18 10-cm radar system was operated in a PPI mode;
a TPS-10 3-cm radar was used to collect RHI data.
Both of these radars were operated to support the
summer field operations during 1971-74, and with the
FPS-18 in a vertical scan mode in 1975, the TPS-10
radar system was not operated.
14 METEOROLOGICAL MONOGRAPHS
l )(I>LA.r.. A J I 0 '4
X A1dleld1.wi'l~reoro,Jttt
o~ir«:r.aft
wttre st•t•on~
• Wttilher "d1rs

I
I
I
I
/

FIG. 14. Aircraft operational area and radar scanning regions.

A second TPS-10 3-cm RHI radar was operated by collected data on air motions over St. Louis and
the University of Chicago at Greenville, approxi- around storms.
mately 55 km northeast of the Arch in St. Louis. Radar The SRI participation in METROMEX largely in-
observations were obtained during the summer field volved use of a portable lidar. The Mark VIII ( 1971)
programs from 1971 through 1975. NOAA brought two and Mark IX (1972-75) mobile lidar systems were used
Doppler radars to two locales (see Fig. 13) for oper- to record and investigate the urban aerosol structure
ations in the summer of 1975. These dual Dopplers over St. Louis, either while moving or from a variety
TABLE 6. Summary of meteorological aircraft usage in METROMEX.

Aircraft Controlling
type institutions Dates of operations
RB-57 (two) ISWS' High-altitude filter sampling in storm localities 1971
RB-57 UCSD' Turbidity and visibility measurements 1971
Piper Navajo Isws• Tracer release; limited aerosol and storm-dynamics 1971-75
characterization
Beech C-45 uw Boundary-layer characterization ; turbulence, aerosols, and 1971
thermodynamic properties
Beech Queen uw Boundary-layer characterization; turbulence, radiation, 1972-76
Air aerosols and thermodynamic properties
Cessna 411 PNL Tracer release; aerosol and trace-gas measurement 1972-75
Douglas DC-3 PNL Aerosol, trace-gas, and solar measurements 1975
DeHavilland ISWS 3 In-storm turbulence measurement 1973
Buffalo
Cessna UMR Aerosol measurements 1973
Beech Queen ISWS3 Turbulence, solar radiation measurements 1974, 1975
Air
Lockheed u c• Cloud physics measurements 1971- 76
Lodestar

'Aircraft and crews were supplied by the U.S. Air Force.

This aircraft and crew were furnished by Atmospherics Incorporated.
2
3These aircraft and crews were furnished by NCAR.
4 Aircraft and pilot leased from Interstate Airmotive Corp.
METROMEX: A REVIEW AND SUMMARY
of urban sites. The ANL had an acoustic sonder in the
urban area in 1972 and it was operated in conjunction
with the SRI lidar.
Meteorological Aircraft
The ISWS, UC, UW and PNL were the principal
groups making extensive use of meteorologically in-
strumented aircraft in the METROMEX field studies
of 1971-75. The 12 aircraft employed and their usage
are listed in Table 6. The ISWS used a Navajo aircraft
to release tracers, to sample aerosols, and to measure
cloud conditions during five summers. The UC flew a
Lodestar as a cloud physics research plane. The UW
used a C-45 H in 1971 and a Queen Air from 1972-76
for various cloud physics research. A Cessna 411 and
Douglas DC-3 were used by PNL during their air
chemistry research.
In addition, Queen Air aircraft from the NCAR
Flight Facility were utilized during METROMEX: in
1971 by ANL; in 1972 by UC; in 1974 by UW and
SRI; in 1975 by ISWS; and in 1976 by UW. The
NCAR-supplied Buffalo was also used by ISWS. The
NCAR aircraft were primarily used for observations
of airflow, radiation and aerosol sampling. The types
of sensors and data collected by each aircraft are de-
scribed in the annual METROMEX operational re-
ports. In summary, just over 1200 hours of flight time
were logged by the research aircraft dedicated to
METROMEX.
REFERENCES
Auer, A. H., 1978: Correlation of land use and cover with mete-
orological anomalies. J. Appl. Meteor., 17, 63~3.
Ashworth, J. R., 1929: The influence of smoke and hot gases from
factory chimneys on rainfall. Quart. J. Roy. Meteor. Soc., 25,
34-35.
Broeck, H. T., 1900: Cumulus clouds above fires. Mon. Wea. Rev.,
16, 549.
Chandler, T. J., 1970: Urban climates: Inventory and prospect.
Urban Climates, WMO Tech. Note 108, Geneva, 43 pp.
Changnon, S. A., Jr., 1962: A climatological evaluation of precipi-
tation patterns over an urban area. Air Over Cities, USPHS
Tech. Rep. A62-5, Cincinnati, 37-67.
- - , 1968: The La Porte anomaly-fact or fiction? Bull. Amer.
Meteor. Soc., 49, 4-ll.
- - , 1970: Reply. Bull. Amer. Meteor. Soc., 51, 337-343.
- - , 1971: Comments on the effect on rainfall of a large steelworks.
J. Appl. Meteor., 10, 165-168.
15
- - , 1976: Inadvertent weather modification. Water Resour. Bull.,
12, 695-718.
- - , F. A. Huff and R. G. Semonin, 1971: METROMEX: An in-
vestigation of inadvertent weather modification. Bull. Amer.
Meteor. Soc., 52, 958--967.
--,and F. A. Huff, 1977: La Porte again: A new anomaly. Bull.
Amer. Meteor. Soc., 58, 1069-1072.
Dettwiller, J., and S. A. Changnon, 1976: Possible urban effects on
maximum daily rainfall at Paris, St. Louis and Chicago. J. App/.
Meteor., 15, 517-519.
Hammon, W. H., and F. W. Duenchel, 1902: A comparison of the
minimum temperatures recorded at U.S. Weather Bureau and
Forest Park Meteorological Observations, St. Louis, Mo., for
the year 1891. Mon. Wea. Rev., 18, 12-13.
Hilberg, S., 1978: Temperature Analysis: Summary of METRO-
MEX, Vol. 2, Causes of Precipitation Anomalies, Bull. 63, Ill.
State Water Survey, Urbana, 11-24.
Holzman, B. G., and H. C. S. Thorn, 1970: The La Porte precipi-
tation anomaly. Bull. Amer. Meteor. Soc., 51, 335-337.
Huff, F. A., and S. A. Changnon, 1970: Urban effects on daily
rainfall distribution. Preprints 2nd Nat. Conf Weather Modi-
fication., Santa Barbara, Amer. Meteor. Soc., 215-220.
--,and S. A. Changnon, 1972: Climatological assessment of urban
effects on precipitation at St. Louis. J. Appl. Meteor., 11,
823-842.
Koepke, R. L., 1970: Industry in the St. Louis Region. East-West
Gateway Coordinating Council, St. Louis, 65 pp.
Kratzer, A., 1956: The Climate of Cities. Translation, Amer. Me-
teor. Soc., 221 pp.
Landsberg, H. E., 1956: The climate of towns. Man's Role in
Changing the Face of the Earth, University of Chicago Press,
584-603.
- - , 1970: Man-made climatic changes. Science, 170, 1265-1274.
Mitchell, W. H., 1901: Cumulus clouds formed by smoke. Mon.
Wea. Rev., 17, 498.
Ogden, T. L., 1969: The effect of rainfall on a large steelworks. J.
Appl. Meteor., 8, 585-591.
Pearson, J. E., 1958: The influence of lakes and urban areas on
radar observed precipitation echoes. Bull. Amer. Meteor. Soc.,
39, 79-82.
Peterson, J. T., 1969: The Climate of Cities: A Survey of Recent
Literature. NAPCA Pub!. AP-59, Raleigh, NC, 48 pp.
Rao, R. A., 1979: Stochastic analysis of annual rainfall affected by
urbanization. J. Appl. Meteor., 18, 1916-1924.
Southwestern Illinois Metropolitan and Regional Planning Commis-
sion, 1976: Air Quality in the Metro-East Area. East-West
Gateway Coordinating Council, St. Louis, 35 pp.
Stout, G. E., 1962: Some observations of cloud initiation in indus-
trial areas. Air over Cities, USPHS SEC Tech. Rep. A62-5,
Cincinnati, 147-153.
Weather Modification Advisory Board, 1978: The Management of
Water Resources. U.S. Dept. of Commerce, Washington, DC,
229 pp.
Wong, K. K., and R. A. Dirks, 1978: Mesoscale perturbations on
airflow in the urban mixing layer. J. Appl. Meteor., 17, 677-688.
 CHAPTER 2

Surface Weather Conditions

RICHARD G. SEMONIN

CONTENTS

ABSTRACT . . • . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . • . . . . . . . . . . . . 17
2.1 Climate of the St. Louis area .......................................................... 17
2.2 Temperature ....................................................................... 20
2.3 Humidity .......................................................................... 23
2.4 Winds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.5 Precipitation ....................................................................... 27
2.6 Severe weather ..................................................................... 33
2.6.1 HEAVY RAINSTORMS . . . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.6.2 THUNDER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.6.3 HAIL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.6.4 WIND GUSTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.6.5 TORNADOES AND FUNNEL CLOUDS . . . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

ABSTRACT

The detailed precipitation pattern observed during the 1971-75 period revealed similar general features found
in long-term records. The summer rainfall increased from west to east across St. Louis, reaching a maximum
east and north of the Mississippi River.
The precipitation maximum was located northeast and east of the urban heat island centered on the com-
mercial district of the St. Louis inner city. The heat island was associated with a humidity deficit in the same
general area.
Annual precipitation patterns showed a great deal of similarity, particularly the east side location of the
maximum. The diurnal distribution of rainfall reflected the diurnal cycle typical of the climatic regime of the
area. The shape of the diurnal curve for stations within the area of maximum rainfall was similar to that for
those located in the rainfall minimum west of St. Louis. Since the diurnal structure of the rainfall distribution
was independent of the rainfall amount, it suggests repeated enhancement of precipitation within storms moving
across the urban area.
The surface winds were perturbed as revealed by the unique signature of wind speed cross-sections showing
increases then decreases across the city. The same signature appeared in both daytime and nighttime winds
indicating that the perturbation was not primarily caused by the heat island which maximized at night.
Severe weather events including heavy rainstorms, hail and wind gusts all showed a maximum in the region
of the east side rainfall maximum. Thunder, a frequent precursor of severe weather, also showed a maximum
in occurrence and intensity over the city and east of the Mississippi River.

2.1 Climate of the St. Louis area
St. Louis lies midway between the Continental Di-
vide and the Atlantic Ocean and some 500 miles north
of the Gulf of Mexico. Its climate is a humid conti-
17
nental type with cold winters, warm summers and fre-
quent short period fluctuations in temperature, humid-
ity, cloudiness and wind direction. The excellent soils
and well distributed annual precipitation of 38 inches
(97 em) favor a very high standard of agricultural pro-
18
duction in the area. The normals, means, and extremes
for selected weather variables are shown in Table 1.
Except for a few low hills of the Ozark Plateau lo-
cated to the southwest of St. Louis and the broad val-
leys of the Mississippi and Missouri Rivers, the terrain
is quite flat. Differences in elevation have no signifi-
cant influence on the climate although the analysis of
Huff and Vogel (1978) suggests 9-14% local increases
of precipitation related to the Ozark foothills and the
Bluffs regions, respectively.
Without the protection of natural barriers such as
mountain ranges, the St. Louis area experiences the
full sweep of winds, which are constantly accom-
panied by a variety of weather. Southeast and easterly
winds are associated with mild and wet weather;
southerly winds accompany warm and showery con-
ditions; westerly winds are dry with moderate tem-
peratures, and winds from the northwest and north are
cool and dry. Winds are controlled by the storm sys-
tems and weather fronts, which move eastward and
northeastward through this area.
Synoptic scale storm systems move through the area
most frequently during the winter and spring months
and cause a maximum of cloudiness during those sea-
sons. Summer-season storm systems tend to be weaker
and to stay farther north, leaving the area with much
sunshine interspersed with thunderstorm situations of
comparatively short duration. The retreat of the sun
in autumn is associated with variable periods of pleas-
ant dry weather of the Indian summer variety. This
season ends rather abruptly with a return to stormi-
ness which usually begins in November.
These large scale climatic features are seen in data
abstracted from the St. Louis Local Climatic Data
summary and shown in Table 1. During the summer
season the sun heats the entire area quite strongly and
uniformly. July is the warmest month in most years
with an average daily maximum of 88.4°F (31.3°C).
The annual average number of days with temperatures
of 90°F (32°C) or higher is near 50. Summer also has
periods of uncomfortably hot and humid weather,
which can be persistent. The highest temperature of
record in the 1941-70 period is ·106°F (41.1°C) ob-
served in July 1966. Highs of l16°F (46.7°C) have oc-
curred in earlier years.
The summer average relative humidity ranges be-
tween 86% at 0600 CST to 57% at 1200 and 1800 CST,
as compared to the annual values of 84% at 0600 and
59 and 61% at the later times, respectively. Of course,
because of the greater temperatures in summer, the
same relative humidity represents greater absolute
water vapor concentration and, therefore, greater la-
tent heat availability.
Latitude is the principal control for both summer
temperature and precipitation. The Gulf of Mexico is
METEOROLOGICAL MONOGRAPHS
a principal source of moisture for summer rainfall.
Mean total precipitation for the 4-month period of De-
cember through March is 9.0 inches (22.9 em). Aver-
age rain for the 6-month period of April through Sep-
tember is 21.7 inches (55 em). The driest month is
usually January, and the wettest month is June. The
summer average rainfall is 11.0 inches (27 .9 em).
Precipitation during fall, winter and spring tends to
fall over large areas. In contrast, summer rainfall oc-
curs principally as brief showers affecting relatively
small areas. The erratic occurrence of summer show-
ers often results in an uneven precipitation distribu-
tion. The high rates of summer thunderstorm rainfall
also cause runoff and soil erosion, and may be accom-
panied by hail or destructive windstorms. Local flash
floods occur in minor streams due to heavy thunder-
storm rains occurring in the warm season.
The annual average snowfall is 15 inches (38 em)
and more than 80% of St. Louis winter precipitation
falls as rain. In a large number of winter storm situa-
tions, only a slight change in the temperature pattern
suffices to change rain to snow or vice versa.
Thunderstorms average 45 days annually, but most
are quite harmless. On occasion they provide the
source for hail, damaging winds and lightning, and,
infrequently, tornadoes. Hail falls at a point on an av-
erage of three days annually, but usually causes little
damage. St. Louis is located in an area of the Midwest
which is particularly favorable for tornado formation.
It was within this climatic regime that a local pre-
cipitation anomaly was detected in the records of co-
operative observers and special rainfall reporting sta-
tions surrounding St. Louis (Huff and Changnon,
1972). The findings of Huff and Changnon were key
to the choice of St. Louis as the location for the in-
tensive field measurement program.
The 5-year sample of surface weather conditions, as
measured during METROMEX, is presented in this
Section. The observed spatial and temporal variability
is contrasted with the long-term climatic values de-
scribed by a single station (St. Louis). Surface weather
is defined to include the temperature, humidity, winds,
precipitation, and severe weather events as recorded
in the METROMEX research area during the 1971-75
field experimentation. The subject of visibility, as a
weather variable, is discussed in the Urban Boundary
Layer Section of this report. The first three of these
weather variables include those which would be di-
rectly sensitive to the development of urban and in-
dustrial components in a large metropolitan area.
There is little question that large metropolitan areas
influence the local temperature and the structure of
the lower boundary layer in a unique way. Although
the urban heat island has been documented for many
years (Hammon and Duenckel, 1902), it was studied
 ~
~
~
~
>
~
<
m
-
~
TABLE 1. The weather normals, means and extremes for St. Louis. The normals and means are determined for the period 1941-70 whereas the >
extremes are established from varying length records for specific elements. Adapted from St. Louis Local Climatological Data, NOAA/EDS, 1977. z
tj
(/)
NORMALS, MEANS, AND EXTREMES
c
s::
Precipitation in Inches Relative Wind Mean Number of Days
Temperatures or Humidity -~ ~ ~
-" ~
"~ -".., ~ ~
V> ""
" ,__.., ~

~ ., ~ Sunrise f ~ f
-~~
:e ~ § to c: ~ ;:::~ V1 -~~
Norma 1 Extremes I; a ter Equ iva 1ent Snow, Ice Pellets ~ ~ (/) Sunset .~ s... cf s...0 0C
, OUO +JO :~
~ ~ '-1 ... a... +-' ttl Q) +-'
V1
... ... 0"
.". I .".
"0 0 0 ·-
"' 0
"" >, +-' ..s:: u..c:
E ,., 4-~QJ ·.-u-us... ~~
,-:;: 3~ 5~ "' "' "' "' "" ·-.., 0 Ill >.
-c Q)
a.. c: .. _1:: "'C
·.--
5 ,_, ~~ "' 5"' U").,... 5-
-" ;;; .!= :5 .;=~ DOl 06112118 "'u
>W • c: 5- ttl ~ u 3 c: ~
o.c ... ow ... ... -~~ ... ... ... ... ~~ ... .+J I'd 1:: Q) 0 Q)r- 0 0 :::::1
c: ......
-51 ~-~>< ~·~
.,._ ~ urn
,._ U3 ~ X c X "'c. ,...._ u w ;::I ,...... ....-- s...o c: • .s::::
0 .. o
"' wo
~0 ·""
- 0 "' ".
:>:E <>-0 0... ::EV'I U U a... • Vh- 1- "'"'
~;:::
~ ~~ C>:>: ~ a:"' ~ .J
a: "'
~ z :>: >- ::.::::E ~"' " ~ "'
~::;::::
~ ~.;: "'
~ ( Loca 1 Time)
""'
J 39.9 22.6 31.3 76 1970 -14 1977 1.85 5.38 1975 0. 22 1970 2.43 1975 23.9 1977 11.2 1958 77 82 65 69 10. 3 NW 54 6. 7 8 6 17 8 2 l 2
F 44.2 26.0 35.1 85 1972 -5 1958 2.06 4.17 1974 0.23 1963 2. 56 1959 12.9 1961 8. 3 1966 76 80 61 64 10.9 NW 53 6. 5 7 7 14 8 2 1 l
M 53.0 33.5 43.3 88 1963 -5 1958 3.03 6.28 1977 1.09 1966 2. 95 1977 22.3 1960 10.0 1958 74 81 58 58 11.8 WNW 56 6. 7 6 9 16 11 1 3 2
A 67.0 46.0 56.5 92 1970 22 1975 3.92 9.09 1970 0.99 1977 2.33 1970 6. 5 1971 6.1 1971 70 79 54 53 11. 3 WNW 57 6. 5 7 8 15 11 0 6 0
M 76.0 55.5 65.8 92 1972 31 1976 3.86 7.25 1961 1. 02 1972 2. 94 1974 T 1944 T 1944 76 83 56 55 9. 3 s 62 6. 1 8 9 14 11 0 6 1
J 84.9 64.8 74.9 98 1971 43 1969 4.42 8.65 1969 0.47 1959 3. 29 1960 0.0 0.0 79 84 57 56 8. 6 s 68 5.9 8 11 11 9 0 7 0
J 88.4 68.8 78.6 106 1966 51 1972 3.69 7.81 1958 0.60 1970 3.16 1958 0.0 0.0 78 86 57 56 7. 7 s 72 5. 5 9 12 10 9 0 7 0
A 87.2 67. 1 77.2 105 1962 47 1965 2.87 6.44 1970 0.08 1971 2.66 1974 0.0 0.0 81 89 57 59 7. 4 s 65 5.4 10 11 10 7 0 6 0
s 80. 1 59.1 69.6 100 1971 36 1974 2.89 6.21 1972 0. 76 1960 2.85 1969 0.0 0.0 83 91 61 63 7. 9 s 63 5. 2 11 8 11 9 0 4 1
0 69.8 48.4 59.1 94 1963 23 1976 2.79 5.77 1969 0. 21 1975 2.14 1969 T 1967 T 1967 77 85 55 61 8. 5 s 63 4. 9 13 8 10 8 0 2 1
N 54.1 35.9 45.0 81 1971 1 1964 2.47 5. 74 1968 0.44 1969 2.87 1972 11.3 1951 10.3 1951 78 84 62 68 9. 9 s 49 6.1 9 7 14 8 l l l
D 42.7 26.5 34.6 76 1970 -10 1976 2. 04 6. 50 1971 0. 32 1958 2. 57 1971 26.3 1973 12.0 1973 81 85 69 75 10.3 WNW 43 6.8 7 7 17 10 1 1 2

JU~ I I I JAN I APR

I I I I I I I I I AUG JUN DEC DEC
YR 165.6146.2155.91106 11966 -14 1977 35.89 9.09 1970 0.08 1971 3.29 1960 26.3 1973 12.0 1973 I 78184 I 59 1611 9.51 S 591 6.0 1103 1103 1159 1109 I 6 145111

T - Trace.
NORMALS - Based on record for the 1941-1970 period.
DATE OF AN EXTREME - The most recent in cases of multiple occurrence.
PREVAILING WIND DIRECTION - Record through 1963.
WIND DIRECTION - Nu"erals indicate tens of degrees clockwise from true north. 00 indicates calm.
FASTEST MILE WIND - Speed is fastest observed 1-minute value when the direction is in tens of degrees.

\0
-
20
in detail during METROMEX as an important element
for the identification of potential causes for the pre-
cipitation anomaly (Changnon et al., 1972).
The humidity distribution in and surrounding a me-
tropolis has been the subject of some controversy in
the recent past (Landsberg, 1974). A metropolitan area
is composed of large areas covered with artificial ma-
terials of a non-transpiring character. It was assumed
that while an urbanized area may not be a humidity
sink it certainly cannot be a humidity source (through
evapotranspiration) in the same way that heavily veg-
etated suburban and rural areas may be considered.
The vegetation normally covering the more rural areas
in the warm season provides a source of humidity as
it transpires throughout the growing season thereby
providing, on the average, a relatively larger humidity
content near the surface as compared to a pure urban
setting (Sisterson, 1975). During the cold season, the
reverse anomaly may be true in response to anthro-
pogenic humidity sources.
Because of its potential direct effect on cloud initi-
ation and development, the urban-rural contrast in hu-
midity content was the subject of measurement at the
surface (Dzurisin, 1978) and aloft by both aircraft
(Spangler and Dirks, 1974) and radiosondes (Acker-
man and Mansell, 1978).
The surface wind speeds and directions are very
sensitive to their location within an urban area. Nearby
obstacles such as buildings can perturb the ambient
wind in particular ways contributing to the creation of
small scale turbulence near the surface. This feature
of the urban weather may also be a contributing factor
to the destabilization of the planetary boundary layer
and consequent impact on convective motions in prox-
imity to the comparatively rough city structural pro-
file.
The final elements of the anomalous urban weather,
the precipitation and severe weather, are the principal
focus of the intense studies that were undertaken in
St. Louis. The observed anomaly of these particular
elements apparently results from a combination of fac-
tors uniquely associated with a large urban-industrial
site, although the quantitative identification of the
physical changes had not been documented prior to
METROMEX. In succeeding chapters of this mono-
graph the relation between various perturbed compo-
nents of the atmospheric structure, as they relate to
the production of clouds and precipitation, are ad-
dressed in a comprehensive manner.
The data for these surface measurements were com-
piled from the instrumentation described previously
(Chapter 1). The data from the surface network oftem-
perature and relative humidity recorders, shown in
Fig. 11, Chapter 1, were used to derive additional vari-
ables such as dew point temperature and equivalent
METEOROLOGICAL MONOGRAPHS
potential temperature. The recorded rainfall data were
processed to obtain 5-minute amounts, and these were
analyzed to delineate raincells, as well as storm (or
rain period) values, and the monthly, seasonal and an-
nual totals. The hailpad data coupled with the rain-
gages (modified to record the time of hail) were used
to reconstruct individual hailstreaks, and volume and
energy of the hailfalls. The locations of the raingages
and hailpads are shown in Fig. 10, Chapter 1. The
thunder detectors (Fig. 10, Chapter 1) defined thunder
occurrence and the frequency as a measure of storm
intensity. In addition to providing the short-term (5-
years) climate during the METROMEX program, the
various data were used by several investigators to de-
scribe case studies of individual storm events and
other characteristics of the urban environment (e.g.,
Lawson, 1973; Grosh and Semonin, 1973; Sisterson
and Dirks, 1978; Changnon and Semonin, 1975; Vogel,
1975; Shea, 1976; Shea and Auer, 1978).
2.2 Temperature
Comprehensive temperature data acquisition began
in the summer of 1972 after the 1971 data indicated
that more spatial data would be extremely valuable in
helping to provide essential information for describing
the surface conditions attending convective cloud de-
velopment. The hygrothermographs used were housed
1.5 m above the surface in Cotton Region shelters;
Jones and Schickedanz (1974) reported the measure-
ment accuracy to be ±0.5°C for temperature and ±5%
for relative humidity. The maintenance of the network
to insure quality records was described by Hilberg
(1978a). The charts were edited and hourly values of
temperature and humidity were extracted and entered
on punch cards for further analysis. The mean tem-
perature over the network for periods of interest was
obtained and the individual station values were ex-
pressed as departures from the mean. This procedure
partially compensates for the monthly temperature
differences during the summer (June-August) season
and for climatological variations between years. No
adjustment for station elevation was attempted since
it would necessitate making an assumption regarding
the near-surface lapse rate.
The 4-year summer mean temperature for the net-
work was 24.5°C and the mean departure is shown in
Fig. 1. The first obvious feature is the St. Louis heat
island with a maximum value of + 1.9°C and greater
than + 1.0°C over the entire inner city area. The ex-
tension of the positive ( +O.SOC) departures to the
northwest reflects the urbanization toward Lambert
Airport. Positive anomalies in the temperature field
are also seen to the east over the East St. Louis area
(compare with the land use pattern shown in Fig. 7,
Chapter 1).
METROMEX: A REVIEW AND SUMMARY 21

The second feature readily seen in Fig. 1 is the area

of negative temperature departure east of East St.
Louis extending north and south. Generally, the area
devoted to agricultural land use (see Fig. 7, Chapter
1) exhibited negative temperature departures. The
greatest negative departure (of -1.4°C) occurred at a
station in the extreme southwest edge of the research
circle and in the forested Ozark foothills region.
The urban summer temperature anomaly was stud-
ied as a function of time of day, sky cover, wind speed
-0 . 5 and wind direction (Hilberg, l978a). The areal patterns
revealed that the maximum heat island of >2.0°C
4- YEAR MEAN
(above the areal mean) was observed between 0000
HfiPERATURE and 0600 COT (Fig. 2a, b). The center was slightly
= 24.s'c west of the commercial district. In the rural area to
the east, the negative departures below the areal mean
SUMMER 1972-1975 TEMPERATURE DEPARTURES were nearly the same magnitude. The maximum ur-
FIG. I. The daily average temperature departure from the areal ban-rural average temperature differences was 4.5°C
mean for the period 1972-75. Mter Hilberg (l978a). at midnight (Fig. 2a), with a minimum difference

KILOl·teTeRS
10 20

a. oooo b. 0600

c. 1200
d . 1800
FIG. 2. The departure from the areal mean of the average 0000, 0600, 1200 and 1800 CDT temper-
atures. After Hilberg (1978a).
22 METEOROLOGICAL MONOGRAPHS

(-1°C) near 1500. These central features during the the exception that at 0000 COT the inner-city depar-
first 6 hours of an average day were surrounded by ture during cloudy conditions was O.SOC less than un-
various centers of positive and negative departures. der fair or partly cloudy skies.
The most notable diurnal fluctuation in the surround- The temperature departure data were stratified by
ing region was the negative departure area over the two categories of wind speed and eight wind direc-
Ozark foothills with a maximum of less than - 2.SOC tions. The wind speed categories were 0-2 m s- 1 and
at 0000 and a minimum of more than -0.3°C at 1800 2-3.5 m s- 1 • It was not possible to study the effect of
(Fig. 2d). higher wind speeds on the temperature distribution
The central heat island diminished between 0900 and due to an inadequate sample size. No great influence
1500 (not shown) with minimum positive departures of wind direction on either the location or magnitude
of -0.1°C. During this period an east-west area of pos- of maximum departure values existed. For example,
itive departure extended from the west edge of the the 2100 COT departure pattern (not shown) for both
research circle to the east edge with an extension speed categories showed a concentration of the posi-
northwest over the floodplain between the Mississippi tive departures over the central urban area and a cen-
and Missouri Rivers. The heat island intensified after ter of negative departures east of the metropolitan
1800 (Fig. 2d) toward its early morning maximum. area. This pattern is similar to that of the 4-year sum-
The urban heat island was most pronounced under mer mean indicating little influence of wind on the
clear skies but with a similar pattern and intensity dur- location of the maximum heat island temperatures.
ing partly cloudy conditions (Fig. 3). A study of the The diurnal cycle of temperature at urban and rural
diurnal fluctuation of the anomaly revealed little influ- sites was studied to determine the influence of site
ence of cloudiness on the pattern at a given hour with location on producing the observed temperature

r:-0.5
0 5 10
'q/
STATUTE MILES
a . Cl ea r b. Partly cloudy

c. Cloudy

FIG. 3. The average temperature departure from the areal mean during clear, partly cloudy, and
cloudy sky conditions. After Hilberg (1978a).
METROMEX: A REVIEW AND SUMMARY 23

ready warmer urban temperature, thus minimizing the

urban-rural difference at this time. In the same way,
(ARC) the greater cooling rates at rural stations easily surpass
the lesser rate of the urban stations accounting for the
3 to 4°C temperature difference at 0000 CDT.
u
0
The spatial pattern and diurnal behavior of the St.
:26
Louis urban heat island is not unlike that of several
""
::::>
1-
~
other large cities studied in the past (Mitchell, 1961).
u..J
These observations, however, were not obtained to
~ 24~-,~-----F-~__M_ea~n-~~--~r---~.-------~ merely document a well established fact, but to pro-
1-
>-
....J
a::
vide quantitative measurements of a function which
::::>
impacts upon other atmospheric variables. Several
METROMEX investigators demonstrated that the
heat island properties extended upward into the
boundary layer, and even to cloud base and higher
20
H during convective conditions (e.g., Dytch, 1972; Auer
Sunset and Dirks, 1973; Semonin, 1974; also section on Urban
I I . I I I I 1 I Boundary Layer).
12 18 00 06
TIME (CDT) 2.3 Humidity
FIG. 4. The diurnal cycle of temperature for an urban and rural
site (see Fig. 3). After Hilberg (l978a). Dew point temperatures were derived from the hy-
grothermograph network to examine urban-rural dif-
anomaly (Hilberg, 1978b). The temperature data were ferences. A detailed study of the spatial and temporal
tabulated once each hour on the hour to produce a variations of dew point in the network was reported
mean hourly value for the years 1972-75. The results by Dzurisin (1978). Hilberg (1978b) presented results
from two representative stations (Fig. 4) show that the of the diurnal cycle of mixing ratios for urban and rural
rural station temperatures increase and decrease more stations.
rapidly than the urban stations, and the amplitude of The dew point and mixing ratio data were prepared
the diurnal change is greater in a rural setting. With in the same manner as the previously described tem-
but one exception at 1100, the urban station is always perature measurements. The mean dew point (or mix-
warmer than the rural. The radiational cooling at the ing ratio) for the network was calculated and individ-
rural station and the great heat capacity of the urban ual station values were expressed as departures from
environment is obvious during the evening and noc- the network mean.
turnal hours. The 4-year summer pattern of dew point tempera-
The rural-urban differences in rate of temperature ture departures (Fig. 6a) shows a deficit over the urban
change, shown in Fig. 5, coupled with the difference area of about -O.SOC. The deficit is persistent through-
in mean temperatures between the two types of sites, out June, July and August, with a -1.0°C maximum
appears to account for the observed heat island. The departure in July. The deficit area extended eastward
greater rate of temperature increase at rural stations from the city in August, dividing the network area in
between 0600 and 1100 CDT nearly equalizes the al- an east-west direction.

-I 2
'-
.<::
:;--' 1

-3~L-L-L-~~~~~~~~~~~~~~~~~~~~~

06 12 18 00 06
TIME, CDT
FIG. 5. The rate of temperature change for an urban and a rural site as a
function of hour of day. After Hilberg (1978b).
24
a. Summer months, 1972-1975
c. Ju l y, 1972-1975
FIG. 6. The average dew point temperature, departure from the network mean for the entire sum-
mer, and each summer month during the years 1972-75. After Dzurisin (1978).
Positive departures were observed over the flood-
plain between the Missouri and Mississippi Rivers
north of St. Louis in June and August with lesser de-
partures in July. A region of positive departures was
observed extending eastward from the Mississippi
River south of the urbanized area.
The most consistent feature of the dew point de-
parture pattern is the deficit area over the city. The
deficit tends to be slightly larger over the western sub-
urbs, generally upwind of major industrial activities.
The departure pattern is not easily identified with par-
ticular land use types. In general, however, the less
evapotranspiration, as a humidity source seems a rea-
sonable explanation for the relatively dry urban area
(Sisterson, 1975). The mean pattern is also consistent
with the aircraft observations of Dirks (1974), obtained
on generally clear days where specific humidity defi-
cits of 1.0-1.5 g kg- 1 were reported at low levels over
the urban area.
Patterns of dew point departures were constructed
METEOROLOGICAL MONOGRAPHS
b. June, 1972, 1973 , 1975
from the 4-year data set for 3-hour intervals. Exami-
nation of these patterns reveals that the urban-related
deficit apparent in the annual and monthly data ex-
hibits a diurnal change. During the early morning
(hours between 0300 and 0600 CDT, Fig. 7) a positive
departure is noted over the city. By 0900 CDT the
pattern of dew point departure is barely discernible.
The urban deficit becomes apparent by 1200 CDT, and
persists through 0000 CDT with the maximum of
- 1.0°C occurring between 1500 and 1800 CDT. The
area of maximum negative departure enlarges to the
west by 1800 CDT, but is very evident verifying the
concept of a " dry island" corresponding to the urban
heat island. However, the dry island maximizes pri-
marily 12 hours prior to the maximum urban-rural dif-
ference in temperature.
Dzurisin (1978) also studied the dew point temper-
ature departure patterns with varying sky conditions.
Little change in the urban pattern was observed re-
gardless ofthe amount of sky cover. A persistent OSC
METROMEX: A REVIEW AND SUMMARY
a . 0600 COT
c. 1800 COT
FIG. 7. Network average dew points and average departures for 1972-75 at 0300, 0600, 0900 and
1200 CDT. After Dzurisin (1978).
negative departure was present over the urban area
under clear, partly cloudy, and cloudy days (Fig. 8).
The network mean values, however, increased from
17.7°C for clear days to 19.6°C on cloudy days and
this obviously related to the prevailing synoptic
weather conditions. A strong positive departure was
noted near Alton under all conditions.
Hilberg ( 1978b) selected one rural and one urban
station to examine the diurnal variation of mixing ratio
derived from the hygrothermograph data. The 4-year
hourly means are shown in Fig. 9 for the diurnal cycles
at single urban and rural sites (shown in Fig. 3). The
urban station in the central business district shows
negative departures from the urban mean beginning at
0600 and lasting until 2000 COT. This urban station
then shows values greater than the mean from 2000 to
0600 COT and maximizing at midnight. The 24 h, 4-
year mean at the urban station is 14.0 g kg-t, and a
positive 0.5 g kg- 1 departure at 0000 COT exactly
equals the rural station at values as seen in Fig. 9. The
25
MEAN = 18.8 •c
d . 0000 COT
positive departure, or increase in urban mixing ratio,
begins at sunset and the reversal occurs at sunrise.
The rural stations show positive departures after mid-
morning with a maximum between 1900 and 2000 COT
followed by a decrease to the minimum between 0500
and 0600 COT. An interesting feature of these curves
is the nearly inverse relationship between the rural and
urban station. While the rural area's near-surface at-
mosphere obviously always contains more water va-
por than the urban area, the contrast is even greater
during the daylight hours. The urban area, on the other
hand, is characterized by greater mixing ratios be-
tween 0000 and 0700 COT.
2.4 Winds
The surface wind data were obtained from 19 loca-
tions (see Fig. 11, Chapter 1) scattered throughout the
research area. The data used for various analyses were
based on 10 min mean values of speed and direction
26 METEOROLOGICAL MONOGRAPHS

KILO~IETERS
0 8 16
o==s==lo

a. Clear days Partly cloudy days

c. Cloudy days
FIG. 8. Average dew point departure patterns as a function of clear, partly cloudy, and cloudy sky
conditions. After Dzurisin (1978).

16 preceding the clock hour, These values are termed

'hourly' in the following discussion.
The average speed data for the 1972-75 were strat-
ified into the three major land use areas of urban, sub-
urban, and rural by Changnon (1978a), The analysis
Rura l revealed a higher average speed in the urban and sub-
............ .........
r~ean

---l
~-4
/ urban areas as compared with the rural values .
Urban r~e an
---------------~------
_./ \ To emphasize these findings , cross-sections of wind
...... speed along the wind direction were constructed and
.......................... 1f............................... ...
/

t ,
Urban Station 6 (ARC) then superpositioned to obtain a single profile centered
at Forest Park which is in the center of the metropol-

~ S~~ot
I , , I
o~6~LJ-L-L~12-L~~~-1Ls~L-~-L~oo~-L~LJ~o6
TIME (CDT)
FIG. 9. The diurnal cycle of mean mixing ratios at an urban site
and at a rural site. After Hilberg (1978).
~1
itan area. Cross-sections were also constructed in a
similar way for daytime (0601-1800) and nighttime
, I , '""'' (1801-0600) periods. The day and night stratifications
(Fig. 10) show little difference in the pattern although
the wind speed is always less at night. The major fea-
tures of this analysis are the pronounced upwind min-
imum (located at the metropolitan boundary) and the
METROMEX: A REVIEW AND SUMMARY 27

o EAST RURAL
t;j Frequency high to and (
e; and lo>t to W and S

5 . 2~-L---L---L--~--~--~--~--~--~~
-50 -40 -30
DISTANCE FROM FOREST PARK , kilometers
S'rATtJ'rE MILf:::S
FIG. 10. Composited wind speed cross-sections determined along
the direction of flow for day and night hours and for aU available FIG. 12. Distinct areas of similar wind characteristics. After
data during 1972-75 summers. After Changnon (1978). Changnon (1978).

maximum located directly over or slightly downwind the 1972-75 period. The predominating wind direc-
of the urban center. A second minimum, more broad tions at all stations were southerly (Changnon, 1978a).
in space, was also observed 20-30 km downwind over The frequency for each of the 16 directions is some-
the rural area with subsequent recovery to upwind what variable although the values in Fig. 11 show the
values indicated at distances greater than 30 km. most frequent directions are between southeast and
Changnon ( 1978a) further stratified the wind data to southwest. The wind speeds were averaged for all pre-
examine the mean values during the 3 h period prior rain periods along cross-sections in the same manner
to the onset of precipitation events. During the 4 years of Fig. 10. The pre-rain wind analysis (not shown here)
for which surface wind measurements were available, revealed that the average speed along the cross-sec-
there were 183 rain periods (storms) in the network. tion had the same signature as for all days as the twice-
Wind roses were constructed for each of the sites for daily stratification (Fig. 10), but was less.
Examination of the 4-summer surface wind direction
values indicated six distinctly different regions. These
wind regime regions appear on Fig. 12 along with their
basic characteristics and a region average speed. The
rural regions are generally distinguished by their up-
wind frequencies ; that is, the west rural has a high
west frequency. The two urban regions differ consid-
erably with flow in the northerly area less disturbed
than in the southerly area which has fewer north and
east winds and more south and west winds than the
undisturbed flow would predict.

2.5 Precipitation
Since the lack of randomized application of precip-
itation modification agent(s) is unavoidable in inad-
vertent rain modification evaluation, the basic ap-
proach to the evaluation of effects was that of 'data
analysis.' In the data analysis approach, the final proof
.;o.u: roR o: Rl.:cr [{;.!,l
~
'"1.0 and acceptance of inadvertent modification of precip-
0 ,
H H A :.TAltl'l't: Hll.ES
itation does not and cannot rest entirely upon statis-
tical evidence and results from tests of hypothesis.
FIG. 11. Summer wind roses constructed from hourly data for the
three hours prior to rain from the period 1972-75. After Changnon The test statistic is treated as an informative summary
(1978). statistic and is to be clearly distinguished from the
28
concept of the test statistic as a strict accept-reject
rule. Thus, the flexibility of attack and the willingness
to study things as they are, rather than as they should
be, were stressed.
In applying the concept of data analysis to non-ran-
domized precipitation and storm comparisons, we
were guided by the general concept of data analysis,
as described by Tukey ( 1962), who indicated that
much of the data analysis must be a matter of judg-
ment. 'Theory,' whether statistical or non-statistical,
must guide, not command. Tukey stated: "Data anal-
ysis must progress by approximate answers, at best,
since its knowledge of what the problem really is will
at best be approximate. It would be a mistake not to
face up to this fact, for denying it, we would deny
ourselves the use of a great body of approximate
knowledge .... "
Thus, the METROMEX analysis of precipitation
proceeded with the knowledge that the treatment ef-
fect, if present, was irregular. Therefore, the definition
of target (effect) and control (no-effect) areas in the
analysis could not be completely rigorous. However,
because of the well-designed field program (Changnon
et al., 1971), there was a great wealth of 'approximate'
knowledge available from the analysis of seasonal,
monthly, rain period (storm), diurnal and individual
raincell data. This knowledge yielded invaluable in-
formation which could be used as a broad basis for
establishing the reality of inadvertent precipitation
modification, as well as to shed light on the causes.
•
KILOMETERS
0 25 50 75 100
FIG. 13. Average summer rainfall (em) based on 1900-44 data.
METEOROLOGICAL MONOGRAPHS
The historical 45-year summer rainfall pattern for
the St. Louis area reveals generally high values,
greater than 30 em, extending N-S through central
Missouri (Fig. 13). There is a relative low rainfall area
extending north of St. Louis and an isolated high in
the Shawnee Hills of southern Illinois. These data re-
flect the general precipitation climate prior to the up-
ward trend east of St. Louis. The trend analysis of
Huff and Changnon (1972) suggests the increased pre-
cipitation was pronounced in the 10 to 15 years before
1968 (see Fig. 3, Chapter 1).
Using all available data for the period 1941-68, Huff
and Changnon (1972) studied the detailed pattern of
summer rainfall over the area involved in the 5-year
measurement project. Their analysis showed a region
of relatively high rainfall from Edwardsville to the
northwest of Belleville with values exceeding 30 em
(Fig. 14). The details shown in Fig. 14 were not ap-
parent in Fig. 13 due to the availability of greater num-
bers of raingages in the recent (1941-68) period.
The total summer rainfall (June, July, August) for
the 5-year period 1971-75 in the research network is
shown in Fig. 15. It exhibited a general west to east
increase as did the long-term pattern (Fig. 14). Key
features to note in the 5-year sample are the pro-
nounced low rainfall region in the western portion of
the metropolitan area and the various high rainfall re-
gions, of which the Edwardsville area exhibits the
most prominence. The "Edwardsville high" was found
to be a consistent feature of the summer-time-total
.' .
St. Louis
\
'
"---
•
• •
KENTUCKY
METROMEX: A REVIEW AND SUMMARY 29

FIG. 14. Average summer rainfall (em) based on 1941-68 data.

After Huff and Changnon (1972).
,.
patterns for each of the five years of study with '
STATUTEY:lt.t:~·

roughly thirty percent greater than the network-mean FIG . 15. Total five-year summer rainfall (em) measured during
rainfall amount. 1971-75. After Huff (1977a) .
Analysis of the 330 individual storms observed dur-
ing the METROMEX summertime periods (Huff, The Edwardsville high rainfall area was observed in
1977a) shows that the Edwardsville high was a result all three individual months as shown in Fig. 16, being
of heavy rains from relatively few storms. Seventy-six most noticeable in June. Highs to the east were closer
of the 330 storms produced rainfall amounts in the to the metropolitan area in July and August. In all
Edwardsville area that were significantly above the three months the Bottomlands high (not seen in the
network means. Storms contributing to the Edwards- long-term analysis) was observed north-northwest of
ville high were generally of an intense nature, reflect- the city, though it was not very impressive in June.
ing the common METROMEX observation that com- The high along the southeast edge of the network oc-
posite rainfall distributions tend to be highly organized curred in all three months. In June a high was ob-
for heavy storms and rather random in nature in the served over the city, but it was less in July and not
case of weaker systems. present at all in August. However, in August a very
Other features are distinguished by the 140 em iso- well defined rainfall maximum was observed extending
hyet; namely, highs located 1) northeast of Alton 2) from Collinsville to south of Belleville and from the
between the Mississippi and Missouri Rivers (Bottom- Mississippi eastward about 20 km. The maximum
lands) NNW of the city, and 3) at the extreme south- value of this high was in excess of 60 em. The relative
east edge of the network. The 5-year patterns (Fig. 15) urban high of June and July is reflected in the 5-year
generally agree with that of the long-term period (Fig. total rainfall pattern as an extension of the Edwards-
14) and with results of the climatological study of Huff ville high to the southwest over the St. Louis city cen-
and Changnon (1972), and provide credibility of the ter, and the Collinsville-Belleville maximum appears
representativeness of the project study period. as the southward extension. The low over the west
The distribution of the mean network summer rain- and southwest suburbs was strongest in August, being
fall is given in Table 2. During the 5-year period the barely evident in June.
months of June, July and August were very dry in Huff(1977b) also studied the precipitation in the fall,
1972, 1974, and 1971, respectively. The wettest month
was June 1973 followed closely by August 1974. Com- TABLE 2. Mean network rainfall (em) for June, July, August and
each summer, 1971-75.
parison between the St. Louis summer point rainfall
during 1971-75 with the long term 1901-75 National Year June July August Summer
Weather Service downtown record, reveals that the
1971 5.99 10.52 1.83 18.34
short term mean was 86% of the long term value . Us- 1972 3.51 7.32 9.14 19.96
ing all possible 5-year periods (moving totals), there 1973 13.88 9.22 6.88 29.97
1974 6.81 2.87 13.79 23.47
were 13 times (19% of possible cases) when equal or 1975 7.06 9.88 12.29 29.24
less rainfall was observed during the summer.
30
a. JUNE 1971 -1975
c . AUGUST 1971- 1975
FIG. 16. Monthly rainfall totals (em) for June, July, August during 1971-75. After Huff (1977a).
winter and spring seasons and found a number of sim-
ilarities to the summer season as regards the place-
ment of high and low values as seen in Fig. 17b. The
fall season showed a low rainfall area beginning in the
western St. Louis suburbs and extending eastward al-
most to the outer edge of the network (Fig. 17a). This
low was the separation between high precipitation
areas generally oriented WNW-ESE to the north and
south of the urban low. During fall and winter, how-
ever, the southeast high was replaced with a relative
low. The winter pattern shows high precipitation from
Edwardsville to the south, in the Bottomlands and Hill
regions, similar to summer. The spring season showed
the greatest deviation from the seemingly established
pattern although a high was present in the Edwards-
ville-Collinsville area. However, the Bottomlands high
was replaced with a low and the southeast summer
high showed no distinguishable feature. Readily seen
in all of the seasons, though, is the relative maximum
of precipitation east of the Mississippi River establish-
ing a seasonally persistent heavy precipitation area
oriented essentially north-south.
The diurnal distribution of summer rainfall was de-
scribed by Huff (1977c) for the entire network and
METEOROLOGICAL MONOGRAPHS
b . JULY 1971-1975
selected smaller areas. Fig. 18 shows the time distri-
bution of rainfall derived from 3-hour moving averages
for the entire (circular) network, an upwind control
area west of St. Louis of 490 km2 , and the 211 km2
area encompassing the Edwardsville maximum. The
diurnal structure is pronounced and apparently re-
flects both the solar heating cycle and urban heat-is-
land phenomenon. The period of maximum rainfall
occurred in the 3 h period ending at 1800 CDT appar-
ently associated with the diurnal peak in solar heating.
This, of course, should be expected owing to the en-
ergetics of storm formation and propagation. The ap-
pearance of a secondary maximum in all three curves
of Fig. 18 suggests that the natural nocturnal thunder-
storm anomaly is a major influence on this portion of
the diurnal cycle. Stations in the Edwardsville high
region showed both primary and secondary maxima
that were significantly higher than those of the total
network and control stations, suggesting that a pos-
sible coupling between local perturbations and natural
phenomena acted to intensify the amplitude in this lo-
cality at all times. The secondary maximum coincides
closely with the nocturnal heat island maximum which
may be a contributing factor. All of the curves show
METROMEX: A REVIEW AND SUMMARY 31

Mean = 97. 1 em
St. Dev . = 6. 0
a. Fa 11 b. Winter

Mean = 100.0 an
c. Spring St . Oev . = 9. 3

FIG. 17. Fall, winter, and spring precipitation (em) for 1971-75. After Huff (1977b).

a well defined minimum between 1200 and 1400 of

- - Ent i re Network
6
- - - Edwards vi 11 e High
nearly equal magnitude coinciding with the minimum
- - Upwind Centro 1 heat-island effect and during a period when the con-
vective temperature is not typically achieved. The first
c:
5 maximum coincides with the period of minimum urban
I
u
I mixing ratio (see Fig. 9) while the second (nocturnal)
....J I maximum in rainfall nearly coincides with the urban
....J
<X: I
LL 4 I maximum mixing ratio.
z
I
Cii I The spatial characteristics of the diurnal rainfall dis-
I
w
I tribution are shown in Fig. 19 (Huff, 1977c). High rain-
""w 3
<X:
\
fall areas essentially surrounded the urban areas in the
""> \.......,/ ''
<X:
--~
3 h period 1200-1500 COT. The St. Louis urban area,
""w 2
<X: including the western part of East St. Louis , was char-
>-
w' acterized by low rainfall. Taken alone, this pattern
>
LL
suggests urban-related diminishment of rainfall and
certainly no enhancement. This 3 h interval, however,
may be regarded as an atmospheric priming period
when the surface is approaching the convective tem-
02L4-~-L--L--L--~_J--~~L__L__ L__L__J
perature and the mixing depth is increasing over the
06 08 10 12 14 16 18 20 22 24
THREE-HOURS ENDING (CDT)
city. The diurnal curve of rainfall (Fig. 18) shows a
rapid increase between 1200 and 1500 suggesting rapid
FIG. 18. The 5-year, 3 h moving average rainfall as a function of
hour for an upwind control area, the Edwardsville area and the changes in the atmospheric environment conducive to
entire network. After Huff (1977c). precipitation.
32 METEOROLOGICAL MONOGRAPHS

KILOMETERS
o a
=
16
0 5 10
STATl1I'E ~HL
ES
Mean = 44 rrm _
St. Oev. = 12.2 rrm
a. 1200-1500 COT b. 1500-1800 COT

c . 1800 - 2100 COT d . 2100-0000 COT

JUNE-AUGUST 1971-1975
FIG. 19. The spatial distribution of average summer rainfall (mm) for selected three-hour periods.
After Huff (1977c).

In the period 1500-1800 CDT the network mean
rainfall attains the 3-hour maximum with a mean of 44
mm. The heaviest rainfall area was observed in and
east of St. Louis. The center of greatest rainfall was
in East St. Louis (more than 80 mm) and a nearby
secondary maximum at Granite City. Low rainfall
( <25 mm), was observed southwest of St. Louis to
the far east, and the northwest edge of the network.
The rainfall pattern is much less organized during
the next 3 h period 1800-2100 CDT (Fig. 19c). How-
ever, the heaviest rainfalls were observed in the Bot-
tomlands and east of the Mississippi. A general area
of below mean network rainfall (35 mm) was recorded
over St. Louis from the river southwestward. This
pattern of precipitation, with numerous maxima and
minima, is suggestive of less organized cloud systems
or cloud systems which were in a decaying state.
The 2100-0000 pattern (Fig. 19d) shows a strong
rainfall maximum in the Edwardsville area discussed
previously in conjunction with the diurnal distribu-
tions. This late evening maximum in the areal distri-
bution extends from Granite City to Edwardsville and
is of equal magnitude to the 1500-1800 maximum in
this region. However, this high did not extend east of
St. Louis as the afternoon peak. The late evening max-
imum in the Granite City-Edwardsville area is likely
associated with urban enhancement of the natural noc-
turnal thunderstorm anomaly through interactions of
the urban heat-island and atmospheric processes.
The results from analyses of the 1971 rainfall data
suggested the need to extend the raingage network
further east and northeast to measure the suspected
extension of urban-induced rainfall changes beyond
the 42 km (26 mi) radius network (Fig. 10, Chapter 1).
A network of gages with lesser spatial density was
installed in 1972 and operated through 1975. Because
the period of data collection was only four years and
the gage density somewhat different, it was not pos-
sible to include the data in Fig. 15. However, to ad-
dress the question of the areal extent of the urban
related rainfall pattern the 4-year data set was studied
independently by Schickedanz and Changnon (1977).
The 4-year circular network pattern of summer rain-
fall (Fig. 20) is very similar to the 5-year pattern (Fig.
METROMEX: A REVIEW AND SUMMARY
Surmer 1972-1975
o.
FIG. 20. The total summer rainfall (em) for the major and the
extended area networks for 1972-75. After Schickedanz and Chang-
non (1977).
15) reflecting representativeness of the sample. The
addition of information to the east reveals a relative
low, with values less than those extending N-S, but
with a magnitude - 90 em, somewhat greater than the
values west of the city. These results suggest that I)
the major rainfall anomaly is one of increase and lim-
ited to within 30 km of the city, and 2) that beyond it,
values return to the background level as found west
of St. Louis. No downwind decrease of any apparent
significance is suggested.
A simple summary of the major features of summer
rainfall patterns for 1971-75 is shown in Fig. 21. The
locations of the three centers of relative high and low
rainfall for each of the five summers are shown. The
high or low areas were defined by mean rainfall values
FIG. 21. The locations of the three major high and low rainfall
centers for each of the 1971-75 summers.
33
for any four adjacent gages and separated by at least
one row of gages from any other high or low area.
Dividing the network into quadrants, it is notable that
the NE quadrant contains nine highs and only two
lows whereas the SW quadrant contains six lows and
no highs. The NW and SE quadrants are very similar
with nearly equal numbers of highs and lows. This
simple analysis supports the findings of Changnon
(1979) where the rainfall total was studied by such a
quadrant analysis. It also reinforces intuitive concepts
of the relationship between average surface wind di-
rection (SW) and storm motion (W), and the expected
location for possible urban affected rainfall (NE).
2.6 Severe weather
The topic of severe weather studied in METRO-
MEX includes an analyses of heavy rainstorms (;:;;.25
mm), thunderstorms, hailstorms, wind gusts and tor-
nadoes. The basic instruments used and their relative
locations in the network are described in Chapter I.
Data for tornadoes and funnels aloft were taken from
the Environmental Data Service (EDS) storm data rec-
ords for the summers of 1971 through 1975.
2.6.1 HEAVY RAINSTORMS
Fig. 22 shows the frequency distribution based on
25 mm rains at each gage during the summers of
1971-75 (Huff, 1977d). A pronounced high was located
in the Edwardsville area, where the recorded maxi-
mum of 23 events at one gage was twice the network
1,[r,c:Y RAWf.AlL ~ 25m- OURHl:; SUH."1E:RSor 1971·1975
Of' S:TOR~
f~EQ
FIG. 22. Frequency of 25 mm storm events for 1971-75. After
Huff (1977d) .
34
mean of 11.5 occurrences and over three times the
standard deviation of 2.9. Least frequency occurrence
of heavy storms was west of St. Louis. In general, the
network pattern is similar to that for total rainfall and
indicative of a major urban effect.
The distribution of total rainfall in these 25 mm
storms (not shown) is also similar to the total summer
rainfall pattern. A maximum of91.90 em (36.18 inches)
was recorded northeast of Edwardsville and this is
nearly twice the network mean of 47.17 em (18.57
inches) and nearly eight times the standard deviation
of 11.56 em (4.55 inches). Again, the primary low was
located west to southwest of St. Louis which is usually
upwind of the urban-industrial area. At the center of
the Edwardsville high, over 60% of the total summer
rainfall occurred in the 25 mm storms. Only 30 to 35%
of the summer rainfall occurred in 25 mm storms im-
mediately west and southwest of St. Louis.
A comparison was made by Huff (1977d) of the dif-
ference in rainfall for 1971-75 between stations in the
Edwardsville high and the network mean rainfall for
total summer rainfall and for the rainfall associated
with storms of various intensity. The purpose was to
determine how much of the rainfall excess (departure
from network mean) in the Edwardsville area could be
explained by the more frequent occurrence of the
heavy rainstorms there.
Five raingage stations that were in the core of the
Edwardsville high were selected. Comparisons were
made for storm rainfall in five classifications of inten-
sity.
Results are summarized in Table 3 where differ-
ences between the total rainfall at the five stations and
the network mean are shown for the five storm inten-
sity categories. Also shown are the total differences
between each gage and the network mean, plus aver-
age differences in each classification for the various
intensities. It is obvious that the rainfall excess in the
core of the Edwardsville high is produced almost en-
tirely by the rainfall excess in the heavy storms (;;.25
mm, or 1 inch). Only gage 37 had a substantial portion
of its excess produced in storms of lighter intensity.
Reference to the five-station averages shows all except
0.11 em (0.04 inch) was contributed by the heavy
TABLE 3. The difference (em) between the five-station Edwardsville high and the entire network mean rainfall in various intensity
categories for summer, 1971-75. From Huff(1977d).
Storm rainfall (mm)
Gage <2.5 2.5-5.9
37 -1.17 -0.30
38 -1.83 -1.45
50 -1.04 -0.86
51 -1.65 -2.13
52 -1.07 +0.33
Average -1.35 -0.88
METEOROLOGICAL MONOGRAPHS
storms. Conversely, part of the rainfall excess in the
25-mm storms was offset by less than average rainfall
amounts in the relatively light storms producing
amounts up to 12.4 mm (0.49 inch).
As part of the investigation of the relationship of
heavy rainstorms to the network anomalies, Huff
(1977d) also analyzed the distribution of hourly
amounts equaling or exceeding 12.5 mm (0.5 inch) and
25 mm (1 inch). These intensities are usually associ-
ated with the heavy rain producers, and, therefore,
occur infrequently as indicated by the network mean
frequencies of 18.8 and 3. 7 occurrences, respectively,
for the 12.5 and 25 mm intensities during the five sum-
mers. The patterns of the 12.5 and 25.0 mm hourly
amounts are shown in Fig. 23.
The spatial distributions are very similar for the two
intensities, and are strikingly similar to the total sum-
mer rainfall pattern (Fig. 20) and the pattern for the
frequency distribution of heavy rainstorms (Fig. 22).
The Edwardsville high, the Bottomlands high, the up-
wind low, and the SE quadrant highs are all outstand-
ing features of the hourly high rate maps.
Changnon (1977a) showed the pattern for the num-
ber of point occurrences of 5 min rain amounts with
2-year or great recurrence interval value ( 11.43 mm or
0.45 inch) for 1971-75. The greater number of high
values (;;;.3) shown in Fig. 24 are in the potential effect
area delineated downwind of the St. Louis and Alton-
Wood River area. The no-effect (west) area has ex-
tensive regions of no occurrences of 5 min amounts of
this magnitude. The difference between the point av-
erages in the effect and no-effect areas indicates an
83% increase. A raincell is defined as a closed isohye-
tal entity within the enveloping isohyet of a synoptic
storm system consisting of one or more rainshowers
or thunderstorms. Each shower or thunderstorm may
be single or multicellular. In a multicellular storm sys-
tem, the raincell incorporates an isolated area of sig-
nificantly greater intensity than the system-enveloping
isohyet. In isolated single-cell storms, the raincell is
uniquely defined by the separation between rain and
no rain. During 1971-75, 8119 such raincells were ob-
served in the network.
Heavy raincells were defined by Huff (1977e) as
6.0-12.4 12.5-24.9 ;;.25.0 Total
-12.12 + 10.34 +28.98 +25.73
-0.33 -7.26 +44.75 +33.88
-3.43 + 1.42 +40.28 +36.37
+2.18 -6.40 +36.53 +28.53
-5.74 +2.46 +35.36 +31.24
-3.89 +0.11 +37.18 +31.17
METROMEX: A REVIEW AND SUMMARY 35

• · Frequency of hourly r ainfall ~ 25mm b. Frequency of hou r ly rainfall ~ 12.5m

FIG. 23. requency distribution of heavy hourly rainfall , 1971- 75. After Huff (1977d).

those having mean rainfall equaling or exceeding 6 mm topographic-effect cells were those which developed
(0.25 inch). This subset of 659 heavy raincells com- over or crossed a distinct topographic feature. The
prised only 8% of the total, but accounted for 51% of two significant features in the METROMEX network
the total water yield. (Fig. 7, Chapter 1) are the Ozark foothills in the south-
The stratification of the heavy raincell data was lim- western part of the network and the Bottomlands in
ited because of the inclusion of only 8% of the total the northwestern quadrant.
sample in the hydrometeorological analyses. The sam- No effect cells were defined as those which had no
ple was divided into three basic types for evaluation. contact with the urban environment or topographic
There were urban-effect, topographic effect and no- features during their lifetime. These were essentially
effect raincells. A cell was placed in the urban-effect rural cells moving across the relatively flat farmlands
group if it developed over one of the two urban-in- within the network.
dustrial areas (St. Louis and Wood River) or passed For hydrologic applications and most other uses,
over one of these areas during its lifetime. Similarly, the major interest is in the net effect of the urban and/
or topographic effects on the rainfall distribution char-
acteristics in the affected area. Therefore, Table 4
shows the combined or net effect of urban and topo-
graphic influences on the distribution of water yield in
heavy raincells. The urban effect is pronounced. Thus,
the median water yield is 70% greater than in the con-
trol cells and ranges from 65 to 72% along the fre-

TABLE 4. Comparison of water yield between urban-effect (U),

topographic effect (T) and no-effect (C) raincells among all cells
with mean rainfall of 6 mm (0.25 inch) or more during summers of
1971- 75. From Huff (l977d).

Cumulative Cell volumes (IO- • m3 )

percent of equalled or exceeded Volume ratios
raincells u T c V IC TIC UIT

5 36.99 26.76 21.45 1.72 1.25 1.38

10 27.37 19.97 15.91 1.72 1.26 1.37
20 18.86 14.06 10.97 1.72 1.28 1.34
30 14.55 10.97 8.51 1.71 1.29 1.33
40 11.47 8.88 6.78 1.69 1.31 1.29
50 9.31 7.28 5.49 1.70 1.33 1.28
60 7.52 5.92 4.50 1.67 1.32 1.27
70 6.04 4.75 3.58 1.69 1.33 1.27
80 4.62 3.70 2.77 1.67 1.33 1.25
90 3.21 2.59 1.91 1.68 1.35 1.24
FIG. 24. Pattern based on point frequencies of 5 min rainfall values 95 2.34 1.97 1.42 1.65 1.39 1.19
of2-year or greater recurrence interval value, 1971- 75. After Chang- N 224 72 331
non (1977a).
36 METEOROLOGICAL MONOGRAPHS

quency distribution curves. The topographic enhance- mtmmums northwest and south of St. Louis. The
ment is less, having a median of 33%, and a range range of values extends from about 25 days with thun-
from 25 to 39% along the frequency curves. As indi- der at the Scott Air Force Base site to 15 days per
cated by the U/T ratios in the last column of Table 4, summer at Waterloo, Illinois.
the urban effect is approximately 20 to 40% greater The addition of three thunder recording stations in
than the topographic effect with a median of 28% in an arc east of the primary network used before 1973
the 1971-75 sample. It should be emphasized that the allowed greater area resolution of the pattern for the
topographic sample is small compared with the other 1973-75 period. It can be seen (Fig. 25b) that the thun-
two cell types, and should be considered only a first der day frequency is more confined to the research
approximation of the average enhancement associated area than was obvious from Fig. 25a. The only major
with the hills and Bottomlands in the St. Louis region. difference in the pattern between the 3-year and 5-year
average maps is the increase of frequency in the urban
2.6.2 THUNDER area of 2-5 days.
A total of 13 points within the METROMEX net- Changnon (1977b) analyzed the total thunder days
work were used by Changnon (1977b, l978b) to in- for all stations to depict the frequency of events for
vestigate the nature and magnitude of thunderstorm each month and the summer season. The results are
activity in proximity to the metropolitan area. These shown in Table 5. The month of June had an average
sites are shown in Fig. 10, Chapter I. of 13.8 thunder days with individual years ranging be-
A thunder day is defined as the occurrence of thun- tween 12 in 1972 and 1974 to 17 in 1971. July averaged
der at least at a single point during a midnight to mid- 12.4 days ranging from 9 in 1971 to 16 in 1975, and
night period. In general, as shown by Changnon August averaged 12 days with a range from 7 in 1973
(1977b), each of the summer months shows a distinct to 18 in 1974. One can see that over the 5 years, the
high frequency of thunder days over the urban center thunder day frequency appeared to compensate for the
extending east and northeastward to the edge of the low or high values in a particular month. The seasonal
research area. Considering the westernmost sites as average was 38.2 days with a range from 34 to 43 days.
a reference, the urban and eastern area show increases Using the thunder recorder data, Changnon (1977b)
of as much as 80% in June but only 50% increases in also studied the duration of thunder. The number of
August. Only in the month of August is there a slight minutes of thunder at each station was tabulated and
shift in the pattern suggesting a west-southwest to sorted into four categories of thunder rate frequency
east-northeast band of relative high frequency extend- (determined by the rate of thunder peals during a pe-
ing across the entire diameter of the research circle. riod of thunder). Average durations were calculated
The summer average thunder days for the 5-year for each of the categories and a comparison between
period during the entire summer (Fig. 25a) indicates urban and rural sites showed a 83% increase for the
the greatest maximum northeast of Belleville with urban stations during intense thunder segments as
[XPLA IIA ! ION
• AUTO.'.,<I I C THUNDER RECORDERS
0 OBSERVATJOt STAT IO.'lS

a . S1.11111er average thunder days 1971- 1975 KI LOMETERS b. Su rt111e r average thunder days 1973-1975

=
0 8 16
0 5 10
STATUTE MILES

FIG. 25. Thunder day pattern for 1971-75 and over the extended area network for 1973-75. After Changnon (1977b).
METROMEX: A REVIEW AND SUMMARY 37

TABLE 5. Number of thunder days within the research circle for Observations were compiled for 2 h periods and plot-
the entire five-summer period 1971-75. From Changnon (1977b). ted as a percentage of the total number of thunder
Summer occurrences. As seen in Fig. 27 the greatest percent-
June July August total age of thunder at both St. Louis University and Tyson
Valley was observed between 1600 and 2400 CDT. The
1971 17 9 11 37
1972 12 10 12 34 St. Louis University station exhibited a single mode
1973 14 14 7 35 distribution whereas the Tyson Valley site was bi-
1974 12 13 18 35
1975 14 16 12 42 modal with one maximum at 1600 and the second at
2000. The Edwardsville station showed more uniform-
Totals 69 62 60 191
ity in the distribution of thunder with two minima oc-
curring at 1200 and 0200 (actually for the periods
more than 60 peals per hour. Changnon (1977b) sum- 1000-1200 and 0000-2000, respectively). The after-
marized these data by showing the durations of the noon and early evening maximum in thunder agrees
intense segments at St. Louis University and Ed- qualitatively with the diurnal rainfall pattern (Fig. 18).
wardsville (urban stations), and Tyson Valley and However, the bimodal distribution found for rainfall
Beaver Creek (rural stations). The marked differences was not so easily recognized in the thunder data. In
between the urban and rural sites are seen in Fig. 26. fact, the relative maximum of thunder in the early
For instance, 20% of the time intense durations at the morning hours at Edwardsville is associated with a
rural stations are 13 min or less compared with 28 period of decreasing rainfall accumulation. It is of in-
min or less at the two urban stations. Similar major terest to note that the Edwardsville curve tends to
differences exist at the longer durations, indicating broaden after maximizing at 1800 indicating a prolon-
that the duration and the characteristics of thunder gation of thunder activity as compared to the Tyson
segments are markedly different at locales where the Valley and St. Louis University sites.
thunder has been potentially affected by the urban- The above results suggest that a simple segregation
industrial areas. of urban and rural stations should provide a useful
The diurnal distribution of thunder occurrences was means of assessing increases in thunder activity as-
reported by Changnon (1977b) for the 1971-75 period. sociated with the urban area. The results of such an
analysis, where the St. Louis University, Edwards-
ville, Scott A.F.B., and Livingston locations were
classified as urban sites, are summarized in Table 6.
These indicate 45% increases in both the frequencies
of thunder period and thunder days between rural and
urban stations, an effect that is reflected in the total
thunder duration periods as well. Statistics for thunder
occurrences as functions of synoptic conditions and
month are presented in Table 6 as well (Changnon,
1978b). In general they are in concordance with the
precipitation trends. Thunder periods are seen to max-
imize in June, with a subsequent decline frequency
throughout the summer.

"'
"'
; 15

u
u
0
>-
-'
"'=>
0
~ 10
,_
<(

:='
...0
~
~ 5~~~~~~~~~-J--~~L-~~~~~_J
2-HOUR PERIOD ENDING AT INOICATEO HOUR
PERCENT TIME DURATION WILL BE LESS THAN THAT INDICATED
Fro. 27. Normalized diurnal distribution of thunder at Tyson Val-
FIG. 26. Accumulated frequency of intense thunder duration ob- ley, St. Louis University and Edwardsville. The data are 2 h moving
tained from recorded thunder peals. After Changnon (1977b). averages plotted on the ending hour. After Changnon (l977b).
38 METEOROLOGICAL MONOGRAPHS

TABLE 6. Urban-related increase in summer thunderstorm characteristics at St. Louis. From Changnon (1978b).

Percent increase

Urban (U) Rural (R) Difference

(u/ x too)
Thunder days• 26.8 18.5 8.3 +45
Thunder periods•·• 33.6 23.1 10.5 +45
East-only versus west-only days2 •5 39.0 22.0 17.0 +77
Duration of thunder (all values in minutes)
Thunder periods• 145 93 52 +56
Very light thunder rates3·• 49 44 5 +II
Light thunder rates 3•4 64 48 16 +31
Moderate thunder rates3•4 56 46 10 +22
Intense thunder rates3.4 86 47 39 +83
Synoptic weather conditions and thunder
Thunder periods for cold fronts• 12.3 9.0 3.3. +33
Thunder periods for warm-stationary fronts• 8.7 5.0 3.7 +74
Thunder periods for air mass• 9.3 4.3 5 .0 +ll6
Thunder periods for squall areas and lines• 53.3 40.4 12.9 +32
Monthly differences (U-R) June July August
Average point thunder periods' + 66% +43% + 34%
East-only versus west-only days 2 + 129% +71 % +38%

'Discrete periods of thunder separated by ;;;. I h from other periods.

2 Based on comparisons of days when there were only thunderstorms to the east of St. Louis (at one or all of four stations), or only to
the west (four stations).
3Thunder rates were defined on peals per hour basis: very light .;:5 h- •, light 6-11 h- 1 , moderate 12-60 h-', and intense > 60 h- 1 •
4Point summer average values, 1973-75.
"Area total values, 1971-75.
"Point average totals, 1973-75.

2.6.3 HAIL
Contours based on hail occurrences for the 1971-75
summer periods are shown in Fig. 28. This 5 year plot
is rather typical of the corresponding individual annual
FIG. 28. Hailfall pattern derived from 1971-75 data. After Chang-
non (1977c).
plots in that high frequency areas occurred just east
of a line between St. Louis and Alton-Wood River,
and in a general north-south line ~ 20-30 km east of
St. Louis. Another prominent high exists northwest of
St. Louis in the river bottomland. These high-intensity
areas can be compared with the general low-frequency
region west and southwest of the St. Louis area.
The network was arbitrarily divided into eastern and
western semi-circles (Fig. 28) to obtain an estimate of
urban effects (east) and no-effects (west). A summary
of the point hailfall values for the 5 years was given
by Changnon (1978b) and is shown in Table 7. For
every parameter presented, the East area (effect) val-
ues exceed the West (no-effect) values by amounts
ranging from 0 to 100%. Every aspect of the point
hailfall characteristics supports the conclusion that
there are local effects on hail activity east of the city.
These effects appear in a more sustained production
of hail and more frequent production of hail east of
the city.
Likewise the average impact energy, mean hailstone
diameter, and average flux of hailstones all were
greater on the eastern side of the study area, features
that were reflected in the hailstreak results as well. A
final noteworthy feature of Table 7 is its indication of
urban enhancement of hail as a function of the diurnal
cycle. From midnight to noon no differences between
effect and no effect areas are apparent, whereas sig-
nificant differences occur in the following 12 h period.
These total hailfall results are somewhat different from
METROMEX: A REVIEW AND SUMMARY 39

TABLE 7. Point hailfall characterization for east and west semicircles divided along the Mississippi River. After Changnon (1978b).

Percent increase
No
Urban effect
value (E)
urban effect
value (NE) Difference
(E~~E X 100)

Mean energy (J per 0.1 m3 ) in effect area versus that

in no-effect area 1.45 0.74 0.17 +96%
Average frequency of point hailfalls, effect area versus
no-effect area 5.5 4.3 1.2 +28%
Median hailstone diameter (em) in effect area versus
that in no-effect area 1.27 0.97 0.3 +32%
Average hailstones per 929 em•, effect hail area versus
no-effect hail area 372 259 113 +44%
Effect area hailfall frequency versus no-effect area
frequency, differences for 6 h periods
0000-0600 CDT 0.6 0.6 0 0
0600-1200 CDT 0.3 0.3 0 0
1200-1800 CDT 1.2 1.0 0.2 +20%
1800-2400 CDT 1.7 1.0 0.7 +70%
Comparison of hail streak characteristics (E vs NE) for
53 hail periods when both effect and no-effect
hailstreaks occurred
Hailstreak durations (min) 14.9 14.2 0.7 +5%
Areas of hailstreak (km 2 ) 37.4 36.4 1.0 +3%
Average point duration (min) 2.9 2.1 0.8 +40%
Maximum hailstone diameter 1.41 1.15 0.26 +22%
Point (929 cm 2 ) number of hailstones 74 54 20 +36%
Energy (J per 0 .I m3 ) 1.35 0.67 0.58 +100%
Mean rainfall in hailstreaks (em) 2.05 1.69 0.36 +22%

the hailstreak data. These suggest that, while an urban
enhancement is apparent in the late afternoon and eve-
ning hours, an attenuation effect may exist earlier in
the day.
2.6.4 WIND GUSTS
Detailed wind gust data were collected at eight sites
distributed around the circumference of a circle with
a radius of about 32 km from the center of St. Louis.
Analyses focused on maximum gusts recorded during
each hour (Changnon, 1977d). One analysis was based
on a count of the strongest gust in each hour. Data
were grouped on the basis of three wind stations lo-
cated to the southwest, west, and northwest of St.
Louis (classed as having no potential urban effect),
and three stations located to the southeast, east and
northeast of St. Louis classed as potentially urban af-
fected sites.
TABLE 8. A comparison between potential effect and no-effect
areas of the rrequency of wind gusts. From Changnon, 1977d.
The average point frequencies of maximum hourly
gust :;:;.48 and :;:;.64 km h- 1 appear in Table 8. The effect
area frequency of hours with :;:;.48 km h- 1 gusts is dou-
ble ( 100% increase) the no-effect area frequency. The
difference in gusts :;:;.64 km h- 1 reveals a 91% increase
in the urban effect area. All the gusts at the effect and
no-effect sites during 1971-75 were associated with
thunderstorms.
2.6.5 TORNADOES AND FUNNEL CLOUDS
The data on tornadoes and funnels aloft from pub-
lished data for the summers of 1971 through 1975 were
studied by Changnon (1977d). The goal was to inspect
for any evidence of urban effects in their patterns.
During the 15-month sample, there were ten tornadoes
and seven funnels aloft recorded within a 97-km radius
of St. Louis. It should be noted that the summer sea-
son is not a primary tornado season in this region and
a low frequency of events in 5 years makes derivation
of conclusions about urban effects on tornadoes very
difficult.

Number of gusts at a point REFERENCES

;;;.48 km h- 1 ;;;.64 km h- 1
Average Range Average Range Ackerman, B., and J. W. Mansell, 1978: Thermodynamic structure
of the PBL at midday. Summary of METROMEX, Vol. 2,
Potential effect 34 27-45 6.3 5-8 Causes of Precipitation Anomalies, Illinois State Water Survey
(east) area Bull. 63, Urbana, 129-164.
Potential effect 17 10-25 3.3 2-5 Auer, A. H., Jr., and R. A. Dirks, 1973: 1973 Annual Report to the
(west) area NSF for Modification of Convective Cloud Activity by an Ur-
Difference 17 3.0
(east - northeast)/ 100 91 ban Area. University of Wyoming, Laramie, 78 pp.
northeast [%] Changnon, S. A., Jr., 1977a: Impacts of urban modified precipitation
in the St. Louis area. Summary of METROMEX, Vol. 1,
40
Weather Anomalies and Impacts, Illinois State Water Survey
Bull. 62, Urbana, 199-248.
- - , 1977b: Thunder analysis for METROMEX. Summary of
METROMEX, Vol. I, Weather Anomalies and Impacts, Illi-
nois State Water Survey Bull. 62, Urbana, 113-141.
- - , 1977c: Studies of urban effects on hail characteristics in MET-
ROMEX. Summary of METROMEX, Vol. I, Weather Anom-
alies and Impacts, Illinois State Water Survey Bull. 62, Ur-
bana, 142-174.
- - , 1977d: Other severe local storm phenomena. Summary of
METROMEX, Vol. I, Weather Anomalies and Impacts, Illi-
nois State Water Survey Bull. 62, Urbana, 193-195.
- - , 1978a: Surface wind climatology. Summary of METROMEX,
Vol. 2, Causes of Precipitation Anomalies, Illinois State Water
Survey Bull. 63, Urbana, 71-99.
- - , 1978b: Urban effects on severe local storms at St. Louis. J.
Appl. Meteor., 17, 578--586.
- - , 1979: Rainfall changes in summer caused by St. Louis. Sci-
ence, 205, 402-404.
Changnon, S. A., Jr., and R. G. Semonin, Ed., 1975: Studies of
Selected Precipitation Cases from METROMEX. Illinois State
Water Survey Report of Investigation No. 81, Urbana, 329 pp.
- - , F. A. Huff, and R. G. Semonin, 1971: METROMEX: An
investigation of inadvertent weather modification. Bull. Amer.
Meteor. Soc., 52, 958--967.
Dirks, R. A., 1974: Urban atmosphere: Warm, dry envelope over
St. Louis. J. Geophys. Res., 79, 3473-3475.
Dytch, H. E., 1972: METROMEX temperature and aerosol data
reduction: Analysis of the urban anomaly. Project METRO-
MEX-1, University of Chicago Tech. Note 45 [NTIS PB
218,678], 73-90.
Dzurisin, G., 1978: Average dew points. Summary of METRO-
MEX, Vol. 2, Causes of Precipitation Anomalies, Illinois State
Water Survey Bull. 63, Urbana, 43-62.
Grosh, R. C., and R. G. Semonin, 1973: Moisture budgets and wind
fields of thunderstorms passing over an urban area in the Mid-
west. Preprints Eighth Conf Severe Local Storms, Denver,
Amer. Meteor. Soc., 130-137.
Hammon, W. H., and F. W. Duenckel, 1902: Abstract of a com-
parison of the minimum temperatures recorded at the United
States Weather Bureau and the Forest Park meteorological ob-
servatories, St. Louis, MO, for the year 1891. Mon. Wea. Rev.,
30, 12-13.
Hilberg, S.D., 1978a: Temperature analysis. SummaryofMETRO-
MEX, Vol. 2, Causes of Precipitation Anomalies, Illinois State
Water Survey Bull. 63, Urbana, 11-24.
- - , 1978b: Diurnal temperature and moisture cycles. Summary of
METROMEX, Vol. 2, Causes of Precipitation Anomalies, Il-
linois State Water Survey Bull. 63, Urbana, 25-42.
Huff, F. A., 1977a: 1971-1975 rainfall pattern comparisons. Sum-
mary of METROMEX, Vol. 1, Weather Anomalies and Im-
pacts, Illinois State Water Survey Bull. 62, Urbana, 13-29.
- - , 1977b: Fall, winter, and spring precipitation on network. Sum-
mary of METROMEX, Vol. 1, Weather Anomalies and Im-
pacts, Illinois State Water Survey Bull. 62, Urbana, 30-36.
METEOROLOGICAL MONOGRAPHS
1977c: Diurnal distribution of summer rainfall. Summary of
METROMEX, Vol. 1, Weather Anomalies and Impacts, Illi-
nois State Water Survey Bull. 62, Urbana, 42-52.
- - , 1977d: Distribution of heavy rainstorms. Summary of MET-
ROMEX, Vol. 1, Weather Anomalies and Impacts, lllinois
State Water Survey Bull. 62, Urbana, 175-181.
- - , 1977e: Distribution of heavy raincells. Summary of METRO-
MEX, Vol. 1, Weather Anomalies and Impacts, Illinois State
Water Survey Bull. 62, Urbana, 182-192.
- - , and S. A. Changnon, Jr., 1972: Climatological assessment of
urban effects on precipitation at St. Louis. J. Appl. Meteor.,
11, 823-842.
- - , and E. E. Schlessman, 1974: 1973 analysis of monthly, sea-
sonal, and storm rainfall. Interim Report ofMETROMEX Stud-
ies: 1971-1973, F. A. Huff, Ed., NSF Grant GI-38317, 17-29.
--,and J. L. Vogel, 1978: Urban, topographic and diurnal effects
on rainfall in the St. Louis region. J. Appl. Meteor., 17,
565-577.
Jones, D. M. A., and P. T. Schickedanz, 1974: Interim Report of
METROMEX Studies: 1971-1973, F. A. Huff, Ed. NSF Grant
GI-38317, 98--120.
Landsberg, H., 1974: Inadvertent atmospheric modification through
urbanization. Weather and Climate Modification, Wiley, New
York, 726--763.
Lawson, R. P., 1973: Urban induced influence on convective cloud
activity. Rep. No. AR107, Dept. Atmos. Resour., University
of Wyoming, Laramie, 53 pp.
Mitchell, J. M., Jr. 1961: The thermal climate of cities. Symposium
Air Over Cities, SEC Tech. Rep. A62-5, U.S. Dept. Health
Education and Welfare, Cincinnati, 131-145.
Schickedanz, P. T., and S. A. Changnon, Jr., 1977: Extended area
rainfall findings. Summary of METROMEX, Vol. I, Weather
Anomalies and Impacts, Illinois State Water Survey Bull. 62,
Urbana, 193-195.
Semonin, R. G., 1974: Urban-induced weather modification. 1974
Earth Environment and Resource Conf Digest Tech. Papers,
Institute of Electrical and Electronics Engineers, Philadelphia,
40-41.
Shea, D. M., 1976: Observations of the regional influence on ther-
modynamic properties and aerosol patterns in the plume down-
wind of St. Louis. Rep. No. AS117, Dept. Atmos. Sci., Uni-
versity of Wyoming, Laramie, 152 pp.
- - , and A. H. Auer, Jr., 1978: Thermodynamic properties and
aerosol patterns in the plume downwind of St. Louis. J. Appl.
Meteor., 17, 689-698.
Sisterson, D. S., and R. A. Dirks, 1978: Structure of the daytime
urban moisture field. Atmos. Environ., 12, 1943-1949.
Spangler, T. C., and R. A. Dirks, 1974: Mesoscale variations of the
urban mixing height. Bound.-Layer Meteor., 6, 423-441.
Tukey, J. W., 1962: Statistical testing. Ann. Math. Statis., 33,
29-38.
Vogel, J. L., 1975: Effects of surface factors on thunderstorms.
Preprints Ninth Conf Severe Local Storms, Norman, Amer.
Meteor. Soc., 113-119.
 CHAPTER 3

Urban Boundary Layer

AUGUST H. AUER, JR.

CONTENTS

ABSTRACT .......................................................................... 41
3.1 Introduction ....................................................................... 42
3.2 Temperature and humidity fields ...................................................... 42
3.2.1 OBSERVATIONAL RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Radiation and Albedo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Vertical Temperature .................................................... 45
3.2.2 MoDEL RESULTs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.3 Wind field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.3.1 OBSERVATIONAL STUDIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Kinematics and Divergence .............................................. 51
3.3.2 MoDEL RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.3.3 TURBULENCE AND FLUXES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
REFERENCES ....................................................................... 60

ABSTRACT

METROMEX studies documented anomalous summer values of radiation components, temperature, humid-
ity, wind fields and aerosol concentrations in the boundary layer over the urban area. Reflected solar radiation,
emitted radiation and net radiation values differed between urban and rural areas, with patterns reflecting land
use types. Solar noontime albedo values varied from 15 to 17% for rural land uses, in contrast to 12 to 13% for
most urban land uses. The metropolitan area experienced a 4% depletion of net radiation throughout the day,
compared to rural areas.
St. Louis had a marked heat island and an identifiable minimum in specific humidity at midday. These effects
were most marked at the surface, but often appeared as height-averaged temperature excesses of I K and
moisture deficits of I g kg- 1 , relative to nearby rural areas, extending through the mixing layer to near cloud
base. The intensity and three-dimensional extent of the urban thermodynamic properties represented a pertur-
bation that could connect the urban surface to an effect in the overriding mixing layer behavior.
In St. Louis, the mixed layer frequently underwent a diurnal cycle including a morning stable layer with a
top close to the surface, a midday convection period with a top typically rising to 2 km, and a late afternoon
or evening return to stable conditions. Daytime mixing heights were domed upward over the urban area.
The thermodynamic perturbations varied with existing weather conditions but, in general, were sizable, being
as large or larger than the city throughout the mixing layer. In several case studies, the airflow over the
metropolitan area was unquestionably perturbed although not in precisely the same manner in all instances.
The perturbations occurred in both fair weather and antecedent rain conditions. The exact nature of the per-
turbations was a function of a number of factors, but in almost all instances, the disturbance was associated
with the metropolitan area. Wind field perturbations were most obvious during light wind conditions with
changes in wind direction. The amount of perturbation in the wind fields, obtained by subtracting the area mean
value, was very similar in both strong and weak winds. Typically, the perturbed airflow extended through a
depth of a kilometer, decreasing in magnitude and changing in form with height.
Although the perturbations in the horizontal and vertical wind fields were directly related to cloud develop-
ment, it appeared likely that the perturbations impacted was more by modifying the storm history, particularly
in view of the fact that the precipitation anomalies were associated primarily with the heavier rain storms.
Thus, the role of urban-induced modifications in airflow within the context of urban modification of the precip-
itation processes was more discernible in individual case studies.

41
42
3.1 Introduction
The anomalous nature of the urban surface and its
concentration of anthropogenic sources significantly
alter the local energy budget and the exchange pro-
cesses for meteorological elements. These alterations
may be recognized in the anomalous values of radia-
tion components, temperature, humidity, wind fields
and aerosol concentrations in the boundary layer over
the urban area.
The boundary layer may be defined as that layer of
the atmosphere that is significantly affected by the sur-
face. In the surface boundary layer of the atmosphere,
of depth up to about 100m, the motion is controlled
predominantly by the presence of the earth's surface,
while within an overlying layer (planetary boundary
layer) with top at about 600 m, effects on air motion
by the boundary remain significant.
There are many independent methods available for
determining the thickness of the planetary boundary
layer; for example, the reader is referred to Hanna
(1969) for a discussion of various meteorological tech-
niques. If the stress is approaching zero asymptoti-
cally as height increases, it is possible to define the
top of the boundary layer as being the height at which
the stress has been reduced to some arbitrary per-
centage, say one percent, of its value at the surface.
Similarly, the average daily thickness of the planetary
boundary layer may be determined from the variation
with height of the diurnal temperature wave. The top
of the planetary boundary layer at any time can also
be estimated by determining the height at which the
wind speed and direction correspond to geostrophic
balance. Methods for predicting the instantaneous
thickness of the boundary layer can be tested when a
vertical temperature sounding through the boundary
layer is available. The observed top of the boundary
layer is defined as the lowest level at which the vertical
gradient of temperature exhibits a discontinuity. For
example, during well-mixed afternoon periods when
an adiabatic layer near the ground is capped by a rel-
atively stable inversion layer, the height of the bound-
ary layer corresponds to the level of the base of the
inversion layer. The top of this well-mixed layer, or
mixing depth, is termed height and is defined as the
approximation to the height of the top of the surface-
based mixed layer through which (ideally) pollutants
readily mix at some distance downstream of the
source. Therefore, during afternoon periods, when
stabilities in the planetary boundary layer are a max-
ima, the terms planetary boundary layer and well-
mixed layer are often used synonymously.
In urban areas, the boundary layer may be expected
to respond to the surface heat island. The intensity
METEOROLOGICAL MONOGRAPHS
and horizontal extent of urban surface thermodynamic
properties can be expected to perturb the overriding
boundary layer behavior.
Table 1 and Fig. 1, after Auer (1978), identify some
certain features of the land use in St. Louis that are
unique and meteorologically significant in explaining
some thermodynamic, kinematic and radiative anom-
alies associated with the overriding atmosphere of the
metropolitan St. Louis area.
3.2 Temperature and humidity fields
3.2.1 OBSERVATIONAL RESULTS
Radiation and Albedo
The relationship between the heat island and hu-
midity deficit over the urban area (see Chapter 2) and
differences in sensible heat and latent heat fluxes has
directed attention to assessing the radiative compo-
nents of the surface energy balance and to influences
which may cause imbalances in the energy budgets
between urban and rural areas.
Observational studies designed for evaluating the
incoming global radiation (the sum of direct and dif-
fuse radiation received by unit horizontal surface)
have shown that the St. Louis metropolitan area re-
ceives 2-5%* less than the surrounding rural areas (Gar-
ret, 1975; White, 1978). Similar studies within RAPS
(Peterson and Flowers, 1973, 1977) revealed a 3% re-
duction over the urban area. While these differences
in incoming global radiation are found to be small,
components of this term are highly influenced by the
change of the aerosol load of the atmosphere. Large
decreases in direct solar beam radiation and increases
in diffuse sky radiation have been observed with an
increase in the aerosol load (White, 1978).
The intensity I of global radiation in both urban and
rural mixing layers was found to vary with height Z
according to Z 0 ·05 for 100 m < Z < 1000 m above the
ground; from 100 m to the surface an empirical linear
relationship was applied (White, 1978). In another field
program in St. Louis, similar trends in filtered global
radiation values were found (Duntely et al., 1973).
Comparisons of longwave incoming atmospheric ra-
diation between urban and rural areas show a slight
increase received by the urban area. This compensa-
tion for the decrease due to aerosols in incoming solar
radiation found over urban areas results in negligible
differences in the total incoming radiation (i.e., global
and atmospheric radiation) between urban and rural
areas (White, 1977). In a study by White et al. (1978),
measurements of the diurnal trend of albedo were
made. At solar noon, the composite metropolitan area
METROMEX: A REVIEW AND SUMMARY
TABLE l. Identification and classification of land use types found in Metropolitan St. Louis
Description
Type Use and structures
II Heavy industrial
Major chemical, steel and fabrication industries;
generally 3-5 story buildings, fiat roofs
12 Light-moderate industrial
Rail yards, truck depots, warehouses, industrial
parks, minor fabrications; generally 1-3 story
buildings, fiat roofs
Cl Commercial
Office and apartment buildings, hotels; > I 0 story
heights, fiat roofs
Rl Common residential
Single family dwellings with normal easements;
generally one story, pitched roof structures;
frequent driveways
R2 Compact residential
Single, some multiple, family dwellings with close
spacing; generally <2 story, pitched roof
structures; garages (via alley), no driveways
R3 Compact residential
Old multi-family dwellings with close ( <2 m)
lateral separation; generally 2 story, fiat roof
structures; garages (via alley) and ashpits, no
driveways
R4 Estate residential
Expansive family dwellings on multi-acre tracts
AI Metropolitan natural
Major municipal, state, or federal parks, golf
courses, cemeteries, campuses; occasional
single story structures
A2 Agricultural rural
A3 Undeveloped
Uncultivated; wasteland
A4 Undeveloped rural
A5 Water surfaces
Rivers, lakes
land use was characterized by albedo values of 12-13%,
and for rural land use by 15-17%.
Dabberdt and Davis (1974), under the auspices of
RAPS, also conducted a similar albedo study of land
use types in the metropolitan area with identical re-
sults. Mapping of the surface temperature field, de-
rived from airborne infrared radiation measurements,
has been conducted during a winter nighttime situation
(Braham, 1974) and clear summer daytime cases (Dab-
berdt and Davis, 1974; White eta/., 1978). The winter
analysis is reproduced in Fig. 2. During both seasons,
the central St. Louis urban area, consisting primarily
of commercial and industrial land uses, had the warm-
est temperatures while the outlying rural (agricultural
and heavily wooded) areas had the coolest tempera-
tures. White et a/. (1978) reported maximum surface
(skin) temperatures of 44.5°C for industrial areas con-
trasted with 29.5°C for undeveloped woods and grass-
lands. Furthermore, between early morning hours and
the hour of peak heating (approximately one hour after
solar noon), surface temperatures showed warming of
23-26°C for urban areas while rural areas showed
43
Vegetation
Grass and tree growth extremely rare; <5% vegetation
Very limited grass, trees almost totally absent; <5%
vegetation
Limited grass and trees; < 15% vegetation
Abundant grass lawns and light-moderately wooded;
<70% vegetation
Limited lawn sizes and shade trees; <30% vegetation
Limited lawn sizes, old established shade trees; <35%
vegetation
Abundant grass lawns and lightly wooded; >80%
vegetation
Nearly total grass and lightly wooded; >95%
vegetation
Local crops (e.g., corn, soybean); >95% vegetation
Mostly wild grasses and weeds, lightly wooded; >90%
vegetation
Heavily wooded; >95% vegetation
warming of 13-16°C. The infrared surface tempera-
ture, and thus the derived emitted terrestrial radiation,
was found to be inversely related to the amount of
vegetative cover during mid-August (White, 1977;
White and Eaton, 1977; White eta/., 1978).
The daytime variation of net radiation and its com-
ponents has been addressed by White ( 1977) and
White et al. (1978). Components contributing energy
to the surface are incoming solar radiation and atmo-
spheric radiation; components representing a loss
from the surface are reflected solar radiation and
emitted terrestrial radiation. Results have shown that
the ordering of land use types according to their net
radiation is correlated to the vegetative cover. It has
also been shown that the reflection of solar radiation
and emission of long wave radiation by differing land
use types are the primary components responsible for
differences of the net radiation. Shown in Fig. 3 is the
general daytime variation of emitted, reflected solar
and net radiation from the St. Louis metropolitan
boundaries and surrounding rural area under mid-Au-
gust clear sky conditions; radiation values for water
 44
FIG . I. Land use mosaic for the Metropolitan
surlaces are included for reference. The metropolitan
area exhibits a slightly lower loss of radiation by re-
flection and a greater loss by emitted radiation than
rural areas. The result is that the metropolitan area
suffers a 4% depletion of net radiation throughout the
day compared to the rural area. It appears that the
metropolitan-rural land uses control the acquisition
and utility of radiant energy and are responsible for
processes related to the documented urban thermo-
dynamic anomalies.
Turbidity measurements provide valuable informa-
tion about visibility, aerosol size distribution and at-
tenuation of solar radiation. This aspect of the total
radiation picture under Project METROMEX has been
most widely studied beginning in 1972.
There have been many different approaches, instru-
mentations and calculations in these turbidity studies.
Turbidity has been most commonly determined for the
0.5 ~-tm wavelength. METROMEX measurements from
the surlace of aerosol optical thicknesses have given
values ranging between 0.12 and 0.50 (Uthe and Rus-
METEOROLOGICAL MONOGRAPHS
CJII
mn r 2
fmC I
D RI
miDR2
~R 3
~R4
EJ AI
~A 2
N A3
~A4
OAS
r
0 5 10 KM
f-----,----1.-r-~
0 2 4 6 N. M I.
t. Louis area. After Auer ( 1978).
sell, 1974; Laulainen et al. , 1978; White, 1978). Other
studies of turbidity in the St. Louis area have indicated
similar values (Dabberdt and Davis, 1974; Wesely,
1975; Bergstrom and Peterson, 1977). Studies of ver-
tical variations of turbidity show consistent decreases
in the aerosol optical thickness with increasing height
(Laulainen et al., 1978).
One other constituent of the turbidity studies dealt
with the investigation of the wavelength dependence
of the optical depths, which provides information on
the size distribution of aerosol particles (see Volz,
1959). For any given time, the wavelength dependence
of the optical depth produces a straight line with slope
-a on a log-log plot. Values of - a for a typical at-
mosphere void of any urban influence are around 1.3
± 0.6 (Curico, 1961); higher values represent greater
concentrations of smaller particles and lower values
represent greater concentrations of larger particles.
Using turbidity values determined from multi-wave-
length measurements in empirical calculations, wave-
length exponents were resolved to 1.8 ± 0.8 (Dabberdt
METROMEX: A REVIEW AND SUMMARY 45

and Davis, 1974; Uthe and Russell, 1974; Laulainen

et al., 1978; White, 1978; Yoksas, 1978). These em-
pirical results of departure from the expected values
of a = 1.3 suggest a variability towards smaller sizes
in the aerosol loading of the urban atmosphere.

Vertical Temperature
The thermodynamic structure of the planetary
TROY
VOR boundary layer is the product of complex energy
transfers between the surface and the free atmosphere.
The physical characteristics of urban surfaces are so
different from those of most rural surfaces that the
"'0> surface heat balance and resultant temperature are
>-
most assuredly different. Indeed, St. Louis as well as
..."'
0
10 many other cities, exhibits the elevated temperature
"'
0
~
~
of the urban surface air which has become known as
u
z
..: the urban heat island. The urban land use with less
15 :;;
c; evapotranspiration is a weaker source of humidity
than vegetative surfaces and even of most soils, at
least during the summer (Sisterson, 1975). However,
the anthropogenic humidity partially offsets this deficit
and during the winter when most vegetation is dor-
FIG. 2. Infrared surface temperature in St. Louis area, 0000--0339
CST, 29 February 1972. After Braham (1974). mant, the urban area may be a better source of atmo-
spheric humidity than rural areas.

LOCAL Tlr·l£ (COT)

08 10 12 14 16 18
1. 6
- - · -- METRO POL ITAN

------ RURAL

1.2 WATER SURFACES

.,
.c
0. 8
'
G
-;;;
v
,: 0.4
=
"'
50
>< Rn
~ 0
,__
"'
..:
0
~ (R0 + R5 )a
-0 . 4

RE
-- .. ---.:._===:------ -- -- ~
-0 .8 SOLAR TIM£ ---· - --- ~
06 08 10 12 14 16 18

20° 40° 60° 64.6° 60° 40° zo•

SOLAR ELEVATION

FIG. 3 Comparison of the daytime variation of net radiation, reflected solar radiation, and
emitted radiation between metropolitan, rural and water land uses for St. Louis during mid-
August. After White et at. (1978).
46
METROMEX measurements revealed height-aver-
aged temperature excesses and humidity deficits, rel-
ative to nearby rural areas, extending through portions
of the mixing layer (Braham, 1972b; 1974; Dirks,
1974a, b; Sisterson, 1975; Lawson and Auer, 1976;
Ackerman and Mansell, 1978). The magnitudes and
vertical extent of the area-averaged temperature ex-
cesses (~8) and humidity deficits (~q) in the mixing
layer have been shown to vary inversely with wind
speed V (m s- 1). Dirks (1974b) observed vertical ex-
tension to be greater than 850 m of the thermodynamic
anomalies in the mixing layer; with >2 m s- 1 , Dirks
found regions of about I K thermal excesses and 1 g
kg- 1 specific humidity deficits displaced a little more
than 10 km downwind from their source regions. With
stronger winds, other case studies (Shea and Auer,
1978) were characterized by less intense thermody-
namic anomalies ( +0.5 K, -0.5 g kg- 1 ) further dis-
placed to 30 km, and extending vertically to only
500-800 m. Combining the results of all available sum-
mertime case studies from METROMEX, Shea and
Auer (1978) found the thermodynamic anomalies, ~8
and ~q (K and g kg- 1 , respectively), to vary inversely
with the mean mixing layer wind speed, V (mps), ac-
cording to
~8 = 1.44- 0.10V} 2 ~ V ~ 8
~q = 1.41 + 0.13V . city was 10-20% less than over the upwind rural area
The results of several METROMEX summertime
case studies (Dytch, 1972; Dirks, 1974b; Ackerman
and Mansell, 1978; Shea and Auer, 1978) have re-
vealed that the thermodynamic perturbations extend
vertically to between 500-1100 m. The variability of
this vertical extent, Hp (m), of the thermodynamic per-
turbations was reconciled by Auer and Changnon
(1977) into a relationship (see Fig. 4) again revealing
a dependence on the mean wind speed V in the mixing
layer according to
Hp = 1265.0 - 87.2V, 2 ~ V ~ 8.
Thus these relationships show that the thermody-
namic anomalies in the mixing layer, and their effect
on the modification of local weather might be dimin-
ished under conditions of stronger mixing layer winds.
The analysis of Ackerman and Mansell ( 1978) sug-
gests that a midday average urban-rural temperature
difference of about 1oc is observed in the lower part
of the planetary boundary layer at about 1 km in fair
weather and at about 500 m in pre-rain periods. An
average humidity deficit, depicted by differences in
the mixing ratio of 0.5 to 1.5 g kg- 1 has been found .
On fair days the urban atmosphere contained, on the
average, less humidity than did the rural atmosphere
through the lower several hundred meters. However,
METEOROLOGICAL MONOGRAPHS
E
12r-·
10
~
~~8~
8Vl 6
z:
LU
;·~2 ~t
0 Dytch, 1972
6oirks , 1974 . a , b
• shea. 1976
A Wong, 1976
OL---~--~----L---~--~--~----~--~
0 2 3 5 6 8
MEAN I·IJXING LAYER \H NO (V) , m s-•
FIG. 4. Summary of METROMEX observations of the vertical
extent of the thermodynamic anomalies versus wind speed in the
mixing layer. After Auer and Changnon (1977).
the upper planetary boundary layer tended to be more
humid over the city than over the country. In pre-rain
conditions this was not the case; the layer of humidity
deficit over the city tended to be much deeper (1-2
km) with urban moisture excesses occurring only in
shallow elevated layers.
Sisterson and Dirks (1978) showed from. extensive
aircraft measurements that specific humidity over the
in the mixed layer. Industrial and urban moisture
sources and sinks were computed to be negligible dur-
ing dry summer days , yet urban surface stations show
a significantly lower absolute humidity compared to
nearby rural stations . Furthermore, their vertical
soundings indicated a significant reduction in moisture
at both the bottom and top of the mixed layer but
somewhat less reduction in the middle. Sisterson and
Dirks explained this phenomenon by two interacting
factors, a drier surface and entrainment. First of all,
the drier urban surface mean's less solar energy is con-
sumed for evaporation. This results in higher surface
temperatures (the urban heat island) and more upward
sensible heat flux. The combined effect of more heat
flux and increased roughness over the city produces
stronger vertical mixing due to increased turbulence;
this results in a detectable, significant deepening of the
mixed layer. The drier overlying air entrained in this
vertical mixing process then dilutes the mixed layer
moisture, decreasing the specific humidity while the
total humidity in the air column is not decreased.
The area of maximum (or core) thermodynamic per-
turbations in the mixing layer show a coverage of
about 150 km2 , similar to that found at the surface.
But the full expanse of the average thermodynamic
anomalies may encompass as much as 300--500 km2
(Sisterson, 1975; Auer, 1978).
 METROMEX: A REVIEW AND SUMMARY
The case studies of Dirks (1974a, b), Changnon and
Semonin (1975) and Shea (1976) point out that the
downwind horizontal extent of the thermodynamic
perturbations from their nonadvective positions are
unlikely to be more than 40 km as shown in Fig. 5.
More specifically, the thermodynamic anomalies should
not be prevalent at more than 1.5 hour transport time,
when the transport time is defined as the downwind
distance with respect to the Gateway Arch (centroid
of activity in metropolitan St. Louis) along the mean
wind direction divided by the mean wind speed in the
mixing layer.
The information above can be combined to yield a
description of the size and magnitude of the urban
thermodynamic anomaly in the mixing layer.
In St. Louis, the mixed layer frequently undergoes
a diurnal cycle including (i) a morning stable layer with
the top close to the surface (typically < 1 km); (ii)
midday convection period with the top typically rising
to 2 km; and (iii) late afternoon or evening return to
stable conditions. For condition (i) the lidar and 100
W acoustic sounder were equally effective sensors of
mixing depth; for (ii) the lidar was superior (but a more
powerful sounder might be able to observe the tem-
perature structure to the tops of the thermals); and for
(iii) the sounder is a superior sensor of the mixing
depths (Uthe and Russell, 1974; Russell et al., 1974).
Local spatial and temporal variations in the height
of the temperature inversion capping the mixing layer
have been examined to assess the urban effect on the
mixing layer depth. Spangler (1972) first showed that
the daytime mixing heights are domed upward over
the St. Louis urban area; this phenomenon has since
been routinely observed and described in METRO-
MEX by other case studies involving lidar (Uthe and
Russell, 1974; Auer and Eaton, 1976) and instru-
mented aircraft and radiosondes (Shea, 1976; Wong,
SOr----.----.---- ,---,,---,----.----.---,
~ 40 o Di r ks , 1974 , a , b
. Wong (1976) and Wong and Dirks (1977) reveal that
• Shea , 1976
c the elevated heat island, or the frictional drag (de-
~ 30
...
UJ
X
UJ
0
::: 20
""z:
50 0 •
10
0~---L----L---~--~----~--~----~--~
0 2 3 4 5 6
MEAN MIXING LAYER WIND (V), m s- '
FIG. 5. Summary of METROMEX observations of the downwind
extent of the thermodynamic perturbations versus wind speed in
the mixing layer. After Auer and Changnon (1977).
47
1976), and by summer climatologies (Ackerman and
Appleman, 1974; Ackerman and Mansell, 1978). Ack-
erman and Mansell (1978), in reviewing average pro-
files of temperature and dew point over urban and ru-
ral stations, concluded that the depth of the planetary
boundary layer was about 300 m greater over down-
town St. Louis; the urban station was also found to
have a slightly deeper moist layer by about 100-200 m
than the rural stations. The dome-like lifting of the
inversion structure occurred over the urban-industrial
center with subsidence of the inversion layer at the
perimeter of the metropolitan area. Distortion of this
inversion structure was observed during both daytime
and early evening cases with the maximum amplitude
of the dome occurring shortly after midday, as illus-
trated in Fig. 6.
In addition to direct measurement by radiosondes,
this boundary could be well-approximated from lidar
data. Aircraft-lidar data comparison showed excellent
correlation between the top of the adiabatic mixing
layer and the top of the haze layer measured by the
lidar, as seen in Figs. 6 and 7 (Spangler and Dirks ,
1974). The lidar observations of Fig. 7 reveal that the
mixing layer extended from 1700 m in the Troy vicinity
to 2200 m over the downtown area.
A review of these mesoscale anomalies in the mixing
layer height, aH(m), from various METROMEX case
study analyses show values ranging from 15~50 m.
Fig. 8 shows the plot of the mesoscale anomaly of
mixing heights against the mean mixing layer wind
speed (after Auer and Changnon, 1977) resulting in the
relationship
an = 468.3 - 42.4V, 2 :,;;; v :,; ; 8.
This description of the mesoscale anomaly in the mix-
ing layer height characterizes the perturbation in a
summertime midday mixing layer with a depth of at
least 1500 m.
The mesoscale anomaly in the mixing height was
first attributed to the thermal influence of the metro-
politan area (Spangler and Dirks , 1974). Studies by
pending on the threshold wind speed of 4 mps), in-
duces horizontal convergence in the wind field at the
lower levels of the mixing layer, as reported by Ack-
erman (1974a, b; 1977), Sisterson (1975) and Kropfli
and Kohn (1978). Thus, it appears that this elevated
heat island, extending as much as 1 km above the met-
ropolitan area, can induce horizontal convergence
through these depths with associated vertical motion
8
of 1-7 em s- 1 , commensurate with the rise of the mix-
ing layer anomaly (Auer, 1975a).
The morning growth rate of the mixing layer over
the city is equal or slightly faster than it is over the
48 METEOROLOGICAL MONOGRAPHS

'"'
0
,_
:::>
,_
--'
«
""g; 1000
"'~
""
0.

! I I I I I I I I I I I ! 9 kg -; 9 kg- I ! I 1 I I I I I I I I I !
'-----''-'-'-'-.__. o K o K ......_.....__.__.'-'-'--'

0' FALLON ARCH BELLEIJ ILLE

FIG. 6. Inversion height over St. Louis on 12 August 1971 , as measured by aircraft
soundings at O ' Fallon (1615 CDT) and Belleville (1530 CDT) and horizontal aircraft flights
shown by arrows. Soundings show potential temperature () and specific humidity q with
scales in unit intervals. After Spangler and Dirks (1974).

14

12
ii
E •o
~
'"
....'5 6
;:: 4
;;_

1-55 SOUTHBOUND 1-55 SOUTHBOUND

(1 4251 1- 55 MISSISSIPPI EAST HORSESHOE VICINITY I - 55

JCT RIVER ST. LOUIS LAKE COLLINSV IL LE JCT.
I 44 1-70
(13551

FIG. 7. Cross section of the aerosol structure in the vicinity of St. Louis, Missouri, 12 August 1971 (1355-1425 CDT) as observed
by SRI/EPA Mark VIII Lidar System. After Spangler and Dirks (1974).

E
5
rural areas (Uthe and Russell, 1974; Auer and Eaton,
;:: 1976).
:0: The distortion of the mixing height and the thermal
,_.
:0:
excesses and specific humidity deficits over and down-
::! :.H = 468. 3 - 42 .4 V
wind of the urban area should be manifested in other
....
:0: 3 • 0
observed and derived properties of the planetary
"'z
>< D. Di rks, 1974, a,b boundary layer; for example, convective condensation
~
2 - & Spang l er & Dirks , 1974 •
level, observed cloud base, horizontal distribution of
>-
;3 o Uthe & Russell, 1974
"':::: • Au er & Ea ton , 1976
equivalent potential temperature, etc. Ackerman and
o Appleman (1974) first suggested higher convective
L •
'-
0 Shea. 1976
~ i4ong , 1976 condensation levels (CCL) over and downwind of the
~ 0 I --'-----L----L---~--- city at midday from radiosonde analyses. Additional
l§o 1 2 3 s 6 8
1·1EAN 11! Xl G LAYER Wl~ D (V) , m s- • confirmation was provided by Ackerman and Mansell
(1978) who found that the CCL over a downtown lo-
FIG. 8. Summary of METROMEX observations of the anomaly
(doming) of midday mixing heights versus wind speeds in the mixing cation in St. Louis averaged 400 m higher and laC
layer. After Auer and Changnon (1977). cooler than in surrounding local locations in fair
METROMEX: A REVIEW AND SUMMARY
weather and antecedent rain conditions. From case
studies, Shea and Auer (1978) found the CCL to be
200-300 m higher over and downwind of the metro-
politan area due to warmer temperatures and drier hu-
midities associated with urban land use; an example
is shown in Fig. 9. Observations of higher cloud cases
by 300-600 m have been reported in a four-summer
survey by Changnon et al. (1976) and in case studies
by Boatman (1974) and Changnon and Semonin (1975)
for convective clouds within urban influences.
The equivalent potential temperature Be is a measure
of the specific energy (sensible plus latent) of an air
parcel and is conservative in both dry and pseudo-
adiabatic processes. On the average, the equivalent
potential temperature was lower by 2-4 K in the urban
atmosphere through the lowest several hundred me-
ters (Ackerman and Mansell, 1978; Shea and Auer,
1978). On fair days, the layer of low Be over the city
was usually confined to 600-1000 m; in the middle and
upper parts of the planetary boundary layer 8e was
higher over the urban area than over the countryside.
In pre-rain cases, low 8e was found over the city in
the lower planetary boundary layer as on fair days,
but Be over the city was not significantly different from
that over the country in the upper planetary boundary
layer. These deficits in 8e reflect a dominating influ-
ence of the specific humidity anomaly. The magnitude
of the Be deficit, !He (K) in the mixing layer has been
I
~ Flight Path
....#Y Sounding
SCALE OF KILOMETERS
10 20 30
0
FIG . 9. lsopleths of midday convective condensation level (CCL)
in the vicinity of metropolitan St. Louis during mean mixing layer
winds of 140° at 5 m s- 1 • After Shea and Auer (1978).
49
reviewed by Shea (1976) and found to be related to
the mean wind speed V in the mixing layer according
to
flOe = -2.96 + 0.195V, 2 ~ V ~ 8.
The result is expected because of the trends in the
magnitudes of !lB and 8q with wind speed.
In a review of thermodynamic stability indices,
Ackerman and Mansell (1978) found that the thermo-
dynamic indices do not indicate any strong differences
between urban and rural locations, with only a weak
suggestion that the urban atmosphere might be slightly
less favorable for thunderstorm development than the
rural atmosphere.
While the influence of wind speed on the magnitude
and vertical extent of the urban thermodynamic anom-
alies in the mixing layer has been suggested, case stud-
ies (Shea, 1976; Shea and Auer, 1978) have cautioned
that the urban thermodynamic plume may be altered
by the presence of synoptic-scale thermal and specific
humidity gradients. For days with similar wind speeds
(e.g., > 5 m s- 1 ) and mixing depths but different ori-
entation of thermal gradients, the extent of the urban
thermal plume would tend to be masked by a down-
wind thermal ascendant and amplified by a downwind
thermal gradient. An estimate of the downwind as-
cendant of potential temperature sufficient to obscure
the urban plume is > 2 K (100 km)- 1 • Similarly, a
downwind gradient of specific humidity in excess of
1.5 g kg- 1 (100 m)- 1 may result in no detectable deficit
of specific humidity in the mixing layer downwind.
Many of the urban modifications of the low level
thermal and humidity stratifications may have impor-
tant implications for the upward transport of heat and
humidity and for the diffusion of materials from the
surface . In the presence of favorable mesoscale dy-
namics, they may' also influence cloud development.
Results shown here point to the fact that studies of
the regional extent of the urban influences upon
weather anomalies, particularly convective rainfall,
must take into consideration a description of the
boundary layer which is characterized by the forcing
of synoptic scale and urban mesoscale features.
3.2.2 MODEL RESULTS
Numerical modeling, while not extensive in the area
of temperature and humidity field specification, has
pointed to the existence of preferential sites of con-
vection associated with the observed thermodynamic
anomalies .
Early in METROMEX, Auer and Dirks (1974) un-
dertook airflow modeling results to ascertain typical
orders of magnitude of the total urban effect in order
to guide program planning. The Lavoie (1968) airflow
50
model was adapted for use because it coupled the dy-
namics of a suitably deep convective mixing layer with
the forcing influence of the surface boundary layer
along with the inclusion of horizontally parameterized
asymmetries required for a three-dimensional descrip-
tion of the urban airflow interaction. In this model,
wind speed and the height of an associated capping
inversion were altered in response to frictional drag,
variation in underlying terrain height and low level
heating. Systematic variation of these input parame-
ters with the inclusion of various wind and tempera-
ture profiles within the boundary layer elucidated the
role played by each in the total urban effect. The in-
tegrated effect of all forcing influences indicate a ver-
tical displacement of the overlying discontinuity sur-
face by some 200 m. Auer and Dirks (1974) also
indicated that thermal excesses were more important
in raising the depth of the mixing layer than surface
roughness.
Lawson (1973) and Lawson and Auer (1976) noted
that the distribution of convective cloud activity was
strongly associated with the differential and vertical
temperature structures and mixing layer depths through
the urban-rural area. Vertical temperature profiles
which suppressed convective cloud activity were
found from case studies to be characterized by ele-
vated stable layers that strengthened through the mid-
day. An entrainment process whereby the air beneath
an elevated stable layer can be warmed was hypoth-
esized from analogous laboratory experiments ad-
dressing fluidal discontinuity interfaces. A comparison
by dimensional analysis of the fluidal and atmospheric
parameters suggested that this entrainment process
was feasible in the atmosphere. During early morning
flights, light-moderate turbulence observed in the vi-
cinity of elevated aerosol layers provided a local
source of turbulent kinetic energy to induce a suffi-
cient forcing mechanism to initiate the entrainment of
air through the top of the mixing layer and account for
the observed elevated heating. The fluidal model in-
dicated a lowering of the density discontinuity as op-
posed to a creation of a buffer zone at the disconti-
nuity. In the urban atmosphere such a lowering of the
stable layer would seem to reduce the total mixing
depth and in turn help inhibit convective cloud activity
in portions of the metropolitan area.
To investigate the role played by surface thermal
and humidity fields on cumulus cloud initiation, Ochs
(1975) employed a two-dimensional vertical slice
model. Diurnal trends of surface temperature and mix-
ing ratio along several rural-urban-rural cross sections
were obtained by interpolating hygrothermograph
readings gathered at 28 urban sites. Observed upper
air rawinsonde data were used (with appropriate pre-
cautions to guard against initial instability and grid
METEOROLOGICAL MONOGRAPHS
edge effects). The model was run to simulate convec-
tive activity for 7 August and 10 August 1973. It was
found that initial cloud development occurred primar-
ily in areas of temperature excesses in spite of mixing
ratio deficits. The model results agreed favorably with
the approximate time and location of initial sightings
of cumulus activity in the vicinity of St. Louis. These
findings and subsequent results obtained by artificially
modifying the input surface temperature and humidity
fields suggested that the surface temperature distri-
bution was more important than the surface humidity
pattern in determining the location of initial cumulus
activity. On one of the two cases studied, the location
of initial cumulus activity and subsequent cloud vigor
occurred quickly exceeding model capabilities; the
simulated cloud activity coincided with the subsequent
isolated thunderstorm. These simulations and corre-
sponding areas of observed cloud development served
to indicate the dominant importance of surface tem-
perature fields in urban cloud production, in spite of
the presence of humidity deficits.
As a step in urban convective cloud modeling,
Hjelmfelt (1980) pursued a three-dimensional, hydro-
static, primitive equation model to determine the rel-
ative importance of such input parameters as topog-
raphy, surface roughness, surface temperature and
humidity fields (keyed to land use categories and al-
lowed to vary in time) and the temperature, humidity
and wind structure of the upstream air in the vicinity
of St. Louis. These parameters were qualitatively ex-
amined and rated to their importance by selectively
deleting them from simulation. For example, the com-
puter simulations with west and southwest winds near
neutral stability show weak downward motions over
the western suburbs, a convective core over the north-
eastern metropolitan area and streaming off to the
east. It was concluded that the urban land use param-
eters are dominant factors leading to positive vertical
motions downwind; and for St. Louis, topography was
judged as a secondary, but not trivial, factor.
In a corollary study involving RAPS data from St.
Louis, Vukovich et al. (1976) have numerically ex-
amined with a three-dimensional primitive equation
model the influence of synoptic wind speed and wind
direction on the intensity and location of the urban
heat island and its internal circulation. It was found
that under conditions of light and variable winds, the
St. Louis heat island remained unchanged. When the
wind speed is increased, however, circulation within
the heat island decreases and the associated heat
plume extends further downstream. This causes the
predominant cyclonic circulation observed under mild
conditions to become a convergent zone. Variation in
the direction of flow is also seen to have a profound
effect on the internal circulation and intensity of the
METROMEX: A REVIEW AND SUMMARY
heat island. Notably, the most dramatic change occurs
with flow from the east, southeast and west. Flow
from the north or northeast produces changes that are
secondary. Local topography (the Ozark highlands to
the southwest of St. Louis) are suggested as influential
mechanisms. Also, for weak west winds, a secondary
upwelling maxima northeast of St. Louis was attrib-
uted to terrain effects interacting with the thermal
plume dynamics.
Vukovich and Dunn (1978) performed a sensitivity
analysis on the same model to determine the more
important parameters affecting the urban heat island
circulation in St. Louis. Their results indicated that
the heat island intensity and boundary layer stability
play the dominant roles in the heat island circulation.
These results have found empirical support from a
study by Shreffler (1978) which showed that weak heat
islands, with attendant conditions of instability and
very weak winds, are helpful in producing centripetal
flows in urban regions.
3.3 Wind.field
3.3.1 OBSERVATIONAL STUDIES
Kinematics and Divergence
Wind measurements made during the course of ME-
TROMEX were used to study the spatial variability of
the planetary boundary layer winds over a region hav-
ing no major natural terrain features but one which
included a sizable metropolitan area. The wind field
over a metropolitan area is a critical factor in air qual-
ity and its control since it is the medium for transport
of a city's many effluents. Moreover, since transport
occurs throughout the mixed layer, the flow through
the depth of the planetary boundary layer must be
considered. The wind field over a mesoscale area con-
taining the urban complex area may be expected to be
perturbed as a consequence of 1) differences in the
density and characteristics of the roughness elements,
2) variations in the vertical transport momentum
through mechanical and thermal mixing, and 3) local
thermally induced pressure perturbations associated
with the urban heat island. The extent to which a ther-
mal perturbation can perturb the wind field is related
inversely to the thermal stability of the boundary layer
air.
Kinematic analyses of boundary layer winds over
St. Louis were based on measurements from single
and double theodolite tracking of pilot balloons, air-
craft equipped with Doppler navigation systems, and
from chaff tracking by dual Doppler radar. A more
comprehensive understanding of the airflow was at-
tempted by relating these measurements of the airflow
51
to thermodynamic properties of the air, the surface
land use and terrain. The review of these kinematic
analyses is partitioned approximately into light (,;;; 5
m s- 1) and moderate (;?; 5 m s- 1) average wind speeds
in the mixing layer.
Afternoon case studies have revealed that below 500
m airflow of ,;;; 5 m s- 1 takes on anticyclonic curvature
as it approaches the main commercial-industrial areas,
followed by cyclonic curvature as it moves over and
out of the urban area (Ackerman, 1974a).
Applying their pibal hodograph climatology of
1971-76, Bryant et al. (1978) developed parameteri-
zation to describe vertical wind profiles characteristic
of rural and urban land uses and their fetches. Findings
were limited to that portion of the adiabatic boundary
layer which extends from the surface to the level of
the gradient wind, around 400 m; only gradient wind
speeds > 5 m s- 1 were considered and the average for
the study was 6.5 mps. It was found that the speed
variations over a given site obeyed a power law of the
form
V = Vu (ZIZujU,
and the directional deviation followed a logarithmic
form
0 = A In (ZIZu),
where V is the wind speed at height Z, Vu the wind
speed at the gradient level Zu, and 0 the deviation from
the gradient wind direction (degrees) defined positive
for clockwise deflection. The power index a and the
directional constant A were determined empirically as
functions of the weighted roughness length Zo, char-
acteristic of a 10 km fetch upwind of the site:
a= 0.054ln Zo + 0.135,
A = 1.94 In zo -
1.99.
Estimates of z 0 , according to the land use classification
of Auer (1978), have been offered by Wong (1976) and
are limited here to values between 15 and 300 em. The
height of the gradient wind level Zu can also be deter-
mined as a function of roughness length, i.e.,
Zu = 52.63 ln Zo + 209.99.
For example, assuming characteristic roughness
lengths of 15 em and 100 em for rural and urban areas,
respectively, approach winds to St. Louis are char-
acterized by a = 0.29 and A = 3.2, while over the
center of the city a = 0.38 and A = 5.0. At 100 m
altitude, just downwind ofthe transition to urban uses,
the wind may be deflected cyclonically by 4° and re-
duced by speed by 19% compared to rural values. The
height of the gradient wind levels averaged 350 m and
450 m for rural and urban areas, respectively. Thus,
52 METEOROLOGICAL MONOGRAPHS

Bryant et al. (1978) concluded that, for wind speeds

of at least moderate strength, the modification of the
airflow resulting when the wind blows from rural to
metropolitan use areas is such that any given height
the wind speed is decreased, the direction is deflected
cyclonically, and the level of the gradient wind in-
creases. These modifications are reversed as the air
blows from the city center to the downwind rural
countryside. Values of the wind vector at any height
may be simply derived where the gradient wind ve-
locity and the roughness length are known.
The observations of accelerating and decelerating
airflow across the urban St. Louis area has been rec-
onciled by Wong (1976) and Wong and Dirks (1977,
1978). Their studies reveal that when elevated heat
islands near 450 m were characterized by thermal ex-
c~o-,
~ 10 15 20 \ '
cesses of ~ 2 K and moderate thermal gradients, along 21 July 1975 ~
with mixing layer winds <4 m s- 1 , the thermal pertur- PERTURBATION STREAML lt!ES
450 m HSL
bation was the dominant force on the urban airflow.
It accelerated the air as it moved into the elevated heat
island. The convergence of the airflow into the ele-
vated heat island can be clearly seen from the pertur- FIG. 10. Perturbation streamline (solid lines) and potential tem-
bation streamline shown in Fig. 10. The analysis perature (dashed lines, K) analyses at 450 m MSL, 21 July 1975,
clearly shows that the perturbation streamlines were between 1345-1600 CDT. Urban areas are outlined by light dashed
lines, high terrain areas are shaded. The analysis is based on data
directed along the thermal gradient, i.e., along the pos- acquired from parallel flight tracks approximately 15 h apart. After
itive pressure gradient, when the thermal gradient was Wong and Dirks (1978).
large. It also shows that there was a convergence cold
air to the north near the Mississippi and Missouri
River junction.
Perturbation streamline fields of the types shown in
Fig. 10 should produce corresponding divergence
fields. In Fig. 11 the divergence field corresponding to
Fig. 10 shows convergence induced by the elevated
heat island over most ofthe boundaries of the elevated
heat island. As the air accelerated into the elevated
heat island, there was divergence northwest of the
metropolitan area associated with the increase of wind
speed downstream. To the north, near the junction of
the Mississippi and Missouri Rivers, there was diver-
gence associated with the diffluent streamlines in the
cooler air.
The evolution of divergence fields in the sub-cloud
layer for northeasterly winds < 5 m s- 1 is shown in
Fig. 12 according to Ackerman (1978). It can be seen ~4. 0

i
that convergence existed around the metropolitan area
from the south through northwest through east north-
east; divergence, equal in magnitude to that of the ~ll":TE~
convergence, occurred over the Illinois urbanized area 21 July 1975
(lo-" s.')
to the east at 1600 CDT. These analyses show that the OIVERGENCE
450 m I•ISL
perturbation streamlines are directed along the ther-
mal gradient and that the divergence shows a good
correlation with the thermal field. Thus the thermal
effect appears to dominate over the terrain and fric- FIG. II. Divergence (I0-4 s-1) and potential temperature (K) anal-
yses at 450 m MSL, 21 July 1975, between 1345-1600 CDT. The
tional effects as the deterministic parameters of the analysis is based on data acquired from parallel flight tracks ap-
urban airflow in the case of light winds ( < 5 m s- 1). proximately 15 km apart. After Wong and Dirks (1978).
METROMEX: A REVIEW AND SUMMARY
FIG. 12. Afternoon evolution of streamlines and divergences (I0- 6 s- 1 ) of the subcloud
transport wind averaged over fair weather days with northeast winds . After Ackerman
(1978).
Wong (1976) and Wong and Dirks (1977, 1978) pre-
sented evidence that weaker elevated heat islands near
450 m with thermal excesses of only 1 K and weaker
thermal gradients, accompanied by slightly stronger
mixing layer winds >4 m s- 1 upwind were associated
with decreases of wind speeds over the metropolitan
area. This decrease of wind speed was best explained
by an increase in frictional drag over the city due to
a combination of increased surface roughness and in-
creased vertical mixing resulting from the higher sur-
face temperature. The relative roles of the thermal
field and the surface drag as a function of wind speed
are explainable from theory. The pressure force ex-
erted by the thermal anomaly is linearly related to the
wind acceleration and is determined by the thermal
gradient. The surface drag, however, is proportional
to the square of the velocity so that its role increases
rapidly with higher wind speeds. A critical wind speed
at which the surface drag begins to dominate ( ~ 4-5
m s- 1 in the summertime) should then depend on the
53
magnitude of the thermal gradient and the complex
characteristics of the surface roughness. A critical
speed would be expected to be higher for winter when
the heat island is more intense, although stability is an
important influence.
Differences have been noted between the mean di-
vergence patterns for light and strong wind speeds
(Ackerman, 1978). The average afternoon evolution of
the divergence field for wind speeds ~5 m s- 1 and >5
m s- 1 regardless of wind direction is shown in Fig. 13.
At noon, the flow over the metropolitan area was di-
vergent when the winds were light. With strong winds
the airflow over the metropolitan area was strongly
convergent, with convergent flow extending to the
southeast. By 1400 CDT, the airflow over most of the
urban area was convergent in both light and strong
wind conditions although the mass convergence was
greater when the winds were >5 m s- 1 • At 1600 CDT,
the air was definitely converging over the metropolitan
area for both light and strong winds, again more
 54 METEOROLO GICAL MONOGRAPHS

1200 COT 1400 COT' 1600 COT

'
.
- .
. ,, o.Jt__/

'
?d
SPEED, >5 m s- 1

SPEED, <5 m s- 1 , NW AND SW WINDS

FIG. 13. Afternoon evolution of divergence (I0-6 s- 1) for all wind direction stratified by wind speed ,;;;5 m s- 1 and >5
m s- 1 • After Ackerman (1978).

strongly for the latter. Extension of the convergence
to the east or northeast also occurred with both weak
and strong winds.
The representativeness of the perturbation stream-
line field can be gleaned from the analysis of 21 fair-
day airflows presented by Ackerman (1978). Fig. 14
shows the principal features of the average perturba-
tion flow and accompanying divergence field con-
structed for the average of all wind directions and
speeds in the well-mixed sub-cloud layer. Although no
thermal data is presented in these analyses, the rela-
tive inflow and convergence in the mixed layer over
the metropolitan area throughout the afternoon is
clearly indicated.
Vertical profiles of divergence averaged over two
urban and three rural areas for 21 fair weather cases
METROMEX: A REVIEW AND SUMMARY
8 1200 COT " "
FIG. 14. Afternoon evolution of the perturbation streamlines and divergence contours
oo-e s- 1) averaged in the sublcoud layer for 21 fair weather cases. After Ackerman (1978).
with winds between 2 and 10m s- 1 are shown in Fig.
15. These areas were chosen on the basis of approxi-
mately homogeneous land use and good delineation by
the pibal station network, and each represents
-300-400 km 2 • Divergence values at various levels
below 1 km correspond well with other reported val-
ues of 2-5 x w-s s- 1 by METROMEX investigators
for similar areas (Ackerman, 1974a, b; Ackerman and
Appleman, 1974; Auer, 1975a; Wong and Dirks, 1978).
Figs. 12 and 14 show that local centers of divergence
can approach I0- 4 s- 1 for small areal extent; even as
much as w-as- 1 has been observed over industrial
areas (Kropfli and Kohn, 1978).
It can be seen in Fig. 15 that the average flow was
more convergent over the city of St. Louis than over
the suburbs to the west by a factor of two in the lower
levels, somewhat less aloft. For the St. Louis metro-
politan area the average horizontal inflow in the lowest
one kilometer must apparently increase significantly
between 1200 and 1400 CDT due to the increase in
both magnitude and depth of the convergence. The
55
Illinois urban area (area B), on the other hand, was
close to nondivergent, on the average, except at 1600
CDT; the convergence at this time, however, was due
solely to the airflow in the Granite City subdivision
which was convergent. Of the three rural areas, only
area D, south of the urban industrial complex, showed
significant convergence when compared to that over
the city in early afternoon. The flow over the river
bottoms between the Mississippi and Missouri Rivers
in the northwest (area E) became divergent as the day
progressed, in opposition to the trend over the met-
ropolitan area. Such features have also been found in
the analyses by Wong and Dirks (1978).
Low-level wind patterns become more complex in
the presence of rain systems because shower clouds
often produce their own local disturbances. These can
mask or alter urban-induced disturbances so that it
becomes difficult to sort out the urban- and cloud-in-
duced perturbations in the wind field. The wind pat-
terns prior to the onset of the rain, however, should
reflect the urban effect on the wind field although it
 56 METEOROLOGICAL MONOGRAPHS

1. 6
A J
1
/. i i '-
].4
i i "
" -
1-•

l-2
j \
I
I'r1
Jl
/ 1! - --1200 • I
...: ].0 1-0 -
~

!.. o. s,-
il ~ o.a ~ //
/
i - -1400
tS i -·- 1200
~
0
::>
r- y
<'I
::> 0.6
:: 0.6f- ;:
~
) i ;;i
<t I /r
I 0.4 \ I
0. 4·-
0.2
O 8 Minouri Mruopolitan Area
--~

.-
1. 4
1
-l B
1.4
1

I
I
~-
T

(
. \ '\_ \ '\ I
1.2
.!i
1. 2
1.0

- '--->
~

j
I
; 1.o1 0.8
//I
~
~
l [
~ 0. 8- 0. 6 -
:t /
/
I
,..

l
0.4 ~ i
v
o.z L
C Ru<al No<lh..SI _I.
- ' I
1.6

1.4 ·
E j
I
.!i J.2•
J.O·
l-
1.2- :g I
I
0.8· --l \ 1
...: 1.0- ~ \\ 1
~
.=. o.e . . . :t 0.6-

§l 0-t
- i
:: 0. 6- I

.
0.2,
~- 0 . 4t-
I
0
~Southeast
l 1
e Mmoun M•n•WPI)I 8ottom\
·10 ·8 -6 ·4 ·2 2 4 -1 0 ·8 ·6 ·4 ·2 0

_._j
0. 2
d Rural Sout hCoil.$11
0 L
-6 -4 -2 0 4 6 ·6
DIV~RGE:fC~
(l o·> s · • )
FIG. 15. Afternoon evolution of vertical profiles of areal-averaged
divergence of five areas, mean of 21 fair weather cases. After Ack-
erman (1978).
may be affected by other processes aloft if the showers
are being triggered by larger-scale systems. Fig. 16
shows the profiles of the average divergence over five
land use areas, averaged over all available antecedent
(1 to 4 hours) rain cases. The transport wind speeds
were about equally divided between ::::::5 m s- 1 and > 5
m s- 1 • The convergence over the Missouri metropolitan
area was largely confined to the lower 600-800 m at
1200 and 1400 CDT. Between 1400 and 1600 CDT, the
mass inflow not only increased in magnitude but also
increased in depth, with convergence greater than 3
x 10 5 s- 1 up through at least 1.5 km, typically at or
near cloud base. The low-level flow over the Illinois
urban area (Area B) was convergent after 1400 CDT
and the depth of the convergence increased and ex-
tended through the lowest kilometer by 1600 COT. In
general, there appears to have been a convergence
01VERGENCE (lo· > s . , ) 01VERGENCE ( 10'' s.')
FIG. 16. Afternoon evolution of vertical profiles of areal-averaged
-4 -2 0 2 4 divergence of five areas, for pre-rain conditions. After Ackerman
DIVERGENCE (10'' s' 1 ) (1978).
over most of the regions at low levels for the five areas
in the 1600 COT sample, with the convergent flow
deepest and strongest over the metropolitan area, and
shallowest in the Missouri-Mississippi Bottoms to the
north.
The average divergence fields for four squall line
and four airmass cases of precipitation occurrence are
shown in Fig. 17. Both show convergence over the
metropolitan area although the magnitude of the con-
vergence in squall line cases is about twice that for
the airmass cases. In addition, squall line cases exhibit
a second center of convergence flow, equal in mag-
nitude to that of the main core, just northeast of Gran-
ite City, and a matching zone of divergence just north
of the city.
Horizontal convergence must be accompanied by
vertical stretching of the airmass. Vertical velocities
calculated from the horizontal divergence are gener-
ally much smaller than those associated with thermals
of relatively small diameter, but are important never-
METROMEX: A REVIEW AND SUMMARY
• •
a ~
Squall line stonns O...._
o
K1 LOMf:'l'ER.I;)
to
FIG. 17. Average divergence (I0-6 s- 1) for subcloud transport wind preceding four squall
line storms and four air mass showers. After Ackerman (1978).
theless since they indicate the rate at which the bound-
ary layer may deepen and destabilize due to converg-
ing air. For mean mixing layer wind speeds of around
4 mps, the evidence provided by Auer (1975a), Ack-
erman (1978) and Wong and Dirks (1978) suggests that
the deepening of the mixing layer over the urban area
is, at least in part, the result of mesoscale airflow con-
vergence, that convergence of flow being into the el-
evated heat island. Calculated vertical velocities of 3-7
em s- 1 were found to be of sufficient strength to ex-
plain the observed mixing layer doming of several
hundred meters over the urban area (Ackerman ,
1974b; Auer, 1975a). The 21-day mean of the area-
averaged vertical velocities at the top of the sub-cloud
layer for the areas in Fig. 15 indicates mean vertical
velocities varying from ~ + 4 em s- 1 in the city to -2.6
em s- 1 in the rural area to the north. In line with di-
vergence patterns, the vertical motion was upward
over the Missouri metropolitan area and downward
over the Bottoms to the north at all three afternoon
times (Ackerman, 1978). On individual days the av-
erage vertical velocity over these areas may be much
larger, and the maximum still larger with reported ver-
tical velocities often reaching 20-30 em s- 1 (Acker-
man, 1977, 1978).
Updraft measurements derived from dual-Doppler
radar show a tendency for a localized steady updraft
of 1 m s- 1 over the southwest corner of Granite City,
a region of heavily industrialized activity (Kropfli and
Kohn, 1978).
Kropfli and Kohn (1978) have also acquired some
unique observations in connection with strong forcing
from an urban heat island. Observations of aircraft-
dispensed chaff detected by dual-Doppler radar sys-
tems has indicated that under certain conditions the
57
20
b
--0 Air mass
horizontal roll mode of convection exists over an ur-
ban area, and it is locked to thermal features generated
by the city, although topography cannot be totally
ruled out. Boundary layer shear did not appear to be
a necessary condition for the existence of these rolls.
This is not to suggest that urban forcing is the only
cause of roll convection, but rather that the urban heat
island may increase the frequency of occurrence of
rolls. From a limited sample it appears that rolls occur
over St. Louis about half the time during clear after-
noon conditions in the summer. Data indicates that
planetary boundary layer winds in excess of about 3.2
mps are necessary for the development of rolls at St.
Louis (Kropfli, 1977).
The data hint, but do not substantiate, that the roll
size varies inversely with wind speed since the strong
boundary layer wind (8 mps) produced a 4 km distur-
bance and the two weaker cases (3-4m s- 1) had a 10
km cross-roll wave-length .
These observations suggest that this mode of con-
vection might be simulated in two dimensions by the
superposition of linear shear and a field of counter-
rotating rolls of equal magnitude. Such a simulation
was performed by adding the streamfunctions for lin-
ear cross-roll shear and for equal-size counter-rotating
rolls. The appropriate spatial derivative of the stream-
function resulted in the two-dimensional wind field
shown in Fig. 18. Dimensions were chosen to be
roughly equivalent to the observed fields on 28 July
1975. The simulated wind field contains the essential
features of the wind field (northeasterly at 3 mps) ob-
served on 28 July 1975. A counter-clockwise roll, cen-
tered at X = 3 km, and a smaller clockwise roll at X
= 7 km, are shown. Although this kinematic analysis
is by no means proof that such superposition is gen-
58
QL___-L----~--~-----L----~--~-----L----~----~~
0 2 3 4 5 6
HORI ZONTAL DISTANCE (X ), km
FIG. 18. Simulated wind field from superposition of uniform counter-rotating horizontal
rolls and linear cross-roll shear. After Kropfli and Kohn (1978).
erally valid, it suggests that heat-island or baroclini-
cally induced rolls can exist.
3.2.2 MODEL RESULTS
Ultimately, the most satisfactory way to detect any
effects of topographic and urban features on clouds
and rain will be through the use of computer models
containing surface heating, surface roughness and ter-
rain height as explicit input variables. The study by
Vukovich et al. (1976) speaks most directly to the
matter of physiographic effects. These authors found
that for winds in excess of about 3 m s- 1 , the wind
convergence induced by the heat island (surface ther-
mal forcing) becomes elongated into a convergence
line extending downwind from the center of the heat
island. This results in an elongated region of upward
vertical motions extending downwind from the city.
Within this region of upward motion, there tended to
be several centers of maximum positive vertical mo-
tion. The center containing the largest positive upward
motions tended to be the one farthest downwind from
the heat island center. The authors concluded that this
pattern could be a result of topography, which except
for narrow bands along the major rivers, rises very
slightly in all directions from the city center. They also
reported that the intensity of the heat island changed
markedly with wind direction. Maps of the vertical
motion field were given for northeast, west and east
winds. The west wind case shows a very pronounced
tongue of upwind motions from the city center toward
the northeast, i.e., toward Edwardsville and Center-
ville , Illinois. This case also shows regions of down-
ward motion over the southwestern, western and
northwestern suburbs of the city. The northeast and
east wind cases both show more localized centers of
METEOROLOGICAL MONOGRAPHS
8 9 10
upward motion to the south and southwest of the city.
Again, these centers of upward motion are attributed
to the rising terrain in what we have called the south-
west hills.
Vukovich et al. indicate the need for a more thor-
ough analysis of interaction between the heat island
circulation and the local topography. Such studies are
currently underway (e.g., Hjelmfelt, 1979). For the
present, our conclusion from the models is that phys-
iographic features may play a major role, through their
interaction with the thermally driven (heat island) cir-
culation, in causing thermally induced effects to be
preferentially located northeast of St. Louis.
3.3.3 TURBULENCE AND FLUXES
The University of Wyoming aircraft was equipped
with a Universal Indicated Turbulence System (UITS)
which yields a measure of turbulence intensity R,
where R = (p/p 0 ) 113 e113 in the inertial subrange (1 em
to 500 m) of eddy sizes. According to Kolmogoroff's
similarity hypothesis, energy first put into a system in
large eddies is extracted, through viscous dissipation,
in small eddies. In the inertial subrange of eddies, tur-
bulence is isotropic and its statistical characteristics
depend only on the rate of energy dissipation, e113 • The
UITS measurement of e113 provides an excellent way
of describing, in relative terms, the urban contribution
to turbulent energy dissipation. Turbulence patterns
analyzed from values of e113 have been determined by
Garrett (1975), Lawson and Auer (1976), and Shea and
Auer (1978). From these studies a consistent picture
of the horizontal and vertical distribution of the eddy
dissipation rate in the mixing layer upwind, over and
downwind of St. Louis is seen. Energy dissipation in-
creases to values characteristic of moderate turbu-
METROMEX: A REVIEW AND SUMMARY 59

turbulent energy in a certain wavelength band. Other

case studies (not shown) reveal that southwest of St.
Louis, where thermal excesses are not generally
found , mechanical turbulence, associated with the
rolling terrain, effects increases in e113 values compa-
rable with those found over St. Louis.
Two studies of turbulent flux measurements of mo-
mentum, heat and humidity utilizing the NCAR Queen
Air equipped with an inertial navigation system were
taken in the St. Louis urban-rural atmosphere to ex-
amine daily variations of turbulent fluxes over various
land uses (Eaton and Dirks, 1977; Ackerman and Hil-
debrand, 1978). In their examination of turbulent flux
measurements, Eaton and Dirks presented the profiles
as shown in Fig. 20, based on one afternoon. For ex-
ample, there are differences in terms of the turbulent
kinetic energy budget between rural and urban atmo-
spheres. Both groups of investigators agreed that
buoyancy plays a stronger role in the production of
turbulence than shear for both the urban and rural
surfaces between the 150 and 500 m flight levels. This
corresponds to the strong thermals observed at these
FIG. 19. Regional turbulence patterns around St. Louis repre- levels which originate at the warm ground surface.
sented at 400 m AGL between l!00-1500 CDT. After Shea and Auer
(1978). Ackerman and Hildebrand concluded that only the
buoyant term showed systematic urban-rural differ-
lence (2.5 ~ e113 ~ 3.2) over the central and eastern ences since the other terms were similar for both areas
portions of metropolitan St. Louis (Fig. 19). Since the on two cases. Energy dissipation was again found to
increases are, for the most part, located both over the be greater over the city than over the rural sites within
higher urban structures and over the surface heat is- the mixing layer, indicating the greater production of
land, it suggests that the increases are most likely as- turbulence also over the city than over the rural area.
sociated with mechanical turbulence and with in- The energy dissipation decreased in magnitude with
creases in the number and intensity of dry thermals. height in agreement with the measurements of Lawson
The relative contribution to turbulent energy increases and Auer (1976). Above the elevated inversion level,
over downtown St. Louis from each source cannot be all turbulent fluxes were found to be very small com-
determined since the instrument used measured total pared to fluxes within the mixing layer.

1500
);yf
~ _____ _ ~V~SI~ ____ -//;/\ ;~~ -- __ _

~ 1000 - city /'' / 1 r\

tl - North of City
~
"'
::>
"'
w
>
g
<t:
....
X
<!)
w
X TIp ~i ..-<:.----'"\
<lz

~7~0----~60~---~50~---4~0--~-3=o--~-2~0----71o~--~0--~10~--2~0~~30
TERMS OF TURBULEtT Kl ETI C ENERGY BALAI CE, em' s-'

FIG. 20. Terms of the turbulent kinetic energy balance for three heights in the
urban and rural atmospheres of the St. Louis metropolitan area.
60 METEOROLOGICAL MONOGRAPHS

REFERENCES ences, Report BNWL-2100, Battelle Pacific Northwest Labo-

ratories, Richland, Washington, 154-156.
Ackerman, B., l974a: Wind fields in St. Louis in undisturbed Auer, A. H., and R. A. Dirks, 1974: Contribution to an urban me-
weather. Bull. Amer. Meteor. Soc., 55, 93-95. teorological study: METROMEX. Bull. Amer. Meteor. Soc.,
- - , l974b: Wind fields over the metropolitan St. Louis area. J. 55, 106-llO.
Air Pollut. Control Assoc., 24, 232-236. - - , 1975a: Calculations of divergence from the wind field over an
- - , 1978: Regional kinematic fields. Summary of METROMEX, urban area. Preprints Second U.S. National Conference on
Vol. 2, Causes of Precipitation Anomalies, Illinois State Water Wind Engineering Research, Wind Engineering Research
Survey Bull. 63, Urbana, 165-209. Council, Ft. Collins, 1-20-1-1-20-3.
- - , and H. Appleman, 1974: Boundary layer program. Interim - - , 1975b: The production of cloud and Aitken nuclei by the St.
report of METROMEX studies, NSF GI-38317, Illinois State Louis metropolitan area (Project METROMEX). J. Rech. At-
Water Survey, Urbana, 121-143. mos., 9, No. I, ll-22.
- - , 1977: Mesoscale wind fields over St. Louis. Preprints Sixth --,and F. D. Eaton, 1976: Final Report for Grant AEN-76-00898
Conf. Planned and Inadvertent Weather Modification, Cham- to the National Science Foundation, Research Applied Direc-
paign-Urbana, Amer. Meteor. Soc., 5-8. torate, Weather Modification Program, Dept. Atmos. Sci.,
--,and P. Hildebrand, 1978: Structure of the planetary boundary University of Wyoming, Laramie, 40 pp.
layer over a complex meso-region. Final report to the National - - , 1977a: Inadvertent modification of visibility in the vicinity of
Science Foundation for Grant ATM 76-15870 to the Illinois a metropolitan area. J. Wea. Mod., 9, No. I, 19--31.
State Water Survey, Urbana, llO pp. - - , 1977b: Evolution of visibility deterioration and aerosol spectra
- - , and J. W. Mansell, 1978: Thermodynamic structure of the in the St. Louis urban plume. J. Air Pollut. Control Assoc., 27,
PBL at midday. Summary of METROMEX, Vol. 2, Causes of 1188--1191.
Precipitation Anomalies, Illinois State Water Survey Bull. 63, --,and S. A. Changnon, 1977: Non-precipitation weather anom-
Urbana, 129--164. alies in St. Louis. Preprints Sixth Conf. Inadvertent and
Alkezweeny, A. J., 1973: Size distribution of St. Louis aerosol par- Planned Weather Modification, Champaign-Urbana, Amer.
ticles. Annual Report for 1972 to U.S. Atomic Energy Com- Meteor. Soc., 1-4.
mission, Division of Biomedical and Environmental Research, --,and M. J. Komp, 1977: The behavior of Aitken nucleus plumes
Vol. 2, Part I: Atmospheric Sciences, Rep. BNWL-1751, Bat- from coal-fired power plants in the Midwest. Preprints Joint
telle Pacific Northwest Laboratories, Richland, WA, 104-106. Conference on Applications of Air Pollution Meteorology, Salt
- - , l974a: Characteristics of aerosols from St. Louis. Annual Re- Lake City, Amer. Meteor. Soc., 29--32.
port for 1973 to U.S. Atomic Energy Commission, Division of - - , 1978: Correlation of land use and cover with meteorological
Biomedical and Environmental Research, Part 3: Atmospheric anomalies. J. Appl .. Meteor., 17, 636-643.
Sciences, Rep. BNWL-1850, Battelle Pacific Northwest Lab- Bergstrom, R. W., and J. T. Peterson, 1977: Comparison of pre-
oratories, Richland, WA, 94-99. dicted and observed solar radiation in the urban area. J. App/.
- - , 1974b: Vertical profiles of trace gases in St. Louis. Annual Meteor., 16, 1107-1116.
Report for 1973 to the U.S. Atomic Energy Commission, Di- Boatman, J. F., 1974: Inadvertent thunderstorm modification by an
vision of Biomedical and Environmental Research, Part 3: urban area. M.S. thesis, Dept. Atmos. Sci., University of Wy-
Atmospheric Sciences, Rep. BNWL-1850, Battelle Pacific oming, Laramie, 129 pp.
Northwest Laboratories, Richland, W A, 105-107. Braham, R. R., Jr., 1971: The University of Chicago in METRO-
- - , 1975: Evolution of the St. Louis aerosol size distribution. MEX. 1971 Operational Report for METROMEX, S. A. Chang-
Annual Report for 1974 to U.S. Atomic Energy Commission, non, Ed., Illinois State Water Survey, Urbana, 29-47.
Division of Biomedical and Environmental Research, Part 3: - - , 1972a: The University of Chicago in METROMEX. 1972 Op-
Atmospheric Sciences, Rep. BNWL-1950, Battelle Pacific erational Report for METROMEX, W. P. Lowry, Ed., lllinois
Northwest Laboratories, Richland, WA 65-69. State Water Survey, Urbana, 25-41.
- - , 1977: S02 to sulfate conversion in the urban plume. Annual - - , 1972b: University of Chicago contribution to Project MET-
Report for 1976 to the Energy Research and Development ROMEX-1. Tech. Note 45, Final technical report to the Na-
Administration, Assistant Administrator for Environment and tional Science Foundation under Grant GA 28190XI, Cloud
Safety. Part 3: Atmospheric Sciences, Rep. BNWL-2100, Bat- Physics Laboratory, Dept. Geophys. Sci., University of Chi-
telle Pacific Northwest Laboratories, Richland, WA, 141-143. cago, 98 pp.
- - , and D. R. Drewes, 1977a: Airborne measurements of pollu- - - , 1974: Cloud physics of urban weather modification. Bull.
tants over urban and rural sites. Annual Report for 1976 to the Amer. Meteor. Soc., 55, 100--106.
Energy Research and Development Administration, Assistant --,and P. Spyers-Duran, 1974: Ice nucleus measurements in an
Administrator for Environment and Safety, Part 3: Atmo- urban atmosphere. J. Appl. Meteor., 13, 940--945.
spheric Sciences, Rep. BNWL-2100, Battelle Pacific Northwest - - , 1975: Overview of urban climate. Proceedings of the Confer-
Laboratories, Richland, WA, 144-145. ence on Metropolitan, Physical, and Environment, Northeast-
- - , and - - , 1977b: Airborne measurement of pollutants over em Forest Experimentation Station, Forest Service, USDA,
urban and rural sites. J. Appl. Meteor., 16, 561-563. Syracuse, 29 pp.
- - , and D. C. Powell, 1977: Estimation of transformation rate of - - , 1976a: The University of Chicago in Project METROMEX.
S02 and S04 from atmospheric concentration data. Annual Re- Tech. Note 48, Report to the National Science Foundation un-
port for 1976 to the Energy Research and Development Admin- der Grant AEN72-03399-A03, Cloud Physics Laboratory, Dept.
istration, Assistant Administrator for Environment and Safety, Geophys. Sci., University of Chicago, 25 pp.
Part 3: Atmospheric Sciences, Rep. BNWL-2100, Battelle Pa- - - , 1976b: CCN spectra in C-k space. J. Atmos. Sci., 33, 343-345.
cific Northwest Laboratories, Richland, WA, 135-140. - - , 1976c: Modification of clouds and weather by a large metro-
--,and R.N. Lee, 1977: Airborne measurements of hydrocarbons politan area. Paper presented at the International Conference
in the St. Louis plume. Annual Report for 1976 to the Energy on Weather Modification, Boulder, 8 pp.
Research and Development Administration, Assistant Admin- - - , 1977: University of Chicago contribution to Project METRO-
istrator for Environment and Safety, Part 3: Atmospheric Sci- MEX-II. Tech. Note 50, Final Technical Report to the National
METROMEX: A REVIEW AND SUMMARY 61

Science Foundation under Grant ENV72-0399, Cloud Physics Louis on boundary layer airflow convection. Ph.D. disserta-
Laboratory, Dept. Geophys. Sci., University of Chicago, 105 tion, University of Chicago, 185 pp.
pp. Johnson, G. B., 1976: Ultragiant urban aerosol particles. Science,
Bryant, H. K., R. T. Glancy and A. H. Auer, 1978: METROMEX 194, 941-942.
urban and rural vertical wind profiles. Proceedings of the Third Komp, M. J., 1977: Visibility reductions and accompanying aerosol
U.S. National Council on Wind Engineering Research, Wind evolution downwind of St. Louis, Missouri. M.S. thesis, Dept.
Engineering Research Council, Gainesville, 11-5-1. Atmos. Sci., University of Wyoming, Laramie, 125 pp.
Cataneo, R., 1973: Aircraft measurements and observations. Sum- --,and A. H. Auer, 1977: Visibility reduction and accompanying
mary Report of MET ROM EX Studies, 1971-1972, F. A. Huff, aerosol evolution downwind of St. Louis, Missouri. Preprints
Ed., Report of Investigation 74, Illinois State Water Survey, Sixth Conf. Planned and Inadvertent Weather Modification,
Urbana, 153-162. Champaign-Urbana, Amer. Meteor. Soc., 25-28.
Changnon, S. A., Jr. 1963: A climatological evaluation of precipi- - - , and - - , 1978a: Assessment of some aerosol properties in
tation patterns over an urban area. Air Over Cities, U.S. Public midwestern power plant plumes. J. Wea. Mod., 10, No. I,
Service Pub!. SEC A62-5, Cincinnati, 37-67. 107-118.
- - , 1968: The LaPorte anomaly-fact or fiction? Bull. Amer. Me- - - , and--, 1978b: Visibility reduction and accompanying aero-
teor. Soc., 49,4-11. sol evolution downwind of St. Louis. J. App/. Meteor., 17,
- - , 1970: Reply. Bull. Amer. Meteor. Soc., 51, 337-343. 1357-1367.
- - , and R. G. Semonin, Eds., 1975: Studies of selected precipi- Kropfli, R. A., 1977: A dual Doppler radar study of the urban
tation cases from METROMEX. Report of Investigation 81, boundary layer: A summary of METROMEX results. NOAA
Illinois State Water Survey, Urbana, 329 pp. Tech. Memo. ERL WTL-26, Wave Propagation Laboratory,
- - , --and F. A. Huff, 1976: A hypothesis for urban rainfall Boulder, 22 pp.
anomalies. J. Appl. Meteor., 15, 544-560. --,and N. M. Kohn, 1978: Persistent horizontal rolls in the urban
Charlson, R. J., A. H. Vanderpol, A. H. Covert, A. P. Waggoner mixed layer as revealed by dual-Doppler radar. J. Appl. Me-
and M. C. Ahlquist, 1974: Sulfuric acid-ammonium sulfate teor., 17, 669-676.
aerosol: optical detection in the St. Louis region. Science, 186, Laulainen, N. S., A. J. Alkezweeny and J. M. Thorp, 1978: Simul-
156--158. taneous aerosol size distribution and turbidity measurements
Curico, J. A., 1961: Evaluation of atmospheric aerosol particle size over St. Louis during METROMEX 1975. J. Appl. Meteor.,
distribution from scattering measurements in the visible and 17, 615-626.
infrared. J. Opt. Soc. Amer., 51, 548--551. Lavoie, R. L., 1968: A meso-scale numerical model and lake effect
Czys, R. R., 1977: University of Chicago measurements of ice-form- storms. Ph.D. dissertation, Pennsylvania State University,
ing nuclei from METROMEX. Preprints Sixth Conf. Planned University Park, 102 pp.
and Inadvertent Weather Modification, Champaign-Urbana, Lawson, R. P., 1973: Urban induced influences on convective cloud
Amer. Meteor. Soc., 29-32. activity. Rep. No. AR107, Dept. Atmos. Resour., University
Dabberdt, W. F., and P. A. Davis, 1974: Determination of energetic of Wyoming, Laramie, 53 pp.
characteristics of urban and rural surfaces in the greater St. --,and A. H. Auer, Jr., 1976: Urban induced influences on con-
Louis area. Preprints Symp. Atmospheric Diffusion and Air vective cloud activity. J. Rech. Atmos., 10, No. 3, 143-156.
Pollution, Santa Barbara, Amer. Meteor. Soc., 133-141. Ochs, H. T., Ill, 1975: Modeling of cumulus initiation in METRO-
Dirks, R. A., 1974a: Urban modifications of urban heat and moisture MEX. J. Appl. Meteor., 14, 873-882.
fields. Preprints Fourth Conf. Weather Modification, Ft. Lau- Peterson, J. T., and E. C. Flowers, 1973: Effects of air pollution on
derdale, Amer. Meteor. Soc., 362-365. solar radiation. Bull. Amer. Meteor. Soc., 54, 910.
- - , 1974b: Urban atmosphere: warm, dry envelope over St. - - , 1977: Interactions between air pollution and solar radiation.
Louis. J. Geophys. Res., 79, 3473-3475. Solar Energy, 19, 23-31.
Duntely, S. Q., R. W. Johnson and J. I. Gordon, 1973: Airborne Russell, P. B., E. E. Uthe, F. L. Ludwig and N. A. Shaw, 1974:
measurements of optical atmospheric properties in southern A comparison of atmospheric structures as observed with mon-
Illinois. Sci. Rep. No. 3, Contract Fl9628-73-C-0013 to Air ostatic acoustic sounder and lidar techniques. J. Geophy. Res.,
Force Cambridge Research Laboratories, Scripps Institution of 79, 5555-5556.
Oceanography, University of California, San Diego, 1% pp. Semonin, R. G., 1978: Urban boundary layer-aerosol patterns.
Dytch, H. H., 1972: METROMEX temperature and aerosol data Summary of METROMEX, Vol. 2, Causes of Precipitation
reduction: analysis of the urban anomaly. University of Chi- Anomalies, Illinois State Water Survey Bull. 63, Urbana,
cago contribution to Project METROMEX-1. Tech. Note 45, 103-110.
Final technical report to the National Science Foundation under Shea, D. M., 1976: Observations of the regional influence on ther-
Grant GA 28190XI, Cloud Physics Laboratory, Dept. Geophys. modynamic properties and aerosol patterns in the plume down-
Sci., University of Chicago, 73-90. wind of St. Louis. Rep. No. AS I 17, Dept. Atmos. Sci., Uni-
Eaton, F. D., and R. A. Dirks, 1977: Turbulent flux measurements versity of Wyoming, Laramie, 63 pp.
in the urban-rural atmosphere of greater St. Louis. Preprints - - , and A. H. Auer, Jr. 1978: Thermodynamic properties and
Sixth Conf. Planned and Inadvertent Weather Modification, aerosol patterns in the plume downwind of St. Louis. J. Appl.
Champaign-Urbana, Amer. Meteor. Soc., 17-20. Meteor., 17, 689-698.
Endsley, K., and D. Knowlton, 1973: Aircraft data acquisition sys- Shreffler, J. H., 1978: Detection of centripetal heat-island circula-
tem for the University of Wyoming research aircraft. Rep. No. tions from tower data in St. Louis. Bound.-Layer Meteor., 15,
AR109, University of Wyoming, Laramie, 18 pp. 229-242.
Garrett, V. F., 1975: Urban-rural differences in the boundary layer Sisterson, D. L., 1975: Studies on the urban moisture budget. Rep.
thermal structure. M.S. Thesis, Dept. Atmos. Sci., University No. ASI14, Dept. Atmos. Sci., University of Wyoming, Lar-.
of Wyoming, Laramie, 129 pp. amie, 52 pp.
Hanna, S. R., 1969: The thickness of the planetary boundary layer. --,and R. A. Dirks, 1978: Structure of the daytime urban mois-
Atmos. Environ., 3, 519-536. ture field. Atmos. Environ., 12, 1943-1949.
Hjelmfelt, M. R., 1980: Numerical simulation of the effects of St. Spangler, T. C., 1972: Dynamically induced-meso-scale variations
62
of the urban mixing layer depth. Rep. No. AR104, Dept. At-
mos. Resources, College of Engineering, University of Wyo-
ming, Laramie, Wyoming, 52 pp.
- - , and R. A. Dirks, 1972: Meso-scale variations of the urban
mixing layer depth. Preprints Conf Urban Environment, Phil-
adelphia, Amer. Meteor. Soc., 7-42.
- - , and - - , 1974: Meso-scale variations of the urban mixing
height. Bound.-Layer Meteor., 6, 428-441.
Spyers-Duran, P., 1974: Cloud condensation nuclei measurements
and estimates of production rates in the St. Louis urban com-
plex. Preprints Fourth Conf. Weather Modification, Ft. Lau-
derdale, Amer. Meteor. Soc., 390-395.
- - , 1975: The effect of repeated activation on depositional ice
nuclei. J. Appl. Meteor., 13, 628--629.
Uthe, E. E., and P. B. Russell, 1974: Experimental study of the
urban aerosol structure and its relation to urban climate mod-
ification. Bull. Amer. Meteor. Soc., 55, 115-121.
Volz, F. E., 1959: Photometer mit Selen-Photoelement zur spektra-
len Messung der Sonnenstrahlung und zur Bewtimmung der
Wellenlangenabhangigheit der Dunststrubung. Arch. Meteor.
Geophys. Bioklim., 810, 100-131.
Vukovich, F. M., and J. W. Dunn, 1978: A theoretical study of the
St. Louis heat island: some parameter variations. J. Appl. Me-
teor., 17, 1585-1594.
--,--and B. W. Crissman, 1976: A theoretical study of the St.
Louis heat island: the wind and temperature distribution. J.
Appl. Meteor., 15, 417-440.
Wesely, M. L., 1975: Measurements of atmospheric turbidity in an
arc downwind of St. Louis. Radiological and Environmental
Research Division Annual Report to ERDA, Contract W-31-
109-Eng-38, Atmospheric Physics, Part IV, Argonne National
Laboratory, 22-30.
White, J. M., 1977: The net radiation budget of the St. Louis met-
ropolitan area. M.S. thesis, Dept. Atmos. Sci., University of
Wyoming, Laramie, 99 pp.
- - , and F. D. Eaton, 1977: The study of net radiation and its
METEOROLOGICAL MONOGRAPHS
components with relevance to inadvertent weather modification
in St. Louis. Preprints Sixth Conf Planned and Inadvertent
Weather Modification, Champaign-Urbana, Amer. Meteor.
Soc., 21-24.
- - , 1978: Global and actinometric measurements within the St.
Louis urban environment. Preprints Third Conf. Atmospheric
Radiation, Davis, Amer. Meteor. Soc., 322-325.
- - , F. D. Eaton and A. H. Auer, 1978: The net radiation budget
of the St. Louis metropolitan area. J. Appl. Meteor., 17,
593-599.
Wong, K. K., 1976: Urban perturbations on the mixing layer inflow
in St. Louis. M.S. thesis, Dept. Atmos. Sci., University of
Wyoming, Laramie, 136 pp.
--,and R. A. Dirks, 1977: Mesoscale perturbations on the airflow
in the urban mixing layer. Preprints Sixth Conf. Planned and
Inadvertent Weather Modification, Champaign-Urbana, Amer.
Meteor. Soc., 9-12.
- - , and - - , 1978: Mesoscale perturbations on airflow in the
mixing layer. J. Appl. Meteor., 17, 677--688.
Yoksas, T. C., 1978: Turbidity measurements in METROMEX,
1976: Dept. Atmos. Sci., University of Wyoming, Laramie, 10
pp. [Unpublished manuscript].
Young, J. A., T. M. Tanner, C. W. Thomas, N. A. Wogman, and
M. R. Petersen, 1974: Concentrations and rates of removal of
contaminants from the atmosphere in and downwind of St.
Louis. Annual report for 1973 to U.S. Atomic Energy Com-
mission, Division of Biomedical and Environmental Research,
Part 3: Atmospheric Sciences, Rep. BNWL-1850, Battelle Pa-
cific Northwest Laboratories, Richland, W A, 102-105.
- - , T. M. Tanner, C. W. Thomas, N. A. Wogman, and M. R.
Petersen, 1975: Concentrations and rates or removal of contam-
inants from the atmosphere in and downwind of St. Louis. An-
nual report for 1974 to U.S. Atomic Energy Commission, Di-
vision of Biomedical and Environmental Research, Part 3:
Atmospheric Sciences, Rep. BNWL-1950, 70--76.
 CHAPTER 4

Cloud Characteristics
RICHARD G. SEMON IN

CONTENTS

ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.I Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.2 Cloud frequencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.3 Cloud morphology and dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.4 Cloud microstructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.4.1 DROPLET SPECTRA .......................... .......................... ........ 69
4.4.2 IN-CLOUD AEROSOL OBSERVATIONS .......................... ................... 71
4.4.3 LIQUID WATER CONTENT (LWC) .......................... ..................... 72
REFERENCES .......................... .......................... ................... 73

ABSTRACT

Studies of convective clouds revealed marked influences on their characteristics and distributions. Surface
cameras and satellite photographs showed first clouds of the day in the St. Louis area initiated most frequently
over the center of the metropolitan area and over the intensely industrialized area of Alton-Wood River. These
areas were also cloudier in the later afternoon. The cloud base heights of cumulus over the St. Louis commercial
district and adjacent downwind area were higher than comparable rural clouds.
Aircraft measurements of many summer clouds showed that the median updraft speed was 2.5 m s- 1 and the
most frequent updraft location was on the right front flank of the storm. The measurements of Aitken nuclei
inside and outside of cumulus with bases extending into the mixing layer showed the direct ingestion of surface
source material into convective clouds. The average downwind increase of the cloud condensation nuclei (CCN)
concentration was 94% measured at 1% supersaturation in a thermal diffusion counter. Measurements of the
cloud droplet size distribution in small cumulus and stratus clouds showed narrower distributions of sizes with
greater number concentrations in downwind clouds compared with upwind clouds.
All of the observations of CCN and cloud microstructure obtained by the various investigators supported the
concept that the urban effect on cumulus clouds is to shift the droplet size spectrum toward smaller median
droplet diameters, and, with a lesser degree of confidence, the observations suggest an increase in the number
of larger droplet sizes. The measured and calculated liquid water content (LWC) in the urban-effected clouds
beyond St. Louis consistently showed increases associated with such clouds.

tive location of cloud initiation in these two areas, as

4.I Introduction
The Illinois Water Survey operated two all-sky cam-
eras and one directional camera to photograph clouds
in the research area during various periods of the proj-
ect. These cameras were located in St. Louis and near
Alton to provide statistical data on the time and rela-
63
well as documented histories of convective storms as
they traversed the area. In addition, GOES (Geosta-
tionary Operational Environmental Satellite) data were
available for 1975 permitting analysis of cloud fre-
quencies for fixed hours during the day.
The University of Chicago, Battelle Pacific North-
64 METEOROLOGICAL MONOGRAPHS

west Laboratories, Illinois Water Survey and the

University of Wyoming operated aircraft during the
METROMEX research efforts from 1971-76. While
the goals for the aircraft operations varied among
these groups, all of them were involved in either sam-
pling the atmospheric conditions prior and during con-
vective cloud formation (frequently near cloud base)
or the penetration of clouds at varying levels to de-
termine their thermodynamic and microphysical char-
acter.
All of these data, both camera and aircraft, were
used on a limited basis as input for initializing and
verifying numerical efforts related to urban-modified
convective clouds. Some of the results are presented
in the chapter on precipitation processes.
The characteristics of convective clouds are de-
scribed in the following sections including a descrip-
tion of cloud. frequencies, cloud morphology and dy-
namics and cloud microstructure. KILOt..,ETERS
0 8 16
0 s 10
4.2 Cloud frequencies S'l'ATUTE MILES

FIG. I. The camera locations used to determine the initiation of

One of the primary concerns of a study of urban cumulus clouds.
weather modification is the possible influence of an
industrialized metropolitan area on the initiation of
cloudiness. A second and related question is then di- port for operation during July-August 1975. This lo-
rected toward determination of the prolongation of cation was quite suitable to detect the influence of the
cloud cover over and downwind of the presumed sur- Alton-Wood River refinery complex on the develop-
face source perturbation resulting from the anomalous ment of initial cumulus clouds. During the June-Au-
initiation of convective cloudiness. gust periods of 1971-73, a pointing 35 mm camera was
The all-sky cloud camera systems used were com- operated at the Pere Marquette site. This camera was
prised of a 16-mm camera mounted over a reflecting oriented to photograph the 120° sector that included
hemisphere 18 em in diameter with a clock and date east through south from the Pere Marquette vantage
indicator in the field of view. Photographs were ob- point. This direction (as can be seen in Fig. 1) included
tained once every five minutes during daylight hours the major industrial areas of St. Louis and Alton-
and within summer operational periods. The geometry Wood River. Photographs were obtained every 5 min
of the all-sky cameras and their photographs was such with this camera providing compatibility with the St.
that detection of clouds inside of 5 km was excellent, Louis University all-sky camera.
but was less quantitative beyond about 5 km. The cloud data were analyzed by Grosh ( 1978a) for
The St. Louis University camera was operated in the 1971-75 period at St. Louis University as well as
August 1971, June-August 1972, June-August 1973, the limited data from the Alton Airport, Pere Mar-
July-August 1974, and July-August 1975. A second all- quette and satellite observations. A summary of the
sky camera was installed at the Alton Me:1torial Air- St. Louis findings is shown in Table 1 where it is read-

TABLE 1. The frequency of cumulus initiation as a function of direction from the St. Louis University all-sky camera site. After Grosh
(1978).

Number
Percent of total in each sector of
N NW w sw s SE E NE cumuli*

First 10 Cu 13.5 17.9 21.9 9.0 9.0 10.6 8.8 9.4 480
First 5 Cu 13.0 17.2 24.4 9.5 7.3 12.6 6.9 9.2 262
First Cu 11.9 14.9 26.9 9.0 4.5 10.4 7.5 14.9 67
First 10 Cu within 4.8 15.3 14.7 23.7 10.7 10.7 9.7 7.0 8.3 300
km of camera

*The numbers of cumuli include multiple clouds on individual photographs and, therefore, are greater than the simple expected numbers
of 40, 200 and 400.
METROMEX: A REVIEW AND SUMMARY 65

TABLE 2. Cumulus initiation frequency within 60-degree sectors as observed with the all-sky camera at Alton site. After Grosh (1978).

Percent per 60° sector Number

w sw of
(Alton) (WWR) 1 SE E NE NW cumuli

First 10 Cu 22.9 22.9 13.7 12.9 13.7 14.0 271

First 5 Cu 21.7 24.8 13.7 14.3 12.4 13.0 161
First Cu 12.5 47.5 15.0 12.5 5.0 7.5 40
Sector involved in first 9.7 41.9 19.4 12.9 6.5 9.7 31 cases
Cu episode (%)

•wood River Refinery Complex.

ily seen that greatest frequency for cloud initiation is
in the west sector. Combining these into larger sec-
tions dealing with upwind (northwest-southwest) and
downwind (northeast-southeast) cloud formation, the
largest frequency was observed to the west with 51.1%
as opposed to 28.7% in the downwind or east sectors.
These results indicate an anomalous initiation of first
clouds to the west of St. Louis University.
The photographic data obtained at Pere Marquette,
48 km northwest of St. Louis, with a pointing 35-mm
camera were studied by sorting the clouds into three
40° sectors to the east, southeast and south. The re-
sults showed frequencies of 38%, 30% and 32% in the
east, southeast and south sectors, respectively. The
difference between these percentages and those that
might be expected if the cloud initiation was uniformly
distributed in these sectors was not statistically sig-
nificant. However, the distance over which clouds
might be discerned with this camera was limited to 24
km primarily by poor visibility which frequently was
observed at Pere Marquette. This limitation on dis-
tance suggests that the observed clouds in the three
sectors occurred over the Bottomlands between the
Mississippi and Missouri Rivers to the south and
southeast, and the river bluffs and higher rural country
to the east. Within this range of the camera the only
obvious anthropogenic causal factor for cloud initia-
tion was the Portage De Sioux power plant located in
the east sector 19 km to the east-southeast. This sector
had the highest frequency of cloud initiation as viewed
from this site. On several days observers visually
noted the initiation of cumulus clouds formed in as-
sociation with the plume from the 200 m tall stacks of
this plant.
The Alton all-sky camera data for 1975 were strat-
ified into 60° sectors. This stratification was dictated
by the desire to include the Alton-Wood River refi-
neries within a sector, and necessitated, therefore, a
different sector definition than used at the St. Louis
University site.
The combined west and southwest sectors showed
a relative frequency of 45.8% opposed to those dia-
metrically opposite, that is, the northeast and east sec-
tors with a total of 36.6% as seen in Table 2. The west
sector with 22.9% was toward the heavily industrial-
ized areas in Alton; and the southwest sector, with
22.9% was directed toward the Alton-Wood River re-
finery complex.
The first cumulus occurred earlier, on the average,
in the southwest sector as compared with the other
five sectors. There were 13 days out of 21 cloud days
on which the first cumulus of the day initiated in the
southwest sector. The average time of appearance was
at 1057 CDT with a range between 0918 and 1350 CDT.
On the remaining 8 days, when the first cumulus ini-
tiated in one of the other sectors, the average time of
appearance was 37 minutes later, at 1134 CDT, with
a range between 1008 and 1353 CDT.
Grosh (1978a) performed a chi-square test to deter-
mine the probability of the observed sector distribu-
tion of cumulus initiation occurring by chance in a
sample from a uniform distribution. These probabili-
ties are shown in Table 3, for all of the available data
obtained at St. Louis University and Alton Airport.
All of the distributions indicated that the probability
of the observed frequencies happening by chance was
::::;3%, and over half the distributions indicated the
probability of chance at ::50.01%. Hence, industrial in-
fluences (or some other effect related to location) on
cloud initiation appeared likely in the Alton area.
The GOES satellite data for 1975 were analyzed by
Grosh (1978b) in an attempt to supplement the fore-
TABLE 3. The probability that the observed initiation of cumulus
by sector was by chance in a sample from a uniform distribution.
After Grosh (1978).
Probability based on
chi-square test
St. Louis University Site
First 10 Cu <<0.0001
First 5 Cu <<0.0001
First Cu =0.02
First 10 Cu within 4.8 km <<0.0001
Alton Site
First 10 Cu =0.03
First 5 Cu =0.02
First Cu =0.0001*
% of days involved in first Cu =0.02*
*Includes Yates correction for small expected values.
 66
Mean = 0.43
St Dev = 0.03
Mean = 0. 76 c. 1600
St Dev = 0. 04
FIG. 2. The departure expressed as standard deviations, from the mean of cloud frequencies derived
from the GOES satellite at 1000, 1300, 1600 and 1900 CDT.
going surface cloud photography. Photographs in-
cluded in the analysis were taken within 30 minutes of
the hours of 1000, 1300, 1600 and 1900 CDT. For the
months of June, July and August there were a possible
92 photographs at each of the hours of which 82, 74,
67 and 38, respectively, were actually used in the anal-
ysis. The decrease in number resulted from various
problems not the least of which was the fact that fewer
photographs at 1900 resulted from darkness and ex-
isting cloud shadows over the Midwest. The photo-
graphs were obtained from the visible channel data
with about 1.6 km resolution. The satellite photo-
graphs were gridded into 81 squares of 10 km on a side
centered on the St. Louis urban center. The occur-
renee of cloudiness within a square was tabulated, and
the total frequency was divided by the number of pho-
tographs to obtain a percentage of occurrences over
METEOROLOGICAL MONOGRAPHS
fll ~ +2 St Dev
0 +1 St Dev
~ -1 St Dev
iJ ~ -2 St Dev
Mean = 0. 76 b . 1300
St Dev = 0.05
Mean = 0.49 d. 1900
St Dev • 0.05
the entire grid. The results from this analysis are
shown in Fig. 2. The average percentage of cloud fre-
quency and the standard deviation over the domain of
the grid is shown for each hour. The shaded areas
indicate grid squares where the cloudiness was one or
two standard deviations from the areal mean value for
the given hour.
The first obvious feature of this analysis is the gen-
eral increase of cloudiness from 1000 to 1600 CDT and
the diminution of cloudiness as indicated at 1900 CDT
following an expected diurnal pattern. The relatively
high frequency (>2 standard deviations) of cloud
cover northwest of St. Louis at 1000 is in an area
where surface cloud camera studies indicated anom-
alous cumulus initiations. This maximum area in a par-
tion of the north-south oriented area of one standard
deviation increased frequency of cloud cover com pen-
METROMEX: A REVIEW AND SUMMARY 67

sated by a low area of 2 standard deviations near

Belleville and south along the Mississippi River. Gen-
erally, there is a low over and southeast of St. Louis
at 1000.
By 1300, when convective cloud activity is normally
increasing, the relative maximum area to the north of
St. Louis observed at 1000 was replaced with a rela-
tively low frequency of cloud cover, and a relative
maximum over the Ozark hill area southwest of the
city at 1000 became more obvious. A minor maximum
of one standard deviation appeared over Belleville in
almost the same location as the negative departure
shown there in the 1000 CDT pattern.
At 1600 a high of one standard deviation exists over
the eastern metropolitan area, and a major high in
cloudiness exists 10 to 20 km east of the urban area.
By 1900 there was a major high in the Alton-Wood
River industrial area. To the south a nearly consistent
low frequency area with one standard deviation below
s
the areal mean value continued as it had throughout 0 10
STATUTE MILES
the entire 9 h period. FIG. 3. Composite cloud base heights (km) observed from an air-
Aircraft observations of cloud base heights were craft on many summer days during 1971-75. The observations were
obtained during the 1200-2000 period of the 5-year obtained randomly over the research area between 1200 and 2000.
project. The geographically random measurements
observations of greater mixing depths over the city on
gathered on many days during 1971-75 were super-
most rain days, combined with visual observations of
imposed and analyzed to show the variation in the
convective cloud bases in the St. Louis area, help es-
base heights of small cumulus over and near the St.
tablish the fact that the urban modified boundary layer
Louis metropolitan area (Semonin and Changnon,
extends to cloud base and subjects the cloud dynamics
1974). These observations showed the average cloud
to these modified properties. The observed difference
base over East St. Louis and along the Mississippi
between urban and rural cloud bases is further sup-
River southward in St. Louis to be approximately 600
ported by the modeling findings of Auer and Dirks
m higher than the surrounding rural clouds (Fig. 3).
(1973a) in that the magnitude of temperature and hu-
Lawson and Auer (1976) reported on the develop-
midity field anomalies can clearly account for the ob-
ment of convective clouds on three days from 1971.
served variations. For further discussion of this point
Aircraft observations of the cloud base on one of these
see the Urban Boundary Layer Chapter.
days certainly confirmed the findings reported by Se-
monin and Changnon (1974) of higher cloud bases over
4.3 Cloud morphology and dynamics
the urban area. No cloud base heights were reported
for the remaining two days. A later case study of 20 Many findings show that the clouds in the vicinity
August 1974 by Shea and Auer (1978) also supports of the urban modified boundary layer are subject to
the higher cloud base observations through measure- the surface-altered properties of that atmosphere.
ments of the convective condensation level (CCL) de- During the five years between 1971 and 1975, many
rived from surface and airborne thermodynamic mea- clouds were studied on a case basis by several inves-
surements. Their determination of the CCL on this tigators using aircraft data. In Chapter 6 the aerosol
single day showed the highest level displaced to the concentrations in the boundary layer are described.
northwest under southeast flow at the surface. The These illustrate the great abundance of nuclei available
maximum height was 1050 m AGL as compared to 750 to convective clouds in the vicinity of St. Louis and
m for clouds in more rural areas. these findings are not repeated here.
These observations of a cloud base height anomaly Semonin (1978) summarized data acquired during
that is closely related to the intensely urbanized St. the flight program of the Illinois Water Survey in
Louis area must be placed in the context of other ob- which cloud updrafts were estimated from aircraft ob-
servations related to urban atmospheric properties. servations near precipitating clouds. The vertical mo-
For example, Beebe and Morgan (1972) and Vogel tion of the aircraft in level flight attitude and constant
(1977) reported on the relationship between high mix- power setting was recorded during each updraft pen-
ing depths and rainfall patterns east of the city. These etration. The lower limit of updraft speed estimated
68
TABLE 4. The characteristics of updrafts observed in proximity to convective storms and their location relative to the axis of storm
motion. After Semonin (1978).
Left Right
front front
Number of updrafts 14 24
Average speed (m s- 1) 2.5 2.3
Median speed (m s- 1 ) 2.5 1.7
Maximum speed (m s- 1 ) 6.1 5.5
Number of updrafts ~ 1.5 m s- 1 6 12
Number of updrafts 1.5-2.5 m s- 1 0 4
Number of updrafts 2.5-3.8 m s- 1 5 3
Number of updrafts 3.8-5.1 m s- 1 0 I
Number of updrafts ~5.1 m s- 1 3 4
by this method was 0. 75 m s- 1 • These data are shown
in Table 4. For the 69 cloud base updraft measure-
ments obtained on 13 days during 1972 and 1973, 70%
of the 51 updrafts with their positions known occurred
on the front sides of echoes, as viewed in relation to
their movement. In addition, most of these updrafts
were observed to occur on the right front of the
storms.
The median updraft speed for all updrafts was 2.5
m s- 1 , and for those found specifically at the right front
of existing radar echoes, the median speed was 1. 7 m
s- 1 • Fifty-eight percent of the front-located updrafts
were less than 2.5 m s- 1 and only 20% were greater
than 5.1 m s- 1 • It was noted by Henderson and Duck-
ering (1973) that the inflow areas were typically of
small diameter and short duration, and they were often
difficult to locate visually because of the presence of
many clouds and poor visibility near cloud base. Weak
or dissipating cloud systems provided no identifiable
base inflow area, whereas the inflow was more easily
discerned at the base of large cumulus clouds or thun-
derstorms with durations greater than 30 minutes.
The observed updrafts were not stratified as to po-
tentially affected or unaffected by the urban air, al-
though the nature of the flight missions was such that
nearly all of the clouds in the measurement set were
obtained in the proximity to the St. Louis urban cen-
ter.
The history of updraft velocities obtained from two
thunderstorms in 1972 as they traversed the metro-
politan area from west-to-east was described by Boat-
man (1974). These observations indicated a decrease
of vertical speed as the storms migrated from the
northwest, and reached minimum values over the city
center followed by increased updraft speed as the
storms moved to the east. The increasing updraft ve-
locity east of St. Louis was associated with an increase
of radar indicated cloud tops from 10.7 km over the
city to 13.7 km 32 km east of St. Louis.
Auer (1974), Grosh and Semonin (1973) and Chang-
non and Semonin (1975) have discussed the transport
of surface-origin materials and atmospheric properties
METEOROLOGICAL MONOGRAPHS
Left Right
rear rear Unknown All
6 7 18 69
2.3 2.0 3.4 2.5
2.5 1.3 3.6 2.5
3.1 3.1 7.1 7.1
0 4 3 25
2 0 3 9
4 3 2 17
0 0 5 6
0 0 5 12
to cloud base and into convective clouds. This direct
link was determined from measurements of Aitken
condensation nuclei, potential temperature, industrial
odors and other tracers of the urban boundary layer.
An elegant example of the direct influence of indus-
trialization on clouds was discussed by Auer (1976)
and will be described later. Extensive aircraft obser-
vations of the thermodynamic parameters, cloud up-
draft, condensation nuclei and turbulence were ob-
tained in and near an isolated, non-precipitating,
stationary cumulus cloud growing over a refinery com-
plex.
Grosh (1978a) has shown that the all-sky camera
located at the Alton Airport photographed an anom-
alous number of clouds developing or initiating to the
southwest in the direction of this refinery. Henderson
and Duckering (1974, 1975) reported qualitative ob-
servations of initial first clouds of the day directly as-
sociated with the refinery, icing at relatively warm
temperatures ( -3°C) in these clouds, as well as fre-
quent lowering of cloud bases in proximity to the re-
finery on days when numerous clouds were present.
Observations were reported by Morris and Ulbrich
(1973) of artifical clouds which may have been pro-
duced by an intense fire at a zinc smelting plant. Iso-
lated rain clouds were observed through the growth
and dissipation stages for about 1 h by 3 em radar.
Although less supporting evidence was available, the
conclusions seem to strongly indicate the rain clouds
were inadvertently caused by the industrial fire. How-
ever, none of these observations provided the insight
into industrial cloud development that the in-cloud
measurements of Auer ( 1976) accomplished.
The measurements by Auer (1976) of a cloud formed
directly over an oil refinery complex demonstrated
that the potential temperature observed below and at
cloud base penetrated into the vertical column of the
cloud. The production rate of active cloud condensa-
tion nuclei (at 1% super saturation) was found to be
5.3 x 1015 s- 1 below cloud base over the refinery. In-
cloud measurements of Aitken condensation nuclei
appeared to be directly associated with the concentra-
METROMEX: A REVIEW AND SUMMARY
tions observed in the lower planetary boundary layer.
Auer also reported that the cloud droplet spectra
broadened with height suggesting droplet interaction
and growth by collection as parcels of cloudy air were
lifted in the updraft. While this study cloud was rela-
tively small, extending from 850 to 2200 m MSL, Auer
demonstrated conclusively that the industrial effluent
can be ingested by and greatly affect cumulus clouds.
The unique stationarity of this refinery-initiated cloud
precludes any direct comparison with similar sized
natural clouds on the same day . However, measure-
ments of cloud spectra parameters, in-cloud aerosol
and cloud liquid water content in affected and unaf-
fected clouds (described later) emphasize the anoma-
lous character of the refinery cloud. This case study
also suggests that cloud-associated convection may be
an important mechanism for the redistribution of sur-
face-source aerosol into the upper layers of the lower
planetary boundary layer.
4.4 Cloud microstructure
4.4.1 DROPLET SPECTRA
The University of Chicago sampled the droplet
spectra in a winter nocturnal stratus cloud at 500 m
above the 1.5 km cloud base (Braham, 1974). The
flight track through continuous cloud was from south-
SCALE OF K1LOHETERS
10 IS
Flo. 4. The flight trace and cloud droplet distributions obtained near cloud base
illustrating an urban perturbation on cloud microstructure. After Braham {1974).
69
east to northwest across St. Louis and nearly perpen-
dicular to the local east-northeast winds. The distri-
bution of droplet sizes is shown in Fig. 4 where the
three distributions to the southeast are considered ru-
ral since there are no known sources of anthropogenic
nuclei to the east-northeast. The concentrations ranged
between 1274 and 1615 cm-3 , and the mean diameters
ranged from 7. 14 to 8.10 fLm. Directly over the city
and to the northwest the three distributions seem to
approach these rural characteristics . However, further
northwest the last two cloud droplet distributions
show an increased number of droplets with a decrease
of the mean droplet diameter in response to the per-
ceived influence of industrial sources. Likewise, im-
mediately to the west of Belleville a similar distribu-
tion, with the largest concentration observed (2984
cm- 3 ), is indicative of a strong urban influence.
These data show that the effect of urban related
cloud condensation nuclei upon the cloud droplet
spectrum is extremely fast and provides a unique sig-
nature when compared with rural values (Fitzgerald
and Spyers-Duran, 1973). Spyers-Duran (1972) and
Dytch (1977) give details on the cloud drop spectra
obtained by the University of Chicago group in over
500 clouds in a variety of meteorological conditions,
particularly during summer convective activity. These
results indicate that clouds formed in urban affected
c 2984
d s. 78
c 1615
d 7 . 52
c 1274
d 8.10
20
70 METEOROLOGICAL MONOGRAPHS

air show evidence for increased droplet concentrations clouds. The downwind clouds consistently possessed
of up to 70% relative to similar rural clouds, and the a higher average concentration of droplets > 10 J.tm
increased concentration is accompanied by a 10% re- in radius, supporting the speculations concerning the
duction of the mean droplet size in the urban clouds. downwind development of precipitation by Braham
There is also evidence that the urban influence in- (1974) and Semonin (1974).
creases the concentration oflarge (r ~ 10 J.tm) droplets According to Dytch (1975), those clouds directly
by a factor of more than two during the early stages over the metropolitan area had 16 to 44% greater drop-
of cloud development in the urban-industrial plume. let concentrations than their rural counterparts on a
The concentrations of drops in the large drop "tail" specific day. The downwind clouds, which are deter-
( r ~ 10 J.tm) was observed to decrease as the cloud mined by Aitken nuclei measurements as urban re-
categories progressed from "cloud water" through lated, exhibited droplet concentrations closer to rural
"light precipitation" to "precipitation." This was ex- values, but with a greater concentration of droplets in
plained by preferred growth of the large tail drops into the large diameter tail of the distribution. Braham
precipitation. (1974) and Semonin (1974) both suggested that the in-
The findings of Dytch confirm the measurements creased frequency of first echo development and the
obtained by Auer and Dirks (1973b) in which a com- observed precipitation anomaly may be linked by the
parison was made between upwind and downwind presence of giant and ultra-giant particles. These par-
clouds exhibiting similar properties to those described ticles, if present in sufficient numbers, may accelerate
above. In a very limited sample of clouds, Auer and the warm rain process. Johnson (1978a) collected giant
Dirks (1973b) found the average diameter for the up- and larger aerosols upwind and over the St. Louis ur-
wind clouds to be 8.9 J.tm with a narrowing of the ban area and found a distinct increase in the downwind
spectrum and a mean of 6.4 J.tm diameter for the down- samples although there was a surprisingly large vol-
wind clouds sampled on five days during 1971 and ume concentration of particles with diameters between
1972. The downwind clouds ranged in concentration 5 and 55 microns upwind. The role of these ultra-giant
from 1500 to 3125 cm-3 which compares very favor- particles in the development of precipitation may be
ably with the measurements reported by Dytch. very important. However, the gap between these ob-
Measurements of cloud microstructure were strati- servational data of nuclei and cloud microstructure,
fied by Dytch (1975) into those clouds observed up- and the ultra-giant particle measurements must be
wind, directly over and downwind of the St. Louis bridged through numerical modeling based upon sound
urban center. The measurements were obtained in the physical arguments. For a discussion of these model-
lower part of individual clouds usually at an altitude ing efforts see Chapter 5. Dytch suggested that the
of 100m to 500 m above the cloud bases. A summary greater concentration of droplets in the large diameter
of these findings is shown in Table 5. Clouds directly tail might stimulate the growth of precipitation by
associated with the urban center showed greater drop- means of the coalescence process, or perhaps, by in-
let concentrations than either the rural or downwind teraction between the droplets and the ice phase re-

TABLE 5. The microphysical properties of rural, city and downwind clouds measured in the lower levels near cloud base. After Dytch
(1975).

Summary of lower cloud measurements

Average Average Average

integrated total concentration
Number liquid cloud of "large"
Group of cloud water droplet (r ~ 10 ~-tm)
classifi- elements content concentration cloud droplets
Flight Date cations in group (gm m- 3) (cm- 3 ) (cm- 3 )

208 5 August 1973 Rural 9 0.24 280 5.4

City 5 0.38 403 6.8
Down- 18 0.40 352 13.6
wind
249 4 August 1973 Rural 27 0.27 536 0.021
City 4 0.26 623 0.003
Down- 21 0.29 573 0.082
wind
254 10 August 1974 Rural 8 0.23 434 0.17
City 5 0.26 569 0.16
Down- 12 0.22 430 0.66
wind
METROMEX: A REVIEW AND SUMMARY 71

suiting in observed increased frequency of first echo
formation and precipitation downwind (see Chapter 6).
He further concluded from these observations that the
St. Louis urban-industrial area is a source of large and
giant condensation nuclei [actual measurements of
these large nuclei were made by Johnson (1976); see
also Chapter 5] in addition to being a prolific source
of smaller cloud condensation nuclei.
Companion measurements to those of Dytch (1975)
of the cloud microstructure at higher elevations in con-
vective clouds was reported by Carrera (1975). A sum-
mary of these data are shown in Table 6 where the
samples were divided into those clouds possessing
greater than 1 g m- 3 liquid water content (L WC) and
those with less than 1 g m- 3 • Again the consistency of
greater droplet concentrations in downwind clouds is
obvious, although it will be noted that the concentra-
tion of large droplets was greater in the upwind clouds
by a significant factor. These clouds were penetrated
at average heights of 2 km above the cloud bases and
within 1 km of the estimated cloud tops.
The clouds formed over the city showed greater
droplet concentrations, especially in the smaller drop-
let sizes, resulting in a smaller modal size in the down-
wind clouds compared with rural clouds. In contrast
to the data presented in Table 5 (Dytch, 1975), there
was a decrease in the large droplet sizes in the down-
wind clouds which led Carrera (1975) to suggest that
the depletion of the larger droplets may be due to an
accelerated process of collection and coalescence at
the upper cloud levels in the downwind clouds.
These observations leave little doubt that the urban
associated aerosol has a measurable effect on the mi-
crostructure of cumulus clouds at all cloud levels.
These natural, but urban affected clouds, as compared
with the "laboratory" cloud associated with the refin-
ery complex (Auer, 1976), show some differences.
There is a clear tendency for the refinery-related
clouds to show lesser droplet concentrations with
broadening of the droplet size spectrum as a function
of altitude. The median droplet diameter for the refin-
ery cloud at 230 m above cloud base was 3 11-m in-
creasing to 6.5 11-m at 120 m below cloud top: The
range of droplet sizes at the refinery cloud base was
1-7 11-m, at mid-cloud 1-13 ~-tm, and near cloud top
1-23 11-m. The droplet concentrations in the refinery
clouds with reported values from 1800 to 2800 cm- 3
certainly exceeded those of natural clouds whether
downwind or not of the urban center.
4.4.2 IN-CLOUD AEROSOL OBSERVATIONS
Aitken condensation nuclei (ACN) were measured
and reported by Auer and Dirks (1973b), Semonin

TABLE 6. The microphysical properties of upwind (UW) and downwind (DW) clouds measured in the upper levels near cloud top. After
Carrera (1975).

Summary of upper level cloud measurements

Average over 3-second interval

Total DW/UW ratio

droplet DW/UW ratio Concentration of concen-
Altitude Cloud location concen- of total of droplets tration Mode
Flight (MSL) and number of LWC tration droplet r > 10.5 ,.,m of droplets radius
No. Date LWC (g m- 3) clouds (g m-3) (cm-3) concentration (cm- 3) r > 10.5 ,.,m (f.Lm)
247 2 August 1974 2130 m uw 8 0.43 549 1.0 5
LWC < 1.0 OW 6 (turrets) 0.70 655 1.19 0.6 0.60 5
2740 m uw 3 1.59 610 7.8 6
LWC;;. 1.0 DW 4 1.85 661 1.08 1.7 0.22 5-6
255 II August 1974 3050 m uw 8 0.74 496 2.3 6
LWC < 1.0 ow 2 0.69 416 0.84 1.4 0.61 6
3050 m uw 6 1.36 612 3.1 6
LWC;;. 1.0 ow 3 1.84 622 1.02 2.3 0.74 6
266 23 August 1974 3050 m uw 10 0.65 309 3.4 6
LWC < 1.0 ow 17 0.74 475 1.54 1.4 0.41 5
3050 m uw 4 1.32 415 4.3 6
LCW;;. 1.0 ow 21 1.41 544 1.31 0.8 0.19 5

Average of all flights 2740 m uw 18 0.69 392 2.9 6

LWC < 1.0 DW 19 0.73 469 1.20 1.4 0.48 5
Average of all except ;;. 2740 m uw l3 1.40 551 4.6 6
those from 247 at LWC;;. 1.0 DW 28 1.52 569 1.03 l.l 0.24 5
2130 m.
72
(1974), Dytch (1975), Braham (1976, 1977), Komp and
Auer (1978) and Shea and Auer (1978). The primary
purpose for many of these measurement programs was
to delineate the urban plume by ACN concentrations.
These measurements, however, were also quite useful
for determination of urban affected clouds as opposed
to natural clouds. This method was used by Dytch
( 1975) in his study of cloud microphysics. The aircraft
penetrations into numerous cloud elements by Dytch
(1975) revealed an average ACN concentration of 9.32
x I ()3 cm- 3 • These cloud penetrations were further
stratified into rural unaffected, downwind affected,
and clouds located directly over the urban-industrial
area. The average ACN concentration for rural clouds
was 4.66 x 10·1 cm- 3 while the urban-related clouds
was 13.98 x 10'1 cm- 3 or a factor of three larger.
In the Auer (1976) study of the industrial cumulus
cloud, the concentration of ACN in the updraft core
was greater than 2 x 104 cm- 3 throughout the depth
of the cloud. This value was, again, close to a factor
of three greater than the surrounding lower boundary
layer concentration at approximately the altitude of
penetrations. Semonin (1978) reported median values
of ACN in these two types of clouds of 3. 75 x I ()3
cm- 3 for 53 affected clouds, and 1.5 x 1()3 cm-3 for 74
unaffected by the urban boundary layer. While these
values are somewhat less than the factor of three re-
ported by Auer and Dytch, they are in reasonable
agreement. Auer (1974) also reported ACN in-cloud
to out-cloud concentration ratios between about 5 and
35 from repeated penetrations at 2100 m of a cumulus
congestus observed in the urban plume. The ambient
concentrations were between 2 x 102 and 8 x 102
cm- 3 , and inside the cloud the range extended from
3. 7 x 1()3 to 9 x 1()3 cm- 3 • The low ambient values
were attributed to the passage of a weak maritime cold
front 15 hours prior to the flight and the vertical mixing
of the surface sources was limited by a stable, low-
level layer. The high in-cloud concentrations, how-
ever, were indicative of the ingestion of air into the
cloud from below the top of the mixing layer observed
at 500 m.
4.4.3 LIQUID WATER CONTENT (LWC)
The relationship between L WC and other cloud en-
vironment and microphysical characteristics is very
complex. The entrainment into a convective cloud, the
stage of its development, and the available water va-
por for condensation, among others, will influence the
measured LWC. Nonetheless, the liquid water content
for clouds both upwind and downwind of the urban-
industrial area were calculated from droplet spectra
and reported by Carrera (1975), Dytch (1975), Auer
and Dirks (1973b), and Auer (1976). Semonin (1978)
reported a number of measurements obtained with a
METEOROLOGICAL MONOGRAPHS
Johnson-Williams hot-wire LWC meter. The measure-
ments are described without supporting observations
of the many factors which may affect the magnitude
ofthe LWC.
Auer and Dirks (l973b) reported a limited number
of measurements obtained near cloud base for upwind
and downwind cumulus clouds. The upwind measure-
ments consisted of only a single cloud because of tech-
nical difficulties encountered in upwind sampling. The
upwind cloud LWC was 0.75 g m- 3 and the median
L WC value of seven clouds observed downwind was
0.40 g m- 3 • The range of values for the downwind
clouds extended from 0.26 to 0.95 g m- 3 • It was un-
fortunate that additional upwind cloud measurements
were not obtained on these same days.
Dytch (1974) showed the LWC from five cities, ten
downwind, and nine upwind clouds with a total of 36
cumulus penetrations on 5 August 1973. A grouping
of the city and downwind clouds showed a median
LWC of0.64 g m- 3 , and for upwind clouds 0.21 g m- 3
as shown in Table 7. The range of L WC values for the
downwind-city clouds was between 0.38 and 1.0 g m- 3
and the range for the upwind clouds was 0.17 to 0.66
g m- 3 . These observations at mid-cloud levels indicate
a three-fold increase in the median value for the down-
wind clouds. Most of the observations reported by
Dytch (1974) were obtained from penetrations at an
altitude of~ 40% of the total cloud depth. The typical
cloud base was 1.5 km with the cloud tops at 3 km and
the average penetration level was 2.1 km MSL.
With one exception, Carrera (1975) showed a similar
difference between upwind and downwind cloud L WC.
The downwind measurements consistently were greater
than the upwind although not by as great a factor as
those in the mid-portion of the clouds reported by
Dytch (1974). The range of upwind LWC measure-
ments extended between 0.43 and 1.59 g m- 3 and the
range for the downwind clouds was between 0.69 and
1.85 g m- 3 • The measurements of Carrera were all ob-
tained near the cloud tops.
In a summary report of many flights, Dytch (1977)
presented L WC obtained by integrating drop spectra
in upwind and downwind clouds. The calculated LWC
values were not preferentially greater in the downwind
clouds. These clouds were stratified into affected and
TABLE 7. The cloud liquid water content (Johnson-Williams) mea-
sured at mid-cloud level in upwind and downwint: clouds. After
Dytch (1974).
Downwind versus upwind L WC of clouds (g m- 3 )
Downwind Upwind
Median Range Median Range
0.64 0.3S-I.OO 0.21 0.17-0.66
METROMEX: A REVIEW AND SUMMARY
unaffected primarily by noting the Aitken condensa-
tion nuclei concentrations whether at cloud base or by
penetration leaving little doubt as to their classifica-
tion. The importance of the opposing results between
observed and calculated L We for describing upwind
and downwind cloud microphysical characteristics is
not known.
Semonin (1978) reported median values for L we
measurements obtained in 74 unaffected clouds and 53
urban perturbed clouds with a median value for up-
wind clouds of0.98 g m- 3 and 1.10 g m- 3 for downwind
clouds. Semonin also combined a number of L we
measurements of natural clouds and normalized the
penetration level as a percent of cloud depth. The re-
sults showed the variation with height of L we in these
small active cumulus clouds. Semonin (1978) indicated
that at an altitude corresponding to about 113 of the
cloud depth, the LWe is approximately 0.6 g m- 3 with
a general increase upward with maximum values near
cloud top of 1.61 g m- 3 •
The refinery cloud documented by Auer (1976) con-
tained lesser L we by comparison with the natural or
urban affected clouds observed elsewhere during
METROMEX. The values reported by Auer near
cloud base were 0.05 g m- 3 increasing with cloud depth
to 0.44 g m- 3 near cloud top.
REFERENCES
Auer, A. H., Jr., 1974: Cumulus congestus growth downwind of St. - - , 1978: Giant nuclei and warm rain initiation. Preprints Conf
Louis, Missouri: Case study. J. Wea. Mod., 6, 229-237.
- - , 1976: Observation of an industrial cumulus. J. Appl. Meteor.,
15, 406-413.
- - , 1981: Urban boundary layer (Chapter 3 this volume).
- - , and R. A. Dirks, l973a: 1973 Annual Report to the NSF for
Modification of Convective Cloud Activity by an Urban Area. Lawson, R. P., and A. H. Auer, Jr., 1976: Urban induced influences
University of Wyoming, Laramie, 78 pp.
- - , and--, l973b: Interim Progress Report Project METRO-
MEX to EPA. University of Wyoming, Laramie, 68 pp.
Beebe, R. C., and G. M. Morgan, Jr., 1972: Synoptic analysis of
summer rainfall periods exhibiting urban effects. Preprints
Conf Urban Environment and Second Conference on Bio-
meteorology, Philadelphia, Amer. Meteor. Soc., 173-176.
Boatman, J. F., 1974: Inadvertent thunderstorm modification by an
urban area. M.S. thesis, Dept. Atmos. Sci., University of Wy- - - , 1978: Cloud characteristics. Summary of METROMEX, Vol.
oming, Laramie, 129 pp.
Braham, R. R., 1974: Cloud physics of urban weather modification:
A preliminary report. Bull. Amer. Meteor. Soc., 55, 100-106.
- - , 1976: The University of Chicago in Project METROMEX.
Tech. Note No. 48, Cloud Physics Laboratory, University of Shea, D. M., and A. H. Auer, Jr., 1978: Thermodynamic properties
Chicago, 25 pp. [NTIS PB26107l/AS].
- - , 1977: Overview of urban climate. Proc. Conf Metropolitan
Physical Environment, 25-29 August 1975, Syracuse, Forest Spyers-Duran, P., 1972: Upwind and downwind cloud-base micro-
Service Tech. Rep. NE-25, Upper Darby, PA, 3-17.
Carrera, N. J., 1975: Urban influences on cloud microstructure
downwind from St. Louis, Missouri: Upper level observations. Vogel, J. L., 1977: Synoptic weather relations. Summary ofMET-
J. Rech. Atmos., 9, 157-165.
Changnon, S. A., Jr., and R. G. Semonin, Eds., 1975: Studies of
73
selected precipitation cases from METROMEX. Illinois State
Water Survey Report of Investigation No. 81, Urbana, 329 pp.
Dytch, H., 1974: An investigation of urban influence on cloud mi-
crostructure several thousand feet above cloud base. Preprints
Fourth Conf Weather Modification, Ft. Lauderdale, Amer.
Meteor. Soc., 386-389.
- - , 1975: Urban influences on cloud microstructure downwind
from St. Louis, Missouri: Lower level observations. J. Rech.
Atmos., 9, 145-156.
- - , 1977: Urban influences on cloud microstructure downwind
from St. Louis. Tech. Note No. 51, Cloud Physics Laboratory,
University of Chicago, 120 pp.
Fitzgerald, J., and P. Spyers-Duran, 1973: Changes in cloud nucleus
concentration and cloud droplet size distribution associated
with pollution from St. Louis. J. Appl. Meteor., 12,511-516.
Grosh, R. C., l978a: Studies of anomalous cumulus clouds. Sum-
mary of METROMEX, Vol. 2, Causes of Precipitation Anom-
alies, Illinois State Water Survey Bull. 63, Urbana, 212-228.
- - , l978b: Satellite-observed urban cloud distributions. Summary
of METROMEX, Vol. 2, Causes of Precipitation Anomalies,
Illinois State Water Survey Bull. 63, Urbana, 229-235.
--,and R. G. Semonin, 1973: Moisture budgets and wind fields
of thunderstorms passing over an urban area in the Midwest.
Preprints Eighth Conf Severe Local Storms, Denver, Amer.
Meteor. Soc., 130-137.
Henderson, T. J., and D. W. Duckering, 1973: Final Report MET-
ROMEX 1973: A summary of operations conducted by Atmo-
spherics Incorporated during the period 7 July through 19 Au-
gust 1973, 51 pp.
- - , and--, 1974: Final Report METROMEX 1974: A summary
of operations by Atmospherics Incorporated during the period
7 July through 17 August 1974, 48 pp.
- - , and - - , 1975: METROMEX 1975: A summary report.
Atmospherics Incorporated, Fresno, 53 pp.
Johnson, D. B., 1976: Ultragiant urban aerosol particles. Science,
194, 941-942.
Cloud Physics and Atmospheric Electricity, Issaquah, Amer.
Meteor. Soc., 31-34.
Komp, M. J., and A. H. Auer, Jr., 1978: Assessment of some aero-
sol properties in Midwest power plant plumes. J. Wea. Mod.,
10, 107-118.
on convective cloud activity. J. Rech. Atmos., 10, No. 3.,
143-156.
Morris, T. R., and C. W. Ulbrich, 1973: Radar observations of fire-
induced rain clouds. J. Appl. Meteor., 12, 551-553.
Semonin, R. G., 1974: Urban-induced weather modification. 1974
Earth Environment and Resources Conf Digest of Technical
Papers, Institute of Electrical and Electronics Engineers, Phil-
adelphia, 40-41.
2, Causes of Precipitation Anomalies, Illinois State Water Sur-
vey Bull. 63, Urbana, 236-239.
- , and S. A. Changnon, Jr., 1974: METROMEX: Summary of
1971-1972 results. Bull. Amer. Meteor. Soc., 55, 95-100.
and aerosol patterns in the plume downwind of St. Louis. J.
Appl. Meteor., 17, 689-698.
structure. Tech. Note No. 45, Cloud Physics Laboratory,
University of Chicago, 53-71. [NTIS No. PB 218678].
ROMEX. Vol. l, Weather Anomalies and Impacts, Illinois
State Water Survey Bull. 62, Urbana, 85-112.
 CHAPTER 5

Urban Precipitation Processes

RoscoE R. BRAHAM, JR.

CONTENTS

ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.2 Diurnal rainfall variations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.3 Synoptic weather relations..... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.3.1 STRATIFICATION BY SYNOPTIC WEATHER CONDITIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.3.2 STRATIFICATION BY SURFACE WIND AND STORM MOVEMENT . . . . . . . . . . . . . . . . . . . . . 82
5.4 Studies of precipitation initiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.4.1 OBSERVATIONAL STUDIES OF FIRST ECHOES..................................... 85
5.4.2 FIRST ECHO LOCATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.4.3 STATISTICAL COMPARISONS OF URBAN-RURAL FE FREQUENCIES................... 88
5.4.4 URBAN-RURAL COMPARISONS OF FIRST ECHO CHARACTERISTICS . . . . . . . . . . . . . . . . . . . 90
5.4.5 INITIATION OF RADAR ECHOES AT GROUND LEVEL............................... 92
5.4.6 NUMERICAL MODELING OF FIRST ECHOES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.5 Life history of radar echoes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.5.1 DATA AND ANALYSIS TECHNIQUES....................................... ...... 93
5.5.2 URBAN AND NON-URBAN ECHO CHARACTERISTICS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.5.3 EFFECTS OF ECHO MERGERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.6 Analyses of surface raincells............................................ ............. 96
5.6.1 THE RAINCELL APPROACH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.6.2 DESIGNATION OF URBAN-AFFECTED RAINCELLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.6.3 TOTAL RAIN PRODUCED BY RAINCELLS . . . . . . . . . . . . . . . . . . . . . . • . . • . . . . . . . . . . . . . . . 97
5.6.4 RAINCELL INITIATIONS AND MERGERS.......................................... 98
5.6.5 INTERPRETATION OF RAINCELL FINDINGS....................................... 99
5.6.6 URBAN EFFECTS ON HAILSTREAKS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.7 Effects of urban area on convective echo heights ....................................... 101
5.7.1 MAXIMUM HEIGHTS REACHED BY INDIVIDUAL ECHOES . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.7.2 FREQUENCIES OF TALLEST ECHOES OVER URBAN AREAS . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.7.3 LOCATION OF THE TALLEST ECHO OF EACH DAY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.7.4 URBAN EFFECTS UPON HI-CU LOCATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.7.5 TIME OF DAY EFFECTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.7.6 HEIGHT DISTRIBUTIONS OF URBAN AND RURAL HI-CU . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.7.7 NUMERICAL MODELING OF MAXIMUM CLOUD HEIGHTS . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.8 Physiographic effects .................................................... ........... 105
5.8.1 THE BOTTOMLAND AREA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.8.2 THE SOUTHWEST HILL AREA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
5.8.3 THE BLUFF LINE 106
75
76 METEOROLOGICAL MONOGRAPHS

5.8.4 THE KASKASKIA RIVER VALLEY ............................................... 108

5.8.5 MODEL CALCULATIONS ....................................................... 108
5.9 Postulated mechanisms for explaining urban precipitation effects . . . . . . . . . . . . . . . . . . . . . . . . . .
108
5.9.1 BOUNDARY LAYER DYNAMICS MECHANISM ..................................... 108
5.9.2 WHAT IS THE EFFECT OF LOW 8e? .............................................
109
5.9.3 THE GLACIATION MECHANISM .................................................
113
5.9.4 WHAT IS THE ROLE OF LOCAL TOPOGRAPHY? ...................................113
REFERENCES ...................................................................... 114

ABSTRACT

Precipitation processes in convective clouds in and around St. Louis were studied by radar, numerical
modeling and surface rainfall data.
Statistically significant differences were found between radar first echoes which developed over the urban
and non-urban areas. Urban first echoes occurred with over twice the frequency (area normalized) and had
lower bases and tops. The area of maximum first-echo frequencies was centered over downtown St. Louis and
extended north and northeast along the Mississippi River. Modeling studies showed that the urban clouds which
gave rise to the first echoes must have had lower cloud-base updraft speeds; a finding which is compatible with
measurements showing that the daytime urban boundary layer was drier than that of nearby rural areas. Based
upon studies of echoes which could be followed through their lifetimes, it is concluded that urban convective
clouds, on average, reach greater maximum heights, last longer, and are more likely to merge with other clouds.
The frequency distribution of heights of the tallest echo in the non-urban area, at each half-hour, was found
to be bimodal with peak frequencies near 6 and 12 km and a frequency minimum near 9 km. In contrast, urban
echoes gave a unimodal distribution with higher-than-rural frequencies at all levels from 5 to 13 km. Evidently,
many urban clouds were able to penetrate a mid-level tropospheric arresting level, which often limited the
growth of non-urban clouds.
Simulations with boundary layer numerical models and comparisons of rainfall amounts received over dif-
ferent areas were used to estimate possible effects of local topography. These studies led to the conclusion that
topography probably does play a role, but definitely is not the major cause, in determining St. Louis area rainfall
patterns.
The possibility that urban clouds may be envigorated through glaciation, and thereby be able to penetrate the
atmospheric condition which frequently restricted rural clouds to 6 km top heights (- -10°C) was examined.
Measurements appear to rule out the possibility that St. Louis was a strong source for ice-forming nuclei.
However, measured cloud microphysical characteristics give rise to the possibility of increased glaciation
through the Hallett-Mossop ice mechanism.
It i~ concluded that the rainfall maximum which was observed to occur over the downwind edge of St. Louis,
during late afternoon hours, resulted primarily from modification of boundary layer dynamics from surface
thermal and frictional forcing. Even though the urban heat island normally had its minimum value during these
hours, the decreased stability of the lowest air layers, associated with afternoon heating, allows even a small
heat anomaly to have a considerable effect on boundary layer motions.
The cause of a secondary rainfall maximum, which was found in the Alton-Edwardsville, Illinois, area around
midnight, was not established.

5.1 Introduction
In this chapter we bring together the findings of a
wide variety of METROMEX studies relating to urban
effects upon precipitation. We begin by examining the
patterns of surface rainfall as a function of time of day,
synoptic weather conditions, and various combina-
tions of surface wind directions and storm motion di-
rections. This is followed in order by studies of pre-
cipitation initiation as revealed by radar, comparative
studies of the complete life histories of urban and rural
echoes, and a study of properties of individual areas
of rain (raincells). Next we summarize an urban-rural
comparison of the tallest radar echoes present at var-
ious times of the day, and discuss various possible
physiographic influences on rainfall at St. Louis. This
chapter closes with a discussion of mechanisms pos-
tulated to account for observed urban precipitation
effects.
5.2 Diurnal rainfall variations
Huff (1977) may have been the first to point out that
two basically different storm types are suggested by
the diurnal variations of rainfall in the St. Louis area.
Fig. 1 shows rainfall amounts, expressed as overlap-
ping three-hour averages, for the raingage network as
a whole and for several areas within it. In most areas
METROMEX: A REVIEW AND SUMMARY
of the network the least rain occurred during the pe-
riod from about 0300 to 1200 CDT. The afternoon and
evening hours show two rainfall maxima, one centered
at about 1700 CDT and the other at about midnight.
The peak at 1700 CDT is identified with convective
clouds resulting from surface heating and low-level in-
stability. Much of our attention in this chapter will be
centered on the storms that gave rise to this peak and
the ways in which modification of the urban boundary
layer may have affected these storms.
The origin of the rainfall peak occurring near mid-
night is less obvious. It could be associated with the
so-called "nocturnal thunderstorm" phenomenon, but
most of the evidence is to the contrary. Nocturnal
thunderstorms are caused by traveling, regional-scale,
dynamical disturbances that originate in the High
Plains during the morning and early afternoon, travel
across the Great Plains, and reach Illinois during the
hours from about midnight to dawn. St. Louis is usu-
ally regarded as being near the southeastern edge of
the region of high frequencies of nocturnal thunder-
storms. Studies of nocturnal thunderstorms in Illinois
[e.g., see Means (1944), Changnon (1968) or Wallace
(1975)] place the time of maximum occurrence some-
j
;':" b.
;:'; 25
~
::; 20
>- St. Louis Urban
t5
~
c.
15
5~~~~~-L-L~~~~
24 12
FIG. 1. Variation in summer season rainfall amounts as a function of time of
day for the METROMEX network and five subareas within it. Amounts are
expressed as 3 h averages, centered on the times indicated, and represent percent
of total rain in the subarea. Based upon five METROMEX seasons. After Huff
(1977).
77
where between 0200 and 0800 local time. The MET-
ROMEX-area diurnal rainfall distributions (Fig. 2 and
Fig. 18 of Chapter 2) and the diurnal thunderstorm
frequency curves (Fig. 27, Chapter 2) clearly show
minor maxima during the early morning hours. Thus,
the rainfall maximum near midnight appears not to be
due to the normal nocturnal thunderstorm phenome-
non.
On the other hand, the clear separation between the
afternoon and late evening rainfall peaks in the diurnal
records, in all areas except the northern part of the
network, suggests that two distinctly different types
of rainstorms are represented in the METROMEX
data. To learn more about these storms we turn to Fig.
I which gives the diurnal rain variations in the raingage
network and five subdivisions of it.
In rural areas to the south of St. Louis, Fig. lc, the
pre-noon rain is ~ 10% per 3 h period, the afternoon
convective peak gets up to at least 20%, and the noc-
turnal rain peak is ~15-17%.
For the city of St. Louis and the industrial areas of
East St. Louis and Granite City, Fig. lb, rainfall cen-
tered at 1700 CDT is very much enhanced relative to
that occurring during the 0300-0800 CDT period
18 24 12 18 24
TIME OF OAY (COT)
78
(28-30% compared to -10%). But, in common with
areas to the south, the nocturnal peak is still-15-17%.
We also note a marked decrease in the amount of rain
over the city and east-side industrial areas-down to
a minimum of - 5%-during the hours from
-0700-1400 CDT.
Roughly east of the east-side industrial areas are the
residential areas of Belleville and Collinsville. Fig. 1a
shows that in this area the late-afternoon peak is still
very intense (- 25%) and has broadened into the time
period from 1300 to 1900 CDT. The nocturnal rain
peak remains - 17% and there is a pronounced late-
morning minimum dropping as low as 6%.
In the Edwardsville area (Fig. 1d), the afternoon and
nocturnal rain peaks are of almost equal magnitude,
23 vs. 21%. The afternoon peak is broadened in time
similar to that for Collinsville-Belleville. The morning
rain shows the early morning plateau and the noontime
minimum.
Fig. 1e shows the diurnal rainfall curve for Alton
and the rural area north and east of Alton. For this
region we note that the nocturnal peak completely
dominates the pattern, reaching a peak value of 23%.
At best, the late-afternoon peak is - 15%, and we do
not find the mid-morning "plateau" that characterizes
the city and rural areas to the south. According to
Huff ( 1978), the Alton curve also characterizes the rain
received on the bottomland area.
Before applying these findings to the urban rainfall
issues, it is useful to recall the areal patterns of rainfall
for different time periods. In Fig. 19 of Chapter 2, it
can be seen that the city of St. Louis and its suburbs
to the west and east were characterized by a relative
minimum of rainfall during the early afternoon. But
during the period from 1500-1800 CDT, this area re-
ceived above-average rainfall, and we note a strong
maximum in the Granite City, East St. Louis and Col-
linsville area. In Fig. 2, which maps the percent of rain
that occurred at each station during the hours of
1500-1800 CDT, we find a strong maximum centered
squarely over the urban area. These data clearly show
enhancement of rain over the urban area during this
time period.
Fig. 19c, of Chapter 2, for the period 1800-2100
CDT, seems to represent the transition between pre-
dominantly afternoon and nocturnal storm types.
Rainfall in the city and its western suburbs was about
equal to that in rural areas to the west and south. The
maximum east of the Mississippi River had moved far-
ther east and heavy rain- up to almost twice map av-
erage- appeared across the northern parts of the net-
work.
These same trends continued into the 2100-2400
CDT period (Fig. 19d of Chapter 2); the city and rural
areas west of it were below map average, while the
METEOROLOGICAL MONOGRAPHS
• 10
STAf UTr t. JLLS
FIG. 2. Percent of total summer rainfall amounts that fell between
1500 and 1800 CDT. Based upon 1971-75 METROMEX data. After
Huff (1977).
eastern rim of the network, the bottomlands, and the
areas northeast of the city (Edwardsville and beyond)
were well above map averages.
These data amply justify Huff's statement about
two different storm types in the METROMEX data.
They also suggest an urban-related decrease in rainfall
over the urban area during midday, and a strong in-
crease in the amount of rain during the late afternoon.
These data also suggest that the very large rainfall
totals measured in the Edwardsville area represent the
combination of nocturnal storms and urban-enhanced
afternoon storms.
5.3 Synoptic weather relations
A major objective of METROMEX was to deter-
mine the atmospheric conditions in which the St.
Louis urban-industrial complex affects precipitation
processes and surface rainfall. To help achieve this
objective, the synoptic-scale and mesoscale weather
conditions that existed during periods of rain over the
St. Louis raingage network were studied by scientists
at the Illinois State Water Survey (Vogel, 1977).
During METROMEX, rain occurred over this net-
work on 228 days in association with major synoptic
weather types. These were:
• Squall line storms: A nonfrontal group of thun-
derstorms accompanied by a trigger mechanism,
usually a shortwave trough. The convective ac-
tivity associated with the storm systems was in-
tense, well-organized, and often times was ar-
rayed in a narrow band or line of active thun-
derstorms.
• Squall zone storms: A mesoscale system of
METROMEX: A REVIEW AND SUMMARY 79

TABLE I. Selected properties of objective storms associated with various synoptic weather conditions. After Vogel (1977).

Average Average rain

rain per Contribution in gages Percent of
Number of storm to total rain with rain network
storms (em) (em) (em) covered

Squall line (SL) 48 1.19 57.12 1.57 75.7

Squall zone (SZ) 83 0.34 28.22 0.64 52.4
Cold front (CF) 46 0.30 13.8 0.9~ 32.6
Warm front (WF) 14 0.16 2.24 0.41 38.1
Stationary front (SF) 19 0.39 7.41 1.09 35.7
Pre-front and post-front (P & PoF) 25 O.o? 1.75 0.33 19.9
Air mass (AM) 90 O.o2 1.80 0.28 6.9
Low centers (LOW) 5 0.13 0.65 0.51 25.2
Five-season total 330 112.99

thunderstorms organized into an area or cluster miles and/or one hour. A total of 330 objective storms
and independent of a frontal zone. These storms, were selected from the 228 days with rain on the net-
like squall lines, tended to move across large re- work. Objective storms encompassed about 94% of
gions of the Midwest, and an upper air impulse total network rain during METROMEX. Selected
was usually discernible. properties of the objective storms, averaged by syn-
• Frontal storms: Precipitation formed within 120 optic type, are given in Tables 1 and 2. Readers should
km (75 mi) of a surface front (cold, static or consult Vogel (1977) for further details.
warm). There was no synoptic evidence that this Table 1 shows that air mass and squall zone storms
precipitation was associated with a squall line or accounted for over half the number, but only one-
squall zone which, on occasion, moved 40 km (25 quarter of the rainfall from objective storms. Squall
mi) or more ahead of the fronts. line storms produced the heaviest rains, covered the
• Pre-frontal and post-frontal storms: Precipitation largest fractions of the network, and accounted for
associated with a frontal structure but at a dis- over half of the total rainfall.
tance of 120-240 km (75-150 mi) ahead or behind Storms associated with cold fronts, squall lines,
a front (cold, static or warm). squall zones and air mass conditions present an inter-
• Air mass storms: A shower or thunderstorm esting sequence of convective storms with different
generated within an unstable air mass. No large amounts of dynamical assistance from synoptic-scale
scale or mesoscale synoptic causes were evident. phenomena. Cold front storms are directly connected
The resulting convective activity was usually with a large-scale weather situation. Squall lines usu-
widely scattered to scattered and weak. ally develop ahead of cold fronts in association with
• Low pressure storms: A cyclonic storm situated a short wave trough. Squall zone storms usually were
so close to the research circle that it was not pos- associated with some upper-air impulse, but much of
sible to associate the precipitation with a frontal the storm dynamics must have been controlled by lo-
or mesoscale weather structure. These systems
TABLE 2. Frequencies of direction from which rainstorms of var-
are rare during the summer months. ious synoptic types were observed to move, 1971-75. After Vogel
(1977).
Separation of rain periods into these various storm
types was based upon hourly sequence data, surface Number of storms
and upper air maps, and radiosonde data from the SL SZ CF WF SF P&PoF AM LOW Total
National Weather Service; radiosonde and surface ssw 5 3 1 4 2 9 24
weather stations maintained by the Illinois State Water sw 9 23 3 3 3 I 21 I 64
WSW 15 24 11 1 6 8 11 1 77
Survey; and the ISWS radar and raingage network. WNW 7 14 12 I 5 4 9 1 53
NW 9 8 8 2 3 5 18 2 55
NNW 0 3 7 0 0 3 6 0 19
5.3.1 STRATIFICATION BY SYNOPTIC WEATHER NNE 2 I 1 4
CONDITIONS NE
ENE
ESE 1 I
The analysis unit used by Vogel for studying the SE 1 I 3
characteristics of rain from individual storms was the SSE 2 I 4
so-called "objective storm." This was defined as a Static 1 1 1 3
In deter- 2 2 3 3 11 23
discrete period of rain in the network which could be minate
identified with one of the synoptic weather classes and Total 48 83 46 14 19 25 90 5 330
which was separated from other rain periods by 20
80
cal conditions. Isolated air mass storms depend en-
tirely upon local air mass stability, moisture and wind
conditions. This sequence then offers a possibility of
detecting the role of urban areas in affecting convec-
tion, and thereby inducing changes in local precipita-
tion.
Average patterns of network rain for these four
weather types were published by Vogel (1977, Figs .
3-6). Use was made of the observed directions of
storm motions which were distributed as given in Ta-
ble 2. One notes that summer storms over St. Louis
move predominantly from SSW through NNW. In
each of these figures, data are given for all storms of
the type and for the two classes of storm movement
which made up the majority of the storm sample. In
discussing these charts, Vogel made reference to the
St. Louis and Alton urban-industrial areas which are
shown by hatched boundaries. In these figures, areas
shown by vertical lines are those in which the rainfall
exceeds one standard deviation above map average.
Areas which received less than one standard deviation
a. All stonns
5TACVTE"''"L' c. Storms from NW
FIG. 3. Rainfall patterns for squall-line storms, 1971-75. After Vogel (1977).
METEOROLOGI CAL MONOGRAPHS
below map average are shaded. Sample sizes, means
and standard deviations are shown for each map.
Since squall-line storms accounted for over half of
the network total, it is not surprising that their rainfall
map (Fig. 3) corresponds closely to the total METRO-
MEX rainfall map (Fig. 15 of Chapter 2). The rela-
tively small number of these storms and their very
heavy average rain increases the risk that the rainfall
pattern was unduly influenced by the chance place-
ment of a few very heavy storms. However, analysis
of two independent, wind direction, subsets of these
storms both place the maximum east and northeast of
the city. This argues for a real effect in the location of
the maximum, but one that may be independent of the
direction of storm movement. The partitions involving
SW-NE moving squall line storms and NW-SE mov-
ing air mass storms are the only partitions providing
above-average rain amounts over downtown St. Louis
and its western suburbs. The cold front storms (Fig.
4) give support for a rain maximum located downwind
of the city in terms of storm movement. The south-
b. Storms from SW
N • 16
ean = 21. 7 em
s.o. = 6.5
 81
METROMEX: A REVIEW AND SUMMARY

N = 46
11ean = 14.0 em
s.o. • 4.2
a . All s t onns b. Stonns from SW

KII.OMFTFR'j.
N • 27
0
=
0
&
~
1•

10
11ean = 4.1 em
s 6
••ATv" M•Ll> c. Stonns from NW .o. • 1.

F rG. 4. Rainfall patterns for cold· fron t storms, 1971-75 . After Vogel ( 1977).

west-northeast moving cold front storms produced the located east and northeast of St. Louis was associated
larger rain amounts, with the result that they too con- primarily with squall line storms and SW-NW moving
tributed substantially to the rain maximum east and cold front storms. These storms gave the largest
northeast of St. Louis. storm-mean rain amounts and dominated total rainfall
Turning now to squall zone storms, we note in Fig. statistics. Squall zone and air mass storms gave less
5 that the all-storm data indicate a rainfall minimum well-defined rainfall patterns , but overall they seemed
over the city and a maximum on all sides from north- to show decreased amounts of rain over the urban
west through north, east, to south. In view of the elon- area. By combining data in Table 3 and Figs. 3-6, we
gated minimum aligned with the storm movement di- get the hint that the midnight peak in rain across the
rectly across the city, this pattern was apparently
controlled by the storm moving from the SW. The TABLE 3. Diurnal variations in the frequencies of METROMEX
rain storms associated with four synoptic weather types (Semonin
NW-SE moving squall zone storms gave a rain mini- private communication).
mum over the northwestern suburbs and a maximum
downwind of the city. Cold Warm Stationary Squall
fronts fronts fronts lines
The rainfall map for air mass storms, Fig. 6, shows
Time interval
a pattern of scattered minor highs and lows , presum- (COT) a b a b a b a b
ably because there was so little rain accumulation from
0001-0600 14 7 9 2 23 7 7 7
these storms. However, rainfall from this type of 0601-1200 18 4 21 4 21 13 7 7
storm seems to be slightly lower over the urban area 1201- 1800 21 6 9 1 19 II 20 20
1801-2400 23 16 9 1 22 9 14 14
than over rural areas.
Thus, on balance, stratification of rainfall by syn- a: Total number of events across raingage network.
optic storm type suggests that the rainfall maximum b: Number of events producing rain in network.
 82
a. All stonns
Klt..OMETERS
=
0 I U
0 ~ 10
STI\TUft: Mit.,~~
c. Stonns from NW
FIG. 5. Rainfall pattern for squall-zone torms, 197 1-75. After Vogel ( 1977) .
north edge of the network may be associated with cold
fronts . Similarly, the afternoon rainfall peak may be
due to squall-line storms moving from the southwest,
plus perhaps a minor contribution from air-mass
storms.
What do these patterns tell us about possible urban
precipitation mechanisms? Since the air mass and
squall zone storms derive their strength mainly from
low-level instability, and since we observed lower-
than-rural values of Oe over the urban area (see Chap-
ter 3), it is not surprising to find urban-induced rainfall
decrease in these storms. The picture is quite different
with squall line and cold front storms. These storms
are less dependent on low-level instability. Also, the
fact that these storms frequently have dynamical
structures favoring upper level divergence would give
them a mechanism for making use of low-level con-
vergence over the urban area to overcome the penalty
of lower Oe.
METEOROLOGICAL MONOGRAPHS
N • 50
Mean • 17 . 7 em
b. Sto nns f rom SW s.o. • 3. 8
N • 25
l·lean • 9 . 5 em
S. D. • 2 .1
5.3.2 STRATIF ICATION BY SU RFACE WIND AND
STORM MOVEMENT
Most of the first-order urban effects , such as the
heat island plume and particle effluents, initially are
transported by winds near the earth's surface. If these
materials affect rainfall, and if the direction of surface
winds differs from that of the storm motion, we should
expect rainfall effects to occur at different places in
response to different combinations of low-level winds
and storm movement.
Table 4 gives the percentage distribution of surface
rainfall for various directions of storm movement
within each of four stratifications of pre-storm surface
wind direction. We note that 45% of the rain was as-
sociated with surface winds from the southeast, 40%
with winds from the southwest; within the latter group
13% came with storms moving from the south-south-
west. From this table we deduce that about 81% of all
METROMEX: A REVIEW AND SUMMARY. 83

a . All stonns b. Stonns from SW

KI LOMETERS
o a u
0= ~ 10
f.tiLES
STATUTE

c. Stonns from NW
FIG . 6. Rainfall patterns for air-mass storms , 1971 - 75. After Vogel ( 1977).

rain came from storms moving from southwest through

TABLE 4. Percentage distribution of network rainfall stratified by northwest, while pre-storm surface winds were south-
surface wind direction and direction of storm movement, summers
of 1971-75. After Vogel (1977). east through southwest. Rainfall for these four com-
binations is shown in Fig. 7.
Percent of total rainfall Superimposed on these charts is a surface-wind-
SE SW NW NE
Storm surface surface surface surface direction arrow equivalent to 16 km length, with its
movement from winds winds winds winds origin at the Arch, at the tip of which is a rainfall-
ssw 13 I 0 0 frequency-weighted storm-motion arrow, also
sw 20 13 2 1 corresponding to 16 km length. If one assumes that
WSW 29 52 58 67 the urban agent affecting rain originated at the Arch
WNW 15 13 25 29
NW 20 9 9 2 and was carried 16 km by surface winds before enter-
NNW 2 I 6 < 0.5 ing a storm, and then was carried an additional 16 km
NNE 0 I <0.5 0
NE 0 0 0 0 by the storm, one would expect the maximum urban
ENE 0 0 0 0 effect to be located at the tip of the second arrow.
ESE < 0.5 0 0 0
SE I 0 0 I
Considering that the urban area is a large and dis-
SSE 0 <0.5 0 0 persed source, that the time and place of storm inges-
Indeterminate <0.5 <0.5 <0.5 < 0.5 tion of urban materials can only be conjectured, that
Total rainfall (em) 48.9 43.9 7.2 8.3 storms are of different sizes and move at different
Percent of all network speeds , and that the 16 km arrow lengths starting from
rainfall 45 40 7 8
the Arch represent a gross simplification, the occur-
 84 METEOROLOGI CAL MONOGRAPHS

a. Winds 135° , b. Winds 135°, storms 304°

c. Winds 225°, storms 248° d. Winds 225°, storms 299°

FIG. 7. Rainfall patterns associated with four combinations of pre-storm winds and movement: (a)
SE surface winds, SW-NE storm movement; (b) SE surface winds, NW-SE storm movement; (c)
SW surface winds, SW-NE storm movement ; (d) SW surface winds, NW-SE storm movement. After
Vogel and Huff (1977).

renee of major rainfall maxima in the general vicinity
of the predicted location is considered to be strong
evidence for urban effects upon rainfall. At the same
time we note major rainfall maxima located in regions
far removed from the tip of the storm-motion arrow.
This is particularly evident in Fig. 7a and 7d.
A somewhat simpler way to look at rainfall stratified
by wind direction is shown in Fig. 8. This analysis
ignores storm movement. Network rainfall was parti-
tioned according to the pre-storm surface wind direc-
tion. It is important to note that in every partition the
downwind quadrant rainfall is greater than that in the
upwind quadrant. The magnitude of the suggested ur-
ban effect from this approach is 11, 37, 45 and 12% for
surface winds of southeast, southwest, northwest and
northeast, respectively. Referring to the overall com-
posite pattern , the downwind quadrant rain exceeds
the upwind quadrant by - 23%. These patterns are re-
garded as strong evidence for urban effects on rainfall.
5.4 Studies of precipitation initiation
The locations and characteristics of radar first
echoes (FE) were used to study possible effects of St.
Louis on precipitation initiation in summer convective
clouds and to map the extent of such effects. These
studies were both observationally and theoretically
oriented. They involve two of the METROMEX co-
operating groups. This section will review the studies
of FE carried out under METROMEX support and put
them into the context of unintended modification of
precipitation by a major urban-industrial complex.
An FE is defined as the initial radar signal (above
a meaningful threshold) from a region of cloud suffi-
METROMEX: A REVIEW AND SUMMARY
FIG. 8. Upper four patterns show summer rainfall (em) in quad-
rants around St. Louis, each associated with one of four pre-rain
surface wind directions during 1971-75; the lower figure is a com-
posite of the total summer rainfall (1975-75) positioned with respect
to the pre-rain winds. After Changnon-(1979).
ciently far away from previously existing echoes to
preclude simple propagation of existing precipitation.
To be useful in studies of precipitation initiation the
threshold of received radar signal, in terms of equiv-
alent reflectivity, should be about 10 dBZ. Fig. 9
shows schematically the growth of a cumulus cloud
and the development of contours of equivalent reflec-
tivity, Z ei < Z e2·
RADAR RETURN INTERVAL-+t-- -.1
<..'-"'"~>"'
'
0 Typically, these conditions were found on days of air
"'
CLOUD
BASE
tl t2
TIME
FIG. 9. Schematic representation of first echo development.
85
In FE observational studies, a radar is programmed
to scan a certain volume of space, then to return to
the starting point and repeat. The elapsed time be-
tween successive "looks" at the same space volume
is known as the radar return interval. Again with ref-
erence to Fig. 9, imagine that a radar with a signal
threshold of Ze 1 scans the cloud at times t 1 and t 2. It
would not detect the growing cloud at t 1 , but at t 2 it
would detect an echo-the first echo-with its top and
base as indicated. A less sensitive radar, such as one
having a threshold of Ze 2 , would detect an echo at a
later time. The longer the return interval the greater
the depth of FEs and the more ambiguous are their
interpretations. For a discussion of the ways in which
radar parameters and FE properties combine to gov-
ern the ability of a radar to detect precipitation initi-
ation, see Braham and Dungey (1978).
5 .4.1 0BSERVATION AL STUDIES OF FIRST ECHOES
Radar first echoes were studied observationally by
scientists from the Illinois State Water Survey and the
University of Chicago, using radars and instrumented
airplanes. Both the Illinois State Water Survey and
the University of Chicago used 3-cm TPS-10 radars,
location of which are shown in Fig. 14 of Chapter 1.
Analysis of radar first-echo data from St. Louis was
reported by Braham et al. (1975), Braham and Dungey
(1978), Changnon (1974, 1976), Dungey (1977), Dun-
gey eta/. (1974), Johnson and Dungey (1978) and Ochs
and Johnson (1979). In the following summary ofthese
papers, attention is given first to the University of
Chicago results since they are based upon the larger
sample sizes and extend across four of the five project
summers.
The University of Chicago study involved 4553 FEs
measured with the TPS-10 radar located at Greenville,
Illinois. This radar was operated in a continuous 3-min
azimuth scan with a sensitivity of 3 dBZ at 16 km, and
20 dBZ at 97 km. First echo data reduction was limited
to range intervals of 16-97 km and to periods in which
all detected echoes in the radar field could be individ-
ually followed, and all new echoes identified through-
out the entire 29 300 km2 radar field. This allowed
meaningful comparisons of FE frequencies in different
areas. However, it tended to bias the FE studies to
periods of relatively scattered echoes and low fre-
quencies of more than one echo on any given azimuth.
mass and squall zone convection, and during onset
periods of more organized and widespread convection.
To maximize the number of days in the FE sample,
within constraints on analysis resources, data reduc-
tion for several days was terminated after about 100
FEs had been evaluated. This resulted in a sample of
86
TABLE 5. Greenville radar first echo data set: days (hours) and
number of first echoes. After Braham and Dungey (1978).
Year June July August
1972 3 (18) 222 6 (20) 495 10 (41)
1973 12 (47) 696 6 (21)
1974 I (I) 4 II (32) 466 9 (29)
1975 4 (10) 178 13 (45) 708 7 (19)
4553 FE from 283 echo-producing hours on 82 days
(see Table 5).
Before reporting on the results of the FE analyses,
it is important to consider the representativeness of
our FE data to the overall METROMEX sample. Cus-
tomarily the Greenville radar was operated continu-
ously from about 0900 until 1900 CDT, six days per
week. Therefore, most of the analyzed FEs occurred
between 1000 CDT and 1900 CDT, with a few as early
as 0300 CDT and as late as 2200 CDT. The Greenville
FE sample was distributed across synoptic weather
conditions as follows: Air mass conditions, 27 periods,
1129 FE; squall zones, 24 periods, 1510 FE; squall
lines, 11 periods, 1021 FE; cold fronts, 12 periods, 492
FE; all others, 8 periods, 401 FE. These compare very
favorably with the mix of synoptic conditions giving
rain in the METROMEX raingage network (Table 1).
We also were concerned whether the restriction on
numbers of echoes present during FE analysis periods
might bias the results away from situations giving im-
portant amounts of rain. Rain was recorded in the rain-
gage network during 65 of the 82 FE analysis periods,
in spite of the fact that the ratio of the two areas is
over 5:1 (29 300 km2 for the radar and 5200 km2 in the
raingage network). Taking 30 min as a reasonable time
for a cloud to deposit rain on the ground after precip-
itation initiation, we summed the total network rain
accumulated during the FE periods plus 30 minutes.
Results are given in Fig. lO where we note that the
pattern is very similar to that of the complete MET-
ROMEX rainfall map (Fig. 15 of Chapter 2), except
that the maximum in the Edwardsville area is less in-
tense, perhaps because the FE studies undersampled
the nocturnal rain periods. The network-average rain,
totaled over the 65 periods, was 13.43 em. This com-
pares with the total METROMEX rainfall of about 39
em accumulated between 1000 and 1900 CDT during
1972 through 1975 (periods equivalent to those used
in FE analysis).
On the basis of the distribution across synoptic
weather conditions and the pattern and amounts of
precipitation, we conclude that the FE study is rep-
resentative of conditions occurring during the after-
noon rainfall peak. There is no way of knowing to
what extent they are representative of the storms that
caused the nocturnal storms which contributed so
strongly to rain in the Edwardsville area.
METEOROLOGICAL MONOGRAPHS
585
304
647
248
0 5 10
STATUTE MILES
FIG. 10. Pattern of total rain in the METROMEX network during
periods of radar first echo analysis.
In the paper by Braham and Dungey (1978), anal-
yses are reported on the complete data set and three
subsets of it. Because of space restrictions in this re-
port, we consider only the analyses of the complete
data and of a 44-day restricted set having 1950 FE.
The latter data are from days which had observed
cloud and weather conditions such as to make it highly
likely, that the FE formed in convective clouds having
roots near the surface. This data set is regarded as the
one most sensitive to influences from the city.
5.4.2 FIRST ECHO LOCATIONS
The actual locations of the 4553 FEs are plotted in
Fig. 11. Fig. 12 is an analysis of these data. Map units
are numbers of FE per 100 mi2 (259 km2) observed
during the 283 hours of data. Fig. 13 gives first echo
densities observed in the 44-day restricted data test.
(The reader is cautioned not to confuse the circular
radar field shown on these maps with the circular rain-
gage network; see Fig. 14 of Chapter 1).
These three figures clearly indicate a region of above
average FE densities centered over downtown St.
Louis and extending north and northeast to encompass
much of the Alton area. An Analysis of FE densities
on days with winds < 3 m s- 1 (not shown here) indi-
cates a separate point maximum over Wood River,
near Alton, Illinois. Fig. 12 and 13 give a clear indi-
cation that St. Louis enhances the frequency of cen-
ters of rain initiation in clouds over it. First echo fre-
quencies in four categories of direction of echo
movement-closely related to wind direction-are
given in Fig. 14. These categories are based upon a
METROMEX: A REVIEW AND SUMMARY 87

FIG. II. Actual locations of 4533 first echoes. After Braham and FIG. 12. Analysis of first echo locations shown in Fig. I I. After
Dungey (1978). Braham and Dungey ( 1978).

mean echo-movement vector (EMV) obtained by the Greenville radar and the edge of its sampling area
tracking three or more widely separated small echoes. in the vicinity of St. Louis are shown in Fig. 15. In
In Fig. 14 we note a distinct tendency for the areas of addition, the FE frequency values in Fig. 15 have been
maxima to shift in keeping with wind direction. How- converted to the same units used in Fig. 12 and are
ever, it is also clear that something in addition to the shown in parentheses. We note that the values ob-
location of the city and wind direction had an influence tained for the Alton and St. Louis-Belleville areas and
in determining the patterns. for the bottomland are generally consistent between
A partitioning of weekday echoes versus holiday the two studies. The Hill area southwest of St. Louis
and weekend day echo frequencies (not shown here) has higher FE frequencies in Fig. 15 than would be
suggests that the urban effect came mainly on week- expected from Fig. 12 but still is lower than the urban
days. This is in keeping with earlier studies of week-
end-weekday variations in rainfall and visibility re-
ported by Summers (1966), Dettwiller (1968) and
Landsberg (1974). Huff and Changnon (1972) also re-
ported such a variation in the 1949-68 rainfall for the
St. Louis area. However, no such variations were
found in the METROMEX total rainfall (Huff, 1977)
or in the METROMEX raincell analysis (Schickedanz,
1978).
The Illinois State Water Survey TPS-10 radar, lo-
cated northwest of St. Louis, (see Fig. 14 of Chapter
1) also was used to study FE locations. Fig. 15, from
a study by Changnon (1978a), shows FE frequencies
for five different geographical areas, based upon a
sample of 811 PEs from 120 hours of observations on
19 days in 1973. Air mass, squall zone, warm front
and stationary front conditions contributed 385 of
these FE; 426 came from cold front and squall line
conditions. Although this sample is smaller than that
from the Greenville radar, it is useful in establishing
the FE frequencies west and south of St. Louis, i.e.,
beyond the limits ofthe other study. To facilitate com- FIG. 13. Analysis of locations of first echoes in 44-day data set.
parison with the Greenville FE data, the location of After Braham and Dungey (1978).
88 METEOROLOGICAL MONOGRAPHS

;.
L-·
(5)
(''\

0
1... )
/ ........
<..\ o· \ I !
'- \ ) ..
) • I I ("'\
~ 1, , '
\ \ I \ \
'
' I .,. < \
' ":.•/
\\
\
-
/
\

.... 0
c-~6~) .
___
-.._ ....... ~/
......
1..

FIG. 14. Maps of first echo densities in four wind direction partitions , Greenville sample : (a) EMV from 20r to 249" ; (b) EMV
from 250° to 269°; (c) EMV from 270° to 350°; and (d) EMV from 312° to 350°. After Braham and Dungey (1978).

areas. This indicates that the urban values based on
the Greenville radar data are indeed localized highs
and not just extensions of higher frequencies from
west and southwest of St. Louis. The FE frequencies
in the rural area east of St. Louis are substantially
below values found in the Greenville sample for the
same area.
5.4.3 STATISTICAL COMPARISONS OF URBAN-RURAL
FE FREQUENCIES
To permit statistical studies of the frequencies of
FE in various areas, the data from the Greenville radar
were partitioned into those that formed over St. Louis,
those that formed downwind of it, and those that
formed in the remainder of the radar field. The tech-
nique for making this partitioning is illustrated in Fig.
16. The metropolitan area was approximated by a sim-
ple compact boundary (labeled City) containing most
of the major local sources of atmospheric pollution as
determined by the Environmental Protection Agency
emissions inventory, the major areas of industrial,
commercial and residential land use as shown by Auer
(1978), the region of boundary layer air-flow modifi-
cation as shown by Ackerman (1978), the area covered
by the St. Louis nocturnal heat island (see Chapter 2),
and the area of doming of the boundary layer reported
by Spangler (1972), Wong and Dirks (1978) and others.
METROMEX: A REVIEW AND SUMMARY 89

•
Greenville Radar

RURAL
5. I
EDGE OF GREENVILLE 1351

~
\ 'O<:J

\ AVERAGE REG!o;;AL VALUES OF

FIRST ECHOES, EXPRESSED AS
NUMBER PER 100 SQUARE KILO-
METERS IN JULY- AUGUST . 1973

Fro. 15. First echo frequencies (Pere Marquette sample) in different land use
physiographic areas. After Changnon (1978a).

Having defined the boundary of the city for this
analysis , its downwind edge was allowed to advect
during each echo period according to the mean echo-
movement vector (EMV). In this way, additional areas
called 1 h, 2 h and 3 h downwind advection areas were
obtained. The city area remained 1522 km 2 on all days,
but the other areas could vary within a day, and from
day to day, depending upon EMV. (Within-day vari-
ation results when an area intersects the 16 km circle
around the radar or the edge of the radar field). A v-
erage values were: rural, 22 702 km2 ; 1 h, 1593 km 2 ;
2 h, 1399 km2 ; 3 h, 1261 km 2 • As suggested by Fig. 16,
the 1 h downwind area almost always lay between St.
Louis and Greenville, well beyond the edge of the
raingage network. Not infrequently, the 2 h area was
as far away as Greenville from St. Louis. The 3 h area
was frequently beyond Greenville.
Urban-rural differences in FE frequencies were
quantified by computing the frequency per unit area
for the city, 1 h, 2 h and 3 h areas relative to that of
the rural area. Summary results for the complete data
set and for the 44-day restricted data set are given in
Table 6. We note that the city and 1 h advection area
show FE frequencies about twice that of the rural
area. The likelihood that this suggested urban effect
might have resulted from sampling biases was checked
by applying the signed rank test to daily values for
each area relative to that of the rural area. Results of

0 Greenville
.&. Radar

16 JLAKE CARLY LE

KILOMETERS
0 20 40 60

FIG. 16. Location of the METROMEX radar sites, outlines of Greenville radar
data field and the dense raingage network, and an example of the analysis scheme
used for Greenville first echo and Hi-Cu data. Area labeled as City represents a
simplified outline of the St. Louis metropolitan area. The downwind advection
areas are based upon the data of 2 August 1974 which had an observed EMV of
240°, 12.7 m s-1 (see text for details). This day was selected as one having nearly
sample-average advection areas.
90 METEOROLOGICAL MONOGRAPHS

TABLE 6. Comparisons of first echo frequencies over the city and cloud formation over St. Louis (see Chapter 4). It is
three downwind areas. Results of signed rank tests (unit normal believed to reflect the fact that the boundary layer
deviates) are given in parentheses. After Braham and Dungey
(1978). destabilizes and deepens faster over urban areas than
over rural areas (Russell et al., 1974). The afternoon
Advective areas maximum corresponds to the time of maximum inter-
Sample City 1h 2h 3h action between the heat island and the unstable air
flowing over it (Shreffler, 1978). The data after 1700
Complete data set 1.94 1.84 l.ll l.ll
CDT are harder to interpret because of the two periods
4553 FE
(1700 and 1830 CDT) when the normalized urban-rural
44-day restricted set 2.28 2.14 l.OO 1.07
ratios were about zero. The buildup in the normalized
1953 FE (2.96) (2.12) ( -0.91) ( -0.74)
ratio after 1930 CDT could be related to the previously
discussed nocturnal rainfall peak, but to give the rea-
this test for the 44-day restricted data set are given in son for urban enhancement during this period would
parentheses as unit normal deviates. For the city and just be speculation.
1 h area we can reject the null hypothesis of equal
ranks at better than the 5% level. However, most of
5.4.4 URBAN-RURAL COMPARISONS OF FIRST ECHO
the statistical strength came from 1972, the year with
CHARACTERISTICS
the largest sample size and the fewest cases of zero
counts over the city. One of the other data partitions Average top and base heights of echoes initiating in
given in the original paper (Braham and Dungey, various parts of the urban and rural areas are given in
1978), viz., for EMV ~3 m s- 1 , showed a city effect Tables 7, 8 and 9. Table 7 gives values from the Chang-
of 1. 87 times rural with an associated unit normal de- non (1978a) study using the Pere Marquette radar. Ta-
viate of 1.96. ble 8 was derived from the Braham and Dungey (1978)
Based upon these studies we conclude that precip- study of FE from the Greenville radar. In these tables
itation initiation (FEs) occurs about twice as fre- differences between values for comparable areas are
quently over and within 1 h of wind travel immediately believed to be due mainly to differences in radar sen-
downwind of St. Louis as over nearby rural areas. In sitivity. In both samples urban FEs were found to have
a later section we consider whether it is due to en- base heights which were lower than those of rural FEs.
hanced echo development in urban clouds or to a In the case of the Greenville data these differences
greater frequency of clouds over the urban area. received strong statistical support. Urban-rural differ-
Urban to rural FE frequency rations, as a function ences in top heights are less clear-cut. In the Pere
of time of day, are shown in Fig. 17. This shows three Marquette sample urban and rural FE top heights were
periods of maximum urban enhancement of FE: 0930- almost identical; whereas in the Greenville data, the
1100 CDT, 1400-1700 CDT, and after about 1930. The urban FE tops averaged about 150m lower than rural
latter maximum was regarded as tentative because of FE tops but the rank test gave little statistical support.
small sample sizes. The enhanced late morning values Ochs and Johnson (1979) reanalyzed some of the
are in keeping with observations of the times of initial University of Chicago FE data to look at FE charac-
teristics in various subareas within the Braham-Dun-
SAMPLESIZES
gey (B-D) city area. For comparison with the urban
Z C+1 0 14 19 18 28 19 23 31 72 65 61 63 68 70 64 42 16 24 16 7
8 2+3 0 1 7 13 11 14 15 24 24 25 41 35 32 8 23 24 11 9 4 3 20 3 5 0
5 7 10 12 1
echoes Ochs and Johnson used only those rural FE
~ RURAL 8 17 67 97100117118146302351316227 49 58 33 26 43 27 14
291210204204157137
which were located at radar ranges within the interval
covered by the B-D city. They also excluded a buffer
zone north and south of the city. These two restric-
tions reduced the rural sample to 2061 FE. To reduce
the effects of day-to-day variability Ochs and Johnson
normalized urban values to average rural values on a
daily basis.
'---- I-l_ ___ , The principal findings of this reanalysis are sum-
I
I I
marized in Table 9. Line 1 in this table can be com-
I I I
10 11 12 13 14 15 16 17 18 19 20 21
pared directly with the city versus rural data in Table
TII~E,COT 8, and we note that the results are almost identical.
FIG. 17. Ratio of area-normalized frequencies of urban and rural Line 2 shows that normalizing urban values on the
first echoes as a function of time of day, based upon the total sample daily average rural values substantially increased the
from Greenville. Solid line represents FE over the city and 1 h wind
travel downwind. Dashed line compares echoes 2 and 3 hours of statistical support for lower-than-rural FE tops in the
wind travel downwind against rural. urban area. Lines 3 and 4 (Table 9) give values sepa-
METROMEX: A REVIEW AND SUMMARY 91

TABLE 7. Average heights (km) of first echo tops and bases in the July-August 1973 Pere Marquette sample. Stratified by degree of
organization of echoes and on land use-physiographic regions. After Changnon (l978a).

Hill Bottomland Urban Rural All

Organized systems*
Frequency of echoes 241 36 54 95 426
Tops 6.4 6.5 6.6 6.7 6.5
Bases 2.9 2.8 2.7 3.2 3.0
Vertical extent 3.5 3.7 3.9 3.5 3.5
Isolated echoes**
Frequency of echoes 113 16 75 181 385
Tops 4.8 4.7 4.9 4.8 4.9
Bases 2.1 2.1 1.7 2.2 2.1
Vertical extent 2.7 2.6 3.2 2.6 2.8
All echoes
Frequency of echoes 354 52 129 276 811
Tops 5.9 5.9 5.6 5.5 5.7
Bases 2.6 2.6 2.1 2.6 2.5
Vertical extent 3.3 3.3 3.5 2.9 3.2

*Cold front and squall lines.

**Air mass and squall zones.

TABLE 8. Summary of first echo characteristics (Greenville sample) in the areas indicated. Results of Student's test (unit normal
deviates) are given in parentheses. After Braham and Dungey (1978).

Base Base Top Top

Sample height temperature height temperature Thickness
Region number (km) ("C) (km) ("C) (km)

City 432 2.16 12.6 4.54 -1.9 2.41

(3.58) ( -4.67) (0.46) ( -1.04) (-3.31)
Ih 358 2.26 11.4 4.39 -1.5 2.10
(1.71) ( -1.34) (2. 76) ( -1.69) (2.25)
2h 192 2.35 11.4 4.51 -1.5 2.14
(0.40) ( -1.25) (1.05) ( -1.53) ( 1.17)
3h !58 2.23 11.9 4.57 -1.9 2.14
(2.68) ( -2.05) (0.29) ( -0.64) (1.19)
Rural 3413 2.38 10.8 4.60 -2.3 2.22
Single sample 4553 2.35 11.1 4.57 -2.2 2.22

TABLE 9. Urban-rural differences in daily average first echo characteristics and associated statistical support (unit normal deviates),
based upon the Wilcoxon two-sample rank test. After Ochs and Johnson (1979).

Urban minus rural values (km)

Urban FE
Urban area sample size Top height Base height Thickness

I. Braham-Dungey City, not normalized 428 -0.10 -0.27 0.17

on day ( -1.52) ( -4.12) (2. 73)
2. Braham-Dungey City, normalized on 412 -0.14 -0.23 0.09
day ( -3.02) ( -4.35) (1.29)
3. North one-third of Braham-Dungey 114 -0.36 -0.45 0.09
City, normalized by day ( -3.82) (-5.15) (0.86)
4. South two-thirds of Braham-Dungey 298 -0.06 -0.15 0.09
City, normalized by day ( -1.37) (-2.18) (1.58)
5. Area where cloud bases between 147 -0.09 -0.24 0.16
1.5-1.8 km, normalized by day ( -1.18) ( -2.84) (1.49)
6. Area where cloud bases above 1.8 39 0.33 0.06 0.28
km, normalized by day (1.25) (0.09) (1.31)
92 METEOROLOGICAL MONOGRAPHS

rately for the northern one-third and southern two-

thirds of the B-D city area. The northern area includes
some of the bottomland of the Missouri River and the
Alton urban-industrial area. We note that in this area
FE bases and tops were markedly lower than in the
rural area and in the entire city area. This may reflect
the greater availability of low-level moisture from the
rivers and refineries which are located in this northern
area. Lines 5 and 6 relate to two areas over St. Louis
where cloud bases were measured to be substantially
higher than rural bases on 19 days in 1973 (see Fig. 3,
Chapter 4). Ochs and Johnson assumed that the data
from these 19 days gave the general pattern of cloud
bases over the urban area. Although the numbers of
FE in these subareas of the B-D city area are quite
small, and almost all statistical strength has been lost,
there is a suggestion that FE tops and bases over the
central part of the city may actually be higher than for
rural FEs.

5.4.5 INITIATION OF RADAR ECHOES AT GROUND FIG. 18. Numbers of near-surface radar echoes initiating in 92
LEVEL km2 areas at various positions in the raingage network. Based upon
a sample of 1550 cell initiations from 35 storms during 1972-75.
Studies of convective radar echoes such as those After Huff (1978) .
that formed near St. Louis during METROMEX have
tion on an initial CCN spectrum rising in a closed par-
shown that most FEs develop measurable rain at
cel (Fitzgerald , 1972; Fitzgerald and Spyers-Duran,
ground level 10-15 min after the FE is first detected.
1973; Ochs, 1974). When initialized with CCN spectra
Therefore, it is appropriate to ask where, in the MET-
measured at St. Louis, these models helped to explain
ROMEX area, the radar showed the highest frequency
the higher concentrations and more narrow spectra of
of ground-level echo initiations. Data obtained from
drops measured in urban clouds compared with rural
the FPS-18, 10 em, Pere Marquette radar operated at
ones. The presence of narrower drop spectra was as-
near oo antenna elevation, and receiver sensitivity of
sumed to imply that precipitation should be delayed
approximately 30 dBZ, are shown in Fig. 18. The max-
imum frequencies of ground level rain echoes during in urban clouds. But this implication seemed to be in
1972-75 were found to be located over downtown St. conflict with METROMEX observations of increased
Louis and just east of the Mississippi River between frequencies of FEs and of reduced separations be-
East St. Louis and Alton. This location is very com- tween cloud bases and FE bases in urban clouds.
parable to the region of maximum FE formation shown Attempts to reconcile these conflicts centered on
on Figs. 12, 13 and 15. A secondary maximum was speculations that urban sources of giant condensation
found over the hill region southwest of the city of St. nuclei might offset the stabilizing influence of urban
Louis. sources of smaller nuclei. Johnson ( 1976) measured
the concentrations of aerosol particles in the size range
from 5-55 J.tm diameter upwind and downwind of the
5.4.6 NUMERICAL MODELING OF FIRST ECHOES
St. Louis urban area. Although there are significantly
One of the long-standing problems in cloud physics more of these particles downwind of the city, even the
has been to explain the rapid development of precip- upwind concentrations are large enough to be poten-
itation embryos in warm-based continental clouds. In tially important in FE formation (see Chapter 6 for
effect this amounts to explaining the formation of a further discussion of these aerosol measurements).
tail of large drops at the upper end of the cloud droplet METROMEX investigators have employed a num-
spectrum (see cloud microstructure discussion in ber of different numerical models of warm cloud mi-
Chapter 4). In clouds which develop in urban atmo- crostructure to investigate the initial development of
spheres the problem is magnified by the higher con- precipitation in urban influenced clouds (e.g., Ochs
centrations of smaller drops resulting from urban con- and Semonin, 1976, 1977, 1979; Dytch and Johnson,
centrations of cloud condensation nuclei (CCN). 1977; Johnson and Dungey, 1978; and Johnson, 1979).
Theoretical models available in the early days of A key aspect of this work was the discovery that it
METROMEX considered only growth by condensa- was necessary to extend the upper size limit of the
METROMEX: A REVIEW AND SUMMARY
initial aerosol distributions to larger sizes than had
ever been used by other investigators. Since large
(millimeter-sized) drops in concentrations of 10-4 to
10-s cm-3 can contribute significantly to the radar re-
flectivity levels present in a first echo (10--20 dBZ), the
initial aerosol distribution must be extended to com-
parable concentrations. For typical continental aero-
sol distributions this means extending the upper end
of the distribution to somewhere between 30 and 70
microns radius. Since the numerical models employed
must be able to accurately describe the evolution of
droplet distributions which typically range over more
than seven orders of magnitude, special attention must
be given to the numerical techniques employed (e.g.,
Ochs and Yao, 1978; Ochs, 1978). The extension of
the initial aerosol data to these larger sizes results in
significant reductions in the predicted time required to
develop detectable reflectivities, and explains the
rapid development of precipitation in seemingly stable
continental clouds (see Johnson, 1979; Ochs and
Semonin, 1979). The increased concentrations of giant
and ultragiant aerosol particles observed by Johnson
(1976) downwind of the city are adequate to offset the
increased stability resulting from higher total concen-
trations of cloud droplets, but are not adequate to pro-
duce significant destabilization relative to upwind
clouds. Reasonable variations in aerosol chemical
composition and partial solubility were found to have
negligible effect on the time required to produce pre-
cipitation (Ochs and Semonin, 1979).
On the basis of numerical modeling of microphysical
processes, we conclude that urban clouds are neither
more nor less stable than their rural counterparts. The
models indicate the importance of giant and ultragiant
CCN for accurate modeling of details of precipitation
initiation. But they do not support the hypothesis that
ingestion of urban pollutants would drastically affect
precipitation development at St. Louis. Therefore, we
conclude that the greater numbers of first echoes ob-
served over and downwind of the city apparently re-
flect the increased numbers of clouds found over the
city rather than any major change in the microphysical
development within them.
Many of the observed features of first echoes, such
as top and base heights and their rates of rise and fall,
can be simulated rather successfully with current
models. These models show that the height of the top
of an FE above cloud base depends primarily on the
speed of the updraft. Separation of an FE base from
cloud base depends both on the updraft speed and the
effect of sedimentation of the largest particles. Com-
bining these model results with the observations that,
on average, FE bases and tops were lower over the
urban area leads to the conclusion that cloud base up-
drafts over the urban area average slightly less than
93
over rural areas. The Johnson (1979) model predicts
this reduction to be about one-third of a meter per
second. Such a conclusion is in keeping with the ob-
servations of lower values Oe over the city. As dis-
cussed earlier, reanalysis of the FE data suggests that
there is a small area over downtown St. Louis and
nearby industrial areas where FE bases are above ru-
ral averages. It follows that in this small area cloud
base updrafts could also be above rural average. The
fact that the average first echo thickness is slightly
larger for urban FE than for rural FE may reflect the
urban increase in concentrations of very large aerosol
particles, and enhanced sedimentation of drops form-
ing upon them.
5.5 Life history of radar echoes
Some very important clues as to possible mecha-
nisms whereby large urban industrial areas cause ef-
fects upon clouds and rainfall came from a study of
the complete life histories of a large number of radar
echoes which occurred over or near St. Louis (Chang-
non, 1976). The TPS-10 radar at Pere Marquette State
Park (see Fig. 14 of Chapter 1), was used to follow the
locations and top-heights of702 individual radar echoes
throughout their lifetimes. These measurements were
used to look for systematic differences between urban
and rural echoes. Data came from 19 periods on 16
days of July-August 1973; they covered 229 hours and
accounted for ~ 70% of the two-month total rainfall on
the network. The 19 periods and 702 echoes were dis-
tributed among synoptic weather types as follows:
squall line, 3 and 162; squall zone, 2 and 92; cold front,
2 and 226; combination cold front and squall line, 4
and 148; air mass, 8 and 74.
5.5.1 DATA AND ANALYSIS TECHNIQUES
The radar field from which data were used in this
study was a 120° sector (80--200°) from the Pere Mar-
quette radar between 16 and 88 km. St. Louis is ap-
proximately centered in this radar field. The radar was
operated with a receiver sensitivity of 30 dBZ at 16
km and 45 at 88 km range. Since this radar is subject
to precipitation attenuation, only those echoes which
were considered to have little or no interference from
intervening echoes were accepted for analysis. As a
result, about one-half of all detected echoes had to be
deleted from the initial sample.
The analysis was made from radar scope photo-
graphs. Individual echoes were identified and followed
from one radar scan to the next at intervals of three
minutes or less. Every time each echo was scanned its
location and top height were recorded along with the
time. Echoes were designated "urban" if, at any time
during their life, they occurred over or passed over
94 METEOROLOGICAL MONOGRAPHS

either of the two cross-lined areas shown in Fig. 14 of 120 I I I I I

Chapter 1. Echoes within the sample area that did not STAGE 2 STAGE 3
touch either of these areas were designated as "rural." •
The approximate size of the urban areas and non-ur-
ban (rural) area used in this study were 960 and 6900
105 ,---
I
•
STAGE 4
-

•
km2 , respectively. New echoes which arose within 4
km of an existing echo were considered to be part of 90 t-
STAGE I
.r URBAN ECHOES THAT MERGED

-
the original echo and not recorded as a "first echo."
·--------·------
2 3

An important part of this study related to the merg- ~

I'
I
II
I

-
ing of echoes. A merger of two echoes was defined to "2 I I RURAL ECHOES THAT MERGED
~ 75 - 1 : ~ -
occur when ''. . . two echoes that had existed for at .:: q ,.------· 3 : 4
least 10 min and had been more than 8 km apart moved ~ 'IIi 2~· -2- · - - ·------·I DISSIPATION
~ ~~~-~3
together such that they meshed to form one or more
cells for at least 5 min .... '' Subsequently the location 60 f-1 !--r +\
-

I
• 1 4 DISSI PAT! ON
and properties of the merged echo were assigned to ~--,
both of the originally separate echoes. This method of tf- INITIATION •r e -4- e
\,URBAN ECHOES THAT DID NOT MERGE
-
analysis tended to accentuate any difference between 45
RURAL ECHOES THAT •
the merged and unmerged echo classes, but the extent DID NOT MERGE DISSIPATION

of this effect is unknown.

3o,___ _j___I_ _ _1
_ L __ __,_1_ ___,1_ _ _,_____.1
20 40 60 80 100

5.5.2 URBAN AND NON-URBAN ECHO TIME (minutes) AFTER ECHO INITIATION

CHARACTERISTICS Fto. 19. Time-height profiles of urban and rural echoes that
merged or did not merge. After Changnon (1976) with revisions.
From the tabulated records on each echo, a number
of echo characteristics were averaged for various echo slightly higher than that of non-urban echoes (6.4 km
partitions to form the basis of an analysis which is vs. 6.0 km). These values are slightly larger and re-
summarized, in part, by Table 10 and Fig. 19. This versed in order from the first echo data reported by
summary concentrates on only two aspects of the Braham and Dungey (1978). However, in both cases
original paper, viz., differences between urban and ru- the differences are small and thought to be not mete-
ral echo characteristics and the effects of echo merg- orologically significant.
ing. In most other characteristics, however, urban echoes
Table 10 shows that on average the heights of the differed substantially from the non-urban ones. Refer-
tops of first echoes that initiated over the urban area, ring first to the complete sample statistics in Table 10
passed over the urban area after initiation, were we see that:

TABLE 10. Summary characteristics of 702 radar echoes studied throughout their life histories, partitioned on whether or not they
passed over an urban area or were involved in echo mergers. After Changnon (1976).

Echoes that merged Echoes that did not merge Complete sample
Urban Rural Total Urban Rural Total Urban Rural Total

Sample size 61 129 190 76 436 512 137 565 702

Number initiating
over urban areas 16 16 40 40 56 56
over rural areas 45 129 174 36 436 472 81 565 646
Average top heights (km)
at initiation 6.9 6.7 6.71 6.0 5.8 5.8 6.4 6.0 6.1
5 min pre-urban 9.0 6.3 7.5
10 min into urban areas 12.3 7.5 9.6
20 min into urban areas 12.8 7.5 9.9
maximum over urban 13.5 9.3 11.2 7.9 8.6
absolute maximum 14.1 10.7 11.6 10.2 7.0 7.5 11.9
Number that grew
300m 61 Ill 172 54 188 242 115 299 414
3000 m 49 45 94 22 28 50 74 73 147
Average maximum 5 min 4.8 2.4 2.7 2.1 0.6 0.9 3.0 0.9 1.5
growth (km)
Average duration (min)
over urban areas 35 35 23 27 28 28
total 110 73 85 44 25 73 36 43
METROMEX: A REVIEW AND SUMMARY
• The fraction of echoes that grew more than 300
m was markedly greater for the urban echoes than
for non-urban ones (84% vs. 53%).
• The fraction of echoes that grew more than 3000
m was very much greater for the urban echoes
(54% vs. 13%).
• The maximum average 5-minute growth for urban
echoes was 3.0 km vs. 0.9 km for non-urban
echoes.
• The average duration was 73 min for urban echoes
versus 36 min for non-urban echoes.
Although on average the urban and non-urban
echoes had about equal top heights at initiation, those
over the urban area showed a greater likelihood for
further growth, and on average grew more, grew
faster, and lasted longer than the non-urban counter-
parts. The two-sample rank test applied to the data on
urban and rural maximum heights showed that these
differences were statistically very significant. Chang-
non also pointed out that they were, in general, smaller
than day-to-day differences. This necessitated nor-
malizing on the daily average values in some parts of
his analysis. It also gives one a rule of thumb for view-
ing the magnitudes of urban effects upon echoes.
Only in one aspect do the results of this complete
echo history study seem to differ substantively from
other METROMEX studies. Considering only the 702
echoes studied, the frequency of echo initiations per
unit area over the urban areas was only about 0.6 that
of the rural area. This contrasts with a 2-2.5 fold en-
hancement of first echo frequencies over urban areas
found by Braham and Dungey and a roughly similar
value reported by Changnon (1978a). Even if that part
of the echo sample which could not be used (roughly
half) were all urban echoes, the ratios of urban to rural
initiations would not have come up to those reported
by Braham and Dungey. The importance of this point
on assessment of the findings of the complete echo-
history analysis is not known.
5.5.3 EFFECTS OF ECHO MERGERS
Examination of the interior values of Table 10 shows
that the largest differences between urban and non-
urban echoes in the Changnon (1976) study related to
echo mergers. For every parameter tabulated the val-
ues for echoes involved in mergers exceeded the cor-
responding value for non-merger echoes. This is
shown graphically in Fig. 19. The four stages of growth
refer to four equal divisions of the echo lifetimes.
We note a systematic progression of echo proper-
ties. Rural echoes that did not merge were the shortest
lived and had the lowest average maximum heights.
This was followed by urban echoes that did not merge,
rural-merged echoes and urban-merged echoes. Thus
95
an urban effect is suggested for all echoes but the big-
gest differences were for echoes involved in mergers.
However, there were marked differences between
urban-merged and rural-merged echoes. On average,
the urban-merged echoes were taller, had greater rates
of growth, and lasted longer than merged echoes over
the non-urban areas. These differences were checked
with the two-sample rank test and found to be statis-
tically very significant. Since in general, taller and
longer-lived echoes are associated with greater amounts
of rain, it is important to understand these data on
merging echoes in seeking understanding of rainfall
enhancement in urban environments (Changnon et al.,
1976).
From Fig. 19 and Table 10 we note that echoes
which subsequently become involved in mergers had
initiation heights substantially greater than those which
did not become involved in mergers. Both Johnson
(1978) and Ochs and Semonin (1977) have shown from
numerical model studies on METROMEX that the top
heights of radar first echoes depend directly upon
cloud updraft speeds. Moreover, mesoscale dynamical
features of the sub-cloud boundary layer are known
to have a strong influence on cloud updraft speeds
during the early life of a cumulus cloud. As far back
as the Thunderstorm Project (Byers and Braham,
1949) it was pointed out that measurable convergence
in the near-surface wind field preceded echo formation
by as much as 30 minutes. Similar findings have been
reported by Ulanski and Garstang (1978) and by Lei-
chter (1974). Therefore, these radar echo data may
suggest that echo merging could be related to pre-
cloud boundary-layer dynamics.
Another interesting aspect of echo mergers can be
found in data from Table 10. The likelihood of merging
of echoes initiated in the rural region and remaining
rural was 129/565 = 23%. The likelihood of merging
of echoes initiated over the city was 16/56 = 29%. The
likelihood of merging of echoes which initiated over
the rural region upwind of the city and then moved
into the city was (45)/(45 + 36) = 56%. Although an
echo which initiated over the city had a slightly better
chance of merger than those over rural regions, the
major difference was found in echoes which initiated
upwind of the city and moved over it. The meaning of
this is not clear. It could be a result of the method of
designating urban echoes. However, it could also be
providing important evidence for understanding urban
weather effects. Some of these mergers may reflect
upwind echoes moving into an enhanced field of
echoes over the city. A case study showing this oc-
currence was presented by Semonin (1975, p. 61).
Yet another indication that something happening
upwind of the city affects each merging can be seen in
Table 10. We note that echoes which merged were
96
FIG. 20. Frequencies of mergers between near surface radar
echoes, based upon a 35-storm sample. After Huff ( 1978).
already substantially taller than echoes that did not
merge five minutes upwind of the urban boundary.
Perhaps this was because some of them had already
undergone merging and had experienced the growth
benefit therefrom. Once over the city. the merger
echoes showed a greater rate of growth and reached
greater heights than did the non-merging echoes.
Changnon also found that the merged echoes spent
more time over the city than did non-merged echoes,
and that both spent only about l/3 of their total life
over the urban area. He also found that merged echoes
reached their maximum heights after exiting from the
urban areas. Thus, a substantial part of the life of ur-
ban echoes is spent downwind of the city where en-
hanced rainfall is observed to occur. Clearly the like-
lihood of merging of urban echoes is intertwined with
the greater frequencies of taller echoes over, and just
east, of the urban area. An analysis of the complete
life histories of 75 echoes on five days of July 1974
allowed determination of echo volume and echo heights
for every 3 min. Echo volumes were strongly corre-
lated to echo heights (r = +0.93), and the relationship
showed that for every 1.46 km increase in height
(above 2.2 km), the echo volume doubled. Thus, taller
echoes also have greater areal extent and greater op-
portunity to interact and merge with other echoes.
Preferred locations for echo mergers was also stud-
ied by Huff (1978) using data obtained with the Pere
Marquette FPS-18 radar operated at a sensitivity of
about 30 dBZ and an antenna elevation of 0.5 to 1.0
degree. In this case, the echoes relate closely to areas
of significant rain reaching the surface. In a 35-storm
sample (1972-75) 138 mergers were detected. Merger
frequencies maximized in the Belleville, East St.
METEOROLOGICAL MONOGRAPHS
Louis area with secondary maxima over the southwest
hill region and along the southeast bluffs (Fig. 20).
5.6 Analyses of surface raincells
One of the many techniques used in our efforts to
understand and evaluate the effects of the St. Louis
urban-industrial complex upon precipitation was the
analysis of surface raincells (Schickedanz, 1972, 1973,
1974, 1978). Data for these analyses were obtained
from the METROMEX raingage network . (See Chap-
ter I for details of the network.) The approach was to
isolate areas of rain such as might be associated with
individual clouds or cloud circulations and then to
seek differences that might be ascribed to urban influ-
ences through a comparison of raincells which had
passed over an urban area and with those confined to
rural areas .
5.6.1 THE RAINCELL APPROACH
Since most of the rainfall in the METROMEX area
in summer arises from convective storms, it was useful
to turn to the thunderstorm analogy in developing a
rationale for studying raincells. The Thunderstorm
Project showed that thunderstorms are composed of
one or more individual units of convection with char-
acteristic patterns of rain at the surface (Byers and
Braham, 1949). Although the concept of identifying
areas of surface rain with individual cloud circulations
is simple and straightforward, in reality rain patterns
often are so complex that the concept is hard to apply
without rather elaborate guidelines. In METROMEX
raincells were rainfall areas within isohyets which
were closed within the raingage network and which
lay within an enveloping isohyet of a synoptic storm
system consisting of one or more rainshowers or thun-
derstorms . Each shower or thunderstorm may be sin-
gle celled or multicellular. In a multicellular storm sys-
tem the raincell incorporates an isolated area of
significantly greater rain intensity than the system-en-
veloping isohyet. In isolated single-cell storms, the
raincell is uniquely defined by the separation between
rain and no rain. This definition was applied to rainfall
charts which had been computer plotted from rain
amounts manually digitized from original raingage
charts . On 5 min isohyetal charts a skilled analyst
made a determination of which rainfall entities con-
stituted raincells . Once the raincells had been delin-
eated, a number of cell parameters were determined
for each cell. These included total rainfall volume,
area covered by rain, mean rainfall, duration, etc. The
cell-defining isohyet varied from cell to cell and with
time on single cells to allow for changing rainfall con-
ditions. For further details of raincell analysis proce-
dures see Schickedanz and Busch ( 1975), Schickedanz
(1978).
METROMEX: A REVIEW AND SUMMARY 97

5.6.2 DESIGNATION OF URBAN-AFFECTED TABLE II. Designations, sizes and descriptions of areas used in
the raincell analyses. After Schickedanz (1978).
RAIN CELLS

For purposes of METROMEX it was desirable to Size

Designation (km2 ) Description
separate raincells likely to have been affected by the
urban areas from those unlikely to have been so af- Urban effect areas
fected. Two different methods were used. In his 1978 L 775 St. Louis-a simplified boundary of the
contiguous St. Louis urban-industrial
paper, Schickedanz used a "factor statistical ap- area; containing known sources of heat,
proach" for determining both the potential areas of cloud and ice nuclei, vertical motions
effect and for separating the types of storms affected and possible sources for moisture.
directly from the storm rainfall patterns. This method Sl 170 St. Louis industrial- a sub-area within
L containing the greatest concentrations
is not discussed in this review; interested readers of industrial activity; frequently near the
should refer to the original papers. The second method center of the nocturnal heat island.
began with the selection of several urban and topo- w 170 Wood-River and Alton, Illinois industrial
graphic areas where local sources of heat, moisture, area, a region of concentrated heavy in-
dustry.
turbulence and/or nuclei might affect clouds and
thereby alter rainfall. These areas are shown in Fig. Topographic effect areas
21 and are described in Table 11 . H 436 Hills area-forested foothills of the Ozark
All raincells developing in, or passing over, one of Plateau.
these source areas were designated as affected cells. B 363 Bottomland of the Missouri and Missis-
Those contained in area C were called control or non- sippi confluence; largely low-lying agri-
cultural and marsh land.
affected cells. This designation was made on the basis
of a cell's base-isohyet. A small number of cells which Non-effect area
occurred in more than one of the source areas were c 3576 Control area- the remainder of the rain-
deleted from the analysis. Many others were not com- gage network; mostly rolling terrain both
forested and in agriculture.
pletely contained in the network and were not used in
the analysis.
Application of these definitions and procedures to in W, 39% in H, 38% in B, and 40% in C. Incomplete
the five-minute rainfall charts resulted in the desig- cells accounted for another 18% ; rainfall not assigned
nation of 7889 raincells that went through their com- to raincells made up the remaining 61%. Raincell sam-
plete life histories within the boundaries of the rain- ple sizes and data on average raincell rainfall are given
gage network. The aggregate rainfall from these 7889 in Table 12.
raincells totaled -21% of the network total. Values
for individual areas were 44% in S I, 40% in L, 38% 5.6.3 TOTAL RAIN PRODUCED BY RAINCELLS

In keeping with the objectives of METROMEX,

priority attention was given to the amounts of rain
produced by raincells. The average rain from 3160
"affected" cells and 4729 "control" cells is given in
Table 12.

TABLE 12. Raincell sample sizes, average raincell rainfall and the
ratio of rain from affected raincells to that from control raincells,
for several areas within the raingage network. See the text for an
explanation of abbreviations . After Schickedanz (1978).

Areas within raingage network

L Sl w H B Control

Sample size 1746 559 449 375 590 4729

Average 24.8 34.2 19.5 19.1 12.3 11.0
rainfall
(hectare-
meters)
Ratio of rain 1.25 2. 11 1. 77 1.74 1.1 2
from
k.m affected
0 - 8 cells to
rain from
control
FIG. 21. Areas within the raingage network used in the analysis cells
of raincells. After Schickedanz (1978).
98 METEOROLOGICAL MONOGRAPHS

TABLE 13. Sample size and average rainfall volume of various categories of raincells according to synoptic storm type. Mter
Schickedanz (1978).

Types of raincell
L Sl w H B c
Sample size
Squall zone 516 !50 163 108 217 1530
Squall line 584 205 117 119 171 1382
Air mass 64 9 19 27 37 237
Cold front 254 69 65 43 76 647
Static front 196 70 51 43 45 538

Volume (hectare-meters)
Squall zone 17.4(93) 28.8(220) 11.7(30) 18.0(100) 10.6(18) 9.0
Squall line 33.3(107) 51.5(220) 37.2(131) 26.3(63) 16.3(1) 16.1
Air mass 11.4(34) 27.1(219) 13.5(59) 9.8(15) 12.3(45) 8.5
Cold front 21.8(86) 29.0(148) 20.7(77) 20.4(74) 11.6( -I) 11.7
Static front 7.9(55) 10.4(104) 8.9(75) 10.7(110) 8.4(65) 5.1

Note: Effect-control differences are expressed as % of control in parentheses.

TABLE 14. Average rainfall volume, in units of hectare-meters, of various categories of raincells stratified by time of day (CDT). After
Schickedanz (1978).

Time of Types of raincells

day L Sl w H B c
0001-0600 15.1(97) 20.1(161) 11.4(49) 12.0(57) 8.1(5) 7.7
0601-1200 13.0(52) 14.5(69) 12.3(43) 13.9(62) 8.8(2) 8.6
1201-1800 40.4(171) 59.5(299) 27.4(84) 28.1(89) 16.9(14) 14.9
1801-2400 22.8(89) 33.8(180) 23.7(96) 22.1(84) 14.4(19) 12.0

Note: Effect-control differences are expressed as % of control in parentheses.

Since properties of cells in the control area were
used for comparing the affected raincells it is useful
first to compare them with data from earlier studies.
Using Thunderstorm Project data from Ohio, Braham
(1952) estimated the mean rainfall from a thunderstorm
cell as 1.24 x 108 kg (about 12.34 hectare-meters).
Huff and Schickedanz (1970) obtained a median value
of 13.57 hectare-meters for raincells over rural south-
ern Illinois. Both of these earlier studies used raingage
networks much smaller than that of METROMEX.
Nevertheless, in view of the fact that the raincell study
included small rainshowers, the value of 11.0 hectare-
meters for the control area cells seems surprisingly
close to the earlier values.
Also shown in Table 12 are the ratios of rain from
affected cells to that of the control cells. The largest
differences were in the L and S 1 area with mean values
of 1.25 and 2.11. Areas W and H averaged 1. 77 and
1.74 but were much more variable across years. Bot-
tomland raincells averaged a few percent above the
control values but were less than the control area cells
for two of the five years. The raincell data were par-
titioned in a number of ways. Results for partitions on
synoptic weather types and time of day are given in
Tables 13 and 14. Approximately 62% of the raincells
came from squall-line and squall-zone conditions.
They accounted for ~ 75% of the raincell rainfall and
showed the largest percent increases over the control
cells. Results for hills and bottomland cells were more
variable across years with the largest percent in-
creases coming with static fronts. Table 14 shows that
cells in all areas had their maximum rain and maximum
percent differences during the hours of maximum
heating (minor exception in theW area). What may be
an important point in interpreting urban effects is the
carry-over of high rainfall into the 0001-0600 CDT pe-
riod in areas L and S 1. The reader will recall a similar
finding from the diurnal rainfall curves.
Not shown here are raincell studies which found no
apparent difference in raincell rain on weekdays com-
pared to weekend days. Schickedanz also found that
maximum urban influences in the St. Louis area oc-
curred in July and August, whereas in the Alton area
they occurred in June.
Another of the important partitions is that reported
by Huff ( 1975) of the properties of the 300 most intense
raincells occurring during the first three summers of
METROMEX. Table 15 gives data on several impor-
tant parameters of these most intense cells. Note that
the average volume of rain from these cells was sev-
eral times that of the all-cell means given in Table 12.
5.6.4 RAINCELL INITIATIONS AND MERGERS
The number ofraincell initiations at individual gages
were mapped as shown in Fig. 22. Areas of maximum
METROMEX: A REVIEW AND SUMMARY 99

TABLE 15. Rain cell means and maxima in 300 cells having
heaviest mean rainfall in 1971-1973 summers. After Huff (1975).

Three-summer
Parameter Mean maximum

Rainfall (mm) 11.4 38.9

Area (km2 ) 114 534
Volume (m3 ) 1.22 X 106 8.8 X !06
Duration (min) 36 160
Path length (km) 8.2 46.7
Rainfall gradient, (mm km- 1) 3.6 13.42
5-min maximum rainfall (mm) 7.1 19.3
5-min maximum area (km2) 54 218
5-min maximum volume (m3 ) 4.14 X !OS 2.9 X 106

frequencies of raincell initiation were roughly over the KEY

urban area with individual maxima over the bottom- ~ >12
lands, over the Alton area and along the south edge ~ 9·121+-+='F

of the St. Louis urban area. D

There also were important findings on raincell merg-
ers. Fig. 23 shows the distribution of mergers of rain-
cells meeting the criteria of having initial separations
of at least 8 km, at least 10 min existence prior to
merging, and at least one of them with a rain intensity
of at least 6 mm h- 1 (0.25 inch h- 1) before merging.
We note a tendency for mergers to occur most fre-
quently to the east and northeast of St. Louis and in
regions of potential physiographic influences. Most
mergers were followed by increases in rain area and
intensity. Merged cells also had longer durations than
the non-merged cells. These findings parallel the re-
sults of the merger of individual radar echoes already
discussed.
5.6.5 INTERPRETATION OF RAINCELL FINDINGS
Early in METROMEX questions of interpretation
of raincells arose because of the complex nature of the
FIG. 22. Pattern of raincell initiation. After Schickedanz (1978).
FIG. 23. Spatial distribution of raincell mergers. After Schicke-
danz (1978).
basic raincell data. The reader will recall that - 81%
of METROMEX rainfall came from storms moving
from the southwest through northwest (Table 4), that
the raincell analysis was limited to cells that under-
went their complete life cycle within the network, and
that a rain cell was assigned to one of the affected-cell
categories if its base isohyet crossed the boundary of
a source (affect) area. With these points in mind, and
referring to Fig. 21, one can easily see that cells with
large horizontal extent and/or long path-lengths will be
undersampled in the control area and to a lesser extent
in areas H , B, and W. The extent of this undersam-
pling and its effect on conclusions from the raincell
analysis can only be guessed at. Previously it was
pointed out that the control cell rain compared favor-
ably with two earlier studies. This argues against a
large sampling effect.
In another check on possible sampling biases, the
data were partitioned on raincell path lengths. Clearly
cells with shorter paths had a better chance of being
sampled uniformly in all areas. Results of this parti-
tioning are given in Table 16. We note that rainfall
from effect-area cells decreases with path length
somewhat faster than it does from control area cells.
This is in keeping with a sampling bias. But even for
the short-path cells, the differences in rain between
cells in the affect areas and in the control area are
substantial- 72% in S 1, 38% in L, 29% in H and 10%
in B. These estimates of urban effects are somewhat
larger, but roughly similar to the estimates of urban
effects found in other parts of the METROMEX anal-
ysis.
100 METEOROLOGICAL MONOGRAPHS

TABLE 16. Average rain from raincells crossing various parts of the raingage network, and the ratio of rainfall for affected cells
to that of control area cells, partitioned according to path length. See the text for an explanation of area designators. Mter Schickedanz
(1978).

Areas within raingage network

L S1 w H B Control

Path length < 19.3 km

Average rain (hectare-meters) 19.5 25.0 14.2 15.4 11.7 10.2
Ratio of rain from affected cells to 1.91 1.45 1.39 1.51 1.15
rain from control cells

Path length < 12.8 km

Average rain (hectare-meters) 15.0 17.7 10.9 12.0 10.6 9.1
Ratio of rain from affected cells to 1.65 1.95 1.20 1.32 1.16
rain from control cells

Path length < 6.4 km

Average rain (hectare-meters) 9.4 11.7 8.8 8.1 7.5 6.8
Ratio of rain from affected cells to 1.38 1.72 1.29 1.19 1.10
rain from control cells

One of the disadvantages of this short-path ap- 5.6.6 URBAN EFFECTS ON HAILSTREAKS
proach is that it could also be working against finding
real urban effects in cases where urban processes tend Hailstreaks, which are surface areas of hail, were
to make storms last longer and have longer path studied on the basis of their frequency and character-
lengths. In summary, the raincell analysis indicated istics in storms that initiated or traveled across the
urban-related increases in precipitation with estimates urban areas and those that did not. For hailstreak stud-
of magnitude ranging from -40-200%, depending ies the urban area was considered to correspond to
upon area and the extent of any sampling biases. The the hatched areas in Fig. 10 of Chapter 1. A study by
largest increases were associated with organized con- Changnon (1978b) showed that storms crossing the
vective systems (squall lines, squall zones and cold urban area produced 50% more hailstreaks than did
fronts). The maximum increases were found during the rural area storms. This apparent urban effect oc-
hours of maximum surface heating with important curred largely in the 1200-2400 CDT period and was
carry-over into post-midnight hours. There seemed to most evident in hail-producing storms associated with
be little difference between weekday and weekend day cold-front and squall-line conditions. Little effect was
results. Cell initiation and cell merger studies gave re- seen in airmass-storm conditions.
sults generally consistent with other studies reported The urban and rural hailstreak characteristics are
in this volume. given in Table 17. The mean values are larger in the

TABLE 17. Frequencies and characteristics of hailstreaks associated with storms crossing the urban area, and with rural storms. After
Changnon (1978a).

Urban storms Rural storms Complete sample

Mean Median Number Mean Median Number Mean Median Number

Sample size 322 303 625

Average point duration (min) 2.7 2.0 301 1.9 1.0 263 2.3 1.0 564
Begin time of streaks (CDT) 1724 320 1525 287 1630 607
Duration of streak (min) 14.9 12.0 106 14.2 12.0 78 14.6 12.0 184
Maximum length (krn) 11.6 9.8 106 11.1 9.3 85 11.3 9.7 191
Maximum width (krn) 3.2 2.1 106 2.7 1.9 85 3.1 2.1 191
Area (krn2 ) 40.4 15.8 106 39.6 15.5 85 40.1 15.5 191
Average point number of stones 796 291 335 584 183 322 694 226 657
(rn-2)
Maximum hailstone diameter in 1.40 1.27 335 1.12 1.07 322 1.27 1.17 657
streak (ern)
Average point energy (rn2 kg s- 2 15.76 1.61 339 8.03 0.58 323 11.97 1.17 662
m-2)
Average rain in streak (ern) 2.03 I. 73 336 1.68 1.27 305 1.85 1.52 641
METROMEX: A REVIEW AND SUMMARY 101

urban hailstreaks for all characteristics. However, the 5.7.2 FREQUENCIES OF TALLEST ECHOES OVER
differences (in the means or medians) are very small URBAN AREAS
in hailstreak areas, width, length and duration. Hail- Braham and Wilson ( 1978) reported results of a
streaks associated with urban areas had longer point study of urban effects on the heights of r~dar .echoes
durations of hail, more hailstones, larger stones, and which were based upon an approach qmte dtfferent
hence more impact energy. These findings suggest that from that used by Changnon, even though it too made
the volumes of hail produced in urban storms are typ- use of echo top heights measured with a TPS-10 radar.
ically not larger than in non-urban storms, but the vol- The radar at Greenville, Illinois, was operated in a
ume inside urban storms tends to have more hailstones continuous azimuthal scan mode during periods of ra-
produced within it. Urban storms also seem to have dar echoes anywhere within a range of 97 km. This
more regions of hail. We can only conjecture expla- created a radar scan area covering a major portion of
nations for these data. They may indicate more nu- south central Illinois with St. Louis near the south-
merous but not more vigorous and sustained updrafts western edge. From film records, data on the location
in urban clouds. They could also support a concept of and top heights of the single tallest convective echo
urban-induced glaciation. detected anywhere within a range of 32-97 km (20-60
mi) at each half-hour were extracted. (Data inside 20
5.7 Effects of urban area on convective echo heights mi were excluded because of the radar's limitation in
accurately observing very tall echoes at close range.)
The effects of the St. Louis urban area upon heights The 140 days of record from the Greenville radar gave
of convective clouds (actually upon radar echoes as- 1573 observation times (on the hour and half-hour) at
sociated with such clouds) was the subject of two which radar echoes were present in the radar circle,
METROMEX studies. We have already mentioned a Table 18. For purposes of discussion each of these
study by Changnon (1976) of the complete life histories half-hourly observations were called Hi-Cu. Because
of 702 radar echoes occurring over an area of over of the way radar operations were scheduled, most of
7800 km2 roughly centered on St. Louis. University of the Hi-Cu observations were between 1000 and 2000
Chicago scientists conducted a study of the highest CDT. Between these two times there were relatively
radar echoes to occur every half-hour anywhere in an few missing observations, though many were echo
area of about 28 000 km2 centered on Greenville, Illi- free.
nois, ~55 km northeast of St. Louis. The result of The Hi-Cu observations were distributed among
these two studies is the subject of this section.
synoptic weather conditions as follows: squall line and
squall zone, 56 periods, 768 Hi-Cu; cold front, I 7 pe-
5.7.1 MAXIMUM HEIGHTS REACHED BY riods, 164 Hi-Cu; air mass, 45 periods, 407 Hi-Cu; all
INDIVIDUAL ECHOES others, 22 periods, 234 Hi-Cu. Of the 140 echo p~riods
used in the Hi-Cu analysis, 110 were accompamed by
We begin by looking again at the study of time his-
rain in the St. Louis raingage network. During the re-
tory characteristics of 702 radar echoes. These echoes maining periods the Hi-Cu echoes were restricted to
occurred on 19 storm days during June-August, 1973.
other parts of the Greenville radar field. The network
The echoes were followed from FE initiation through average rain accumulated during the 110 periods was
their maximum development to the point of dissipa- about 60 em or ~90% of the total project rain falling
tion. Data about the heights of these echoes over ur- between 1000-2000 CDT. The pattern of rain during
ban and rural areas are given in Table 10 and Fig. 19. Hi-Cu periods (Fig. 24) shows a maximum east of the
Changnon found that 137 echoes initiating over, or city. We conclude that the Hi-Cu data are represen-
passing over, the St. Louis area or the Alton area had
average maximum tops of 11.9 km compared to 7.9 TABLE 18. Number of days and observations making up the
km for 565 rural echoes. More importantly, he showed Hi-Cu sample. After Braham and Wilson (1978).
that when echoes merged, growth was accelerated and
the maximum top heights increased considerably. June July August
Echoes that did not merge had average maximum top Obser- Obser- Obser-
heights of 10.2 km, urban; and 7.0 km, rural. Echoes Year Days vations Days vations Days vations
involved in mergers had maximum top heights of 14.1 3 21 10 67
1971
and 10.7 km for urban and rural, respectively. This 1972 11 123 16 127 II 139
study of complete histories of echoes concluded that 1973 16 216 13 170
1974 I 2 13 121 13 199
urban echoes can become much taller than rural ones, 1975 6 72 15 184 12 132
provided they are involved in echo merging.
 102
0 5
STATUTE MILES
FIG. 24. Pattern of total rain (em) in the METROMEX network
during periods of Hi-Cu analyses.
tative of conditions during the afternoon rain peak, but
perhaps not of the nocturnal peak.
The analytical procedure for estimating urban ef-
fects was similar to that used in the analysis of first
echoes. This resulted in stratifying the data into five
groups, those over the city (see Fig. 16), those over
one of three downwind areas based on 1 h increments
of echo movements, and those over the rural area. The
rural area represented the entire 360 degree radar field
between 32 and 97 km, minus the city and three down-
wind areas. The Hi-Cu data for the city and the three
advected areas were then expressed as area-weighted
frequencies relative to rural values. (See the original
paper for details).
5.7.3 LOCATION OF THE TALLEST ECHO OF EACH
DAY
Measurements of the single tallest echo of each day
form an attractive data set since these observations
may be regarded as statistically independent. Their
height distribution is given in Fig. 25. (Recall that the
majority of the Hi-Cu data were taken between 1000
and 2000 CDT.) Table 19 gives the area-normalized
frequencies with which the daily maximum echo was
found over the city and downwind regions. We note
a value of 1.34 for the city and about 1.7 for the first
2 h downwind. The third-hour advection region gave
essentially rural values.
The 22 days with EMV <3 m s- 1 gave a relative
frequency of 2.55, suggesting a substantial urban ef-
fect. A binomial test applied to the 22 daily values
allowed rejection of the null hypothesis of random
METEOROLOGICAL MONOGRAPHS
ECHO TOP HEIGHT, f t x 10- 3
FIG. 25. Top-height distribution for the tallest echo on each day
in the Hi-Cu analysis. After Braham and Wilson (1978).
echo locations at a level of0.14 (one-tail). This is taken
for evidence of a real urban effect upon echo top
heights but not at a very convincing significance level.
On days with EMV :;;.:3 m s- 1 the suggested urban
effect upon the tallest echo of the day appears to be
10
limited to the 1 and 2 h advected areas.
5.7.4 URBAN EFFECTS UPON HI-CU LOCATIONS
Table 20 gives data for a similar analysis of the com-
plete sample of 1573 Hi-Cu observations. We note that
in all partitions the third hour advected area has values
very close to those of the rural region. Data for the
city and first two hours downwind give evidence for
an urban effect. This evidence is strengthened when
the data are partitioned on wind speed. With EMV <3
m s- 1 the relative frequency of Hi-Cu over the city is
about twice that of the rural area. In the 3.0 .;;; EMV
.;;; 7.9 m s- 1 partition, this effect is reduced and spread
over the city and the first hour downwind. With still
greater wind speeds, we find the urban effect decreas-
ing in magnitude and shifting to the first two downwind
areas while the city has returned to near rural values.
It is well to recall that with EMV = 10 m s- 1 the 2 h
wind travel distance would reach almost to Greenville,
Illinois.
In assessing the statistical significance of these rel-
ative frequency values , we encounter several prob-
lems. The observations cannot be regarded as statis-
tically independent. Moreover, the city and the three
downwind areas are always much smaller than the ru-
ral area, and on any given day, more often than not,
TABLE 19. Area-normalized frequencies ofthe daily maximum Hi-
Cu in selected areas for two partitions based on the echo movement
vector (EMV). Mter Braham and Wilson (1978).
Relative frequency
Number
of days City Ih 2h 3h Rural
EMV < 3 m s- • 22 2.55 1.00
EMV ;;. 3 m s- • 118 1.07 1.57 1.61 0.86 1.00
Total sample 140 1.34 1.67 1. 72 0.92 1.00
METROMEX: A REVIEW AND SUMMARY 103

TABLE 20. Area-normalized relative frequencies of the half-hourly 5.7.5 TIME OF DAY EFFECTS
Hi-Cu observations in various parts of the urban area for the total
sample and four partitions on echo movement (EMV). After Braham Having found that the effect of St. Louis on the
and Wilson (1978).
relative frequencies of the tallest echoes was limited
Relative frequency to the areas designated as city and 1 and 2 h down-
Number wind, data from these three areas were combined into
of days City Ih 2h 3h Rural a single sample. In the remainder of this section when
All data 140 1.19 1.46 1.30 0.96 1.00 we refer to urban Hi-Cu we mean the combined city,
Wind partition 1 h and 2 h data. Fig. 26 shows the relative frequency
EMV <3m s- 1 22 1.99 1.00
3.0-7.9 m s- 1 42 1.42 1.43 1.09 1.01 1.00 with which the tallest echo was found in the urban
8.0-10.9 m s- 1 37 0.98 1.69 1.41 0.92 1.00 area as a function of time of day.
11.0-19.9 m s- 1 39 0.80 1.38 1.37 1.01 1.00 We find that consistently throughout the late morn-
ing and early afternoon the frequency with which the
do not contain a Hi-Cu event. Therefore, the analysis tallest echo occurred over the urban region was sub-
must cope with a large number of zeros. In addition, stantially higher (up to 2.7 x) than that expected from
the downwind areas and the rural area vary in size a random distribution. From about 1400 to 2130 CDT
among themselves on any given day and from day to the relative frequencies are about equal, though there
day depending upon EMV. may be a tendency for urban excesses until about 1830
On 22 days in the Hi-Cu analysis the winds were and urban deficits from 1830 to 2130 CDT. There also
very light and the EMV was less than 3 m s- 1 • On is a suggestion of urban enhancement prior to 1000
these days we are willing to assume that the urban and after 2130 CDT, but the samples are small and
effect is restricted to the area designated as city, while caution is advised about drawing conclusions from
the remainder of the radar field remained unaffected. these parts of the data.
Under this assumption computations were made of the
probability that a random distribution of Hi-Cu over
the radar field would give the same or greater number 5.7.6 HEIGHT DISTRIBUTIONS OF URBAN AND
RURAL HI-CU
of city echoes as was actually observed. Using the
Poisson approximation to the binomial expansion, ap- Fig. 27 is a plot of the actual frequencies with which
plied to the daily data, a standardized normal deviate Hi-Cu in the urban and rural areas had the indicated
of 3.65 was computed. This suggests rejection of the heights. Also shown is an adjusted frequency curve
null hypothesis at a significance level of 0.0001 (one- for urban Hi-Cu. This adjustment is used because the
tailed). But this result assumes independence of the differences in size of the urban and rural areas would
observations, an assumption that is almost certainly result in apparent frequency differences even if there
unwarranted. were no real differences. The adjustment consists of
Previously, it was shown that for the single tallest multiplying the urban frequencies by the ratio of the
echo of each day, where independence was more sizes of rural and urban areas.
likely, the null hypothesis of no urban effect could be Looking first at the data for rural Hi-Cu we note
rejected at a level of only 0.14. Thus, although the city that the distribution is distinctly bimodal with peaks
effect was about the same in the two cases (1.99 vs at 20 000 ft (6.1 km) and 40 000 ft (11.6-12.2 km) above
2.55), the differing sample sizes markedly affect the ground level (agl). The frequency of rural Hi-Cu with
calculated significance. tops at 28 000-34 000 ft (8.5-10.4 km) is only about

SAMPLE
URBAN = 1 1 1 2 3 7 12 8 11 1~ 18 14 11 10 15 16 13 16 17 14 157 5 8 9 6 4 3 5 4 2 1 1
RURAL "' 1 1 4 420 22 26 27 27 38 43 50 60 68 74 77 83 79 74 7263 57 59 47 41 35 30 24 18 8 6 4 2
7.0 rr-.~_,.,-,,,.-,~-,.,,.,-,.-,.,-,,-,.-,~-.,

~ s.or-~----------------------------------------~
u>- "'
:::>
r5 b
:::> u.J
3.0
ON
~ ~ ~2.0 ~4-~-=~--~-+------------------------4-~~
"""'"'
"":::>
=>e::::c:::
ua
...'. ":' 1. 0 ~---+--1----------=~--~-----+--...--L--L..-r--+---------j
:I:<(

~ 0 0 7 ~------------------------------"'=t-+------'.,._---------j
<(
0.5 ~~~~~--L..~~~~_L~~~~~~--L..~~~~~
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
TH1E, COT
FIG. 26. Ratio of area-normalized urban Hi-Cu frequencies to area-normalized rural Hi-
Cu frequencies as a function of time of day. Numbers of urban and rural Hi-Cu given above
the diagram. After Braham and Wilson (1978).
104 METEOROLOGICAL MONOGRAPHS

a glaciation effect, pre-moistening of the cloud envi-

10 12 14 16 18 ronment, and an increase in size of the updraft area
at cloud base.
Fig. 28 gives the joint time-height probabilities for
100 the occurrence of the Hi-Cu echoes. Solid isolines de-
r-'
r~
I
lineate zones in the time-height field where urban fre-
~ 1.. _,.-, AREA
I :
I r I I : NORI·1ALI ZED
quencies exceed rural values. Parts of the field where
.._,
L..J
BO L..J I 1 ,...- URBAN area-weighted urban/rural ratios exceed three are
I
cross-lined. Stippled areas indicate zones where the
ratios are less than unity. The isoline labels, cross-
60 hatching, and stippling help to delineate the field
where we had sufficient urban and rural data to permit

5 40
this analysis.
Several general features stand out in the analysis
~ shown in Fig. 28. The urban effect is strongly positive
=> up to about noon. This is the period when nocturnal
::f 20
inversions in rural areas would tend to delay convec-
I

tion. The urban effect is positive at almost all hours

10
15
Hl-CU TOP HEIGHT, ft x 10- 3
FIG. 27. Frequency with which Hi-Cu had the top heights indi-
cated. Actual frequencies are given for rural Hi-Cu in the top part
of the diagram; actual frequencies for urban Hi-Cu are shown below.
Also shown in the upper part of the diagram is a curve for urban
frequencies adjusted by the ratio of the average size of the rural
area to the average size of the urban area. After Braham and Wilson
(1978).
half that of the peak values. The upper height mode
obviously contains storms that have reached the tro-
popause. This bimodal distribution for rural Hi-Cu was
found on 4 of 5 individual years and in a series of
random samples drawn from the data. It seems to be
a real feature of convective clouds over south central
Illinois in summer and represents their response to
some frequently occurring mid-tropospheric feature
such as a stable layer or more frequently the top of
the low-level moisture layer.
In contrast with the height distribution of rural Hi-
Cu, the urban curve shows little of a bimodal char-
acter. It reaches a maximum at about 18-22 000 ft
(5.5-6.7 krn) AGL, decreases slowly through 42 000
ft (12.8 krn), and then drops sharply. The urban values
are well above rural from 16 000 ft to about 44 000 ft
(4.9-13.4 krn) and about equal to rural values for
heights below and above these values. Apparently one
of the actions of the urban area is to enable clouds
which would have been restricted to a top height of
6-10 krn to overcome this restriction and grow to
greater heights. Possible explanations for this include
in the 25 000-35 000 ft (7 .&-10. 7 krn) height interval.
This interval corresponds to the saddle in the rural Hi-
Cu heights distribution. There is a large zone of neg-
ative urban effects from about 1600 CDT until 2000
CDT, except for the previously mentioned saddle in-
terval and above about 50 000 ft (15 krn). This time
period corresponds to, and immediately follows, the
afternoon peak in rainfall in the urban area. Evidently
rain during this period carne preferentially from clouds
with tops in the 30 000-35 000 ft (9-10.7 krn) height
interval and from clouds exceeding 50 000 ft (14 krn).
After ~2100 CDT the urban effect is positive at all
heights, but most strongly so at levels below 35 000 ft
(10.7 krn). The reader will recall that the rainfall data
showed a peak during the late evening but there was
no corresponding peak in thunderstorms. The latter
observation would be in keeping with clouds of modest
heights.
5.7.7 NUMERICAL MODELING OF MAXIMUM CLOUD
HEIGHTS
The finding of a unimodal distribution of Hi-Cu fre-
quencies over the urban area, in contrast to the bi-
modal one over the rural region, prompted a study
based upon a thermodynamic model for predicting
cloud heights. The model selected was that of Simpson
and Wiggert (1969, 1971). Tops of convective clouds,
and models used for predicting them, are very sensi-
tive to details of the surrounding environment. Since
the radiosonde station closest to the Greenville radar,
and the only one from which data were routinely avail-
able during METROMEX, was the NWS station at
Salem, Illinois, it was decided to limit the study ini-
tially to 30 days in which the atmosphere showed rel-
atively little change between the morning and evening
soundings from Salem. This study was still in progress
at the time of this report. Therefore, only preliminary
METROMEX: A REVIEW AND SUMMARY
oorr~-,~.-.-.->r-r~-,~.-.-~--~--~-,~.
!3 30
>-
=>
u
'
:r: b)
10
9 11
FIG. 28. Joint time-height distribution of the ratio of urban Hi-Cu frequencies
(area adjusted) to rural Hi-Cu frequencies. Average temperatures at various
heights, for the days entering the Hi-Cu analysis, are given on the right. After
Braham and Wilson (1978).
results can be given, based upon a paper by Mueller
(1978).
In all such cloud models the growth-inhibiting action
of entrainment enters the equations as a statement
about cloud updraft radius. Since actual updraft radii
were unknown, Mueller ran each sounding for three
different radii, bracketing the most probable natural
values (750, 1500 and 2500 m). To test for possible
effects of cloud glaciation all day-radii combinations
were also run with and without glaciation. In keeping
with METROMEX airplane measurements, when gla-
ciation was included in the model, it was assumed to
begin at - soc and extend uniformly to -20°C.
Mueller showed that for all 30 days in his sample
the use of the small radius and no glaciation kept cloud
tops in the height range of the low top-height mode in
Fig. 28. The largest radius put most cloud tops into
the upper height mode regardless of glaciation. The
intermediate radius gave tops in the high or low mode
depending upon whether or not glaciation was used.
The major value of this study is as a reminder that
glaciation and cloud dimensions can be just as impor-
tant as thermal stability and environmental moisture
in controlling the heights of convective clouds. The
increased heights of clouds over urban areas could, in
principle, have been caused from increased horizontal
dimensions of urban clouds, moistening of the urban
cloud environment, or anthropogenic effect on cloud
glaciation.
5.8 Physiographic effects
The preceding sections in this chapter and preceding
chapters document many kinds of evidence establish-
ing that the urban area at St. Louis exerts a strong
influence on many meteorological variables, not the
105
'\AVERAGE THIPERATURE , •c 18
~ -69.9 16
3 -66.9
-63. 2 14
12 ~
>--
10 ~
..,
:r:
8
-11.1 6
-2.1
15 17 21
THIE, COT
least of which are rainfall and thunderstorms. But
when we come to making estimates of the magnitudes
of these effects, and recall that long-term weather pat-
terns frequently exhibit variations over areas the size
of the raingage network, we wonder to what extent
the METROMEX patterns might be due to non-urban-
related factors, such as local physiographic features.
Some of the METROMEX scientists are of the opinion
that the downwind rainfall maximum at St. Louis first
became detectable in the 1950's. If further study bears
out this view, then the issue of physiographic influ-
ences will become moot. For the present it seems de-
sirable to examine possible ways in which local phys-
iographic features might affect local cloud and rainfall
patterns.
One of the advantages of St. Louis as the site for
METROMEX was the absence of any large nearby
body of water or major elevation differences. But
there are three relatively minor terrain features within
the raingage network that might influence clouds, and
hence precipitation. As shown on Fig. 7 of Chapter 1
and on Fig. 29, these areas are 1) the low, relatively
wet bottomland between the Mississippi and Missouri
Rivers in the northern and northwestern part of the
raingage network; 2) a region of rolling hills in the
southwest part of the raingage network; and 3) a line
of bluffs, mostly less than 100 m high, extending
roughly eastward from near Pere Marquette State Park
to a point just north of Edwardsville , and then south-
ward to rejoin the left bank of the Mississippi near the
south edge of the network.
5.8.1 THE BOTTOMLAND AREA
The bottomland area around the Mississippi, Mis-
souri and the Illinois Rivers is a combination of agri-
106
cultural farmland and wetlands. It is obviously a
source for moisture flux into the atmosphere. It is a
region of above average dew-points (Fig. 6 of Chapter
2). A study of GOES data taken during the summer of
1975 showed a maximum of cloudiness over the north-
western part of the city and the surrounding bottom-
land at 1000 CDT. This changed into a region of min-
imum cloudiness at 1300 and 1600 CDT and back into
an area of maximum cloudiness at 1900 CDT (Grosh,
1978a). This area was observed to have above-average
frequency of the earliest cumulus clouds of the day
(see Chapter 4).
As a source of abundant low-level moisture it is not
surprising that it would also be the preferred location
of the first cumulus clouds of the day, as seen by the
ground camera at Pere Marquette State Park. The
abundance of moisture and the prevalence of tilled soil
in the bottomlands resulted in surface temperatures
that frequently were below those in the city. Mea-
surements showed that the equivalent potential tem-
perature in this area tended to be below average for
the metropolitan area during pre-rain periods (Vogel
and Dzurisin, 1978). One would expect this region to
have its maximum effect in increasing the number of
convective clouds a few miles downwind, after the
moist low-level air has been heated over more favor-
able surfaces. A similar effect was found near the
north end of Lake Carlyle (southeast of Greenville,
Illinois) where a local maximum in radar first echoes
was found during wind conditions parallel to the long
axis of the lake (Braham and Dungey, 1976).
From an analysis of boundary layer winds, Acker-
man (1978) showed that the bottomlands experienced
predominantly downward air motions during fair
weather conditions (16 of 24 cases) and about an equal
frequency of upward and downward air motions dur-
ing pre-rain periods.
The METROMEX total rainfall map (Fig. 15 of
Chapter 2) shows a region of above average rainfall
located in the bottomlands north and northwest of St.
Louis. In Section 2 of this chapter we found that the
rainfall maximum in the bottomlands occurs during the
early evening and nighttime hours as part of a general
high-rain condition across the north side of the net-
work. Moisture from the bottomlands could play a role
in the evening and nocturnal rainstorms although, as
yet, this has not been demonstrated. Moisture from
the bottomlands could also fuel the cumulus clouds
that develop downwind of the bottomland-namely in
the Alton, Edwardsville area for commonly observed
wind directions.
5.8.2 THE SOUTHWEST HILL AREA
The hilly terrain southwest of the metropolitan area shine hours, the south- and west-facing rock walls
is over 50% forested with intermixed fields devoted to probably act as a line oflocal hot spots that will initiate
METEOROLOGICAL MONOGRAPHS
animal grazing and row crops. The hills range up to
~ 100 m above the height of the downtown area and
almost 160 m above the height of the Mississippi
River. With south and southwest winds, this provides
a downslope of about 100 m in about 15 km. It is
conceivable that downslope effects contribute to the
general rainfall minimum observed over the western
suburbs of St. Louis (see Figs. 14 and 15 of Chapter
2) by systematically drying out the very lowest cloud
layers which are so important for raindrop collection
in widespread general rains. However, this minor low
is seen as a large scale feature in the area (Fig. 13 in
Chapter 2) and hence does not appear to be closely
related to physiographic features.
The satellite data for 1975 showed that the south-
west hill area had a local maxima in cloudiness at 1000
and 1300 CDT. Later in the day this area had cloud
cover about equal to the remainder of the area. The
surface network showed this region to be a little
cooler, and the dew points about average in compar-
ison with the rest of the network. (See Chapters 2 and
3 for details of these observations.)
The hills in the southwestern part of the raingage
network are essentially similar to the Shawnee Hills
in extreme southern Illinois which are known to cause
a local precipitation maximum of~ 15% more rainfall
than nearby flat rural land. [See Fig. 13 of Chapter 2;
for additional details see Jones et al. (1974), Changnon
et al. (1975) and Huff et al. (1975).] The total rainfall
maps for METROMEX show only a minor rainfall
maximum in the southwest hill region during summer,
but it is a region of maximum rainfall during fall
months.
5.8.3 THE BLUFF LINE
The location of the line of the bluffs is shown in Fig.
15 of Chapter 2 and in Fig. 29. We note that the bluffs
are roughly parallel to the Mississippi River. From the
northwest edge of the raingage network to about Al-
ton, the bluffs are nearly vertical, a south-facing rock
wall north of which the terrain is hilly and tree cov-
ered. The bluffs also are sharp, nearly vertical rock
walls in the southern part of the network. But from
about Alton to Edwardsville, then southwest through
Collinsville to the south side of Belleville, the bluffs
have been eroded into a steep slope which is tree cov-
ered in most places. Maps of total project rainfall (Fig.
15 of Chapter 2), rain echo initiations (Fig. 18), raincell
initiations (Fig. 22), and raincell mergers (Fig. 23) all
show minor maxima along, or slightly east of the bluff
line. The shape and orientation of the bluff line is such
that it might contribute in two different ways to an
increase in clouds, and perhaps rainfall. During sun-
METROMEX: A REVIEW AND SUMMARY
FIG. 29. Ratios based on rainfall in selected subareas of the raingage network,
compared with that in a control area in checking the western suburbs and nearby
area southwest of St. Louis. After Huff and Vogel (1978).
upward air motion or enhance any that may otherwise
be present. Grosh (1978b) noted that the all-sky cam-
era, located at Pere Marquette State Park, often
showed a line of clouds generated along the sharp bluff
to the east of the park. Ackerman (1978) showed that
the southeast bluffs were second only to the down-
town area of St. Louis and the western suburbs, at
1200 and 1400 CDT, in the strength of updraft gener-
ated by near-surface wind convergence during fair
weather conditions. At 1600 CDT the southeast bluffs
were found to be a region of low-level convergence
during pre-rain conditions (Fig. 15 of Chapter 3). Soar-
ing pilots have long recognized that sharp ridges, even
though of modest height, are loci for frequent and sus-
tained upward air motions .
The second possible way in which the bluffs could
affect clouds would be through their overall shape,
which could tend to funnel low-level airflow from the
south through west toward the apex at Edwardsville.
This could lead to an effect similar to that measured
by Hoecker and Angell (1969) over a 100m high bluff
along the California coast. These investigators found
that the bluffs produced a sustained, but fairly local,
upwelling averaging 6 em s- 1 over and slightly down-
stream from the bluff line. It is not the magnitude of
this effect, but its persistence that one would expect
to be most important.
107
ST. LOUIS
EAST
1.19
16
5 10
STATUTF: MILES
There is some slight evidence for a funneling effect
in the wind rose for Alton, which shows a slightly
greater frequency of southwest winds than is found at
nearby stations to its south and west (Fig. 11 of Chap-
ter 2). The analysis of low-level winds showed con-
vergence and upward motions in the Edwardsville area
at all three analysis times (1200, 1400, 1600 CDT) dur-
ing fair weather, although this upward motion was not
as strong as measured over the urban area and over
the southeast bluffs (Fig. 15 of Chapter 3). Several
case studies of individual rain periods seem to show
that wind field patterns were tied primarily to the ur-
ban area, suggesting that physiographic features were
not the controlling factors in boundary-layer flow.
Huff and Vogel (1978) reported an effort to compute
the possible contribution of each of these three topo-
graphic regions to local rain and to use the results to
adjust rainfall values in regions where urban effects
and topographic effects could be superimposed. They
used as a reference (control area) an area in the west-
ern suburbs of St. Louis. Against this they compared
rainfall amounts in 16 other areas thought to contain
topographic and/or urban effects. Several of these
areas are shown in Fig. 29, along with ratios of the
five-season rainfall referenced to that in the control
area. We note that this calculation indicates a topo-
graphic effect of - 9% for the southwest hills, 34% for
108 METEOROLOGICAL MONOGRAPHS

the bottomlands and 24% for the southeast bluffs. If
one takes the latter figure as representing the magni-
tude of bluff effects throughout the network-re-
garded as a very conservative view-the urban effect
on rainfall in the Edwardsville area would be adjusted
approximately 20%. This value is close to the value of
22% found for the St. Louis industrial area where no
topographic effects were expected.
The legitimacy of choosing as a control an area cor-
responding to the minimum area rainfall and located
largely in the western suburbs where other analyses
suggest possible urban effects, was questioned by Bra-
ham (1979). However, these questions related more to
the magnitudes computed for topographic and urban
effects than to the existence of such effects. They also
suggested an opportunity for use of historical data and
data beyond the limits of the project raingage network
for assessing magnitude of topographic and urban ef-
fects.
suppression of, the thermally driven heat island cir-
culation depending upon wind direction. Hjelmfelt
found that at St. Louis the topography enhanced the
downwind thermal dome with southwest winds while
with northwest winds the effect was one of suppres-
sion.
5.9 Postulated mechanisms for explaining urban
precipitation effects
Several mechanisms have been advanced to explain
METROMEX observations of localized rainfall pat-
terns at St. Louis. This section will outline some of
these mechanisms and assess the evidence for and
against them. We will note that none fit all the evi-
dence, nor are they mutually exclusive. It is possible
that each of them plays a role whose magnitude de-
pends on initial conditions.
5.9.1 BOUNDARY LAYER DYNAMICS MECHANISM

5.8.4 THE KASKASKIA RIVER VALLEY It is well known that convective clouds are driven
by the positive buoyancy of their warm updrafts, that
Another physiographic feature which has drawn the
most of the updraft air originates near the earth's sur-
attention of St. Louis area weather forecasters is the
face, and that the thermodynamic parameters of the
Kaskaskia River Valley located just east of the MET-
storm's environment, from cloud base to cloud top,
ROMEX raingage network. This valley slopes upward
set a lower threshold for the properties of the bound-
from southwest to northeast. The north side of this
ary layer air required to sustain convection. In the
valley, along with one of its southward-flowing tribu-
case of large storms, the energetics of the warm up-
taries (Silver Creek), can be seen in the pattern of the
draft may be supplemented by the negative buoyancy
500 ft contour in Fig. 7 of Chapter 1. During south and
of a cold downdraft.
southeast wind conditions, this valley would cause
It also is an axiom of cloud forecasters that thermal
slight upslope motions along the east side of the net-
instability alone is not adequate for causing large con-
work. In Table 4 we noted that 45% of all St. Louis
vective storms. What is needed is some feature of the
area rain comes with southeast surface winds. One
atmospheric dynamics, or of the underlying surface,
wonders how much upslope motion in the Kaskaskia
which can serve to focus or localize the response of
River Valley contributes to this percentage. We also
the atmosphere to the thermal instability and thus
note in Fig. 12 that the first echo frequencies show
cause a few large storms to develop in place of many
secondary maxima along a north-south line centered
small ones.
approximately 25 mi east of downtown St. Louis. One
A number of investigators, both within and outside
wonders whether it is just a coincidence that this line
of METROMEX, have indicated the presence of low-
lies along Silver Creek.
level air convergence and upward air motions over
urban heat islands. The strength of these motions de-
5.8.5 MODEL CALCULATIONS
pends not only on the intensity of the heat island but
Computer simulations with well-tested models which also on the low-level instability of upstream air when
use as input data surface heat and moisture fluxes, it crosses over the urban boundary. Since summer
surface roughness and terrain height, in addition to convective clouds are primarily driven by low-level
upstream air thermodynamic and kinematic proper- instabilities, any effect of the city that can localize and
ties, may ultimately be the most satisfactory way to strengthen boundary layer convergence patterns should
separate urban effects from physiographic effects. Of result in more and larger clouds and consequently in
the several models which have been applied to the St. more rain (perhaps in spite of low Oe values in the
Louis situation, those of Vukovich et al. (1976) and urban area).
Hjelmfelt (1980) speak most directly to the issue of METROMEX observations clearly show low-level
physiographic effects. These authors conclude that wind perturbations at St. Louis. Although details of
thermal forcing is the main cause of urban effects, wind patterns varied greatly with meteorological con-
while topographic features play a secondary role by ditions, the resultant wind field consistently showed
interacting with, and causing an enhancement or patterns of convergence and divergence (see Chapter
METROMEX: A REVIEW AND SUMMARY
3). Between noon and 1600 CDT a definite conver-
gence pattern developed over the metropolitan area-
up to 2.5 x 10-s s- 1 when winds were < 5 m s- 1 and
up to 10- 4 s- 1 when winds were > 5 m s- 1 • The re-
sulting computed vertical motions at the top of the
subcloud layer were -10 em s- 1 over areas of a few
100 km 2 • A study by Shreffler (1978) showed that for
summer conditions at St. Louis the urban circulation
reached its maximum in organization and intensity
during the late afternoon when the inflowing air was
most unstable.
Further details on the kinematics of the urban
boundary layer were obtained from the dual-Doppler
radar measurements (see Chapter 3). From a rather
limited sample it was concluded that a dominant mode
of convection over the city was a horizontal roll
aligned with the wind and locked in position by the
thermal features of the city.
The action of the city in localizing convection in the
planetary boundary layer has been, and still is being,
studied by numerical models. In general these provide
theoretical understanding for the wind flow observa-
tions (see Chapter 3).
On the basis of both observations and theory, we
have documented an interaction between the city and
regional airflow which would tend to localize low-level
convection over and downwind of the city. This effect
maximizes during the afternoon at a time when, on
average, there is a peak in rainfall over downtown St.
Louis and the urban-industrial areas just to the east.
Modification of the dynamics of the boundary layer
seems to offer the most reasonable explanation for the
urban enhancement of convective clouds. Although
the theories and models of urban circulations have not
yet been extended to include wet convection, it seems
important that the urban boundary layer plume is
where radar first echoes, and mergers of radar cells
and of raincells, were found to maximize.
It has been apparent for some time that large con-
vective clouds grow by drawing into the parent storm
a number of smaller clouds that develop on the inflow
side. In the technology of hail suppression seeding,
these clouds are known as feeder clouds. Presumably
cloud elements merge because they are embedded in
a mesoscale convergence region. If so, the increased
frequency of merging over the city follows rather nat-
urally from the thermal and roughness forcing which
brings about convergence of low-level winds over the
urban area.
The most serious problem with the boundary-layer
mechanism is the observation that 8e in the urban
boundary-layer is consistently less than that of rural
areas. This means that urban clouds have less buoy-
ancy available to drive convection at least in the lower
levels. For urban clouds to become dominant over ru-
ral clouds, as the study of radar first echoes (Section
109
5.4.1) and Hi-Cu (Section 5.7.1) suggest, this defi-
ciency in buoyancy must be overcome. Possibilities
for this include pressure work done by the thermally
driven urban circulation, energy from urban enhanced
glaciation and a reduction in entrainment drag because
of increased cloud sizes and/or pre-moistening of the
air around clouds.
5.9.2 WHAT IS THE EFFECT OF LOW (}e?
Early in METROMEX it was observed that some
storms moving across the St. Louis area underwent a
weakening ofthe updraft, lifting of the cloud base, and
decline of the radar top, while at the same time in-
creasing their rate of surface rainfall. Sometimes,
these storms would undergo rejuvenation after reen-
tering rural conditions downwind of the city. These
observations prompted a study by Boatman (1974),
which we will now consider. The equivalent potential
temperature 8e is a derived thermodynamic property
of air that represents the combined sensible and latent
heat energy available for driving convection.
Boatman argued that ingestion of warm, dry air with
relatively low 8e, characteristic of urban areas, causes
a decrease in updraft buoyancy and its ability to carry
previously condensed liquid water. This results in a
premature weakening of the storm and a release of
stored water leading to a local increase in surface rain
rate. General concepts of convective storms would
lead one to expect that the surface rain maximum
should lag the point of initial ingestion of the lower 8e
air by 20-40 min.
Boatman presented detailed case history analyses
for five afternoon storms (3 and 21 August 1972, and
9, 10 and 13 August 1973). All five cases had relatively
light wind shear aloft. Three of the five storms were
associated with cold fronts, and two were classified as
air mass. One storm originated over the city, did not
tap a surface "pool" of low 8e air, and continued to
develop as it crossed the urban area. The other four
storms studied originated upwind of the urban area
and moved over it. In the process, all four storms
tapped regions of lower 8e values 5-8 K below that of
surrounding rural areas. Radar echoes for all four
storms exhibited either decreases in horizontal area or
lowering of echo tops, or both, some 10 to 15 minutes
after arrival of the lower 8e air at cloud base. Accom-
panying the changes in echoes were increases in the
surface rainfall rates.
Fig. 30 summarizes Boatman's findings for the 3
August 1972 case. Measurements in the updraft at
cloud base showed that inflow of lower air was accom-
panied by a rise in the cloud base by about 300 m and
a decrease in updraft speed by about 2 m s- 1 . Other
manifestations of an urban origin for the air were odors
and increased concentrations of Aitken nuclei. Fol-
110 METEOROLOGICAL MONOGRAPHS

SURFAC( 000R

I I I TIME. min I
sus LAMBEU Ft(LO FMEST GAH~AY SOUTHEAST
PARK ARCH RURAL

FIG . 30. Time sections of various thunderstorm intensity parameters , including

o. in the updraft at cloud base, updraft speed and air flux, radar echo height and
rainfall rate . Based upon data from 3 August 1972 case. After Boatman (1974).

lowing the inflow of lower 8e air, the echo top de-
creased in height and there was an increase in surface
rainfall rate.
Three of the four storms continued to dissipate and
showed maximum surface rain intensities after passing
across the city. The fourth case (3 August) rejuvenated
after reentering the downwind rural area.
Acceptance of a 8e mechanism as a primary expla-
nation for rainfall maxima northeast of St. Louis ob-
viously depends first and foremost on the prevalence
of lower 8e air over the city just prior to the passage
of ongoing storms . This point was examined using data
from the surface network. Vogel and Dzurisin (1978)
reported a study of surface 8e patterns for the three-
hour period prior to rain in 45 storms from 1972-75.
Storms selected for this study were distributed over
years and synoptic types, although there was a
"deliberate attempt to choose storms with large rain-
fall amounts in squall line situations'' because it was
felt that urban effects maximized in the heavy-rain
squall-line storms. This resulted in fewer air-mass
storms than was experienced in the total METRO-
MEX data. The 8e departures from network average
and the rainfall patterns for these 45 storms are given
in Fig. 31. We note that the lowest values of 8e are
over the city and western suburbs and over the rural
area across the north. Positive 8e departures are noted
in the bottomland. Although the paths and inflow re-
gions for these storms were not given, the location of
the rainfall maximum east of the city is consistent with
the 8e mechanism if one assumes a typical west to east
storm movement.
Ackerman and Mansell (1978) studied the 8e profiles
through the depth of the boundary layer on a number
of occasions and found significantly lower 8e over the
city. In 23 fair weather cases and 15 pre-rain cases of
1973 and 1975, the 8e deficit over the city was generally
confined to the lowest km (Figs. 32 and 33). (We
should recall that general concepts of convective
storm dynamics call for most of the storm inflow to
METROMEX: A REVIEW AND SUMMARY 111

0
=
0 5 10
STATUTE MILES
em
a.
FIG. 31. Equivalent potential temperature departures and rainfall for 45' selected storms from
1972-75. (a)o. departures from network average for the 3-h period prior to beginning of rain. (b)
Aggregate rainfall from the 45 storms. After Vogel and Dzurisin (1978).

FAIR CASES
5 ~----~----,---~~-----.-----,197~3----~-----.--~rT~----.-----.
ARC-PMQ ARC-BCC
a N = 13 b N = 10

....
>
_)

u.J +
_)

<
:
+
w +
VI
z +
+
+
·~+
""
w
:>::
+
++ ++
+
++
+
....
> ++
+
+
+
+
+
;
+
0
+ +
+ t+ +

"""'
+ + + +
+ +
t-
t t
+
+
+
+

.·•.
:t: + +
~ + + + +
+
w + +
:r + + + +
+
+
+

~1~5~--~~----~----~~----~----.~10 -~1~5----~----~~----~----~----~.~10
URBAN-RURAL 01FFERENCE

FAIR CASES
5 r-----~----,-~~-------.----_,197rs-----.-----,----~r-----.-----.
RXS-NDS
c N= 10 d
.!2
_; 4
w
>
w
_)

<
~ 3 +
+
{

..·.
+

·:
+ +
w +
+ +
>

.
0

"""'
+
+

.·
+
+ +
t-
:t: +
+
+
•
::::! +
+ +

~
UJ +
::r: + .... +
+
+ +

~~15~--~~----~~--~~----~--~.~
10 -~1~5----~----~~--~~----~~--~.~1~
0
URBAN-RURAL DIFFERENCE IN ee• °K

FIG. 32. Average profiles of urban-rural differences in equivalent potential temperature observed
in 23 fair-weather (no-rain) situations. ARC and RXS were the urban stations. After Ackerman and
Mansell (1978).
112 METEOROLOGICAL MONOGRAPHS

PRE-RAIN CASES
1973

ARC -PMQ ARC- BCC

e N• 9 f N = 10

jj 4
_s
..
UJ +
> +
+
..+ +
UJ
--'
<(
UJ
Vl
3 +
+
+
+
+
+
..t
z + +
+ +
c(
w +
+
:0: +
w
> 2
+
+ ..+
0
00
<(
,_ +
+
+
+
+
+
+
......
:r + ... +

.. ..
~ + +
UJ
:r 1
+
+ +
+

+
+
+
+
.. +

++
+

0
.
+
+

- 20 +5 -10 +10
URBAN-RURAL DIFFERENCE IN ee, °K

PRE-RAIN CASES
1975
5
t RXS-NDS
+
+
+ .. +
{ N • 4 h

E 4
+
.. ++ +
+
+
+
-"
.
--'
+
+
+ :+
w +
>
w .t+
-'
+
+
+ +
+
:'5 +
+ +
+
+
+
Vl + +
+
....
:z +
<(
w
"'w ++
+
>
0 +
00
c( +
+
,_ + +
:r
~ + •
+
w
:r •+
+
+
+ +
+
+
0
-20 +5 - 10 +10
URBAN-RURAL DIFFERENCE IN ee , °K

FIG. 33. Average profiles of urban-rural differences in equivalent potential temperature observed
in 15 pre-rain situations. ARC and RXS were the urban stations. After Ackerman and Mansell (1978).

come from the lowest levels of the boundary layer reported that, on average, radar echoes increased in
except in cases where that air is trapped by a strong height as they crossed the urban area (Table 10, page
inversion.) 190). Changnon (l978b) also found that maximum fre-
On balance the ee data speak favorably for the ee quencies of thunderstorms and lightning reached a
mechanism. Other data favorable for this theory are maximum immediately downwind of the city (see
observations that the hail, maximum wind gusts, and Chapters 2 and 8). Moreover, the studies of Hi-Cu
maximum 5 min rainfall rates, often maximized im- (Braham and Wilson, 1978) found an increased fre-
mediately downwind of the city. These weather ele- quency of very large echoes over and one to two hours
ments could reflect a weakened updraft or downdraft of wind travel downwind of the city. The findings
as a result of decreased storm buoyancy. hardly seem compatible with a mechanism calling for
There are other METROMEX observations that are a systematic decrease in intensity of storms crossing
less compatible with this mechanism. Changnon (1976) the city.
METROMEX: A REVIEW AND SUMMARY 113

5.9.3 THE GLACIATION MECHANISM 2511-m is well documented (see Chapter 4). Thus urban
clouds seem to be uniquely favorable for operation of
Another mechanism is one based upon the possibil- the Hallett-Mossop process. Also well documented is
ity that cumulus clouds of the urban boundary layer the existence of urban aerosol particles large enough
might overcome the 8e penalty through urban en- to initiate drop collection without the necessity for a
hanced glaciation. The elements of this mechanism are long period of condensation growth (see Chapter 6).
as follows. Summer convective clouds in the Midwest These large liquid drops, when supercooled, were the
are known to initiate precipitation through collision very ones observed to freeze and initiate the growth
and coalescence of cloud drops. It is well established of rimed particles (Koenig, 1963).
that large drops freeze more readily than small ones; The presence of large concentrations of rimed ice
in fact, that probability of freezing increases exponen- particles is well established as an important aspect of
tially with drop size. Thus, urban clouds have the ad- precipitation development in continental clouds such
vantage over rural clouds in the formation of ice as those at St. Louis in summer.
through cloud-drop freezing. (For more detail about As discussed in the section on Hi-Cu (this chapter),
drop freezing the reader is referred to Mason, 1971; one of the key METROMEX findings to be explained
Rogers, 1976; Bigg, 1953; and Vali and Stansbury, by any theory of urban rain enhancement is the for-
1966.) After large drops in summer continental clouds mation of a unimodal top height frequency distribution
freeze, further growth is through the interception and for urban echoes, in contrast to a bimodal distribution
freezing of cloud drops, a process known as riming. of rural echoes. Obviously, some mechanism operated
Rimed hydrometeors have been observed to develop during the METROMEX period to allow urban clouds
in small summer cumuli of Missouri in concentrations to penetrate the atmospheric condition which fre-
up to 104 m- 3 prior to the time cloud tops reached quently restricted rural clouds to maximum top heights
-10°C (Koenig, 1963; Braham, 1964; Hoffer and Bra- of ~6 km (temperature ~ -10°C). (See p. 104 for fur-
ham, 1962). ther discussion ofthis point.) Preferential glaciation in
One of the recent developments in cloud physics urban clouds was one of two possibilities discussed by
has been the discovery that the process of riming can Braham and Wilson (1978) for causing this additional
produce secondary ice particles (splinters) provided growth of urban clouds.
that the temperature is between ~- 3 and -soc, and Another major finding of METROMEX was the
that the cloud contains drops larger than 25 11-m in downwind maximum of thunderstorms and hail
diameter. A large number of drops less than 13 11-m in (Changnon, 1978b). Scientists generally agree that
diameter also was found to enhance splintering (Mos- thunderstorm charging requires a mixed phase cloud
sop and Hallett, 1974; Hallett and Mossop, 1974). All containing supercooled cloud drops and riming hydro-
of these criteria were met in the summer clouds of meteors. Riming hydrometeors are the embryos, or
METROMEX. early growth phase, of all hailstones. Thus the findings
These secondary ice particles in tum initiate freez- by Changnon of a maximum in thunderstorm frequen-
ing of all drops which they contact. The end product cies, electric power outages to lightning, hail and crop-
is the rapid conversion to a totally glaciated cloud, hail loss-cost values east and northeast of St. Louis
unless a vigorous updraft supplies water vapor at a (see Chapter 8) are all compatible with increased de-
rate greater than can be consumed by deposition on velopment of the ice phase in urban clouds.
the ice particles. Koenig (1963) describes observations On the other hand, there are two observations which
of small cumuli in Missouri which completely glaciated do not favor the glaciation mechanism. Ice nucleus
in only a matter of minutes. measurements using membrane filters and the expan-
The creation of ice in a supercooled cloud liberates sion chamber were made in great detail on METRO-
heat of fusion. The more ice created, the more heat MEX (see Chapter 6). These generally indicated fewer
released. When this heat release occurs in an updraft, ice-forming nuclei in urban areas, with a few specific
that updraft will be strengthened resulting in a larger localized nucleus sources in the industrial activities
and more vigorous cloud. This chain of reasoning along the Mississippi River. The University of Chicago
gives one a plausible theory explaining why urban group made an effort to detect differences in the ice
clouds could be larger and more vigorous than rural particle content of urban and rural clouds but had little
clouds-local rainfall maxima downwind of the city success. The reason given was the limited capability
would be a logical result. of available sampling equipment.
METROMEX observations reported in other sec-
tions offer several findings in support of, or at least 5.9.4 WHAT IS THE ROLE OF LOCAL TOPOGRAPHY?
consistent with, the glaciation theory. The urban ex- The numerical modeling studies of Vukovich et al.
cess of cloud drops in sizes less than 13 11-m and over (1976) and Hjelmfelt (1980) clearly show that local ter-
114 METEOROLOGICAL MONOGRAPHS

rain affects local-area boundary layer convection, and - - , 1978b: Urban effects on severe local storms at St. Louis. J.
presumably local precipitation patterns. These studies Appl. Meteor., 17, 578--586.
- - , 1979: Rainfall changes in summer caused by St. Louis. Sci-
agree that topographic effects are not the major cause
ence, 205, 402-404.
of rain enhancement of St. Louis. Patterns of land use --,D. M.A. Jones and F. A. Huff, 1975: Precipitation increases
which give rise to differences in surface heat and mo- in the low hills of southern Dlinois: Part 2, Field investigation
mentum fluxes are thought to be much more impor- of anomaly. Mon. Wea. Rev., 103, 830-836.
tant. However, both studies also show that terrain ef- Dettwiller, J., 1970: Incidence possible de l'activite industrielle sur
les precipitations a Paris. Proc. WMO Symp. Urban Climates
fects at St. Louis tend to cause a focusing and
and Building Climatology, Brussels, 361-362.
augmentation of boundary layer effects northeast of Dungey, M., 1977: The effects of St. Louis on first echo frequencies.
St. Louis and a reduction in boundary layer effects Preprints Sixth Conf. Inadvertent and Planned Weather Mod-
southeast of St. Louis. Possible effects of topography ification, Champaign-Urbana, Amer. Meteor. Soc., 49-52.
in localizing the nighttime storms along the north edge - - , T. R. Morris and R. R. Braham, 1974: Radar data analyses in
of the raingage network have not been carried out. METROMEX, 1971-72-73. Preprints Fourth Conf. Weather
Modification, Ft. Lauderdale, Amer. Meteor. Soc., 358--361.
Dytch, H. E., and D. B. Johnson, 1977: Urban influences on warm
cloud microstructure. Preprints Sixth Conf. Inadvertent and
REFERENCES
Planned Weather Modification, Champaign-Urbana, Amer.
Ackerman, B., 1978: Regional kinematic fields. Summary of MET- Meteor. Soc., 37-40.
TROMEX, Vol. 2, Causes of Precipitation Anomalies, Illinois Fitzgerald, J. W., 1972: A study of the initial phase of cloud droplet
State Water Survey Bull. 63, Urbana, 165-205. growth by condensation: Comparison between theory and ob-
- - , and J. W. Mansell, 1978: Thermodynamic structure of the servation. Ph.D. dissertation, University of Chicago, 144 pp.
PBL at midday. Summary of METROMEX, Vol. 2, Causes of [NTIS PB 211522].
Precipitation Anomalies, Illinois State Water Survey Bull. 63, --,and P. A. Spyers-Duran, 1973: Changes in cloud nucleus con-
Urbana, 129-164. centrations and cloud droplet size distribution associated with
Auer, A. H., Jr., 1978: Correlation of land use and cover with me- pollution from St. Louis. J. Appl. Meteor., 12, 511-516.
teorological anomalies. J. Appl. Meteor., 17, 636-643. Grosh, R. C., 1978a: Satellite-observed urban cloud distributions.
Bigg, E. K., 1953: The formation of atmospheric ice crystals by the Summary of METROMEX, Vol. 2, Causes of Precipitation
freezing of droplets. Quart. J. Roy. Meteor. Soc., 79, 510-519. Anomalies, Illinois State Water Survey Bull. 63, Urbana,
Boatman, J. F., 1974: Inadvertent thunderstorm modification by an 229-235.
urban area. M.S. thesis, Dept. Atmos. Sci., University of Wy- - - , 1978b: Studies of anomalous cumulus clouds. Summary of
oming, Laramie, 126 pp. METROMEX, Vol. 2, Causes of Precipitation Anomalies, Il-
Braham, R. R., Jr., 1952: The water and energy budgets of the linois State Water Survey Bull. 63, Urbana, 212-228.
thunderstorms and their relation to thunderstorm development. Hallett, J., and S. C. Mossop, 1974: Production of secondary ice
J. Meteor., 9, 227-242. particles during the riming process. Nature, 249, 26-28.
- - , 1964: What is the role of ice in summer rain-showers? J. Hjelmfelt, M., 1980: Numerical simulation of the effects of St. Louis
Atmos. Sci., 21, 640-645. on boundary layer airflow and convection. Ph.D. dissertation,
- - , 1979: Comment on urban, topographic and diurnal effects on University of Chicago, 185 pp.
rainfall in the St. Louis region. J. Appl. Meteor., 18, 371-375. Hoecker, W. H., and J. K. Angell, 1969: Effect of a sudden change
--,and M. J. Dungey, 1976: Effects of a large city on convective in terrain height on the three-dimensional low-level air flow, as
clouds and coalescence rain. Preprints Int. Conf. Cloud Phys- estimated from tetroon flights. Mon. Wea. Rev., 97, 845-849.
ics, Boulder, Amer. Meteor. Soc., 275-278. Hoffer, T., and R. R. Braham, Jr., 1962: Laboratory study of atmo-
--,and--, 1978: A study of urban effects on radar first echoes. spheric ice particles. J. Atmos. Sci., 19, 232-235.
J. Appl. Meteor., 17, 644--654. Huff, F. A., 1975: Urban effects on the distribution of heavy con-
- - , - - , T. R. Morris, S. A. Changnon and F. A. Huff, 1975: A vective rainfall. Water Resour. Res., 11, 889-896.
summary of Project METROMEX radar findings. Preprints - - , 1977: Diurnal distribution of summer rainfall. Summary of
16th Radar Meteorology Conf., Houston, Amer. Meteor. Soc., METROMEX, Vol. 1, Weather Anomalies and Impacts, Illi-
404-409. nois State Water Survey Bull. 62, Urbana, 45-52.
--,and D. Wilson, 1978: Effects of St. Louis on convective cloud - - , 1978: Radar analysis of urban effects on rainfall. Summary of
heights. J. Appl. Meteor., 17, 587-592. METROMEX, Vol. 2, Causes of Precipitation Anomalies, Il-
Byers, H. R., and R. R. Braham, Jr., 1949: The Thunderstorm. linois State Water Survey Bull. 63, Urbana, 265-273.
U.S. Weather Bureau, Washington, DC [Govt. Printing Of- --,and S. A. Changnon, Jr., 1972: Climatological assessment of
fice], 287 pp. urban effects on precipitation at St. Louis. J. Appl. Meteor.,
Changnon, S. A., Jr., 1968: Climatology of hourly occurrences of 11, 823-842.
selected atnospheric phenomena in Illinois. Illinois State Water - - , S. A. Changnon, Jr., and D. M. A. Jones, 1975: Precipitation
Survey Circ. 93, Urbana, 28 pp. increases in the low hills of southern Illinois. Part 1, Climatic
- - , 1974: RHI First echo study. Interim Report on METROMEX and network studies. Mon. Wea. Rev., 103, 823-829.
Studies, 1971-1973, F. A. Huff, Ed., lllinois State Water Sur- --,and P. T. Schickedanz, 1970: Rainfall evaluation studies. Final
vey, Urbana, 71-76. Report, Part II, NSF Grant GA 1360, Dlinois State Water Sur-
- - , 1976: Effects of urban areas and echo merging on radar echo vey, Urbana, 224 p.
behavior. J. Appl. Meteor., IS, 561-570. --,and J. L. Vogel, 1978: Urban, topographic and diurnal effects
- - , 1978a: Vertical ch~acteristics and behavior of radar echoes. on rainfall in the St. Louis region. J. Appl. Meteor., 17,
Summary of METROMEX, Vol. 2, Causes of Precipitation 564--577.
Anomalies, Illinois State Water Survey Bull. 63, Urbana, Johnson, D. B., 1976: Ultragiant urban aerosol particles. Science,
274--279. 194, 941-942.
METROMEX: A REVIEW AND SUMMARY
- - , 1978: Giant nuclei and warm rain initiation. Preprints Conf.
Cloud Physics and Atmospheric Electricity, Issaquah, Amer.
Meteor. Soc., 31-34.
- - , 1979: The role of coalescence nuclei in warm rain initiation.
Ph.D. dissertation, University of Chicago, 119 pp. [Released
as Tech. Note No. 55, Cloud Physics Lab., University of Chi-
cago.]
- - , and M. J. Dungey, 1978: Microphysical interpretation of radar
first echoes. Preprints 18th Conf. Radar Meteorology, Atlanta,
Amer. Meteor. Soc., 117-120.
Jones, D. M.A., F. A. Huff and S. A. Changnon, Jr., 1974: Causes
for precipitation increases in the hills of southern Illinois. Rep.
74-RI75, Illinois State Water Survey, Urbana, 36 pp.
Koenig, L. R., 1963: The glaciating behavior of small cumulonimbus
clouds. J. Atmos. Scie., 20, 29-47.
Landsberg, H. E., 1974:. Man-made climatic changes Proceedings
of Symposium on Physical and Dynamical Climatology, Len-
ingrad, World Meteorological Organization, Rep. 347, 262-303.
Leichter, 1., 1974: Moisture !tux and precipitation studies of con-
vective storms in western South Dakota using pibal and radar
data. Rep. 74-11, Bureau of Reclamation, Contract 14-06-D-
6796, South Dakota School of Mines and Technology, Rapid
City, 63 pp.
Mason, B.)., 1971: The Physics of Clouds. Clarendon Press, Ox-
ford, 671 pp.
Means, L. L., 1944: The nocturnal maximum occurrence of thun-
derstorms in the midwestern United States. Department of
Meteorology, University of Chicago, Misc. Rep. 16, 37 pp.
Mossop, S. A., and J. Hallett, 1974: Ice crystal concentration in
cumulus clouds: Influence of the drop spectrum. Science, 186,
632-633.
Mueller, S. F., 1978: An application of the Simpson-Wiggert cloud
model to METROMEX Hi-Cu data. Preprints Conf. Cloud
Physics and Atmospheric Electricity, Issaquah, Amer. Meteor.
Soc., 490-495.
Ochs, H. T., III, 1974: Cloud modeling in METROMEX. Preprints
4th Conf. Weather Modification, Ft. Lauderdale, Amer. Me-
teor. Soc., 396-400.
- - , 1978: Cloud modeling. Summary of METROMEX, Vol 2,
Causes of Precipitation Anomalies, lllinois State Water Survey
Bull. 63, Urbana, 240--264.
--,and D. B. Johnson, 1979: Urban effects on the properties of
radar first echoes. J. Appl. Meteor, 19, 1160-1166.
--,and R. G. Semonin, 1976: Microphysical computations in ur-
ban and rural clouds. Preprints Int. Conf. Cloud Physics, Boul-
der, Amer. Meteor. Soc., 22-26.
- - , and--, 1977: The sensitivity of cloud microphysics to an
urban environment. Preprints Sixth Conf. Inadvertent and
Planned Weather Modification, Champaign-Urbana, Amer.
Meteor. Soc., 41-44.
- - , and--, 1979: Sensitivity of a cloud microphysical model to
an urban environment. J. Appl. Meteor., 18, 1118-1129.
--,and C. S. Yao, 1978: Moment-conserving techniques for warm
cloud microphysical computations. Part 1: Numerical tech-
niques. J. Atmos. Sci., 35, 1947-1958.
Rogers, R. R., 1976: A Short Course in Cloud Physics. Pergamon
Press, 227 pp.
Russell, P. B., E. E. Uthe, F. L. Ludwig and N. A. Shaw, 1974:
A comparison of atmospheric structure as observed with mon-
115
ostatic acoustic sounder and lidar techniques. J. Geophys.
Res., 79, 5555-5566.
Schickedanz, P. T., 1972: The raincell approach to the evaluation
of rain modification experiments. Preprints Third Conf.
Weather Modification, Rapid City, Amer. Meteor. Soc., 88-95.
- - , 1973: A statistical approach to computerized rainfall patterns.
Preprints Third Conf. Probability and Statistics in Atmospheric
Science, Boulder, Amer. Meteor. Soc., 4 pp.
- - , 1974: Inadvertent rain modification as indicated by surface
raincells. J. Appl. Meteor., 13, 891-900.
- - , 1978: Surface raincell analysis. Summary of METROMEX,
Vol. 2, Causes of Precipitation Anomalies, Illinois State Water
Survey Bull. 63, Urbana, 280--341.
- - , and M. B. Busch, 1975: Data processing and analytical pro-
cedures for urban precipitation studies. Proc. Nat. Symp. Pre-
cipitation Analysis for Hydrologic Modeling, Davis, Amer.
Geophys. Union, 101-110.
Semonin, R. G., 1975: Isolated storms of 14 July 1973. Studies of
selected precipitation cases from METROMEX, S. A. Chang-
non and R. G. Semonin, Eds. Illinois State Water Survey Re-
port oflnvestigation 81, Urbana, 68-89.
Shreffler, J. H., 1978: Detection of centripetal heat-island circula-
tions from tower data in St. Louis. Bound.-Layer Meteor., 15,
229-242.
Simpson, J., and V. Wiggert, 1969: Models of precipitating cumulus
towers. Mon. Wea. Rev., 97, 471-489.
- - , and - - , 1971: 1968 Florida cumulus seeding experiment:
Numerical model and results. Mon. Wea. Rev., 99, 87-118.
Spangler, T. C., 1972: Dynamically induced meso-scale variations
of the urban mixing layer depth. Rep. AR104, Dept. Atmos.
Resour., University of Wyoming, Laramie, 30 pp.
Summers, P., 1966: The seasonal, weekly and daily cycles of atmo-
spheric smoke content in central Montreal. J. Air Pollut. Con-
trol Assoc., 16, 432-438.
Ulanski, S. L., and M. Garstang, 1978: The role of surface diver-
gence and vorticity in the life cycle of convective rainfall. J.
Atmos. Sci., 35, 1047-1069.
Vali, G., and E. J. Stansbury, 1966: Time-dependent characteristics
of the heterogeneous nucleation of ice. Can. J. Phys., 44,
477-502.
Vogel, J. L., 1977: Synoptic weather relations. Summary of MET-
ROMEX, Vol. 1, Weather Anomalies and Impacts, Illinois
State Water Survey Bull. 62, Urbana, 85-112.
- - , and G. Dzurisin, 1978: Equivalent potential temperature pat-
terns associated with rainstorms. Summary of METROMEX,
Vol. 2, Causes of Precipitation Anomalies, Illinois State Water
Survey Bull. 63, Urbana, 63-70.
- - , and F. A. Huff, 1977: Relation between surface winds, storm
movement, and rainfall. Summary of METROMEX, Vol. I,
Weather Anomalies and Impacts, Illinois State Water Survey
Bull. 62, Urbana, 61-66.
Vukovich, F. M., J. W. Dunn III and B. W. Crissman, 1976: A
theoretical study of the St. Louis heat island: The wind and
temperature distribution. J. Appl. Meteor., 15, 417-440.
Wallace, J. M., 1975: Diurnal variations in precipitation and thun-
derstorm frequency over the conterminous United States. Mon.
Wea. Rev., 103, 406-419.
Wong, K. K., and R. A. Dirks, 1978: Mesoscale perturbations on
airflow in the urban mixing layer. J. Appl. Meteor., 17, 677-688.
 CHAPTER 6

Atmospheric Chemistry and Source-Receptor Relationships

JEREMY M. HALES

CONTENTS

ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
6.2 Emission inventory and source identification ........................................... 118
6.3 Atmospheric aerosols and their gaseous precursors ...................................... 118
6.3.1 AEROSOLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Aitken Condensation Nuclei. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Cloud Condensation Nuclei ............................................. 121
Ice Forming Nuclei .................................................... 121
Aerosol Particle Size Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
6.3.2 GASEOUS PRECURSORS AND AEROSOL COMPOSITION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6.3.3 VISIBILITY AND HAZE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
6.4 Dry deposition and dustfall .......................................................... 129
6.5 Precipitation scavenging ............................................................ 130
6.5.1 SCAVENGING OF AIR POLLUTANTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
6.5.2 SCAVENGING OF TRACERS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
6.6 Modeling of convective-storm scavenging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138

ABSTRACT

The chemical and physical attributes of pollutants and potential weather-modification agents in and around
the St. Louis/METROMEX region were carefully studied. Both airborne and precipitation-borne pollutants
were examined with special emphasis on the transformation and natural-removal processes, which occurred as
these materials drifted downwind of the metropolitan area. Precipitation scavenging of airborne pollutants was
of particular interest owing to its intrinsic relationship with many potential weather modification processes.
A definite urban plume of Aitken condensation nuclei was emitted by the St. Louis area. The general urban
plume is a composite of individual sources, and often was characterized by concentrations of several tens of
thousands of nuclei per cubic centimeter. The urban plume tended to approach regional concentrations after
4-5 h of downwind travel, an effect which obstensibly was caused primarily by a combination of coagulation
and dilution mechanisms. The St. Louis area was also a general source of CCN during all seasons of the year.
In contrast, the city seemed to act as a sink or "deactivator" for ice-formation nuclei, at least during winter
months. Importantly, aerosol particles in the accumulation mode tended to increase in concentration immedi-
ately downwind of the urban complex, and with subsequent decreases after about 3 hours of travel time.
Importantly, sulfur dioxide emitted from the metropolitan area on clear days was removed by a combination
of dry deposition and chemical conversion. Chemical reaction rates within the urban plume were typically of
the order of 10% per hour. Thunderstorms were effective scavengers of St. Louis-emitted SOx and NOx, and
a definite urban deposition pattern occurred immediately down-storm of the metropolitan area. Free-hydrogen
ion deposition patterns showed no easily discernible evidence of the city's presence, however. Convective
storms were also effective scavengers of particulate tracer materials released into clouds. Tracer deposition
patterns resulting from these tests testify to the complexity of flow patterns in convective-storm systems.

117
118 METEROLOGICAL MONOGRAPHS

6.1 Introduction
A number ofMETROMEX studies dealt specifically
with pollutants, including their sources, transport,
transformation and deposition. There were several
reasons why these pollutant-related studies were per-
formed under METROMEX. These included:
• Inadvertent weather modification effects were ex-
pected to be caused to some extent by air pollu-
tion; thus aerosol and trace-gas characterization
is an important requirement for the basic pro-
gram.
• The extensive atmospheric data base generated
by the weather-modification research efforts at
METROMEX provided an attractive substrate
for the simultaneous conduct of independent air
pollution research studies, even though some of
these studies had no direct association with
weather-modification processes.
• Many potential weather-modification processes KILOMETERS

are intrinsically related to pollutant fate (e.g. pre- 0 10 20 30 40 50

F3 E*'i E"'3
cipitation scavenging); thus, the simultaneous
performance of such studies generally provided FIG. I. Counties of the Metropolitan St. Louis interstate air qual-
ity control region (shaded). After Littman (1978).
substantial benefit on an interpretive basis.
The purpose of this chapter is to describe these pol-
terization of background and urban-induced pollution
lutant-related studies. The material is arranged here in
were conducted as components of the METROMEX
a progression as the pollutant moves from its sources
effort. Most of these studies involved aircraft sampling
to its receptors, experiencing the sequence of events
(cf. Table 7, Chapter 1), although a number of ground-
that lead ultimately to its removal from the atmo-
based measurements were performed as well. These
sphere.
studies are classified in the following text according to
6.2 Emission inventory and source identification the type of aerosol or gaseous precursor.
Emission assessments were not performed as a part
of the METROMEX program. Because of the basic 6.3.1 AEROSOLS
importance of emission sources in the context of pol-
lutant transport and fate studies, however, it is appro- Aitken Condensation Nuclei
priate at this point to discuss briefly the distributions One aspect of understanding the air mass modifi-
and magnitudes of emissions within the study area. cation induced by an urban area is the determination
Several emission inventories have been prepared for of condensation-nucleus input to the air mass during
the St. Louis area since the early 1960' s. The most its lifetime over an urbanized-industrialized area. Ex-
current and useful inventory for present purposes, tensive mappings of Aitken Condensation Nuclei
however, is that prepared by Littman (1978) as part of
the EPA Regional Air Pollution Study (RAPS). This
inventory pertains to the St. Louis Air Quality Control
TABLE I. Total emissions for the St. Louis air quality control
Region, shown in Fig. 1. Annual emissions, subdi- region. Values are in tons per year and percent of regional total.
vided into point-source and area-source categories ,
are shown for specific pollutants in Table 1. For a Particulate so. NOx* HC co
comprehensive discussion, including temporally and Point 45 224 I 007 530 322 730 47 610 164 331
spatially resolved emission rates of pollutants, the sources (3%) (97%) (72%) (23%) (II%)
reader is encouraged to consult Littman's original re- Area l 299 782 30 813 125 567 157 204 l 325 556
port. sources (97%) (3%) (28%) (77%) (89%)
Regional l 345 006 l 038 334 448 297 204 814 l 489 887
6.3 Atmospheric aerosols and their gaseous precursors total

Numerous individual studies addressed to charac- *Calculated as NOz.

METROMEX: A REVIEW AND SUMMARY
(ACN) have been acquired by various METROMEX
investigators. 1
From these investigations, a consistent description
of the general characteristics of ACN upwind, over
and downwind of the St. Louis metropolitan area in
the summer months emerges. First, the regional back-
ground (rural) concentration of ACN ranges from
5000-7000 cm- 3 • A general urban plume, with a width
comparable to the metropolitan source region and
identifiable by ACN concentrations greater than 10 000
cm- 3 , exists in the afternoon hours. Within the general
urban plume, distinct localized plumes with concen-
trations greater than 30 000 cm- 3 can be identified from
industrial sources within and around metropolitan St.
Louis. These characteristics are demonstrated in Figs.
2 and 3 showing representative examples of CN
plumes. Braham (l976a) believes that Fig. 2 is fairly
typical of most days when ACN plumes from many
point sources can be identified before they merge into
a more general urban plume. Shea and Auer (1978), in
comparing urban plume widths (with concentrations
~ 10 000 cm- 3 ) with the diameter of the metropolitan
'Auer, 1975a, 1977a, b; Garrett, 1975; Sisterson, 1975 ; Braham,
1976a; Shea, 1976; Wong, 1976; Auer and Komp, 1977; Komp and
Auer, 1977, 1978a, b; Semonin, 1978; Semonin and Changnon, 1974;
Shea and Auer, 1978.
762 m, 1300 COT, 3 Aug 74
FIG. 2. AW plumes from flight data, 3 August 1974. In upper right are mea-
surements for so. overburden as obtained from ground vehicle . After Braham
(1976a).
119
area and its uniform layout of arterial streets and high-
ways, point to the automobile as the principal anthro-
pogenic contributor to ACN in the general urban
plume. The data in Fig. 2 are also especially interesting
in that they clearly show the association of ACN with
S0 2 from two large power plants and a refinery com-
plex northeast of St. Louis.
According to Auer and Changnon (1977), the lateral
spreading 8 (degrees), of the general Aitken nucleus
plume is related to the height H (m) of the capping
stable air in the upwind rural regions by the relation-
ship 8 = 24.3 - 1.48 x w-z H. For rural mixing
heights > 1500 m, little widening of the ACN plume
downwind of the metropolitan area occurs. Average
midday mixing heights in St. Louis during June, July,
and August are 1407, 1901, and 1484 m, respectively.
Thus, on summer days without appreciable horizontal
wind shear in the mixing layer, only small downwind
lateral spreading ( <3°) of the urban Aitken nucleus
plume is expected (Auer and Changnon, 1977; Shea
and Auer, 1978).
Distinct plumes have been detected from smelters,
refineries, chemical plants and coal-fired power plants.
The Alton-Wood River industrial-refinery complex
produced narrow plumes with Aitken nucleus concen-
trations exceeding 40 000 cm- 3 (Fig. 3). These plumes,
Aitken Nucleus
Concentrations
120
10
KILOMETER>
0 10 20 30
FF""S3
AUGUST 9, 1976
FJG . 3. ACN concentrations ( 103 cm- 3 ) at 450 m MSL between
1400 a nd 1800 COT. After Komp and Auer (1978b).
as well as definite refinery odors, have been detected
as far as 65 km from the stacks. Coal-fired power
plants located around the metropolitan area (all of
which are fitted with electrostatic precipitators and
whose maximum electrical outputs range from 500 to
2400 MW) produced ACN plumes with concentrations
up to 50 000 cm- 3 at distances up to 30 km. Fig. 4
shows the plume of ACN from the Labadie Power
Plant (2400 MW) as mapped during the afternoon of
9 July 1975. In this case, the plume was traceable with
concentrations well above background out to dis-
tances of over 100 km. Computation of ACN fluxes
showed essentially constant values from 20 to 50 km,
or in terms of wind travel, up to - two hours. Beyond
50 km there was a gradual decrease in the flux of ACN
out to about 80 km (Braham, 1977).
Braham (1977) suggests that in a neutrally buoyant
atmosphere, such as prevails on summer days at St.
Louis, the ACN plumes mix vertically through the
depth of the mixing layer and spread horizontally
- 10°. Other METROMEX studies found that the ACN
METEROLOGICAL MONOGRAPHS
Fl ight Path
plumes emanating from 1800-2400 MW fossil-fueled
power plants spread horizontally - 16° up to down-
wind distances of 25 km (or wind travel times of
- l.5h) in neutral stabilities (Auer and Komp, 1977;
Komp and Auer, 1978a). Beyond this distance, the
plume width remained unchanged at 7-10 km out to
at least 50 km.
The maximum concentrations of ACN from large
power plants (2000 MW capacity utilizing electrostatic
precipitators with > 99.5% efficiency) were found to
exceed midwestern regional background concentra-
tions for up to 4 hours transport time downwind. For
the medium-sized power plants (- 1000 MW capacity)
the time decreased to - 2 hours. Smaller power plants
( < 500 MW) induce insignificant perturbations on the
Aitken nucleus loading of the regional background
concentrations. Often, the plumes from these plants
are completely obscured by the ACN output of the
general metropolitan area.
While St. Louis and the adjacent Alton-Wood River
industrial complex produced distinct plumes out to 100
METROMEX: A REVIEW AND SUMMARY
KILOMETERS
0 10 20
JULY 9, ! 975
FIG. 4. The plume of ACN from the Labadie Power Station during the afternoon
of 9 July 1975. After Braham (1976a).
km or more on some days, the longer plume length
appeared to be related to stronger mixing-layer winds
during neutral stabilities (Shea and Auer, 1978). The
processes of ACN formation appear to be very strong
in the first hour or two of travel time from the centroid
of urban sources, as concentrations of ACN maximize
to 4-7 fold over concentrations found in nearby rural
areas (Auer, 1975, 1977b; Garrett, 1975; Semonin ,
1978; Semonin and Changnon, 1974; and White, et al. ,
1978). However, depletion of ACN , presumably by
self-coagulation and/or dispersion, dominates over the
production of ACN after - 2 h and continues such that
ACN concentrations return to regional background
values after 4-5 h of travel time beyond the urban
source region (Braham, 1977; Komp and Auer, 1978b;
and Shea and Auer, 1978).
There is no doubt that urban, industrial, and other
human activities supply a copious quantity of particles
to the atmosphere. Auer (1975) has prepared maps of
metropolitan St. Louis showing preferred areas of
ACN production from surface sources; he concludes
that the integrated metropolitan production rate is - 44
x 1()4 nuclei cm- 2 s- 1 • For the most part, the areas
with highest production rates correspond to the areas
of greatest industrial activity (also Semonin, 1978),
while some portions of the residential suburb show
very little enhancement of Aitken nucleus production
121
AITKEN NUCLEUS
CONCENTRATIONS
over the surrounding rural countryside. The two high-
est local production rates, > 2 x 106 nuclei cm- 2 s- 1 ,
were found over a small areal extent corresponding to
the site of the only two steel plants in the metropolitan
area.
Vertical distributions of ACN showed little varia-
tion, with point concentrations remaining within a fac-
tor of three of the means within the well-mixed layer
for rural, urban and downwind regions (Auer, 1975;
White, 1977).
Cloud Condensation Nuclei.
Aerosol particles capable of serving as condensation
nuclei at supersaturations of less than 1% are com-
monly referred to as Cloud Condensation Nuclei
(CCN) . Their concentration is an important factor in
determining the initial size and concentration of the
cloud droplet spectra. Atmospheric CCN concentra-
tion is an important factor in cloud formation and pre-
cipitation processes, and because of this detailed stud-
ie s of CCN modification were undertaken within
METROMEX. The results of these measurements
during summer months have been summarized by
Spyers-Duran (1974) and Auer (1975); in addition,
Spyers-Duran (1975) had described CCN measure-
ments taken in the winter months of 1972 and 1973.
122
A somewhat consistent description of the summer-
time CCN population upwind, over and downwind of
the metropolitan St. Louis area has emerged. Concen-
trations of CCN at activating supersaturations from
0.3 to 1.5% were derived and expressed in the form
N = c(s)k where N is the CCN concentration (nuclei
per cm3 ) at supersaturation s. The constants c and k
determine the properties of the spectra. Table 2 shows
the CCN spectra for summer months. It should be
pointed out that there is considerable variation in the
average concentration of CCN shown in Table 2; in all
of the daily input data pairs, however, the downwind
concentrations of CCN exceeded the upwind concen-
trations. On the average, the downwind CCN concen-
trations of s = 1% were 61% higher than the upwind
samples and at s = 0.3% about 58% higher, indicating
that the city acts as a source for CCN of all activation
supersaturations.
The University of Chicago summer 1972 data show
a somewhat lower enrichment in the downwind sam-
ples; this is thought to be due to the fact that these
samples represent averages through the depth of the
mixing layer, whereas samples during other flight pe-
riods were taken at a constant altitude within the mix-
ing layer.
Spyers-Duran (1974) reported CCN spectra for win-
ter months. For the 1972 data the mean spectra were
N = 848s 0 •83 upwind, and N = 1678s 0 ·97 downwind.
These give downwind increases of CCN of 98% at 1%
and 49% at 0.3% supersaturation. For the 1973 winter
data the upwind spectrum was N = 977s 0 · 96 and the
downwind spectrum was N = 1451sL07 • This gives
downwind increases of CCN of 67% at 1% and 30%
at 0.3% supersaturation. The two sets of winter spec-
tra are surprisingly similar and indicate lower CCN
concentrations than any of the summer spectra.
TABLE 2. Average upwind (UW) and downwind (DW) CCN supersaturation spectra observed in Project METROMEX.
Number of
Season Location comparisons
Summer 1971 uw 19
DW
Winter 1972 uw 13
DW
Summer 1972 uw 23
DW
March-April 1973 uw 10
DW
Summer 1973 uw ? N = 1356s 0 ·58
DW
August 1971-73 uw 17
DW
METEROLOGICAL MONOGRAPHS
For both summer and winter measurements, the
percent of downwind increase in CCN concentration
is greatest at the largest supersaturations, indicating
relatively more urban nuclei of lower activity. The
downwind increases at large supersaturations were
more pronounced in winter than in summer, indicating
that the concentrations of low-activity nuclei in the
urban plume are even more pronounced in winter than
in summer.
The pattern of production of CCN for summer
months has been developed for the metropolitan area
(Auer, 1975). The areas with the greatest CCN pro-
duction rates correspond with the location of the in-
dustrialized portions of the metropolitan area, espe-
cially in the southeastern part. The most copious
producer of cloud nuclei appears to be the Granite
City, Illinois, area at which there is located a steel
plant. The lowest CCN production rates were found
principally in the residential areas located in the south-
western, western, and northern portions of the met-
ropolitan area. For the metropolitan area, the inte-
grated average CCN source strengths are 1 to 4 x I0- 4
cm- 2 s- 1 for activating supersaturations of 0.5 and 1%
(Spyers-Duran, 1974; Auer, 1975). Urban production
rates of CCN are about twice as large in summer as
in winter (Spyers-Duran, 1974).
Limited CCN measurements in the urban plume at
distances out to 100 km downwind have been ac-
quired. Spyers-Duran (1974) reported that the highest
concentrations of CCN active at 1% supersaturation
were found immediately downwind of St. Louis. On
some occasions a secondary maximum appeared at
50-55 km downwind. Also noteworthy were measure-
ments over Greenville Airport ( ~65 km downwind of
St. Louis) which occasionally showed higher concen-
trations of CCN than either the upwind or downwind
CCN
concentration
spectra
(nuclei cm-3 ) Reference
N = 2210s 0 ·69 Braham (1974a)
N = 3890s 0 · 75
N= 848s 0 ·83 Spyers-Duran (1974)
N = 1678s 0 ·97
N = 1167s 0 ·63 Spyers-Duran (1974)
N = 1625s 0 ·68
N = 977s 0•96 Braham (1974a)
N = 1451s 1· 08
Braham (1974a)
N = 2149s 0 ·53
N = 1300so.s2 Auer (1975)
N = 2200s 0 ·57
METROMEX: A REVIEW AND SUMMARY
samples near the St. Louis urban area. Increases over
regional upwind values (930 cm-3 ) in the number of
CCN activated in the broad general downwind plume
of St. Louis have been reported to peak at 77% at -3
h wind travel time (Auer, 1977b).
The vertical distribution of CCN observed at and
over coincident urban locations shows little variation
in the vertical within the first 500 m above ground level
during neutral stabilities (Auer, 1975a). Vertical pro-
files of CCN concentrations immediately downwind of
the city, through a greater depth of the mixing layer,
show that CCN concentrations increase somewhat
with height, reaching the highest concentrations about
the middle of the mixed layer (Spyers-Duran, 1974).
Spyers-Duran (1974) has employed the Twomey equa-
tion (Twomey, 1959) in conjunction with multi-super-
saturation CCN data obtained from METROMEX to
calculate the expected concentrations of drops in
clouds forming on these nuclei. These results were
compared with concentrations of CCN at various sin-
gle supersaturations. Results based upon winter 1973
data (Spyers-Duran, 1974) show that a thermal diffu-
sion device operating at a supersaturation of 0.7%
would have overestimated cloud drop concentrations
95% of the time with an average error of 58%. The
same device, if used to quantify the urban effect
through comparison of upwind and downwind mea-
surements, would have overestimated the average ef-
fect of the city by -100%. This suggests that single-
supersaturation CCN counters should be used with
great care if meteorologically significant data are to be
obtained.
Ice Forming Nuclei
In an effort designed to ascertain anthropogenic ef-
fects on relative concentrations of Ice Forming Nuclei
(IFN), measurements ofiFN have been taken upwind,
over, and downwind of St. Louis using membrane fil-
ter techniques during both summer and winter months.
Data from March 1973 (Braham and Spyers-Duran,
1974) and February-March 1975 (Braham, 1976a)
show average IFN concentrations obtained with mem-
"'z
0.2 0.4 0 0.4
ICE NUCLEI PER LITER
FIG. 5. The distribution of daily average concentration of IFN upwind, over, and down-
wind of St. Louis for summer. Nuclei active at -16.6°C. After Czys (1977). IFN concen-
tration measurements exceeding abscissa scales are cited as numbers in figures.
123
brane filters in the mixed layer of -1 per liter ( -l7°C),
and rather strong evidence for urban deactivation of
IFN (Braham and Spyers-Duran, 1974).
Fig. 5 gives the distribution of daily average con-
centration of IFN upwind, over, and downwind of St.
Louis based on the summer data (Czys, 1977). These
distributions represent the number of days N having
average concentrations falling within the various in-
tervals of 0.05 nuclei per liter. Individual cases where
concentrations were larger than the range of plotted
data are noted within the figure. The mean summer
(1972-74) concentrations of IFN upwind, over, and
downwind of the metropolitan area were found to be
0.25, 0.19, and 0.25 L - 1 respectively; winter (1973)
values were 1.91, 1.60, and 1.43 L- 1 •
Special flights focused on some of the industrial ac-
tivities around St. Louis to identify possible point
sources or sinks of IFN. Czys (1977) reported three
industrial areas found to be sources of IFN: the Gran-
ite City area, an industrial area 8 km south of the Gate-
way Arch, and an industrial area 15 km south of the
Gateway Arch. Average concentrations in these in-
dustrial plumes were found to be greater than relative
upwind concentrations by factors ranging from 1.1 to
nearly 3. In contrast, the Wood River area was found
to be a sink. The concentration of IFN measured up-
wind of St. Charles was found to be the same as the
concentration of IFN measured downwind. Semonin
( 1978) reported high concentrations of IFN, estimated
by a portable coldbox technique, extending south from
central St. Louis but confined to a narrow band along
the river where heavy industry is located. In addition,
isolated areas of high IFN concentrations were ob-
served near Waterloo, Belleville, and Collinsville, Il-
linois, and a large area of high concentration encom-
passed the Alton-Wood River industrial center along
the floodplain between the Mississippi and Missouri
rivers.
A study by Spyers-Duran (1975) showed that a ma-
jority of IFN collected over and downwind of St.
Louis on membrane filters were reactivated during a
second processing. This suggests that if an ice crystal
OVER CITY DOWNWIND
o. 91' 1.13
0.6
124
should evaporate in natural cloud, the remaining nu-
cleus could be reactivated, and serve again to grow a
new ice crystal.
Aerosol Particle Size Distributions
Aerosols, as a group, constitute a pollutant of major
importance. Because of this it is essential that their
size distributions and variations in time and space be
known for an understanding of their formation, re-
moval and aesthetic consequences. Several studies in
METROMEX have documented the characteristics of
the regional and downwind aerosol-particle size dis-
tributions and their variations in space and time (Bra-
ham, l972a; Alkezweeny, 1973, 1974a, 1975; Auer,
1977b; Komp, 1977; Komp and Auer, 1978b); from
these works a rather consistent description of evolving
aerosol distributions emerges.
The particle number distributions between 0.1 and
5.0 11-m show a regular decrease in concentration with
particle size, conforming to the approximation dN/
d(log D) = cD- 13 where {3 = 3 (Alkezweeny, 1973,
1974a; Auer, 1977b; Komp, 1977; Komp and Auer,
1978b).
Alkezweeny (1973) presented observations to sug-
gest that particle size distributions could be charac-
terized by larger concentrations at D > 1 11-m during
high relative humidity (>60%); Komp (1977) provided
confirming qualitative support from his case studies.
Measurements of particle size distributions have
shown that concentrations of aerosol particles of all
sizes generally decrease with increasing altitude (Al-
kezweeny, 1974a, 1975). Fig. 6 gives an example of
the differences in particle surface distributions below
and above the top of the mixing layer; the ACN con-
centration also drops from 2000 cm- 3 to less than 200
cm- 3 through the inversion (Alkezweeny, 1975).
Alkezweeny, 1974a, 1975; Auer, 1977b; and Komp
1400
1200
(")
I
1220 m over St.
E
u 1000
N
E
;::i 800
Cl
VI m 600
<l 0
<l 400 1525 m
over St.
200
o~~~~~~WL~~WW
0.001 0.1 1.0
PARTICLE DIAMETER, ~m
FIG. 6. Particle surface distributions over St. Louis measured on
20 August 1974. After Alkezweeny (1975).
METEROLOGICAL MONOGRAPHS
and Auer, 1978b have observed pronounced changes
in particle size distributions as a function of distance
downwind from the urban area. In general, concentra-
tions of aerosols in the nucleus mode (D < 0.05 ~-tm)
attained maximum values at times up to one hour from
release. Although particle concentrations typically de-
creased rapidly after the maxima were reached, con-
centrations were higher than regional background at
all times out to five hours. While decreases of the con-
centration of smaller aerosol particles were taking
place downwind, increases in the concentration of ac-
cumulation mode aerosols were occurring. Concentra-
tions of aerosols between 0.1 and 1.0 ~-tm seem to in-
crease through 3. 5 hours trans port time before
returning to background levels after 4-5 h. Maximum
increases in the number of particles <2 11-m diameter
were noted to occur at -2.5 h wind travel time.
The modal nature of aerosol size distribution be-
comes more apparent in both the surface-area and vol-
ume distributions of the aerosols. Such an example is
Fig. 7. Changes in this aerosol-volume distribution are
centered in the accumulation-mode size range of par-
ticles, with the largest increases within this mode po-
sitioned around 0.2-0.3 11-m diameter.
During the 1975 summer expedition, "giant" and
larger aerosols were collected over and upwind of St.
Louis urban area. These data provided for the first
time direct airborne measurements of ultragiant par-
ticles upwind and downwind from major urban-indus-
trial cities (Johnson, 1976; Braham, 1977). Fig. 8
shows upwind/downwind aerosol-volume distributions
for diameters between 5 and 55 11-m averaged from 50
samples spread over 11 flights. Relatively large con-
- · - 1.0 hr
30 -•oo- 1.6 hr
- - - - 2.3 hr
2.8 hr
- - - 4.0 hr
<1
...._
>
<1
10
1.0 10
DIAMETER, D, llm
Fro. 7. Representative example (9 August 76) showing evolution
of volume distributions for aerosols sampled downwind of St. Louis
at specified times. After Komp and Auer (l978b).
METROMEX: A REVIEW AND SUMMARY 125

UPWIND DOWNWIND downwind. Particle concentrations showing the great-

80 3 3 -3
vtota 1 = 20.3 ~m cm- est increase above those numbers found directly over
'(
3 vtotal = 36.1 ~m em
E
u
Labadie (0 h) lie in the size range of 0.2 to 1.0 JLm.
""' E 60
The mode of the distribution is seen to shift to larger
""-
0 40 sizes at points downwind. The conversion rate in the
en
::::.
0
first 3.4 h of transit time strongly suggests active aer-
~ 20 osol growth mechanisms in power-plant plumes.
>
"0

6.3.2 GASEOUS PRECURSORS AND AEROSOL

DIAMETER, ~m COMPOSITION
FIG. 8. Volume distributions of large areosol particles upwind and
downwind of St. Louis. After Braham (1977). Contaminants, once released into the atmosphere,
can be altered in many ways. Reactive volatile gases
centrations of large particles at 300 m above ground may undergo gas-to-particle conversion to form par-
level upwind of the city were unexpected since there ticles of low volatility. Gases may become adsorbed
is no appreciable urbanization within several hundred on particles, and particles may coagulate. Many chem-
kilometers upwind. These data must be considered as ical processes occur, frequently promoted by light and
typical of background aerosol loadings encountered in catalyzed by trace substances. Understanding of
the Midwest in summer. Combining the information transformation processes is essential in making ma-
from Figs. 7 and 8 and assuming a particle density of terial balances of air pollutants and accounting for ob-
2 g cm- 3 , these METROMEX measurements indicate served downwind concentrations .
that the concentration of particles between 0.05 and The atmospheric concentrations of trace elements,
55 11-m in diameter provide average mass loadings of sulfates, S0 2 , NO, N0 2 , NOx, CO and halocarbons
44 and 86 11-g m- 3 upwind and downwind, respectively. were measured upwind, in, and downwind of St. Louis
These are to be compared with present clean air stan- to determine the rates at which these contaminants are
dards of 75 11-g m- 3 • removed from the atmosphere by dry deposition and
A study of aerosol evolution in the 2400 MW La- chemical reaction (Alkezweeny, 1974b, 1977; Young
badie power plant plume has been published by Auer et al., 1974, 1975; Braham, 1976a; Alkezweeny and
and Komp (1977) and Komp and Auer (1978a). Similar Drewes, 1977a, b; Alkezweeny and Lee, 1977). During
to Fig. 7, Fig. 9 shows the evolution of the aerosol selected days of July 1973-74, concentrations of atmo-
volume distribution at transit times up to 3.4 hours spheric contaminants produced by St. Louis were

DOWNWIND TRANSIT TIMES:

- - - 0 hr
- - - - 2 . 6 hr
- - - 3 . 4 hr

DIAMETER, D, llm
FIG. 9. Volume distributions for aerosols sampled in the center
of Labadie Power Plant plume at specified downwind times, 16 Au-
gust 1976. After Komp and Auer (1978a).
126
sampled in a pseudo-Lagrangian frame of reference at
the surface downwind, generally on arcs of 30, 65 and
95 km on nonconvective afternoons (Young et at.,
1974, 1975). The lead and bromine concentrations in
downtown St. Louis averaged twenty times the up-
wind concentrations. The concentrations decreased
very rapidly between downtown St. Louis and the 30
km arc. This large decrease was attributed to dilution
caused by vertical transport through the mixing layer,
since particle size analysis indicated that only a small
fraction of the lead and bromine would be expected to
settle out in this distance, and because this decrease
in concentration further downwind of St. Louis was
much smaller. Seventy percent of the lead and bro-
mine was present on particles smaller than 0.31-Lm and
only 10% was present on particles larger than 3
1-Lm. The lead and bromine concentrations at the outer
arc averaged 68 and 74%, respectively, ofthe concen-
trations at the 30 km arc. Upwind concentrations of
lead were reported as 0.26 x 10-7 g m- 3 • Coincident
airborne samples to 1 km above the surface revealed
that the greatest concentrations of lead and bromine
were found at altitudes near 300 m, with maximum
concentrations 30 km downwind.
CChF concentrations measured at the 30 km arc
averaged 60% of the downtown concentrations. The
fact that the decrease in CC13 F concentrations be-
tween downtown St. Louis and the 30 km arc was
considerably less than the decrease in the lead and
bromine concentrations may be explained by the fact
that CChF is released primarily in the suburbs while
lead and bromine are produced in large amounts of
downtown traffic; although deposition of lead and bro-
mine on the earth's surface by gravitational settling
and turbulent diffusion certainly contributed to the
decreases in their concentrations. CChF concentra-
tions at the outer arc (95 to 125 km) averaged 90% of
those observed at the 30 km arc. Comparison of the
decreases of lead, bromine and CChF concentrations
between the 30 km arc and the outer arc indicate that
dilution due to atmospheric mixing was responsible
for approximately a 10% concentration decrease,
while deposition on the earth's surface caused ap-
proximately a 20% decrease in the lead and bromine
concentrations in surface level air (Young et at., 1974,
1975).
The downtown S0 2 concentrations averaged 20
times the upwind concentrations. However, the S02
concentrations at the 30 km arc averaged higher than
the downtown concentrations, presumably because of
elevated releases of this material and the presence of
large S0 2 sources along the Mississippi riverfront.
Average upwind concentrations of S02 were -51-Lg
m- 3 • The S02 concentrations decreased much more
rapidly between the 30 km arc and the outer arc than
METEROLOGICAL MONOGRAPHS
did the concentrations of the other contaminants mea-
sured. The S0 2 concentration at the outer arc aver-
aged only 27% of the 30 km arc concentration; similar
supportive evidence for this reduction in so2 over-
burden was given by Braham (1976a). A relatively
large decrease in so2 concentration has been attrib-
uted to the conversion of so2 to sulfates in the at-
mosphere and to the reaction of so2 with vegetation.
The half-residence time for S02 in the atmosphere was
calculated to be 4.5 hours. Since selenium is generally
emitted into the atmosphere by the same sources that
emit so2, the fact that very little decrease in selenium
concentrations occurred downwind was especially
useful in demonstrating that S02 is removed from the
atmosphere more rapidly than particulate materials.
Upwind concentrations of selenium are quoted at
0.0019 1-Lg m- 3 •
The sulfate concentrations in downtown St. Louis
averaged only a few percent higher than the upwind
concentrations. This can be attributed to the appre-
ciable periods of time required for the so2 produced
by the St. Louis metropolitan area to be oxidized.
Eighty-five percent of the sulfate in downtown St.
Louis was present on particles smaller than 0.3 1-Lm
and only 2% was present on particles larger than 3/Lm.
The sulfate concentrations increased with distance
with the maximum concentrations being at the 95 km
arc, indicating a continual conversion of S0 2 to sul-
fates. Limited data between 95 and 125 km downwind
indicate that sulfate concentrations start to decrease
at this distance. The increase in sulfate concentration
between the 30 km arcs and the 95 km arcs was only
- 10% of that necessary to account for the decrease
in the S02 concentration, even if it is assumed that the
sulfate concentration was decreased by 30% by dilu-
tion and deposition, as was the case for lead and bro-
mine. Therefore, the large decrease in S02 concentra-
tion with distance downwind of St. Louis was attributed
primarily to the reaction of so2 with vegetation
(Young et at., 1974, 1975). The transformation rate of
so2 to sulfate has been estimated from aircraft data
in the St. Louis urban plume by two methods outlined
in Alkezweeny and Powell (1977a, b) and Alkezweeny
(1977). Rates between 8 and 14% per hour have been
reported for the St. Louis plume.
Arsenic, bromine, chlorine, copper, mercury, lead,
sulfur, antimony, selenium and zinc are greatly en-
riched relative to iron in St. Louis air, indicating that
they are primarily of anthropogenic origin. Elements
of anthropogenic origin often can be identified by their
particle-size distribution, since they are generally
present on smaller particles. Calcium and iron are pri-
marily of crustal origin; 50% of the calcium and 45%
of the iron were present on particles larger than 10
/Lm, while only 7% of the calcium and 14% of the iron
METROMEX: A REVIEW AND SUMMARY
were present on particles smaller than 0.3 p.m. Tita-
nium and vanadium are two elements which did not
appear to be greatly enriched relative to iron in the St.
Louis air, but whose particle-size distributions indi-
cated large anthropogenic contributions. Twenty-two
percent of the titanium and vanadium were present on
particles smaller than 0.3 p.m (Young et al., 1975). The
copper concentrations at the 30 km arc were some-
what higher than in downtown St. Louis. Also, the
peak concentrations occurred downwind of Alton
rather than St. Louis. The copper concentrations de-
creased with distance beyond the 30 km arc at about
the same rate as did the other trace elements.
Gatz et al., (1978) have attempted to identify sources
associated with aerosol loadings at specific sites, using
a factor analysis of elemental composition. The factor-
analysis method is essentially a correlation procedure
where a matrix of correlation coefficients between ele-
ments is derived, and factors are expressed as linear
combinations of these coefficients, chosen to maxi-
mize explanation of the variance of the data. Using
the eleven elements, AI, Si, S, K, Ca, Ti, Mn, Fe, Zn,
Brand Pb, these investigators have isolated a number
of source groups or factors, which are common
throughout the METROMEX area. These include a
"soil and flyash" factor, a "sulfur and metals" factor,
an "auto-exhaust" factor, and a "metals without sul-
fur" factor. In addition, some individual sources, in-
cluding a Ti pigment plant and a secondary Pb smelter,
were identified.
Airborne and ground measurements of NO, N0 2
and NOx have been carried out on a case-study basis
in METROMEX during the summers of 1973 and 1975
(Alkezweeny, 1974b, 1977; Alkezweeny and Drewes,
1977a, b). Within downtown St. Louis, surface con-
centrations of NO, N0 2 and NOx were found to be in
the range of 8-24 ppb, 11-36 ppb and 2~ ppb, re-
spectively. Concentrations of NO and NOx at 300 m
250
o- SOo
200 +• 0)
.0
g: 150
o'"' lOO
50
1.0 . . . . . - - - - - - - - - - ,
e o.s : ..
~ 0 .6
!::. ""''·
~ 0.4
".
. ~ · "-...
~ 0 .2
~~""'
0 ~~-~~~-~~
0 25 50 75 100 125 150
OISTAHCE OOWNWI 0 FROM ALTON, km
FIG. 10. Concentrations of 0 3 , S02 , NO, N02 and aerosols northeast of St.
Louis on 16 August 1975. After Alkezweeny and Drewes (l977b).
127
above the ground were observed to be higher by fac-
tors of 4 and 2, respectively, than those on the ground,
and both decreased with altitude above 300 m. Fur-
thermore, the concentrations in the afternoon were
higher than in the morning for all altitudes (Alke-
zweeny, 1974b). In studies of the trace gases just
downwind of St. Louis in the urban plume, Alke-
zweeny (1977) found concentrations of NO at 5 ppb
and N02 at 15 ppb at 1000 m, compared to the range
of values of 25-50 ppb for NO and 14-15 ppb for N02
at the same altitude directly over the city (Alke-
zweeny, 1974b). Investigations oftrace-gas concentra-
tions in the urban plume out to 15 km (Alkezweeny
and Drewes, 1977a, b) revealed concentrations of NO
of 25 ppb and N02 at 40 ppb just downwind of the city
at 600 m, compared with the urban-profile concentra-
tions at the same altitude of 26-60 ppb for NO and
14-20 ppb for N02 (Alkezweeny, 1974b). In Fig. 10
(Alkezweeny and Drewes, 1977a, b), the concentra-
tions of NO, N0 2 , S02 and Aitken nuclei are seen to
decrease with distance from the city, reflecting the
effect of physical and chemical transformation and re-
moval processes. Ozone is observed to increase to
more than 200 ppb at 150 km downwind. Fig. 10 is
primarily concerned with pollutant levels downwind
of a major urban source; at 105 km, however, plumes
from the Lakeside and Dallman power plants near
Springfield, Illinois, were apparently intercepted,
causing a sharp decrease in the ozone level and cor-
responding significant increases in the levels of NOx,
S02 and Aitken nuclei.
Results have been reported for two hydrocarbon
sampling methods used during the airborne study of
the St. Louis urban plume in August 1975 (Alke-
zweeny and Lee, 1977). The results of two experi-
ments, conducted in a Lagrangian frame of reference,
are presented in Table 3. It should be noted that 17
August was a day characterized by weak and variable
150 . - - -- -- - - - - r 30
125 25
100-8_
0.
75
N
50~
25
0
. - - - - - - - - - - - - - . - 60
50
.~ . ..
:- ·.
. .......
. • • :. • • ' ~ ~;'
- , .~ •••·•• ·,.:...
. r ~:' * "''l
.'\
:•
'~~.. llf
:...
10
o
0 25 50 75 100 125 150
128 METEOROLOGICAL MONOGRAPHS

TABLE 3. Hydrocarbons in bag samples (ppb) in the St. Louis urban plume (after Alkezweeny and Lee, 1977).

12 August 17 August

1235-1247 1604-1612 1350-1400 1450-1500 1600-1608 1647-1655

I. Total hydrocarbons 2120 2160 2560 2260 2130 2540

II. Carbon monoxide 830 760 530 470 400 550
III. Saturated hydrocarbons
A. Alkanes
Methane 1700 1710 1680 1670 1660 1690
Ethane 12.1 13.6 15.0 13.5 8.5 14.0
Propane 7.9 9.0 30.1 21.4 10.3 23.1
!sobutane 3.3 3.7 19.6 15.6 3.8 24.9
N-Butane 5.9 7.0 58.3 40.8 16.3 64.3
Isopentane 5.6 6.8 61.7 52.0 17.6 56.1
N-Pentane 4.0 4.0 24.3 19.4 9.6 23.3
2, 2-Dimethylbutane 0.2 1.9 1.2 0.6 1.4
2, 3-Dimethylbutane + Trans-4-
methylpentene-2 0.5 1.1 8.8 5.5 3.3 7.8
2-Methylpentane 1.5 2.9 19.5 13.5 6.6 17.2
3-Methylpentane + (Hexene-1 + 2-
Ethylbutene-1) 0.7 1.8 13.5 9.3 3.9 11.1
N-Hexane 5.9 45.4 20.2 12.6 18.1
B. Cyclic Compounds
Cyclopentane 0.5 1.1 3.0 2.8 2.3 3.3
Methylcyclopentane + (3, 3-
Dimethylpentene-1) 20.1 19.7 36.1 28.9 27.5 24.3
Cyclohexane + (Cis-4-Methylhexene-2) 2.9 18.8 12.6 12.6
IV. Unsaturated compounds
A. Olefins
Ethylene 2.9 2.8 8.7 17.9 4.6 4.7
Propylene 0.4 0.5 1.9 1.9 1.2 1.1
Isobutylene + Butene-! 0.7 1.7 1.2 1.6 0.5 0.5
Trans-Butene-2 0.2 0.5 0.9 0.8 1.3
Cis-2-Butene 0.2 0.1 0.5 0.2 0.5 0.5
3-Methylbutene-1 0.4 0.3 0.2 0.3 0.4 0.1
Pentene-1 1.6 0.6 0.8 1.1
2-Methylbutene-1 0.2 OJ 0.3 0.2 0.1
2-Methylpentene-1 + 4-Methylpentene-1 0.2 0.1 0.4 0.1
2, 4-Dimethylpentene-2 3.6 42.0 19.6 23.7 47.4
2, 4-Dimethylpentene-1 40.0 21.3 67.4 12.1
Trans-Hexene-3 19.8 21.2 8.2 15.8
V. Aromatic
Benzene 67.4
VI. Acetylene 0.9 1.3 1.0 1.6 1.8 1.4

winds; the data were obtained within a few miles of ample of visibility deterioration downwind of St.
the Wood River refinery complex, and the dominant Louis. A tendency for visibility anomalies to persist
influence of this source is reflected in the dramatic over greater distances downwind on days when the
increase in hydrocarbon concentration in the last sam- ambient relative humidity exceeded 90% was also
ple due to fresh hydrocarbon input. Table 3 presents noted. Downwind of the Alton-Wood River industrial
complex visibility reductions of 40% were noted.
tentative identification of organics in the St. Louis ur-
ban plume. The regions of minimum visibility did not coincide
with locations of maximum Aitken nucleus concentra-
tions (compare Fig. 3 vs Fig. 11). Aerosol surface and
6.3.3 VISIBILITY AND HAZE
volume distributions indicate a general growth in the
Estimates of horizontal visual range, along with accumulation mode (0.1-1.0 JLm) with maximum val-
standard meteorological data, and aerosol distribu- ues in a much narrower size range within the mode of
tions within the mixing layer have been acquired up- 0.2-0.4 JLm diameter, as shown in Fig. 9. Komp (1977)
wind, over and downwind of the metropolitan area and Komp and Auer (1978b) have reviewed possible
(Shea, 1976; Auer, 1977b; Komp, 1977; Komp and mechanisms for explaining the growth of light-scatter-
Auer, 1977, 1978b; Shea and Auer, 1978). A 50% re- ing aerosols in the downwind plume of St. Louis. They
duction from regional upwind visibilities consistently conclude that, because of time and size coincidence,
occurred at a distance corresponding to 2.5-3.0 h the resultant aerosol distributions result either from
travel time downwind for an air parcel moving with generation by gas-to-particle reactions or from the rel-
the mean transport wind. Fig. 11 shows a typical ex- atively subdued effects of gravitational sedimentation
METROMEX: A REVIEW AND SUMMARY 129

haze particles were the result of repeated cycling of

S02 , Aitken particles, CCN, and water vapor through
r. ""'
_.. - ----+-- Fl ight Path
the shallow clouds (Fig. 12). With each cycle of con-
densation and evaporation, the particles grow larger
4 )-- Sounding
until finally they are capable of S_!lrviving at relative
humidities well below 100%. Once formed, these so-
lution droplets can survive vertical mixing through the
depth of the mixed layer and advection over great dis-
tances from the S02 sources. Charlson et al. (1974)
have shown that sulfates often account for at least half
of the mass of particles in the accumulation-mode di-
ameter range at St. Louis.

6.4 Dry deposition and dustfall

Two separate studies related to dry deposition of air
pollution were conducted during METROMEX. The
first of these involved the measurement of "dustfall"
into polyethylene jars during dry periods (Gatz et al.,
1978). The second was centered on pollutant-ratioing
and material-balance measurements (Young, 1978).
The dustfall measurements were obtained primarily
KILOMETERS
for the purpose of assessing possible errors in precip-
0 10 20 itation-chemistry samples, which were acquired in
identical polyethylene jars during rain periods. Al-
AUGUST 9, 1976
though it was recognized that aerosol deposition to
these collectors would not necessarily provide a cor-
FIG. 11. Isopleths of prevailing visual range (in statute miles)
downwind of St. Louis during 1400-1800 CDT on 9 August 1976. rect characterization of deposition to natural surfaces,
[After visual estimates of Komp and Auer (1978b).] the resulting data are still of some interest.
Plots of areal distributions of specific metals in dust-
and coagulation on 0.1-1.0 p,m sized aerosols. Alke- fall samples for the total 1972-74 summer network
zweeny (1975) suggested also that gas-to-particle measurements appear in Fig. 13. Using measured air
transformations involving existing nuclei play an im- concentrations of these elements, the averaged depo-
portant role in the growth of particles in the St. Louis sition velocities given in Table 4 are obtained, which
plume. St. Louis' influence on the aerosol accumula- are compared with those determined by Cawse (1976),
tion mode were not significant at 4-5 h travel time again with an artificial deposition substrate.
downwind, and corresponding improvement in the vi- The pollution-ratio studies have been mentioned
sual range at 4-5 h downwind has also been noted
(Komp and Auer, 1978b). TABLE 4. Summary of mean deposition velocities.
The process of urban CCN production and nuclea-
tion in urban clouds appears to be closely related to Deposition velocity1 (em s- 1)
one of the most serious of all air pollution problems; METROMEX
viz., the formation of stable haze and low visibility
Element 1972- 1974 1975 United Kingdom, 19752
during periods of stagnant anticyclonic circulation. At
St. Louis, Braham (1975, l976c, 1977) found that dur- Li 1.69"
ing periods of widespread restricted visibility, asso- Na 2.864 0.20-1.3
Mg 2.584 0.59-> 3
ciated with air pollution alerts, the upper part of the K 4.07' 4.11 0.58->2
mixed layer contained scattered shallow cumuli which Ca 5.255 3.55 0.38-1.4
Zn 0.933 1.14 0.28-0.99
had short lives and evaporated without precipitating.
Between and· below these clouds, haze particles were 1METROMEX concentrations in air were estimated from mea-

found with modal sizes of several microns and con- surements of Gatz (1977).
2 Range of values observed at inland stations by Cawse (1976),
centrations of - I cm- • To be stable at the subsatur- with filter paper as a collection substrate for dry deposition.
3

ation humidities involved, these haze particles must 3 Assumes same ratio of total/soluble dry deposition as found in
1975 network.
have been composed of droplets having rather high 4Assumes aerosol is 65% soluble.

solute concentrations. Braham hypothesized that these 5Assumes aerosol is 90% soluble.
130
Sunlight
Water
Catalytic Surfaces
Background
Water Vapor
Aitken Condensation
Nuclei-ACN -100-1000/cc---J'-------~~
Cloud Condensation
- 500/cc,---1-------~
Nuclai-CCN-200
FIG . 12. Repeated cycling of S02, ACN, CCN and water vapor through many small clouds can lead to micron size haze
droplets. After Braham (1977).
previously in the preceding section. These studies
were conducted by perlorming ground-level sampling
from arrays of points in and downwind of St. Louis,
timing sample acquisition in a sequential manner such
that a quasi-Lagrangian reference was maintained.
Specific materials sampled included CO, CC13 F, SOz,
S04, Pb, Br and Zn. By computing ratios of a given
pollutant's concentration to that of a comparatively
inert species (e.g., CO or CC13 F), the effects of dilu-
tion and reaction could be compensated, and an esti-
mate of deposition rate could be established.
Examples of the results are shown in Fig. 14, which
is a composite of data acquired for the four-day period
3-6 August 1975. Central points in these plots pertain
to averages of all samples obtained at approximately
similar downwind distances, while the error limits de-
note standard deviations computed from the scatter of
the individual experiments . Deposition velocities com-
puted from these data using the ratio method were 1.8
em s- 1 for Zn, Br, and Pb, and 6.2 em s- 1 for S0 2 •
These compare with similar measurements for 1973 ,
where the values were 2.2 em s- 1 and 3.7 em s- 1 , re-
spectively .
6.5 Precipitation scavenging
Because of its intrinsic relationships with storm be-
havior and weather modification processes, the sub-
ject of precipitation scavenging received considerable
attention under the METROMEX program. Field
studies of precipitation scavenging included measure-
ments of ambient air pollutants as well as analyses
METEOROLOGICAL MONOGRAPHS
-
--+CCN
HAZE
small
solution
droC
~
using intentionally-released tracer materials. These
two types of studies are summarized individually in
the following text.
6.5.1 ScAVENGING OF AIR POLLUTANTS
METROMEX studies of air pollutant scavenging in-
cluded both metallic and nonmetallic materials. The
nonmetallic scavenging studies consisted largely of
material-balance measurements for individual storms
(Hales and Dana, 1978), utilizing a precipitation-chem-
istry network deployed over the area shown in Fig.
15. Chemical species of particular interest were sulfite,
sulfate, nitrite, nitrate and ammonium ions.
Typical results for a storm experiment are shown in
Fig. 15, which illustrates spatial distributions of rain
amount and rainborne pollutant concentration ob-
served for the storm of 23 July 1973. This particular
storm system moved across the network from the
southwest, providing maximum rainfall in the northern
portion of the grid. As can be noted, sulfate-ion and
nitrate-ion concentrations appear to have reflected the
presence of the metropolitan source rather strongly,
while ammonium ion concentration did not. Total
amounts of sulfate and nitrate deposited by this storm
were calculated as 1.7 x 106 and 7.8 x 105 gram-moles,
respectively, which may be compared with the cor-
responding hourly emission rates for St. Louis of 1.6
x 106 and 1.1 x 106 gram-moles.
Statistical analyses of the nine storms investigated
during this study of inorganic, nonmetallic pollutant
scavenging are summarized as follows:
METROMEX: A REVIEW AND SUMMARY 131

pg cm- 2 hr- 1

0
KILOM.E TERS
0 4 8 12

FIG. 13. Distribution of median dry deposition rates in 1972- 74 network (higher rates are shaded for emphasis). After
Semonin, et a/. (1978).
132 METEOROLOGICAL MONOGRAPHS

where c and A denote concentration and rainfall

amount at a particular site, respectively, and the
overbars denote average values for a particular
experiment.
100
• A similar correlation between rain amount and
z
2 NHt concentration is not readily apparent.
.....
<( • A weak positive correlation exists between NHt
"'.....zw 10 and the species N03 and S04. A positive corre-
u
z
0
u
lation should be expected in this case, owing to
the acid-base relationship of these species.
10 • Free hydrogen ion concentration does not corre-
late well with N03, S04 or NHt; this is some-
what surprising and apparently reflects the abun-
dancy and mix of other ionic species in the St.
Louis rainwater.
• The strongest, most consistent correlation is that
• A negative correlation usually occurs between between N03 and so.;. A positive correlation
produced the relationships should be expected between these variables ow-
ing to their mutual correlation with rain rate; how-
• A negative correlation usually occurs between ever, the observed relationship appears too strong
rain amount and concentration for the species to be explained simply on this basis. This suggests
N03 and S04. A least-squares fit of all data for either that both species are scavenged by a com-
the study produced the relationships mon mechanism, or that their source configura-
cso4 = 0.894 cso:t (AIA)-o.t4s, tions tend to be similar and are sufficiently local
to introduce variability in the washout patterns.
CNOa = 0.932 CNOa (A/A)-0· 24 1,

a. Rain amount, em b. so~ concentration, ~m/liter

c. NOj concentration, ~m/liter d. NH~ concentration, ~/liter

Fro. 15. Concentrations and rain amounts for storm of 23 July 1973: (a) rain
amount (em); (b) S04 concentration (j.Lm L -•); (c) NOa concentration (j.Lm L -•);
(d) NH:j concentration (p.m L -•). Distributions are shown on computation grid
used for integrating deposition amounts.
METROMEX: A REVIEW AND SUMMARY 133

The above statistical evidence combined with the

magnitudes and distribution of the washout patterns,
as shown in Fig. 15, suggests that much of the SO 4
and N03 recovered in precipitation originated from
the city. In contrast to these indications, however, a
fundamental material-balance analysis indicates that
sufficient airborne sulfate from sources other than St.
Louis existed in the ambient air to account for the
sulfate observed in rainwater. This raises the question
of the relative importance of direct scavenging of sul-
fate versus reactive scavenging of S02 gas. This aspect
is still being studied in the context of the METRO-
MEX data base.
The lack of correlation between hydrogen ion con-
centration and other measured attributes is of partic-
ular interest in view of the more extensive pH mea-
surements reported by Semonin (1976). Based upon
data obtained from a separate METROMEX precipi-
tation chemistry network for the summers of 1972 and
1974, Semonin notes a rather disorganized areal dis-
tribution of pH in rainfall, which is exemplified by the
plot in Fig. 16. In concordance with the statistical
analysis of Hales and Dana, Semonin finds little basis
for simple or obvious relationships between hydrogen
ion concentration, and source configurations and rain-
fall rate. Moreover, the frequency distribution of pH
measurements appears to be bimodal in nature, as in-
dicated in Fig. 17.
The above results are a somewhat surprising con-
trast to the apparently strong relationships between
source configuration and the rainborne anions S04
and N03 indicated in Fig. 15. In the context of the
overall ion balance, these features suggest that exten- FIG. 16. Areal pH (a) and rainfall (b) for storm of 28 July 1972,
sive variability must exist in the concentration pat- rainfall amount in mm. After Semonin (1976).
terns of ionic species other than those measured here.
This characteristic is borne out by additional METRO- One portion of the metals-scavenging research of
MEX measurements involving rainborne metallic sub- METROMEX was conducted using data from a lim-
stances (Young et al., 1976; Changnon and Semonin, ited number of stations on the network, where se-
1975). quential samples or air and precipitation were care-
An example of the areal distribution of normalized fully collected and analyzed to obtain scavenging
metal deposition for the 23 July 1972 rain period is ratios, i.e.,
given in Fig. 18. As expected, these deposition pat-
terns tend to follow the rainfall trend, but exhibit in- Scavenging ratio
dividual variabilities as well. Aside from zinc, which Concentration of pollutant in rainwater
has an apparent source in the Granite City region, Concentration of pollutant in air
there is no consistent or obvious source-configuration
relationship for the metal-deposition patterns (cf. Measurements in this portion of the program typi-
Gatz, Semonin, and Peden, 1978). These features pro- cally were conducted at five sites surrounding St.
vide an explanation for the seemingly independent be- Louis. Air samples were obtained using a high volume
havior of rainborne hydrogen ion concentration. The sampler, and rain samples were collected with poly-
sources of variability in the areal patterns of the pri- ethylene-funnel sampling devices. Rain samples were
mary ionic constituents affecting these hydrogen ion filtered using 0.5-micron membrane filters, and both
levels, however, will require extensive additional the collected and aqueous fractions were analyzed,
study before an adequate level of understanding can along with the air samples, using atomic absorption
be attained. techniques.
134 METEOROLOGICAL MONOGRAPHS

tracers into storm inflow regions from stacks and

ground level positions, these developed into compar-
853 RAIN SAMPLES
atively sophisticated, coordinated multiple aircraft re-
leases during subsequent years (1972-75) (cf. Gatz,
1977). The surface-release experiments provided some
indication of the efficiency of scavenging in specific
cases, but the aircraft experiments were much more
valuable as indicators of the pertinent mechanisms of
storm behavior. A listing of the tracer-release exper-
ffi
u
10
iments is given in Table 5.
~ The tracers used included lithium, indium, ruthen-
ium, tantalum, europium, iridium, rehnium, gold and
cesium, all relatively rare elements. These were re-
leased in aerosol form from pyrotechnic flares and
from acetone generators, and the resulting rain sam-
ples were analyzed using atomic-absorption, flame-
emission, and neutron-activation spectroscopy. Dur-
ing several storm events aircraft operated by the Illi-
nois State Water Survey (ISWS) and Battelle Pacific
Northwest Laboratory (PNL) were coordinated tore-
lease materials into the same convective system, es-
sentially in a simultaneous fashion. During these co-
ordinated experiments the ISWS aircraft released
material into the cloud base updraft regions, while the
PNL aircraft released various tracers at multiple levels
pH
adjacent to the convective tower at suspected inflow
FIG. 17. Frequency distribution of pH measurements from 853
samples of 22 storms in summers of 1972 and 1974. After Semonin
(1976). TABLE 5. Summary of METROMEX tracer release studies.

Type of
Two results of this research are of special interest. Date release Tracers deployed Laboratory*
The first of these is that, for the elements studied, the 14 July 1971 Stack Li ISWS
observed scavenging ratios were relatively constant
29 June 1972 Surface Li ISWS
between sampler sites and in time. The second result 4 July 1972 Surface Li ISWS
is that significant differences in scavenging ratio oc- 15 July 1972 Surface Li ISWS
curred between various elements. Hypothesizing that 18 July 1972 Aircraft Li & In ISWS
24 July 1972 Aircraft Li ISWS
this difference may be caused in large part by differ- 3 August 1972 Aircraft Li & In ISWS
ences in particle size, Gatz (1972) has plotted the ob- 3 August 1972 Aircraft Ru,Ta,Eu,IR,Re & Au PNL
6 August 1972 Surface Li ISWS
served scavenging ratios versus particle size for se- 11 August 1972 Aircraft Li & In ISWS
lected elements. The results (Fig. 19) offer encouraging 19 August 1972 Surface Li ISWS
evidence on behalf of the washout-ratio concept, at 9 July 1973 Aircraft Li ISWS
least as a gross indicator of scavenging behavior. 23 July 1973 Aircraft Li & In ISWS
30 July 1973 Aircraft Li & In ISWS
9 August 1973 Aircraft Li & In ISWS
6.5.2 SCAVENGING OF TRACERS 10 August 1973 Aircraft Li & In ISWS
12 August 1973 Aircraft Li & In ISWS
In addition to investigations pertaining to scaveng- 13 August 1973 Aircraft Li ISWS
ing of ordinary pollutant materials, METROMEX 2 August 1974 Aircraft Li & In ISWS
studies also involved sizeable scavenging efforts using II August 1974 Aircraft Li& In ISWS
11 August 1974 Aircraft Ir,Re,Au,Eu & Ta PNL
tracers. The primary objectives of the tracer studies 17 August 1974 Aircraft Ir,Re,Eu & Ta PNL
were 1) to elucidate basic scavenging mechanisms and 18 August 1974 Aircraft Ir,Re,Au & Ta PNL
capture efficiencies, and 2) to investigate influx and 18 July 1975 Aircraft lr,Ta PNL
transport phenomena within individual storm systems. 19 July 1975 Aircraft Cs ISWS
These studies represent the most comprehensive ef- 19 July 1975 Aircraft Ir ,Re,Au,Eu & Ta PNL
23 July 1975 Aircraft lr,Re & Ta PNL
forts to date in the field of multi-element tracer appli- 30 July 1975 Aircraft Ir,Au,Eu,ln & Ta PNL
cations for scavenging research. Beginning in the sum-
*ISWS =Illinois State Water Survey.
mer of 1971, with rather modest attempts to inject PNL =Battelle, Pacific Northwest Laboratory.
METROMEX: A REVIEW AND SUMMARY
tr::!LWTERS
0 4 8 12
I RA E'3
FIG. 18. Deposition (mass/area) of non-tracer elements and rain, each normalized to its network mean for 23
July 1973. After Changnon and Semonin (1975).
areas. An example of tracer deposition from a single
aircraft tracer experiment is given in Fig. 20.
Major problems with the tracer experiments arose
from inadequate surface network coverage. Owing to
unpredictability of the target positions there were sev-
eral circumstances where significant amounts of tracer
were deposited locally, but off of the major grid-cov-
erage areas. Also, even when good targeting was
achieved, the deposits tended to exhibit a highly lo-
calized character with significant variability occurring
between adjacent sampling points. This local variabil-
ity introduced rather large uncertanties in numerical
135
,
integration of the data, which was necessary to esti-
mate total deposition amounts and scavenging effi-
ciencies.
Regardless of the above difficulties, a number of
valuable qualitative observations could be elucidated
from the tracer-scavenging results. The first of these
is the general conclusion that the METROMEX storms
were rather effective scavengers of tracer material in-
jected adjacent to the sides of the convective towers,
as well as that injected into the cloud-base regions.
This conclusion was drawn by 1) comparing the de-
position amounts obtained by integration across the
136
.~------
2
!;;: 600
""<.!)
z ,,,
Mn~
<.!)
z
~
""~ 400
FIG. 19. The precipitation scavenging ratio as a function of the mass median diameter (MMD). The
triangles represent values for St. Louis from the literature and the range for other urban locations are
indicated by the horizontal bars. After Gatz (1972).
network with known release amounts, and 2) by noting
the tendency for the deposition to exhibit a highly
structured pattern on the sampling network (see Fig.
20). A further significant finding from these studies is
the indication that tracer material was being ex-
changed between individual convective cells within a
multicellular system before its ultimate deposition.
This evidence for intercell transport is obtained from
comparison of tracer release and deposition data with
radar analysis and rain deposition patterns.
6.6 Modeling of convective storm scavenging
An additional component of the METROMEX scav-
enging research has been addressed to the modeling
of convective storms, both for the interpretation of
field data and for the formulation of parameters to be
used in more general types of scavenging calculations.
This work was conducted principally by PNL, and
centered around the two-dimensional convective storm
model developed by Hane (1978).
In its earlier form, Hane's model predicted the dy-
namical character of a storm system based upon pro-
files of ambient winds, temperatures and humidities.
During the METROMEX effort, the model was ex-
panded to incorporate pollutant concentration esti-
mates for the gas and liquid phases within the storm,
thus allowing computations of scavenging. An exam-
ple case is offered to demonstrate the character, ca-
pabilities and extent of this modeling effort. The ex-
ample considered here is the storm that occurred in
the St. Louis area on 21 August 1972, with a rainfall
pattern as shown in Fig. 21. The storm model was
METEOROLOGICAL MONOGRAPHS
executed in conjunction with ambient sounding data
for this period, and the developed model storm was
placed in the vicinity of a simulated groundlevel pol-
lution source, which approximated the emissions of
the St. Louis area. This consisted of pollutants with
a uniformly-mixed initial vertical distribution of 2.4
km, and extending over a downwind distance of about
22 km.
In Fig. 22a, the storm has just encountered the
source, and traces of pollutant are beginning to appear
in the rainwater aloft. In subsequent times the pollu-
tant is being deposited on the ground; in Fig. 22d the
pollutant source has been passed over by the storm
and no more gas-phase material is being entrained.
Gas-phase isolines to the left (in Fig. 22d) correspond
to pollutant that has been released by evaporation of
the falling raindrops; pollutant appearing in rain aloft
occurs primarily from raindrops being elevated in the
storm's updraft.
A more quantitative assessment of computed wash-
out is given in Fig. 23. This gives the areal deposition
of pollutant material as a function of distance of storm
travel, for a pollutant whose initial mixing ratio is 0.17
x I0- 6 g g- 1 distributed evenly over an initial 2.4 x 22
km area. The data points correspond to washout mea-
surements performed for arcs of samplers which were
oriented normally to the direction of storm travel.
These values are normalized by multiplying by the ra-
tio of estimated airborne mixing ratios to 0.17 x 1o-6 .
The calculations performed by Hane admittedly in-
volve some crude approximations, including the as-
sumption of initial plume configurations, the existence
of steady-state storm dynamics and assumed attach-
METROMEX: A REVIEW AND SUMMARY 137

l e(@-~.
400
• 200

KILOMETERS
0 5 10 15 20

FIG. 20. Indium and rain deposition patterns for storm of II August 1972.
Indium deposition (solid lines) in picograms em- •; rainfall (dashed lines) in inches
for 165~1705 CDT. Indium release from 1646--1652 CDT. After Changnon and
Semonin (1975).

21 August 1972

, ........,
.," I
0.25 -~" ,'
I
I
I
0.25

FIG. 21. Isohyetal pattern (em) for 21 August 1972. After Dana et al., (1974).
138 METEOROLOGICAL MONOGRAPHS

Sciences, Rep. BNWL-1850, Battelle Pacific Northwest Lab-

10 a . t • 10 min b. t • 30 min
oratories, Richland, WA, 94-99.
- - , 1974b: Vertical profiles of trace gases in St. Louis. Annual
Report for 1973 to the U .S. Atomic Energy Commission, Di-
vision of Biomedical and Environmental Research, Part 3:
Atmospheric Sciences, Rep. BNWL-1850, Battelle Pacific
Northwest Laboratories, Richland, WA, 105-107.
- - , 1975: Evolution of the St. Louis aerosol size distribution.
Annual Report for 1974 to the U.S. Atomic Energy Commis-
sion, Division of Biomedical and Environmental Research, Part
3: Atmospheric Sciences, Rep. BNWL-1950, Battelle Pacific
Northwest Laboratories, Richland, WA, 65--69.
- - , 1977: S02 to sulfate conversion in the urban plume. Annual
Report for 1976 to the Energy Research and Development
Administration Assistant Administrator for Environment and
Safety. Part 3: Atmospheric Sciences, Rep. BNWL-2100, Bat-
telle Pacific Northwest Laboratories, Richland, WA , 141-143.
- -,and D. R. Drewes, 1977: Airborne measurements of pollutants
15 10 over urban and rural sites. Annual Report for 1976 to the En-
HORIZONTAL DISTANCE , km ergy Research and Development Administration Assistant Ad-
FIG. 22. Vertical cross section of model storm moving over a ministrator for Environment and Safety, Part 3: Atmospheric
pollution source at approximately 6 m s-•. Pollutant distributions in Sciences, Rep. BNWL-2100, Battelle Pacific Northwest Lab-
air or cloud water are given by solid lines; those in rainwater are oratories, Richland, WA, 144-145.
given by dashed lines. After Hane (1978).
- - , and - - , 1975: Airborne measurements of pollutants over
urban and rural sites. J. Appl. Meteor. , 16, 561-563.
- - , and D. C. Powell, 1977a: Estimation of transformation rate of
SITE 1 SITE 2 S02 and S04 from atmospheric concentration data. Atmos. En-
viron., 11, 179--182.
SAMPLING - - , and--, 1977b: Estimation of transformation rate of S02 and
ARCS S04 from atmospheric concentration data. Annual Report for
0. 375
1976 to the Energy Research and Development Administration
'":'E
Sciences, Rep. BNWL-2100, Battelle Pacific Northwest Lab-
oratories, Richland, WA, 135-140.
z
0 --,and R.N. Lee, 1977: Airborne measurements of hydrocarbons
!: 0.250
in the St. Louis plume. Annual Report for 1976 to the Energy
"'0
0. Research and Development Administration Assistant Admin-
"'
c
istrator for Environment and Safety, Part 3: Atmospheric Sci-
ences, Rep. BNWL-2100, Battelle Pacific Northwest Labora-
0.125
tories, Richland, WA, 154-156.
I Auer, A. H ., Jr. , 1975: The production of cloud and Aitken nuclei
'(
URBAN PLU11E by the St. Louis metropolitan area (Project METROMEX). 1.
R ech. Atmos., 9, No. I, 11-22.
0~--~~--~~--~PO~S~IT~I~ON~~~--~--~~----~
30 40 - - , 1977a: Inadvertent modification of visibility in the vicinity of
DISTANCE ,
OOWNSTOR~1
a metropolitan area. J. Wea. Mod., 9, 19--31.
FIG. 23. Comparison of observed deposition amounts with model - -, 1977b: Evolution of visibility deterioration and aerosol spectra
calculations. After Hane (1978). in the St. Louis urban plume. J. Air Pollut. Control Assoc. , 21,
1188-1191.
- -, and S. A. Changnon, Jr., 1977: Non-precipitation weather
ment efficiencies for incorporation of pollutant into anomalies in St. Louis. Preprints Sixth Conf. Inadvertent and
cloud and rainwater. The rather good agreement be- Planned Weather Modification, Champaign-Urbana, Amer.
tween model predictions and observations, however, Meteor. Soc., 1-4.
suggests that this is a fruitful approach for future inter- --,and M. J. Komp, 1977: The behavior of Aitken nucleus plumes
from coal-fired power plants in the Midwest. Preprints Joint
pretation of scavenging data.
Conference Applications of Air Pollution Meteorology, Salt
Lake City, Amer. Meteor. Soc., 29--32.
Braham, R. R., Jr., 1972a: The University of Chicago in METRO-
MEX. 1972 Operational Report for METROMEX, W. P.
REFERENCES
Lowry, Ed., Illinois State Water Survey, Urbana, 25-41.
Alkezweeny, A. J ., 1973: Size distributions of St. Louis aerosol - - , 1972b: University of Chicago contribution to Project MET-
particles. Annual Report for 1972 to U .S. Atomic Energy Com- ROMEX-1. Tech. Note No. 45, Final Technical report to the
mission, Division of Biomedical and Environmental Research, National Science Foundation, Grant GA 28190Xl, Cloud Phys-
Vol. 2, Part 1: Atmospheric Sciences, Rep. BNWL-1751 , Bat- ics Laboratory, Dept. Geophys. Sci. , University of Chicago,
telle Pacific Northwest Laboratories, Richland, WA, 104-106. 98 pp.
- - , 1974a: Characteristics of aerosols from St. Louis. Annual Re- - -, 1974a: Information content of CCN spectra vs. measurements
port for 1973 to U.S. Atomic Energy Commission, Division of at a single supersaturation. Preprints Conf Cloud Physics,
Biomedical and Environmental Research, Part 3: Atmospheric Tucson, Amer. Meteor. Soc. , 9-12.
METROMEX: A REVIEW AND SUMMARY
- - , 1974b: Cloud physics of urban weather modification. Bull.
Amer. Meteor. Soc., 55, 100-106.
- - , 1975: Overview of urban climate. Conference on Metropoli-
tan, Physical and Environment Sciences, USDA Forest Service
General Technical Report NE-25, Syracuse, 29 pp.
- - , 1976a: The University of Chicago in Project METROMEX.
Tech. Note No. 48, Report to the National Science Foundation,
Grant AEN72-03399-A03, Cloud Physics Laboratory, Dept.
Geophys. Sci., University of Chicago, 25 pp.
- - , 1976b: CCN spectra in C-k space.]. Atmos. Sci., 33,343-345.
- - , 1976c: Modification of clouds and weather by a large metro-
politan area. Preprints Second Sci. Conf Weather Modifica-
tion, Boulder, WMO, 435-442.
- - , 1977: University of Chicago contribution to Project METRO-
MEX-11. Tech. Note No. 50, Final Technical Report to the
National Science Foundation, Grant ENV72-0399, Cloud Phys-
ics Laboratory, Dept. Geophys. Sci., University of Chicago,
105 pp.
- - , and P. Spyers-Duran, 1974: Ice nucleus measurements in an
urban atmosphere. J. Appl. Meteor., 13, 940-945.
Cawse, P. A., 1976: A survey of atmospheric trace elements in the
U. K.: Results for 1975. Report, United Kingdom, Atomic En-
ergy Authority, Harwell, Oxfordshire, AERE-R 8398.
Changnon, S. A., and R. G. Semonin, 1975: Studies of selected
precipitation cases from METROMEX. Report of Investigation
81, Illinois State Water Survey, Urbana, Ill.
Charlson, R. J., A. H. Vanderpol, A. H. Covert, A. P. Waggoner
and M. C. Ahlquist, 1974: Sulfuric acid-ammonium sulfate
aerosol: optical detection in the St. Louis region. Science, 186,
156-158.
Czys, R. R., 1977: University of Chicago measurements of ice-form-
ing nuclei from METROMEX. Preprints Sixth Conf Planned
and Inadvertent Weather Modification, Champaign-Urbana,
Amer. Meteor. Soc., 29-32.
Garrett, V. F., 1975: Urban-rural differences in the boundary layer
thermal structure. M.S. thesis, Dept. Atmos. Sci., University
of Wyoming, Laramie, 129 pp.
Gatz, D. F., 1972: Washout ratios in urban and non-urban areas.
Proc. Conf Urban Environment, Philadelphia, Amer. Meteor.
Soc., 124-128.
- - , 1977: A review of chemical tracer experiments on precipita-
tion systems. Atmos. Environ., 11, 945-953.
- - , R. G. Semonin, and M. E. Peden, 1978: Deposition of Aero-
sols. Summary of METROMEX, Vol. 2, Causes of Precipi-
tation Anomalies, lllinois State Water Survey, Bull. 63, Ur-
bana, 345-376_
Hales, J. M., and M. T. Dana, 1979: Precipitation scavenging of
urban pollutants by convective storm systems. J. Appl. Me-
teor., 18, 299-316.
Hane, C. E., 1978: Scavenging of urban pollutants by thunderstorm
rainfall: numerical experiment. J. Appl. Meteor., 17, 699-710.
Johnson, D. B., 1976: Ultragiant urban aerosol particles. Science,
194, 941-942.
Komp, M. J., 1977: Visibility reductions and accompanying aerosol
evolution downwind of St. Louis, Missouri. M.S. thesis, Dept.
Atmos. Sci., University of Wyoming, Laramie, 125 pp.
- - , and A. H. Auer, Jr., 1977: Visibility reduction and accom-
panying aerosol evolution downwind of St. Louis. Preprints
Sixth Conf Planned and Inadvertent Weather Modification,
Champaign-Urbana, Amer. Meteor. Soc., 25-28.
- - , and - - , 1978a: Assessment of some aerosol properties in
midwestern power plant plumes. J. Wea. Mod., 10, 107-118.
139
--,and--, 1978b: Visibility reduction and accompanying aero-
sol evolution downwind of St. Louis. J. Appl. Meteor., 17,
1357-1367.
Littman, F. E., 1978: RAPS Emission Inventory Summarization.
Final Report to EPA by Rockwell International, Task Order
108J, Contract 68-02-2093.
Semonin, R. G., 1976: The variability of pH in convective storms.
Water, Air, Soil Pollut., 6, 395-406.
- - , 1978: Urban boundary layer-aerosol patterns. Summary of
METROMEX, Vol. 2, Causes of Precipitation Anomalies, Il-
linois State Water Survey Bull. 63, Urbana, 103-110.
- - , and S. A. Changnon, Jr., 1974: METROMEX: Summary of
1971-1972 results. Bull. Amer. Meteor. Soc., 55, 95-100.
Sisterson, D. L., 1975: Studies on the urban moisture budget. Rep.
No. AS114, Dept. Atmos. Sci., University of Wyoming, Lar-
amie, 52 pp.
Shea, D. M., 1976: Observations of the regional influence on ther-
modynamic properties and aerosol patterns in the plume down-
wind of S<. L,ouis. Rep. No. AS117, Dept. Atmos. Sci., Uni-
versity of Wyoming, Laramie, 63 pp.
- - , and A. H. Auer, Jr., 1978: Thermodynamic properties and
aerosol patterns in the plume downwind of St. Louis. J. Appl.
Meteor., 17, 689-698.
Spyers-Duran, P., 1974: Cloud condensation nuclei measurements
and estimates of production rates in the St. Louis urban com-
plex. Preprints Fourth Conf Weather Modification, Ft. Lau-
derdale, Amer. Meteor. Soc., 390-395.
- - , 1974: The effect of repeated activation of depositional ice
nuclei. J. Appl. Meteor., 13, 628--629.
Twomey, S., 1959: The nuclei of natural cloud formation. Part II.
The supersaturation in natural clouds and the variation of cloud
droplet concentrations. Geofis. Pura Appl., 43, 227-249.
White, J. M., 1977: The net radiation budget of the St. Louis met-
ropolitan area. M. S. thesis, Dept. Atmos. Sci., University of
Wyoming, Laramie, 99 pp.
--,F. D. Easton, A. H. Auer, Jr., 1978: The net radiation budget
of the St. Louis metropolitan area. J. Appl. Meteor., 17,
593-599.
Wong, K. K., 1976: Urban perturbations on the mixing layer airflow
in St. Louis. M.S. thesis, Dept. Atmos. Sci., University of
Wyoming, Laramie, 136 pp.
Young, J. A., T. M. Tanner, C. W. Thomas, N. A. Wogman and
M. R. Petersen, 1974: Concentrations and rates of removal of
contaminants from the atmosphere in and downwind of St.
Louis. Annual Report for 1973 to the U.S. Atomic Energy
Commission, Division of Biomedical and Environmental Re-
search, Part 3: Atmospheric Sciences, Rep. BNWL-1850,
102-105.
- - , - - , - - , - - , and--, 1975: Concentrations and rates or
removal of contaminants from the atmosphere in and downwind
of St. Louis. Annual Report for 1974 to the U.S. Atomic Energy
Commission, Division of Biomedical and Environmental Re-
search, Part 3: Atmospheric Sciences, Rep. BNWL-1950,
70-76.
- - , - - , - - , and - - , 1976: The entrainment of tracer near
the sides of convective clouds. PNL Annual Report for 1975 to
USERDA/DBER, Rep. BNWL-2000, Part 3, 179-184.
- - , 1978: The rates of change of pollutant concentrations down-
wind of St. Louis. PNL Annual Report for 1977 to USDOE/
DBER, Rep. BNWL-2500, Part 3, Ch. 1, 35-39.
 CHAPTER 7

Summary of Urban Effects on Clouds and Rain

R. R. BRAHAM, JR., R. G. SEMONIN, A. H. AUER,
S. A. CHANGNON, JR., and J. M. HALES

CONTENTS

ABSTRACT ......................................................................... 141

7.1 Introduction ...................................................................... 142
7.2 St. Louis rainfall patterns ........................................................... 142
7.3 Thermodynamic structure of the urban boundary layer ................................... 143
7.4 Boundary layer wind patterns ........................................................ 145
7.5 Cloud frequencies and morphology ................................................... 146
7.6 Radar studies of urban clouds ....................................................... 146
7.7 Urban effects on maximum echo heights .............................................. 148
7.8 METROMEX data on thunderstorms and hail .......................................... 149
7.9 Possible effects of physiography ...................................................... 149
7.10 Possible mechanisms for urban rain anomalies ........................................ 150
7.11 What about anthropogenic nuclei? ................................................... 150
7.12 Recapitulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152

ABSTRACf

METROMEX observations relating to urban weather modification are summarized and examined against
background knowledge and theory.
The diurnal variations of the urban heat island, low-level moisture over the urban area, boundary-layer wind
field and air stability combine to produce low-level convergence and vertical air motions over and downwind
of the city during daylight hours.
Observations show that, during the forenoon hours, the urban area experiences earlier cloud buildup and
enhanced frequencies of radar first echoes due to the rapid deepening of the boundary layer. During the period
from about 1500 to 2100 CDT there was a rainfall maximum centered over the eastern metropolitan area and
adjacent rural areas. This rainfall maximum, with an area-average amount~ 25% above background, is attributed
directly to urban enhancement. It apparently comes from urban-induced invigoration of convective clouds
which would naturally have been restricted to top heights less than about 6 km. Associated with these storms
was an area of increased frequency of thunderstorms and hail over St. Louis, and to the east.
A secondary rainfall maximum occurring between 2100 and 0300 CDT was found north and northeast of St.
Louis, in the Edwardsville-Alton area. Evidence about these storms is incomplete. The mechanism by which,
and the extent to which these nocturnal storms were modified by the city, were not determined.
Possible mechanisms of urban-weather modification are discussed. These include possible effects of anthro-
pogenic nuclei, physiography and lower urban 0., as well as boundary-layer convection, and enhanced cloud
glaciation. Enhanced boundary layer convection, perhaps with help from enhanced cloud glaciation, are believed
to be the major urban weather mechanisms operating at St. Louis.

141
142
7.1 Introduction
The foregoing chapters of this monograph document
many of the ways a large, industrial city, and man's
activities within it, can change meteorological param-
eters. This chapter attempts to summarize and inter-
relate some of the detail of previous chapters and to
look beyond them to a general understanding of urban
weather modification. Our major objective is to un-
derstand urban effects on summer convective clouds
and convective storm rainfall in the St. Louis area.
This attempt cannot be completely successful be-
cause, in spite of our best efforts, there still are gaps
in our understanding of some important parts of this
problem. Nevertheless, since METROMEX was the
most comprehensive urban weather project so far
completed, a synthesis of some of its findings should
provide an important contribution to meteorology.
Undoubtedly, some of these findings will have to be
modified as further research points out new relation-
ships and provides better theoretical foundations.
We will begin our summary and synthesis by re-
viewing the METROMEX rainfall observations, plac-
ing greatest emphasis on those aspects that appear to
be urban related. We then proceed to an examination
of METROMEX data taken in the subcloud layers,
since these layers strongly influence summer convec-
tive clouds. Moreover, it permits us to proceed from
the more certain to the less certain knowledge in our
discussion. Then we consider our measurements of
clouds and cloud properties to see to what extent they
can be related to the boundary layer. The chapter
closes with a discussion of precipitation mechanisms
for explaining the observed rainfall patterns.
7.2 St. Louis rainfall patterns
The pattern of rainfall around the St. Louis area,
based upon long-term climatological data (Figs. 13 and
14, Chapter 2), shows a SW-NE trending trough of
relatively low rain amounts ( <28 em year- 1) extending
across the western St. Louis suburbs, the bottomland
of the Mississippi and Missouri rivers, and northeast
across Illinois. Almost parallel to this trough, east of
the Mississippi River, was a ridge of relatively high
summer rainfall amounts (>30 em year- 1). This ridge
extended from the Mississippi River on the south edge
of the city, northeastward across Belleville and Cen-
terville, to Edwardsville and beyond. The location of
the ridge of high rain amounts coincides generally with
the downwind side of St. Louis. (Recall from Table 4,
Chapter 5, that the majority of St. Louis area rain
during METROMEX came from storms with move-
ments from the southwest through northwest.)
The total rainfall pattern over the five summers of
METROMEX (Fig. 15, Chapter 2) was broadly similar
METEOROLOGICAL MONOGRAPHS
to the long-term pattern except for features which a
denser network would detect, whereas the bottomland
received high rain amounts. The rainfall ridge east of
the Mississippi is clearly evident as a SW-NE trending
line of maxima ending in a major maximum northeast
of Edwardsville. The latter rainfall maximum is re-
ferred to as the Edwardsville high in earlier chapters
of this monograph.
Rainfall patterns tend to be dominated by the rain
from the largest storms. Normally these storms make
up only a relatively small fraction of all rain events.
Thus, in trying to deduce cause-effect relationships
between St. Louis area rainfall and urban forcing func-
tions, it is important to consider the persistence,
within independent time blocks, of features suspected
of being related to the urban area. The persistence,
across years, of features in the METROMEX rainfall
were studied both on a monthly and seasonal basis
(Huff, 1977; Braham, 1979). It was found that two fea-
tures of the total rainfall map could be regarded as the
most persistent. These were 1) above average rain
amounts in the northern part of the network-roughly
the Bottomland, and the area from Alton to Edwards-
ville and north and east from them-and, 2) a north-
south trough of low rain amounts extending from the
western suburbs to the south across the southwestern
hills. The region directly east of St. Louis-East St.
Louis, Granite City, Belleville, and Collinsville (see
Fig. 2, Braham, 1979)-was substantially above av-
erage during two seasons, slightly above average two
seasons, and substantially below average during one
season. The same distribution was found for the rain-
fall high in the Bottomland area. The Edwardsville
area was well above area average on four of five sea-
sons. The rain in the western suburbs was substan-
tially below area-wide average for three seasons and
just about average on the other two.
From a consideration of persistence, the two fea-
tures that command the most attention are the rain
maximum in the Edwardsville area and the rain min-
imum in the western suburbs. One of the tasks of
METROMEX scientists was to determine to what ex-
tent either, or both, of these could have been causally
related to the urban area. We recognize that they
might in part represent effects due to natural meteor-
ological processes interacting with local physiographic
features, or result from sampling vagaries. But the lat-
ter seems unlikely in view of the fact that these fea-
tures were also found in the long-term climatological
data.
Through an analysis of diurnal variations in rainfall
patterns over the METROMEX network it was shown
that the urban area was characterized by reduced rain-
fall during the late morning and early afternoon, and
by a greatly enhanced rainfall during the late afternoon
METROMEX: A REVIEW AND SUMMARY
and early evening hours. The afternoon-evening max-
imum consisted of two distinct parts, with one peak
in the 1500-1800 CDT period and another in the
2100-2300 period. The 1500-1800 CDT peak maxi-
mized over the city of St. Louis and its suburbs to the
east and west. The late evening peak maximized in the
northern and northeastern parts of the network. For
a variety of reasons which will be discussed, we be-
lieve that these maxima are associated with two dif-
ferent types of rainstorms.
As our discussion of METROMEX data continues,
we will note that there is a substantial body of evi-
dence associating the enhanced afternoon rainfall peak
to increased convection over the urban area. The pre-
midnight peak is less well understood and not so easily
associated with the urban area. The distinct separation
found between the 1500-2100 and 2100-0300 CDT
rainfall peaks suggests that two different storm types
are involved. It is unlikely that the second of these
peaks is due to the so-called nocturnal thunderstorms
because we did not find an associated increase in the
occurrence of thunder or hail. Moreover, most studies
of nocturnal thunderstorms put the time of maximum
occurrence in Illinois as 0400-0600 local time. Indeed,
careful inspection of the METROMEX rainfall data
and thunder data show minor maxima during these
hours. As of this writing, we are not sure of the origin
of the pre-midnight rainfall peak, and we do not know
the extent to which it could have been caused by the
urban area. We will return to these issues following
summarization of other METROMEX observations.
Analyses of the frequencies of rain and total rainfall
amounts associated with storms producing various
categories of total rain at each raingage also provide
important clues for establishing a cause-effect rela-
tionship for urban rain enhancement. As shown by
Huff (1977), storms which gave raingage amounts of
less than 0.25 em showed a maximum in frequency
and in total amounts over the city, the industrial area
east of the city, and in the river bottomland. Storms
which produced raingage amounts between 0.25 and
1.25 em were shown to give a SW-NE trending trough
of low frequencies and low rain amounts which ex-
tended from near the center of downtown St. Louis
across Belleville, Granite City, Collinsville to Alton,
and Edwardsville. This pattern was almost exactly re-
versed for storms producing raingage amounts be-
tween 1.25 em and 2.5 em of rain. With these storms
there was a SSW-NNE trending ridge of high fre-
quencies and amounts which extended from south of
Belleville to northeast of Alton. These two patterns,
taken together, suggest that many clouds which would
naturally have produced between 0.25 em and 1.25 em
of rain were invigorated to a size capable of giving
more than 1.25 em of rain by some action of the city.
143
We will return to possible mechanisms by which the
city could have so influenced the clouds in a later par-
agraph.
Storms which produced raingage amounts of 2.5 em
or more received considerable attention in METRO-
MEX. As discussed in Chapter 2, these heavy rain-
storms showed a strong maximum in frequency and
total rain amounts in the Edwardsville area. The fre-
quency of these storms in the Edwardsville area was
about twice that found over most of the city [20 vs 10,
(see Fig. 22, Chapter 2)]. Because of the major con-
tribution of these heavy rainstorms to total rainfall, it
is not surprising that the pattern of frequency total rain
and the pattern of these storms were remarkably sim-
ilar (compare Figs. 15 and 22, Chapter 2). The excess
rain in the center of the Edwardsville high (about 40
km northeast of downtown St. Louis) could be more
than accounted for by these relatively infrequent
heavy rainstorms. Previously we found that rain in the
Edwardsville area was made up almost equally by the
late afternoon and nocturnal-type storms. An impor-
tant question which has not yet been answered con-
cerns the contribution of these two storm types to the
heavy rainstorms in the Edwardsville area.
7.3 Thermodynamic structure of the urban boundary
layer
Like most other major cities, St. Louis was found
to have a well-defined heat island centered approxi-
mately over the downwind commercial district. The
St. Louis heat island showed a typical diurnal varia-
tion, reaching its maximum size and intensity between
midnight and about 0600 CDT. For explaining urban
effects associated with afternoon and evening convec-
tive clouds, we are most interested in the heat island
properties during these periods when the average shel-
ter-height, urban-rural temperature differences were
~0.5 to 1°C.
It was also found that the air over the urban area
usually was drier than that over nearby rural areas.
(Only during the period from midnight to 0600 CDT
were reverse conditions found.) Dew point deficits
over the urban area reached maximum values of~ 1oc
between 1500 and 2100 CDT. Translated into specific
humidity, this corresponds to a deficit of about 1 g
kg-1.
The heat island and moisture patterns are mainly a
consequence of the different physical properties of the
land-air interface materials in the urban and rural
areas, and the different ways in which these materials
respond to diurnal variations in incident solar radia-
tion. The end result is a diurnal pattern of low-level
air temperature, moisture, stability and vertical mixing
that is slightly, but characteristically, different in ur-
144
ban and rural areas. [Details of these patterns, as ob-
served at St. Louis, are discussed at several places in
the preceding chapters. A recently published theoret-
ical model of heat islands is that of Atwater (1972).]
We will begin our summary of the St. Louis boundary
layer by considering conditions at midday.
Typically during midday hours, ground surfaces are
warm, low-level lapse rates are unstable, and evapo-
ration of available surface moisture proceeds at the
maximum possible rates. In the urban area, tempera-
tures are slightly higher, the unstable air layer thicker,
and large amounts of thermal energy are stored in
pavements, building walls and roofs, and in bare soils.
By late afternoon, the surface radiation balance re-
verses sign, viz., from a net surface gain to a net sur-
face loss of radiant energy. In both the urban and rural
areas, surface temperatures begin to cool and low-
level lapse rates begin to stabilize.
In the rural areas a surface-based radiation inversion
usually begins to form before sunset. This results in
a rapid cooling of surface air temperatures (see Fig.
4, Chapter 2). By sunrise, the rural-area surface in-
version may have developed to a depth of several
hundred meters. This surface-based inversion tends to
decouple the winds from the surface, resulting in
drastically reduced wind speeds and vertical mixing in
rural areas.
However, in the urban area during evening hours
conduction of heat from the ground limits the rate of
cooling and the development of stable lapse rates. At
the same time, the enhanced surface roughness pro-
motes vertical mixing. Often this leads to the forma-
tion of an elevated inversion, or stable layer. The
strong nocturnal heat island over the urban area re-
sults mainly from strong radiational cooling of rural
surfaces compared to the relatively slower cooling of
surfaces in the city.
With the rising sun, temperatures begin to rise in
both urban and rural areas. Air temperatures in rural
areas rise rapidly because heating is restricted to shal-
low air layers beneath the radiation inversion. Over
urban areas, air temperatures start from a higher base
and rise more slowly because deeper air layers are
involved in heating (Fig. 4, Chapter 2).
During the morning hours, warming surfaces pro-
mote the upward flow of heat and moisture by turbu-
lence and convection from ground level sources. This
causes a deepening of the mixing layer over both urban
and rural areas. However, deepening begins earlier
and proceeds more rapidly over the city where it is
less impeded by a surface-based inversion, where
evaporation of surface moisture (dew) consumes less
of the incoming solar energy, and where enhanced sur-
face roughness helps to promote vertical mixing.
During the morning and early afternoon, the top of
METEOROLOGICAL MONOGRAPHS
the boundary layer takes on an everchanging hum-
mocky character which reflects the effects of individ-
ual rising air currents-thermals-feeding on the warm
moist air below and mixing into the drier air aloft.
These and other features of the boundary layer are
easily seen when viewed by vertically pointed lidar.
Thermals which are strong enough to lift air parcels
to their condensation levels become the roots of cu-
mulus clouds. These too can be seen by lidar and by
radar once the clouds are large enough, or last long
enough for precipitation to be initiated. We will return
to the METROMEX data on cloud and rain initiation
in later paragraphs.
Previously we noted that the boundary layer was
slightly drier in the urban area than in the rural areas
except during the pre-dawn hours. In a general way
these observations can be understood by reflecting on
the sources and sinks of airborne water vapor. The
major sources of vapor in the lower atmosphere are
evaporation and transpiration of ground level mois-
ture. These processes are sensitive both to tempera-
ture and moisture availability. Minor moisture sources
to the lower atmosphere are anthropogenic water va-
por release and evaporation from clouds and rain.
Once introduced into the near-surface air layers,
moisture is mixed upward by turbulence and convec-
tion. Thus, the moisture concentration in the lower
parts of the boundary layer represents a balance be-
tween the rate vapor is supplied from below and the
rate it is diluted with drier air aloft. Because there is
less surface moisture available in the urban area, and
because that which is imported with rural air as it en-
ters into the city is mixed aloft and diluted, the urban
air moisture concentration is usually slightly less than
that of the rural (Fig. 6, Chapter 2). A study of the
diurnal variation of dew-point patterns (Dzurisin,
1978) showed that urban dew point deficits were most
pronounced ( -1°C) and covered the largest areas from
1500 to 2100 CDT. During the pre-dawn hours
(0300-0600 CDT), when suppressed vertical mixing
allowed anthropogenic moisture sources to become
more evident, urban dew points were slightly higher
than rural dew points (Fig. 7, Chapter 2).
The combined effect of enhanced convection and
decreased surface evaporation is to produce a dome
of relatively warm dry air over the urban area during
the daylight hours. At St. Louis the vertical extent of
the urban mixed layer was typically > 1200 m in light
winds. With stronger winds the height was smaller and
it was observed to trail off downwind as an elevated
plume of warmer, drier, and aerosol-laden air. Height-
averaged temperature excesses were I to l.5°C and
specific humidity deficits were 1 to 1.5 g kg- 1 with
winds <2 m s-t, becoming respectively about 0.5°C
and 0.5 g kg- 1 when winds were 8-10 m s- 1 • On typical
METROMEX: A REVIEW AND SUMMARY 145

afternoons, the urban boundary layer was 100--400 m wind field divergence data. Data for 1200 CDT on 21
deeper than that over nearby rural areas. Both air tem- fair-weather (no-rain) days showed average diver-
perature and moisture concentration are important in gence values up to -2 x I0- 5 s- 1 (convergence) up to
determining atmospheric thermodynamic stability and about 1 km over the city, the western suburbs, and
the potentially available energy for convective clouds. the industrial and residential suburban areas east of
The appropriate parameter for indicating the combined the city. Beyond these areas the low-level divergence
effects of heat and moisture is the equivalent potential values were small and of both signs. This convergent
temperature (Oe). Observations at St. Louis showed flow deepened to 1.4 km and strengthened to -3 to
that the effects of the moisture (dew point) deficits -4 x I0- 5 s- 1 by 1400 CDT. Above the lower-level
overbalanced the effects of temperature excesses with convergence the mean flow was divergent with mag-
the results that the urban Oe was lower than rural Oe nitudes up to 2 x 10-5 s- 1 •
by 2-4°C through the lowest several hundred meters A similar analysis of data from 16 pre-rain cases
of the atmosphere. This urban Oe deficit was observed showed even stronger low-level convergence (up to
to maximize during late afternoon and evening hours -9 x I0- 5 s- 1 in the lowest 400 m) extending to even
(recall the diurnal variation of dewpoint tempera- greater heights over the entire urban area. The reason
tures). for the greater low-level convergence in the pre-rain
Thus we found that the urban atmosphere at St. cases was not determined. However, one can specu-
Louis is thermodynamically less favorable than nearby late that it reflects regional weather conditions favor-
rural areas for the development of convective clouds. able for convergence in which the urban area tends to
But theory also tells us that it is the combined effect localize, or focus, the upward motion.
of the intensity of the heat island and the low-level air The basic reasons for net upward air motions over
stability, which leads to upward air motions over the the city are the low-level temperature excesses in the
heat island. These upward air motions are centered presence of unstable lapse rates at low levels. Using
over the heat island during calm winds and displaced data from the RAPS network in St. Louis, Shreffler
downwind from it during windy conditions. In either (1978) showed that the maximum upward motions due
case it leads to net air convergence in the lower level to the heat island occurred during the afternoon when
and to divergence in upper levels of the boundary unstable low-level lapse rates combined with a small,
layer. Therefore, we will next consider the METRO- but positive heat island. We recall that at St. Louis
MEX observations of the wind field. there is an afternoon rainfall maximum.
Calculations based on pibal measurements of con-
7.4 Boundary layer wind patterns vergence call for area-average upward motions of 3-7
em s- 1 over the city. Individual measurements by pi-
Since the boundary layer wind patterns are impor- bals traced with double theodolites, airplanes, and
tant in the development of clouds and precipitation, those deduced from dual-Doppler measurements show
during METROMEX, wind patterns were studied by that the upward flow was concentrated in relatively
multiple releases of pilot balloons, airplanes and lidar. small areas where upward velocities often exceeded
Observations at St. Louis showed that wind field 30 em s- 1 and occasionally were a few m s- 1 • Bound-
perturbations caused by the city were observed in al- ary layer model simulations (Hjelmfelt, 1980 and Vuk-
most all meterorological conditions; although details ovich et al., 1976) give similar values.
of the perturbations depended strongly on ambient Chapter 3 reports data obtained from dual-Doppler
wind speed, heat island temperature and time of day measurements on chaff released in the boundary layer.
(surrogate for air stability). Perturbations were most These showed that horizontal roll convection was a
obvious during light wind conditions when they were common flow pattern. Often the upward moving part
manifest in obvious changes in wind directions. How- of the roll seemed to be locked-in over the central part
ever, when the mean flow was vectorially subtracted of the city (in general the region showing the maximum
from each observation, the resulting perturbation afternoon heat island and skin temperatures). Both
flow-field was found to be relatively independent of wind-parallel and cross-wind rolls were observed. The
wind speed (up to a mean flow of about 10 m s- 1). observed upward motions were not uniform along the
Among the more general findings about low-level length of the rolls but seemed to be concentrated in
winds we note that 1) there was increased turbulence regions of 5 to 15 km horizontal scale.
over the city due to greater surface roughness and Thus the METROMEX data show two different
enhanced bouyancy, 2) low-level winds moving into scales of motion in the wind field during afternoon and
the urban area are accelerated and turned cyclonically; evenings. Overall, the urban area was found to be a
the reverse occurs upon leaving the urban area. region of low-level convergence and upward motion.
Multiple pibal releases (12~1600 CDT) provided Frequently the convergence and upward motion take
146 METEOROLOGICAL MONOGRAPHS

the form of horizontal rolls. Within this broad field of 1900 the maximum had moved closer to the city and
convergence were localized updrafts representing in- was located in the Alton-Wood River area. It is rea-
dividual thermals large enough to be detected. These sonable to associate the locations of maximum cloud-
often had updrafts or more than a meter per second iness at 1600 and 1900 CDT with a downwind elevated
and served as the roots for small convective clouds. plume from the heat island but there is no direct proof
The combined effects of the thermodynamic and of this.
kinematic features of the urban boundary layer leads The ground-based all-sky camera at St. Louis Uni-
to a maximum of urban convection during afternoon versity (5 km west of the Arch) showed that the ear-
and early evening hours. The fact that this period co- liest cumulus clouds of each day were more frequently
incides with a maximum of rainfall over the urban area located to the west over the surburbs, than to the east
was previously pointed out. The urban convergence over the downtown. The higher frequencies to the
field should result in an enhanced number of clouds, west could relate to the satellite observations of a
whereas the lower average urban Oe should mean that morning cloud frequency maximum over the Bottom-
these clouds would have higher bases and be less vig- lands (northwest to north of St. Louis University). But
orous than rural clouds. Some of these predictions in view ofthe limited range of the St. Louis University
were borne out by observations, some were not, as camera, it seems more likely that it was seeing differ-
we recall in the next sections. ent clouds. Perhaps these clouds were due to the ac-
tion of the heat island acting on air which had recently
7.5 Cloud frequencies and morphology been over the rural areas and still had essentially rural
moisture values.
Cloud-base heights were measured from airplanes The camera at Pere Marquette State Park verified
and calculated from thermodynamic soundings. Both the early cloud frequencies over the Bottomland as
methods showed that bases of urban clouds were from seen by the satellite. In addition, it occasionally de-
600 to 700 m higher than those of rural clouds (see tected small cumulus clouds anchored to the Portage
Fig. 3, Chapter 4). The higher bases were in agreement de Sioux power plant. Two important findings came
with the observations of a dooming of the mixed layer, from the Alton all-sky camera. It showed that clouds
and the low moisture values over the urban area. Ex- formed preferentially over the Wood River refinery
ceptions were clouds over the Wood River refineries area, and that these clouds preceded, by about 30 min-
where bases were frequently below rural values, ap- utes, the formation of similar clouds over rural areas
parently reflecting anthropogenic moisture from the north and east of Alton. It seems likely that anthro-
refineries. Aircraft measurements of cloud-base up- pogenic moisture, as well as localized anthropogenic
drafts, on 69 clouds on 13 days, gave a median value heat releases are involved in the refinery and power
of 2.5 m s- 1 • This value is compatible with balloon, plant clouds.
airplane and lidar measurements on the more vigorous
subcloud thermals. 7.6 Radar studies of urban clouds
The satellite cloud data on amounts and locations
of maximum cloudiness are only partially understood. METROMEX studies of the initiation of precipita-
The satellite data showed an increase in cloudiness tion were based mainly on RHI radar measurements
after 1000 CDT, reaching a maximum at 1600 CDT. of first echoes (FE). The frequency of PEs (per unit
This is in keeping with the preceding discussions of area) achieved a maximum in a SW-NE oriented
diurnal variations in convective cloudiness, and with plume along the Mississippi River from south of the
Shreffler's (1978) calculations of the strength of the Arch to north of Alton (Fig. 12, Chapter 5). The FE
urban convection cell. frequency over the metropolitan St. Louis area, and
According to the satellite observations, the maxi- in the area represented by 1 h of wind travel downwind
mum cloudiness at 1000 CDT was over the Missis- from it, was a factor of 2 higher than that of rural
sippi-Missouri River Bottomland and at 1300 CDT was areas. The urban enhancement of FE frequencies ex-
over the southeast hills. It seems improbable that tended throughout the period of observations, roughly
either of these was related to processes in the urban from 0900 to 2400 CDT, but showed three maxima: 1)
area; more than likely, they reflected greater amounts 0930-1100 CDT, which corresponds with the time of
of moisture in the near-surface air layers of these rural maximum rate of heating of urban surfaces and max-
areas. By 1600 CDT, the cloudiness was above aver- imum rate of thickening of the urban mixing layer, 2)
age over the city, in keeping with the maximum de- 1400 CDT, which corresponds to the time of maximum
velopment of the heat island circulation. However, at interaction between the heat island and the unstable
this time the maximum cloudiness was found along the air entering it from upstream, and 3) after about 1930
edge of the raingage network east of Edwardsville. By CDT. The urban enhancement of FE frequencies in
METROMEX: A REVIEW AND SUMMARY
the first two of these periods appears to be directly
related to urban effects in the boundary layer as pre-
viously discussed. The third period of high FE fre-
quencies corresponds in time with the nocturnal rain-
fall maximum. However, any cause-effect relationship
between this third group of FE and the urban bound-
ary layer has not been determined.
First echo bases and tops over most of the St. Louis
urban area were found to average about 200 m lower
than FE in rural areas (see Table 8, Chapter 5). These
differences were judged to be statistically significant.
One small area over, and just south of, the main com-
mercial district was found to have FEs with both bases
(90 m) and tops 320 m above corresponding rural val-
ues. The area in which the higher FEs were located
contains most of the high-rise buildings in St. Louis.
It was this same area which contributed most impor-
tantly to airplane observations of higher-than-rural
cloud bases (see Fig. 3, Chapter 4).
Interpretation of the FE top and base heights is
aided by the results of first echo modeling studies.
These showed that the time required to initiate pre-
cipitation was about the same in urban and rural
clouds, in the St. Louis environment, in spite of
marked differences in cloud-base microstructure. As
a consequence, the space interval between cloud base
and FE top depends mainly on the strength of the
updraft. Furthermore, under typical initial conditions,
FE thicknesses are calculated to be about the same,
i.e., FE bases and tops move in unison. Therefore, we
infer that our measurements of higher-than-rural FE
bases and tops over the commercial district reflected
the higher cloud bases and perhaps stronger base up-
drafts. Similarly, we infer that the lower-than-rural FE
bases and tops over the remainder of the urban area
must reflect weaker-than-rural updrafts, since lower-
than-rural cloud bases were not observed in these
areas. Such an inference would be compatible with the
Be measurements.
These arguments then lead to the conclusion that
the strong maximum in FE numbers found along the
Mississippi River at St. Louis (recall Fig. 12, Chapter
5) represented increased numbers of individual cu-
mulus clouds growing to the cumulus congestus stage
required for rain initiation. The fact that this enhance-
ment in FE frequencies maximized during the after-
noon hours, indicates that the convergence field is
more important than the Be field in controlling the for-
mation of individual clouds over the urban area. But
with lower values of Be in the updraft, urban clouds
would tend to be less energetic than their rural neigh-
bors.
At this time this is about as far as we can carry the
synthesis based only on boundary layer and FE data.
Any significant amount of rain from convective clouds
147
requires considerable cloud growth beyond the con-
gestus stage. This leads us to studies of echoes fol-
lowed through their entire life history, and from case
studies of individual storm situations.
A study was made of 702 radar echoes followed
from initiation through their subsequent growth and
decay. In this study, echoes were grouped into urban
and rural categories (see Fig. 19, Chapter 5). These in
tum were divided into those that did, or did not,
undergo merging with other echoes. Urban echoes that
did not undergo merging had slightly greater maximum
heights and substantially greater durations than rural,
non-merger echoes, but the greatest urban-rural dif-
ferences were found for echoes which underwent
merging with other echoes (see Fig. 20, Chapter 5).
Although mergers were found in both groups they
were more frequent in the urban group, and on the
average, the urban merged echoes developed to greater
heights and lasted longer than did the merged echoes
over rural areas. Since, in general, taller and longer-
lived echoes are associated with greater amounts of
rain, these findings are compatible with an urban en-
hancement of rain.
Merging with smaller clouds has frequently been
noted as an important process in the growth of large
convective clouds. Experiments have shown that
tracer chemicals released into the base updraft of one
cloud were found in the rainfall of neighboring clouds
at a later time. These observations support the concept
of mergers. As yet, however, cloud dynamicists do
not have good theories as to how and why merging
occurs. Therefore, the METROMEX data on urban
echo mergers were examined for other features which
might shed light on the merger process in general, and
to mergers over St. Louis in particular.
A model of echo mergers can be postulated from
the METROMEX FE studies. The likelihood of echoes
interacting is a function of echo densities. The FE
studies showed a plume of high FE density over the
Mississippi River, roughly from south of the Arch to
Alton, Illinois. We also found, in the study of702 com-
plete echo histories, that on average FEs subsequently
involved in mergers had greater FE top heights than
those which did not become involved in mergers.
Based on the results of the numerical models, these
FEs probably had above-average initial updraft speeds.
If these clouds also had above-average movement
speeds-perhaps because they had grown more rap-
idly into upper levels where winds are higher-they
would tend to move into the slower-moving echoes.
Such a model would predict mergers to be more fre-
quent over the urban area than over the rural area.
However, some of the data in Table 10, Chapter 5,
places constraints on the applicability of this model.
We note that of 61 urban echo mergers only 16 in-
148
volved an echo which originated over the urban area,
while 45 involved echoes originating in rural areas up-
wind of the city. This suggests that regardless ofwhere
the echo merging took place, the factors most impor-
tant in setting up conditions favorable for merging
were operating preferentially in rural areas upwind of
the city! As of the writing of this monograph, we do
not understand what this factor might be, or how it
operates.
Closely related to the merging of discrete radar
echoes is the merging of discrete rain areas. These
were studied in METROMEX using data from low-
elevation-angle PPI radar data and surface raingages.
A 35-storm sample seen on low-angle PPI gave the
maximum number of mergers in the Belleville and East
St. Louis area (Fig. 20, Chapter 5). Secondary maxima
were found in the SW Hills region and along the
Bluffs, southeast of St. Louis. Analysis of raingage
data showed that maximum frequency of raincell
mergers (Fig. 23, Chapter 5) occurred just east of East
St. Louis and Granite City, in the Bottomlands north
of St. Louis, and in rural areas south and southeast of
St. Louis. These two studies focus our attention on
the area immediately to the east of the city as that
where echo merging occurs most frequently. As pre-
viously indicated, this region is most often downwind
ofthe city during convective storm situations. We also
found that this was the location of greatly enhanced
rainfall during the late afternoon, and the area of in-
creased thunderstorms and hail.
7. 7 Urban effects on maximum echo heights
Another set of data that seems to have an important
place in our quest for understanding of urban effects
on clouds and rain came from a study of top heights
and location of the tallest echo within the surveillance
areas of the Greenville radar at every half hour of its
operation. These data, called Hi-Cu, were grouped
according to their location over the urban area, or
downwind from it by 1, 2 or 3 hours of echo move-
ment. Echoes outside the urban and downwind areas
were called rural.
It was found that the urban Hi-Cu frequency, per
unit area, was about twice that of the rural area during
light winds. With increasing winds the maximum fre-
quency shifted downwind by an amount equal to 1 to
2 hours of wind travel and decreased to a factor of
~ 1.5. These urban-rural ratios were found to be sta-
tistically significant and led to the conclusion that the
urban area acts in some way to increase convective
cloud echo heights. (It is interesting to note that Ait-
ken nucleus-formation processes are sufficiently strong
to maintain concentrations of 4 to 7 fold in excess of
ambient levels in the first 1 to 2 hours of plume travel
METEOROLOGICAL MONOGRAPHS
from primary emission sources. Used as a tracer,
these observations indicate the presence of the urban
signature in the environment of the Hi-Cu.)
The top height distributions of urban and rural Hi-
Cu were quite different (Fig. 27, Chapter 5). Rural Hi-
Cu showed a distinctly bimodal top height distribution
with maximum frequencies at~ 18 000 ft (5.48 km) and
38 000 ft ( 11.58 km), and a distinct minimum of echoes
in the 26 000-36 000 ft (8-11 km) height interval. It is
speculated that, in summer, the natural atmosphere
near St. Louis frequently contains some mid-tropos-
pheric feature (5-8 km msl) which acts to arrest the
growth of many cumulus clouds. This could be either
a stable layer or the top of the low-level moisture
layer-observations suggest that the latter is more fre-
quent.
The urban Hi-Cu showed little or no tendency for
a bimodal height distribution. This led to the conclu-
sion that the action of the urban area was such as to
destroy the arresting layer or to cause urban cumuli
to overcome (penetrate) it, thus obliterating the bi-
modal distribution. With the aid of numerical model
simulations, it was concluded that the increased heights
of urban clouds could have resulted from any of three
causes: a) increased horizontal dimensions of urban
clouds, b) moistening of the urban cloud environment,
and c) anthropogenic effects on cloud glaciation.
There is no direct evidence, recognized thus far, to
favor any one of these causes although the echo and
raincell mergers appears to support the second factor.
Some indirect evidence favoring cloud glaciation will
be discussed in a following section.
When the urban/rural Hi-Cu frequency was exam-
ined in a joint time-height framework (see Fig. 28,
Chapter 5), we found that urban enhancement of cloud
heights was most pronounced during the period from
prior to 0800 CDT until about 1400 CDT. Urban en-
hancement during this period involved clouds of all
top heights from 15 000 ft (4.57 km) to over 45 000 ft
(13.72 km) but was most pronounced in the interval
from 25 000 to 40 000 ft (7.62 to 12.19 km), which
corresponds to the minimum in rural Hi-Cu heights.
Thus the city was acting to produce big convective
clouds early in the day. The correspondence in time
of day of urban enhancement of Hi-Cu and FE again
suggests that the urban-induced convergence and net
upward airflow overcomes the effect of a slightly drier
boundary layer in the growth of clouds.
There also were indications of an urban enhance-
ment of clouds of intermediate size, 20 000 to 35 000
ft (6.10 to 10.67 km) height, after 2100 CDT. As of this
writing we have not been able to relate this observa-
tion to a specific property of the urban area. However,
the time period involved corresponds to the time of
the nocturnal rainfall maximum.
METROMEX: A REVIEW AND SUMMARY
7.8 METROMEX data on thunderstorms and hail
Meteorologically closely related to large radar echoes
are the phenomena of thunderstorms and hail. As re-
ported in Chapter 2, during METROMEX observa-
tions were made on both of these weather elements.
We found that the frequency with which these weather
elements were observed over, or downwind of the ur-
ban area exceeded that over corresponding rural
areas.
The frequency of days with thunder showed a sharp
maximum over downtown St. Louis and extended east
from it as far as Scott Air Force Base and Centerville,
and perhaps also to the northeast beyond the network
(see Fig. 25, Chapter 2). We also found that electrical
power outages due to lightning strikes showed a max-
imum in about the same area (Fig. 4, Chapter 8). These
provide strong evidence that St. Louis is increasing
the occurrence of thunderstorms. The question is
how? The diurnal variation of hourly thunder occur-
rences showed a minor maximum during the 0200-1000
CDT period and a broad afternoon peak from - 1400
to midnight at all stations.
The early morning maximum is evidently associated
with the so-called nocturnal thunderstorms. This time
period agrees with the data of Means (1944) and Wal-
lace (1975) who have previously studied nocturnal
thunderstorms. The relative frequency of these early
morning storms was somewhat greater at Edwardsville
than at Tyson Valley, which is a rural station south-
west of St. Louis. This could indicate an urban en-
hancement of the early morning storms or it could
reflect the natural north to south gradient previously
discussed. In Fig. 1, Chapter 5, we also note a minor
maximum in Edwardsville rainfall during this early
morning period. The afternoon-evening peak is rela-
tively similar for the three stations shown in Fig. 27,
Chapter 2. An analysis based upon six stations, but
only for three years, supports this description of the
diurnal distribution of thunder (Changnon, 1977).
Studies of hail in the METROMEX area showed
increased numbers of hail occurrences, larger average
hailstone sizes, and greater average numbers of stones
(per occurrence) in areas east and northeast of St.
Louis and in the Bottomlands to the north. The MET-
ROMEX hail data shows a very strong peak in fre-
quency of occurrence between about 1500 and 1900
CDT. These times correspond closely with those
found earlier for maximum hail frequencies in Illinois
(Changnon, 1968). Both the time and place of maxi-
mum occurrence of hail correspond with the METRO-
MEX data on thunderstorms and electrical power out-
ages. Previously, we noted that there was a strong
peak in rainfall for the same time and place. All of
these factors show that St. Louis increased the oc-
149
currence of hailstorms. Unfortunately, however, me-
teorologists know all too little about the formation of
hail in natural clouds, yet alone how the process could
be modified by a city, other than the obvious sugges-
tion that hail is a natural consequence of the increased
frequency of large convective clouds. The existence
of larger hailstones well beyond the city has been cited
as evidence for stronger updrafts in urban effected
clouds. This would be contrary to the observations of
average Oe deficits, but one must recognize that we
are discussing relatively few occurrences of hail and
that these could represent non-average conditions.
In considering the METROMEX data on thunder-
storms and hail it is important to note the absence of
even a suggestion of an increased frequency during
the period from 2000-2400 CDT. This means that the
pre-midnight rainfall maximum in the northern parts
of the network probably was not associated with large
convective clouds. One should also recall Fig. 24,
Chapter 2, which shows that very high rainfall rates
occurred with greater frequency in the area immedi-
ately east of the city than in the northern parts of the
network. These factors suggest that the majority of
the pre-midnight rainfall maximum in the northern part
of the raingage network must have come from rela-
tively small clouds, or clouds in a dissipating state.
7.9 Possible effects of physiography
Meteorological literature contains many examples
of studies showing how local physiography played a
substantial role in determining local climate. The
SW-NE trough of low rainfall amounts across the
western St. Louis suburbs, as seen both in METRO-
MEX and pre-METROMEX data (Figs. 13, 14, and
15, Chapter 2) could be related to physiography. One
of the difficult tasks facing METROMEX scientists
has been to determine the extent to which local phys-
iographic features might have contributed to the local
rainfall patterns-both the west side rainfall trough
and the east side rainfall ridge. Three physiographic
features-the Bottomland, the Southwest Hills and
the Bluff line-were considered as possible contribu-
tors to the rainfall pattern at St. Louis.
Moisture from the low, relatively wet, river Bottom-
land undoubtedly contributed to the formation of cu-
mulus clouds in northern parts of the urban area. Per-
haps some of the FE along the Mississippi River were
in clouds which were assisted by moisture evaporated
from the river. We do not yet know how to assess the
importance of this moisture source on area rainfall.
The possibility that the topography around St. Louis
could influence the downwind location of urban effects
must also be considered. Numerical model studies by
Vukovich, et al. (1976) and Hjelmfelt (1980), suggest
150 METEOROLOGICAL MONOGRAPHS

that there may be important interactions between the An example of the favorable conditions at St. Louis
topography and the downwind thermal plume. In west affecting strong convection are reflected in the Hi-Cu
and southwest wind situations, calculations showed an observations. Radar observations showed that almost
elongated urban plume extending to the northeast (to- half of the time rural convective echoes were limited
ward Edwardsville). With northwest winds terrain ef- in growth by an arresting layer located near the - 10°C
fects tend to cancel thermal effects. However, these level. However, echoes over and immediately down-
studies apply only to mid-day conditions. They do not wind of the urban area were able to penetrate this
apply to nocturnal conditions and provide no insight layer and grow to substantially greater heights. This
into the nocturnal rain maximum at, and northeast of, urban-induced increase in the numbers of very large
Edwardsville, Illinois. Funneling of moist surface air echoes was measured by radar and can be inferred
up a valley east of the raingage network might be re- from the downwind increases in heavy rain, thunder-
sponsible for two secondary maxima in radar first storms and hail.
echoes east of the city. Urban-induced cloud glaciation, and the attendant
In view of the considerable amount of theoretical heat-release, was proposed as a mechanism which
and modeling evidence, linking the afternoon rain would allow urban clouds to penetrate the arresting
maximum to urban effects, in contrast to only weak level. However, measurements of anthropogenic ice
evidence linking it to physiography, we conclude that nuclei suggested that St. Louis was, at best, only a
the latter effects are minor. The potential for a phys- weak source. This tends to argue against the glaciation
iographic effect contributing to the 2100-2300 CDT mechanism. However, urban-induced changes in cloud
rain maximum is not so easily dismissed. Until we drop spectra appear uniquely favorable for promoting
have better theoretical insight into the nature and ice particle multiplication through the Hallett-Mossop
cause of this second rainfall peak, the possibility of (1974) mechanism. Since enhanced glaciation of urban
important interactions between physiography and the clouds was not directly observed, this mechanism re-
urban thermal plume, during stable boundary layer mains unproven.
conditions, must be considered undetermined. Another mechanism proposed to explain rainfall in-
creases downwind of the city was one based upon the
7.10 Possible mechanisms for urban rain anomalies
presence of lower Oe airmasses over the city. Recall
Several mechanisms have been proposed to account that the average Oe values over St. Louis were 2 to
for the increase in rainfall observed downwind of St. 4°C lower than over nearby rural areas. Ingestion of
Louis. The most fundamental is one ascribing en- lower Oe air into storms entering the upwind side of
hanced urban rainfall to an increased number of con- the city would decrease the strength of updrafts and
vective clouds as a result of the warm temperatures allow suspended precipitation to fall to the ground. If
and aerodynamic roughness of the urban surface. Cer- this were to happen preferentially over the urban area
tain aspects of this mechanism were observed at St. it should produce a localized rainfall maximum down-
Louis in a wide variety of observations, particularly wind of the city. Observations compatible with the Oe
in such fair weather phenomena as the heat island, mechanism were obtained on several occasions. It is
doming of the top of the mixed layer, downwind ther- not obvious, however, how this mechanism could ex-
mal plumes, wind convergence patterns and first echo plain the observations of increased echo heights and
patterns. more thunderstorms and hail downwind of the city.
Although our data indicates an increased number of However, this mechanism is a candidate for explaining
first echoes over the urban area, it does not follow the peak in rainfall in the 2000--2400 CDT period which
automatically that this would increase rainfall and maximized farther from the city and had less thunder
thunderstorms, especially when the lower average ee and hail than did the afternoon rain maximum. Recall
of the urban boundary layer argues for urban clouds that the maximum urban deficits in low level moisture
being weaker than their rural neighbors. To obtain were observed between 1500 and 2100 CDT.
clouds large enough to account for the rainfall, and
other observations, there is a further requirement that 7.11 What about anthropogenic nuclei?
there be favorable interactions between the urban
boundary layer and the larger environment. Anthropogenic condensation nuclei and anthropo-
Obviously the extent of modification possible from genic gases that feed gas-to-particle reactions are re-
any given heat island and surface roughness pattern sponsible for many unintended weather modification
depends entirely on the properties of the upstream air- effects. They contribute to the formation of hazes that
mass. It appears likely that airmass conditions at St. characterize urban-industrial regions and to the well-
Louis in summer are particularly well suited for urban publicized acid rainfall.
weather modification. Like all major industrial cities, St. Louis is a source
METROMEX: A REVIEW AND SUMMARY
for condensation nuclei. The major source points for
Aitken condensation nuclei are large fossil-fuel power
plants and certain industrial plants. Against a back-
ground of-104m-3 , plumes of Aitken particles from
these sources stand out with their concentrations of
up to 105 cm- 3 (see Figs. 2, 3, and 4, Chapter 6). Aitken
condensation nuclei were observed to have a close
association with gaseous S02 and are thought to be
small sulfuric acid drops and small sulfate particles
(Fig. 12, Chapter 6). The METROMEX tracer exper-
iments demonstrated that surface-released materials
can be transported vertically, scavenged, and returned
in precipitation in the rainfall anomaly area. There-
fore, it seems likely that the fraction of the urban aero-
sol suitable for cloud condensation nuclei are active
in urban-related clouds.
Source points for cloud condensation nuclei were
not specifically identified although one expects that
most of them developed from Aitken particles through
aggregation and continued condensation. The overall
source strength of St. Louis for CCN was -104 cm- 2
s- 1 , or -10- 1 cm- 3 s- 1 since they were formed within
the boundary layer.
Anthropogenic CCN caused marked increases in
drop concentrations, with associated decreases in
mean drop size, in clouds forming within the St. Louis
urban plume. We also found that urban clouds initially
had increased numbers of drops larger than about 20
~-tm diameter. But these drops have appreciable falling
speeds and are consumed in the precipitation process
so that they frequently were not detected in the upper
levels of young clouds or in precipitating clouds.
Wettable particles larger than those conventionally
called CCN play a dominant role in initiating these
very large cloud drops. Therefore in METROMEX a
special effort was made to collect and measure very
large aerosol particles in the boundary layer upwind
and downwind of St. Louis. We found surprising num-
bers of particles out to 60 ~-tm diameter, giving mass
loadings that often exceeded current clean air stan-
dards.
Our conclusions as to the importance of CCN and
large aerosol particles in urban precipitation formation
are based mainly on the results of numerical model
simulations. When aerosol and thermodynamic con-
ditions typical of St. Louis were used to initialize these
models, the calculated times required to initiate pre-
cipitation were only a few minutes different between
upwind and downwind clouds. On this basis we con-
clude that anthropogenic condensation nuclei do not
play a major direct role in the formation of the urban
rainfall anomaly.
However, there is a possibility that these nuclei may
play an important indirect role. Since the days of Proj-
ect Whitetop (Braham, 1964) cloud physicists have
151
noted that ice particle concentrations in clouds often
exceed ice nucleus concentrations by large factors.
The biggest differences are found in warm season,
moist airmass, cumulus clouds such as typify St. Louis
summer cloud patterns. Efforts to isolate the cause for
the observed ice particle "multiplication" has led to
the so-called Hallett-Mossop mechanism. This mech-
anism is thought to be one in which there is mechanical
fracturing of ice splinters from graupel that are grow-
ing through the combined processes of riming and va-
por deposition when such a particle collides with a
water drop larger than about 25 ~-tm. Large concentra-
tions of drops less than about 13 ~-tm favor vapor de-
position growth while drops larger than 25 ~-tm provide
the needed impact energy to cause splintering. As pre-
viously pointed out, urban-affected clouds seem to be
uniquely favored for operation of this model, with
their large concentrations of small drops and the pres-
ence of a large-drop tail. Whether or not urban clouds
actually contain increased numbers of ice particles
was not determined in METROMEX in spite of con-
siderable effort.
Measurements of anthropogenic ice nuclei at St.
Louis were made using both the membrane filter tech-
nique and an expansion chamber technique. These
measurements suggested that certain industries along
the Mississippi River probably are weak sources for
ice nuclei; however, the overall urban enhancement
of ice nucleus concentrations appears to be too small
to have meteorological importance.
Many atmospheric chemistry studies related to air
and rain quality occupied a large measure of effort on
METROMEX. The results are not included in this
chapter because they do not appear immediately rel-
evant to the issue of how the urban area alters the
timing or placement of rainfall. Readers are encour-
aged to consult Chapters 6 and 8 where these studies
are reported in considerable detail.
7.12 Recapitulation
Based upon a number of cohesive lines of evidence,
both observational and theoretical, we conclude that
there is an urban-induced increase in summer rainfall,
thunderstorms and hailstorms over St. Louis and the
area immediately downwind from it. The primary
causal agents for these weather effects are the warm
and aerodynamically rough urban surfaces which pro-
mote increased mixing and convection within the
boundary layer. Some of the clouds which are initiated
by this convection outgrow competing rural clouds
which often are restricted in their growth by a mid-
tropospheric arresting level. We speculate that the
ability of some urban clouds to overcome the effects
of a low Be updraft and the presence of a mid-tropo-
152
spheric arresting level is due to urban-enhanced cloud
frequencies, resulting in mergers, and possibly urban-
enhanced cloud glaciation.
Urban enhancement of thunderstorms and hail-
storms occurs mainly between 1500 and 2100 CDT in
an area extending from approximately the downtown
commercial district eastward to include Scott Air
Force Base and Collinsville. The enhancement of
thunderstorms and hailstorm periods is estimated at
45 and 31%, respectively.
Urban enhancement of total rainfall during the pe-
riod from 1500 through 2100 CDT, over the city and
the close-in area east and northeast, appears well es-
tablished (see Figs. 1 and 2, Chapter 5). Rainfall en-
hancement during this period is centered near East St.
Louis and extends over the Alton, Collinsville, Belle-
ville and Edwardsville (see Fig. 19, Chapter 2) area.
The area-average urban-related increase is estimated
to be 25%. These urban-related weather effects appear
directly related to enhanced convection over the urban
area, which reaches its maximum intensity during the
same time interval.
The rainfall observations indicate a secondary max-
imum extending from -2100 CDT until 0300 CDT with
a peak near midnight. For several reasons, none really
compelling, we believe that this period of rain repre-
sents storms of different character than those which
produced the rain, thunderstorms and hailstorms of
the late afternoon. These midnight storms show a
strong north-south gradient with a maximum across
the north edge of the raingage network. Compared
with the afternoon storms, there is much less evidence
that these storms were modified by the urban area,
although the low Oe mechanism offers a possibility.
But it is clear that the maximum in seasonal rainfall
at Edwardsville represents a combination of rain from
the midnight storms and the afternoon convective
storms. If one accepts the view that all of the storms
in the Edwardsville area were urban-affected, it leads
to a point estimate of the urban rainfall of -49%; if
only the afternoon were urban-effected the corre-
sponding figure is -30%.
Some important issues relating to St. Louis area
rainfall remain unresolved. The role of local physiog-
raphy in influencing the location and amount of rain
is thought to be secondary but not insignificant during
the day. The amount of such influence on nocturnal
storms has yet to be determined. We did not establish
whether local increases in rain may have been asso-
ciated with wide-area decreases. No such decreases
were found. But it is clear that the net energy released
to the atmosphere by the urban area is rather small
compared with the net latent heat of condensation rep-
resented by the increased rainfall. Therefore, viewing
METEOROLOGICAL MONOGRAPHS
the problem in a broad context, one wonders whether
the main effect of the urban area is to trigger and lo-
calize the release of instability from the air passing
overhead.
Other important meteorological issues raised by
METROMEX include 1) a more complete explanation
of the bimodal top height distribution for rural-area
convective clouds, 2) better resolution of questions
dealing with anthropogenic ice nuclei and ice particle
concentrations in urban clouds, 3) better understand-
ing of the nature of the storms causing the rainfall peak
occurring near midnight, and 4) an explanation of ur-
ban effects on convective storm mergers.
This chapter, and especially this closing section, has
focused only on a few key topics relating to urban
effects on clouds and rain. A large body of the MET-
ROMEX findings dealing with other urban meteoro-
logical effects was not included in this summary.
Readers are urged to consult the body of this mono-
graph, and its references, for other effects and the
details thereof.
REFERENCES
Atwater, M. A., 1972: Thermal effects of urbanization and indus-
trialization in the boundary layer: a numerical study. Bound.
Layer Meteor., 3, 229-245.
Braham, R. R., Jr., 1%4: What is the role of ice in summer rain-
showers? J. Atmos. Sci., 21, 640--645.
- - , 1979: Comment on urban, topographic and diurnal effects on
rainfall in the St. Louis Region. J. Appl. Meteor., 18, 371-375.
Changnon, S. A., Jr., 1968: Climatology of hourly occurrences of
selected atmospheric phenomena in Illinois. Circ. No. 93, Illi-
nois State Water Survey, Urbana, 28 pp.
- - , 1977: Thunder analysis for METROMEX. Summary of MET-
ROMEX, Vol. I, Weather Anomalies and Impacts, Illinois
State Water Survey Bull. 62, Urbana, 113-141.
Dzurisin, G., 1978: Average dew points. Summary of METRO-
MEX, Vol. 2, Causes of Precipitation Anomalies, Illinois State
Water Survey Bull. 63, Urbana, 43-62.
Hallett, J., and S. C. Mossop, 1974: Production of secondary ice
particles during the riming process. Nature, 24, 26-28.
Hjelmfelt, M., 1980: Numerical simulation of the effects of St. Louis
on boundary layer airflow and convection. Ph.D. dissertation,
University of Chicago, 185 pp.
Huff, F. A., 1977: 1971-1975 Rainfall pattern comparisons. Sum-
mary of METROMEX, Vol. I, Weather Anomalies and Im-
pacts, Illinois State Water Survey Bull. 62, Urbana, 13-36.
Means, L. L., 1974: The nocturnal maximum occurrence of thun-
derstorms in the midwestern United States. Misc. Rep. 16,
Dept. Meteor., University of Chicago, 37 pp.
Shreffler, J. H., 1978: Detection of centripetal heat-island circula-
tions from tower data in St. Louis. Bound. Layer Meteor., 15,
229-242.
Vukovich, F. M., J. W. Dunn III and B. W. Crissman, 1976: A
theoretical study of the St. Louis heat island: The wind and
temperature distribution. J. Appl. Meteor., 15, 417-440.
Wallace, J. M., 1975: Diurnal variations in precipitation and thun-
derstorm frequency over the conterminous United States. Mon.
Wea. Rev., 103, 406-419.
 CHAPTER 8

Impacts of Urban Modified Precipitation Conditions

STANLEY A. CHANGNON, JR.

CONTENTS
ABSTRACT 154
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
8.2 Water resources impacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
8.2.1 WATER SUPPLY ..•••..•••.•.•••...•.•.•..•.•••.•••.•..........•.............. 155
8.2.2 SEWAGE TREATMENT . . • . . . • • . . . . . . • • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
8.2.3 SURFACE WATER POLLUTION • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . • • . • . . . . . . . . . . . . . . 157
8.2.4 GROUNDWATER QUALITY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . . . 158
8.2.5 FLOODING . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
8.2.6 DESIGN OF HYDROLOGIC STRUCTURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
8.3 Agricultural impacts ................................................................ 160
8.3.1 EFFECTS OF URBAN-RELATED PRECIPITATION ON CROP YIELDS . . . . . . . . . . . . . . . . . . . . 160
8.3.2 EFFECTS ON AGRICULTURAL PRACTICES . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . 162
8.4 Impacts on business and industry ..................................................... 162
8.4.1 AGRIBUSINESS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
8.4.2 PLANNED WEATHER MODIFICATION INDUSTRY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
8.4.3 IMPACTS ON UTILITIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
8.4.4 OTHER IMPACTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
8.5 Ecological effects .................................................................. 163
8.5.1 PLANTS AND TREES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
8.5.2 SoiLs . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . • . • • . . . . . . . . . . . . . . . . . . . . . . . 164
8.5.3 ANIMALS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . 164
8.6 Impacts on human health and activities ............................................... 165
8.6.1 HEALTH AND SAFETY . . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
8.6.2 EFFECTS ON HUMAN ACTIVITIES AND RELATED ECONOMIC IMPACTS . . . . . . . . . . . . . . . 165
8.7 Impacts on atmospheric sciences ..................................................... 167
8.7.1 FORECASTING OF WEATHER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
8.7.2 PLANNED WEATHER MODIFICATION AND CLOUD PHYSICS . . . . . . . . . . . . . . . . . . . . . . . . 167
8.7.3 NETWORK DESIGN AND OPERATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
8.7.4 INFORMATION FOR NUMERICAL MODELING . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . 168
8.7.5 SETTING RESEARCH PRIORITIES . . . . . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . • . . . 168
8.7.6 PREDICTION OF URBAN ANOMALIES . . . . . . . . . . . . . . . . . . . . . . • • . . . . . . . . . . . . . . . . • . . . 169
8.7.7 UNDERSTANDING CLIMATE CHANGE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
8.8 Institutional impacts ................................................................ 170
8.8.1 URBAN AND REGIONAL PLANNING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
8.8.2 LOCAL AND REGIONAL INSTITUTIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
8.8.3 FEDERAL AGENCIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . • . . . . . . . . . . . . . . . . . . . . . 173
8.9 Review of impacts and implications ................................................... 174
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . • . • . . . • . . . . • • . • • . • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
153
154 METEOROLOGICAL MONOGRAPHS

ABSTRACT

The direct environmental impacts due to St. Louis were the summer weather changes, including increased
local cloudiness (up to 1090), increased total rainfall (up to 3090), and increased severe storm activity (up to
100%). These increases occurred over but also east of St. Louis and within a 4000 km2 area.
The many indirect, second- and third-order impacts resulting from these direct (weather change) impacts
were complex and interwoven. Changes in streamflow, flooding and crop yields appeared around 1950, indicating
that the urban-induced anomalous precipitation had become sufficiently large by that time to induce detectable
changes in such conditions.
Impacts on water resources of the more intense and polluted urban rains included more runoff (+II%), more
local flooding (up to+ 100%), and more stream and groundwater pollution (up to +200%). The aggregate water
resource impacts revealed that the net effect was a considerable economic disbenefit, particularly for the
suburban floodplain area east of St. Louis. Costs due to flooding and water management were translated to
local, state and federal agencies. The findings relating to altered weather conditions influence the atmospheric
sciences including the forecasting of urban weather and in revealing how to better address weather modification
efforts.
Urban-altered precipitation resulted in a net local-area average increase of 3-4% in grain crop yields, which
was reflected in the value of agricultural lands and planting strategies of local farmers. The increases in severe
storm activity led to 100% increases in crop-hail losses, costing both farmers and the insurance industry.
However, the net effect on agriculture appeared to be beneficial. Various shifts of agricultural practices and
recreational behavioral patterns indicate some local perception of the anomalous impacts. Transportation,
including increased vehicular accidents in rains, was adversely affected by the anomalous weather changes.
The myriad of impacts resulting from the St. Louis precipitation changes suggest those produced by a major
city represent a net dis benefit or loss when all factors are considered. However, derivation of all the social and
economic impacts is as yet incomplete. There are economic winners and losers in the effect area.
The impact analyses suggest that the urban-related cloud, rain and storm alterations in and east of St. Louis
are sufficiently harmful to deserve consideration in urban planning efforts aimed at wise land use. Although the
urban anomalies affect taxable income and many other governmental activities, little is yet being done to address
these because they are largely unrecognized at this time.

8.1 Introduction
This section deals with the fourth objective of MET-
ROMEX, the study of atmospheric, societal and en-
vironmental impacts identified as relating to the St.
Louis altered precipitation. Primary topics of impact
research in METROMEX have concerned the effects
of the urban-altered precipitation anomalies on water
resources, agriculture, institutions and the atmo-
spheric sciences. Very little has been done to measure
the effect of urban weather changes on man and his
activities at other cities (CEM, 1977).
The effects of the St. Louis urban-industrial com-
plex on atmospheric processes that produce weather
impacts are those most clearly defined. Since these
weather impacts are the direct (immediate) results of
urban effects on atmospheric processes, they are the
direct, or first (in a sequence) environmental impacts.
The alterations in both the frequencies and magnitude
of various weather conditions already described in
prior chapters result in a locally different climate that
exists in and directly above the city and extends in a
plume-like shape beyond the city.
The ensuing impacts from the climate changes that
have been identified as existing or likely (defined as
second and then third order impacts) have been
grouped according to seven major topics. These in-
clude impacts on water resources, agriculture, busi-
ness and industry, ecology, human health and activi-
ties, atmospheric sciences, and institutions. In some
instances, the impacts listed are actually human re-
sponses to the impacts of the changed climate. These
classification areas include some separation of impacts
and some overlap. For example, the impacts of added
hail damage are treated in the agricultural economic
context and in the context of business (insurance in-
dustry). The consequences of more occurrences of
heavy rain rates (a direct impact), which produce in-
creased soil erosion (a secondary impact), and then
silting of catchment basins and removal of silt (tertiary
impacts) are treated within the context of agricultural
impacts, water resource impacts, and as institutional
response.
Lowry (1972) questioned whether the impacts of
urban altered weather at St. Louis (or anywhere) were
serious. He implied they were not, but proceeded to
elaborate on some impacts in the human bioclimate
including comfort (heat exchange), medical effects,
and aesthetic effects. This pioneering effort did not
treat the myriad of second order impacts resulting
from changes in precipitation or storminess.
Although the urban-induced weather changes are
the most direct, well-measured environmental im-
pacts, this section does not elaborate on these, other
than to present information that generally supports the
reality of the weather changes. This section focuses
METROMEX: A REVIEW AND SUMMARY
on the impacts resulting from the urban-induced
changes in clouds, rainfall, and storminess (see Fig.
1). It is important that impacts on many human activ-
ities are very difficult to isolate and quantify; hence
many impacts noted are qualitative.
8.2 Water resources impacts
The increases in total average rainfall, in the number
and intensity of heavy rainfall events, and in pollutants
in the rainfall, all impact on most local water charac-
teristics and activities related to water resources.
These impact areas include water supply, sewage
treatment, water quality, flooding, and the design of
hydrologic structures.
8.2.1 WATER SUPPLY
An analysis of streamflow data for two small basins
downwind of St. Louis (see Fig. 2) was performed to
examine for possible changes due to the increased
summer rainfall. This was accomplished by comparing
the warm-season values of runoff for one small basin
(Canteen Creek) located in the rain anomaly area with
that of another small basin (Indian Creek) located on
the edge of the altered rainfall area. Neither basin
underwent any major land use changes in the 1941-75
period. The basin curves, based on 5-year moving av-
erages of warm season (May-September) flow (Fig. 2),
reveal a near-continuous positive departure in the
Canteen Creek (major rain increase area) values in the
early 1950's. During the 1941-55 period, the average
May-September runoff at Canteen Creek was 136 mm
(5.34 inches) and that of Indian Creek was 108 mm
(4.27 inches). Their difference (27 mm) was 20% of the
Canteen Creek flow. Since 1955, the Canteen value
was 111 mm and the Indian value was 76 mm; and
their difference of 35 mm was 31% of the Canteen
flow. The warm season runoff increase on Canteen
Creek that is related to the urban-induced rainfall in-
crease is 11%. Thus, one secondary impact of more
total rain is greater runoff (Fig. 1).
The increased runoff enhances supplies in local farm
ponds, in local reservoirs, and in the Mississippi
River. In general, these increases are slight, being
about 10% of the total seasonal input into these storage
facilities from the added rain in the affected area
(-3100 kml, or 1200 mi2).
Effect of the summer season rain and subsequent
runoff increases on groundwater supplies is difficult to
ascertain since long-term groundwater data are gen-
erally not available and because the effect of warm
season rainfall on groundwater resources is often ame-
liorated by evaporation, loss to use by plants and tran-
spiration.
Immediately downwind (east) of St. Louis is the 454
155
km2 (175 mi2) floodplain of the Mississippi River. This
area is effectively a closed aquifer, composed of un-
consolidated sands and gravels 37 m (120 ft) thick
which are underlain by impermeable rock. Ground-
water recharge comes largely from rainfall and infil-
tration from local streams (Schicht and Jones, 1962).
This essentially closed hydrologic system is very de-
pendent on direct rainfall for recharge. Unlike many
areas, groundwater in this area is responsive to heavy
(> 12.5 mm) summer rainfall (Changnon, 1973b) and
thus should reflect some of the urban-increased sum-
mer rainfall. Huff (1975a) shows that -50% of the in-
creased anomalous rainfall comes from increases in
heavy rains (>25.0 mm). However, the heavy and
complex pumpage pattern in the floodplain negates
any specific determination of the rainfall impact on
groundwater levels. This is the second most heavily
populated and industrialized area in Illinois, and it de-
pends almost entirely on the local groundwater for its
supply with an aver•:tge pumpage of 100 million gallons
per day (mgd), or 400 000 m3 • Knowledge of the local
hydrologic cycle, coupled with the runoff and heavy
rainfall results, indicates the urban-produced rainfall
increase results in about a 5% increase in shallow
groundwater recharge in the summer (Changnon et al.,
1977a).
8.2.2 SEWAGE TREATMENT
In general, summer season increases in runoff have
a beneficial impact on sewage treatment. Increased
streamflow means more dilution and lower concentra-
tions of pollutants in the streamwater (Huff, 1975b).
However, the 30-80% increases in heavy rain events
and rain rates east of St. Louis (a direct impact, Fig.
1) have led to other secondary impacts including an
increased frequency of supercharges, or bypasses of
wastewater treatment plants in the area. The Illinois
EPA has reported a considerable problem with com-
munities in the Mississippi River floodplain east of St.
Louis. Detailed records of bypasses of the sewage
treatment plant at Belleville show a 40% increase has
developed in recent years as the anomaly developed
and intensified (Changnon et al., 1977a). These by-
passes lead to added stream and river pollution down-
stream of local treatment plants.
USEPA regulations call for increased levels of
wastewater treatment when streamflow (for waste-
water release) is 2.5 times or more above dry weather
flow. The Illinois EPA reports (Helfand, 1975) that
defining dry weather flow in itself is a problem in the
anomaly area because the urban-related shift in sum-
mer rainfall has created new and higher levels of low
flows.
Each Illinois community measures and reports its
 VI
0'\
-
~ ~
t;
~ <
~ ~
MORE, HEAVIER RAIN RATES MORE THUNDERSTORMS AND LIGHTNING t;
::!
~ 0

~
~
~ INCREASED GREATER CROP
~ FLOODING OF FIELDS, AND PROPERTY
1;:: BASEMENTS, AND LOSSES AND
~
1;::
STREETS PAYMENTS TO
~ INSURED ;'§
~ ~

EPA ~
COMPLAINS
~ WITH
t; DELAYS IN
~
~ ACTION
>
~
AGAINST "'
> SEVERAL <
~
"'
::; COMMUNITY ~
CONSTRUCTION SYSTEMS 1-
§ OF DRAINAGE
INEQUITABLE FACILITIES
SHARING OF (CANALS AND
REMOVAL COSTS STORAGE
AND LOCAL BASINS)
DISPUTES POLLUTION
LOAD
INCREASES t--------+------~
IN LOCAL
STREAMS

s::
INCREASE OF
STORAGE IN SURFACE
IMPOUNDMENTS
~
~
5
g
n
FIG. 1. Interrelated impacts of certain urban-induced precipitation anomalies at St. Louis. After Changnon et al. (1977a). >
t""'
s::
~
~
;g
Vl
METROMEX: A REVIEW AND SUMMARY
,,
I\
I \.'"""'
J \
I \
,J \ concentrations (Huff, I976b). Only minor changes of
CAIITEEN CREEK~ f \ I
J INDIAN CREE~\.)
1950 1955 1965 1970
ENDING YEAR OF 5-YEAR PER IOD
FIG. 2. Runoff of two basins east of St. Louis for the warm season
(May-September), based on five-year moving averages. After
Changnon eta/. (1977a).
daily effluent (water) quality to the EPA who checks
this against the regulations established for each com-
munity. These community records for the 1973-75 pe-
riod indicate that the regulation levels are exceeded
about twice as often (97% increase) in the anomaly
area as in communities located to the east of the anom-
aly. Disputes between adjacent communities and the
regional water management entity over wastewater
treatment and flooding problems have occurred (Wil-
marth, 1977).
The sewage and water quality problem in the flood-
plain communities is further compounded by the sandy
soils and near surface levels of the shallow ground-
water. The heavy rain-produced fluctuations in the
level of the unusually high groundwater often occur at
the depth of the sewer lines, causing them to shift, and
resulting in breaks that cause underground spillage
into the groundwater (SIMRPC, I975).
The heavy rain anomaly has helped cause many
communities east of St. Louis to have difficulty in
meeting state and federal standards for waste treat-
ment. However, the EPA has generally delayed action
against several communities because of the complex-
ity of the issue (Changnon et al., I977a).
8.2.3 SURFACE WATER POLLUTION
Surface water pollution relating to the rain anoma-
157
lies was extensively investigated as part of a project
within METROMEX (Huff, 1975b). One phase of this
I972-73 study focused on the effects of atmospheric
effluents of St. Louis and Alton-Wood River on the
water quality in the stream waters of two small basins,
one located in the center of the region that is fre-
quently downwind of these urban-industrial areas
(Canteen Creek), and the other less frequently down-
wind (Indian Creek; the basins are shown in Fig. 2).
Sampling of streamwater and atmospheric deposition
was done on a weekly basis and sequentially in se-
PERIOD
lected storms. Urban-generated surface pollutants
OF
ALTON
were not a major source of stream contamination in
All~MALY these essentially rural basins.
Analysis showed that weekly antecedent rainfall
was an important factor in establishing streamwater
chemical concentrations occurred during and imme-
diately following convective rainstorms on these small
basins.
Separation of the dry and wet components of the
weekly deposition samples indicated that over a month,
1975 the volume of wet (rainwater) and dry deposition were
about equal. However, the rate of deposition was
much greater in rain, which occurs in only a small
percentage (6 to IO%) of the total time. Analyses in-
dicated that the dry deposition rate was usually less
than IO% of the wet rate, and with some elements the
dry rate was only I to 2% of the rainfall deposition
rate.
In the streamwater of both basins the highest av-
erage concentrations occurred with sulfates (403 mg
L - 1) and calcium ( 134 mg L - 1); the lowest with zinc
(I mg L- 1), potassium (4 mg L- 1), and nitrates (10 mg
L - 1). In most cases, concentrations were higher in the
Canteen Creek (more in the anomaly area) stream-
water than in Indian Creek streamwater, especially
with respect to S04 and Ca which were more than
IOO% greater than in Indian Creek. The sulfate con-
centrations were more affected by the rainfall than the
other chemical constituents analyzed.
Computations were made of total basin deposition
of wet and dry fallout (atmospheric deposition) and
the total stream load of selected chemical constituents
in a 12-month period. Table I presents the atmospheric
deposition of each constituent expressed as a percent
of its stream total. The atmospheric deposition was
generally small (1 to 10%) compared with the annual
stream load on both basins. However, atmospheric
deposition of nitrates was more than 43% of the stream
load, and the zinc deposition on Indian Creek was
223% greater than stream load.
Since rainfall produces about half of the deposition,
the average urban increased summer rainfall of ~ 5%
(Indian Creek) and 20% (Canteen Creek) can be esti-
158 METEOROLOGICAL MONOGRAPHS

TABLE I. Atmospheric deposition expressed as a percent ofgroundwater storage with most evapotranspirating
stream load for selected chemical constituents during 12-month pe-
(Schicht, 1965). As shown in Table 2, concentrations
riod (June 1972-May 1973). After Huff (1975b).
of atmospheric deposition for most constituents were
Percent for given basin less than 10% of the concentrations found in ground-
and constituent water or river water. The zinc concentrations in rain-
Chemical
constituent Canteen Creek Indian Creek fall (-50% of total deposition) and dry fallout (the
other 50%) are large compared to the concentrations
Sulfates 14.5 17.4
Nitrates 51.6 43.2
in the groundwater and river; hence soil pollution by
Chlorides 2.2 5.2 zinc is indicated. The Changnon et al. (1979) mea-
Calcium 4.6 6.4 surements confirm this with very high zinc values in
Magnesium 1.5 1.7
Sodium 1.5 0.6 the floodplain soils. The potassium content in atmo-
Potassium 10.2 8.3 spheric deposition is also significant. However, potas-
Zinc 15.7 223.0
sium is also moderately attenuated by soil materials.
The atmospheric nitrate deposition is also large rela-
tive to its content in groundwater and river water.
mated to be responsible for a small fraction (2.5% in However, plant uptake of nitrates in soil is significant.
These results suggest most atmospheric pollutants do
Indian and 10% in Canteen) of the atmospheric de-
position percentages shown on Table 1. not reach the primary groundwater storage areas, al-
though high zinc values were found in soils at the 45
8.2.4 GROUNDWATER QUALITY em depths.
Long-term changes in groundwater quality in a rural
Another METROMEX project involved 1) study of well field (within the floodplain) indicated a gradual
the general effect of precipitation scavenging of urban- increase in sulfates since the early 1900's. However,
industrial pollutants on groundwater quality, and 2) it appeared to correspond to a gradual increase in
investigation of the possibility that groundwater qual- pumpage and possibly the resulting upward movement
ity had deteriorated more rapidly in the region where of high sulfate water from the bedrock (Schicht, 1977).
urban-induced precipitation enhanced total rainfall Chloride concentrations did not show a gradual in-
(Schicht, 1977). Groundwater was studied in the flood- crease but increased significantly when the precipita-
plain of the Mississippi River where the urban-induced tion anomaly began to appear. Shifts in the chloride
rainfall shift occurred. and sulfate contents of groundwater in the industrial
There have been reports of an increase in the min- sectors of the floodplain were dominated by pollutants
eral content of the groundwater. For example, because infiltrating from direct industrial sources.
of the gradual increase in the mineral content of The complex factors affecting the groundwater qual-
groundwater, two municipalities in the area east of St. ity, as well as quantity, in and near the Mississippi
Louis relocated their well fields. Changnon (1973b) River and the heavy industrial areas, make it very dif-
suggested, from a study of historical groundwater ficult to measure the exact change in groundwater
chemistry data, that rainfall may be instrumental in quality due to the precipitation-borne pollutants.
increasing the mineral content of groundwater. The However, at the more rural sites, it appears that cer-
chloride and sulfate contents in groundwater measured tain pollutants in groundwater such as chlorides, zinc
in a well field in the anomaly area showed a sizeable and potassium have increased as a result of atmo-
temporal increase during the 1957-63 period, a period
when the rain also increased by 8% (relative to upwind
values). However, it could not be discerned how much TABLE 2. Relationship between chemical constituents in ground-
water east of St. Louis and the Mississippi River with that in atmo-
of this increased groundwater pollution was due to dry spheric deposition.
fallout captured by surface water and subsequent in-
filtration; how much was due to direct rainfall scav- Atmospheric deposition as a percent of
constituents in
enging in the atmosphere followed by infiltration; or
the quantity from surface pollutant sources. River water Groundwater
Changes in groundwater quality from atmospheric
Ca 9 4
pollutants are partially dependent on the percent of Mg 4 2
total precipitation that reaches the water table Na 3 1
K 37 22
(groundwater recharge), on the percent of precipita- Zn >100 950
tion that evaporates and transpires, and to some extent NOa 20 120
on the hydrogeology of an aquifer. Only about 20% of Cl 4 2
the annual average precipitation at St. Louis goes into
so. 13 4
METROMEX: A REVIEW AND SUMMARY
spheric deposition. The amount contributed by the
urban-increased rainfall is very slight, estimated at 5
to 10% of the total groundwater concentration of any
pollutant (Changnon et al., 1977a).
8.2.5 FLOODING
The frequency of short-duration heavy rainfall rates
and heavy rainstorms is increased in a fan-shaped
(4000 km 2 ) area east of St. Louis. Dettwiller and
Changnon ( 1976) have shown a 37% increase from
1871 to 1954 in the warm season maximum daily rain-
fall values in downtown St. Louis, and Huff (1976a)
used METROMEX data to show that in the Edwards-
ville area east of St. Louis the number of heavy (;?;25
mm or 1 inch) rainstorms were 93% greater than else-
where in the area (see Fig. 22 of Chapter 2). Huff and
Changnon ( 1972) studied regional historical data on
point frequencies of storms that produce ;?;25 mm in
24 hand found a 41% increase since 1950 in the Mad-
ison-St. Clair County area east of St. Louis.
The difference between the point averages in the
effect and no-effect areas for excessive 5-min amounts
indicates an 83% increase. A similar analysis of the 1
h values of 2 year or greater recurrence interval values
(38.1 mm or 1.5 inches) indicated a 37% increase in
the effect area (Fig. 23, Chapter 2).
The occurrence of more and heavier, short-duration
rainfall rates and storm events leads to more high run-
off rates producing soil erosion and flooding, both
identified as secondary impacts (Fig. 1). A portion of
the urban-related summer rainfall anomaly occurs in
the extremely flat and poorly drained floodplain east
of St. Louis. Helfand (1975) has reported on the in-
creased incidences of local flooding and ponding in
floodplain communities. Situations leading to flooded
streets and basements are twice as frequent now in
the floodplain area as they were abut 20 years ago
(-1955). Clearly, the increases in heavy storm events
related to urban effects are an important factor, along
with growth in impervious surfaces, in the 100% in-
crease in local flooding. Separation of these two influ-
ences is difficult and many urban hydrologists unaware
of urban rain increases likely blame the shift in flood-
ing totally on increases in impervious surfaces. The
increased flooding delays traffic and damages property
(SIMRPC, 1975). Complaints by citizens to their city
governments have occurred (Wilmarth, 1977), leading
to the passage of a bond issue in one anomaly-located
community (Belleville) for an improved storm and
sewer drainage system.
The urban-related increases in runoff from the more
frequent heavy rainfall events also affect the operation
and management of hydraulic structures in the flood-
plain. Major canals, storage basins, and flood gates
159
exist to drain and/or control the flow of water within
the floodplain and between the floodplain and the Mis-
sissippi River (SIMRPC, 1975). The management of
high water is made more difficult by the fact that the
anomaly-related runoff from the higher ground to the
east of the floodplain is accelerated as it descends
down the bluffs into the floodplain (Wilmarth, 1977).
The floodwater management problem in the flood-
plain is not resolved, and major canals and floodwater
storage basins are being planned for construction in
the floodplain by the U.S. Corps of Engineers and by
the Southwestern Illinois Metropolitan and Regional
Planning Commission. This planning and construction
are tertiary impacts (Fig. 1) but if pursued will be very
costly ones. Total cost of proposed projects is $73.6
million (1973 dollars). The sizeable increase in heavy
rain events, coupled with urban expansion, have made
many hydrologic facilities that were adequate in 1950
now inadequate.
The urban-related increases in heavy rainfall rates
have also affected soil erosion, especially in the rolling
farmlands east of the floodplain. First, 55% of all the
erosion in this area occurs in summer (ARS, 1972),
the time of the rain anomaly, and the factor most af-
fecting erosion is the magnitude of short-duration
(5-30 min) high rainfall rates. Rainfall values from the
local area without the higher rates produce an average
summer estimate of 8.3 tons (7530 kg) per acre (4047
m2 ) of soil eroded. The heavier rainfall regime in the
St. Louis-produced anomaly area yielded an estimate
of 10 070 kg per acre per summer. Thus, the added
erosion (2540 kg) represents a 34% increase in erosion,
much of which is transported in local streams, thus
degrading local water quality (Changnon et al., 1977a).
The added soil load in the streams, particularly that
from the higher elevation rural lands within the rain
anomaly area, has considerable impact on stream-
water quality and on the floodplain. The streamwater,
carrying loads of eroded soil particles during high run-
off periods, descends into the floodplain where, due
to the extreme flatness of the terrain, the streamflow
is greatly slowed. Furthermore, this excess flow is
often stored in special basins because the major drain-
age canals cannot handle it. The change in the flow,
either slowed (in the canals) or stopped (in catchment
basins), causes much of the sediment to be dropped.
Silting in the storage basins and the costs of its re-
moval have become a major problem for the Eastside
Levee District which manages and maintains many of
these hydrologic facilities. Estimated sediment depos-
ited in the rivers and canals located within the south-
ernmost 80% of the floodplain is 585.8 million kilo-
grams per year (SIMRPC, 1975). The lack of local, on-
farm erosion control has led to planning for either con-
struction of 369 sediment catchment basins (at a cost
160 METEOROLOGICAL MONOGRAPHS

of $6.6 million), or a sediment removal program for
the minor drainage channels (at an annual cost of $1.8
million).
8.2.6 DESIGN OF HYDROLOGIC STRUCTURES
These flooding-related problems are reflected in an-
other area, the role the heavy rainfall anomalies have
in indicating shifts in design values for hydrologic
structures and their operations. In the major rain effect
areas of large urban-industrial areas, urban storm
drainage and sewer systems, which are usually de-
signed on the basis of historical rainfall records col-
lected at airports or other nearby climatic stations, are
very likely to be under-designed for the loads they will
be required to carry. The increase in the frequency-
intensity regime of heavy rainfalls east of St. Louis is
most pronounced in short-duration events which pro-
duce excessive rainfalls for storm periods of 3 h or
less. Unfortunately, these short-duration, extreme
rainfall events are the major cause of urban flash-
flooding and the associated property damage and
health hazards created by these events. Adjusting for
the urban effect is becoming increasingly important
with the present emphasis on improved storm drain
and sewer design and engineering. There is increasing
national concern with sewer supercharging and such
allied problems as basement inundations from storm
water, particularly with current requirements to meet
federal, state and local environmental regulations.
Failure to consider urban effects is causing problems
in meeting the established requirements for urban
water resources systems.
8.3.1 EFFECTS OF URBAN-RELATED PRECIPITATION
ON CROP YIELDS
The urban-related increase in summer rainfall im-
mediately downwind of St. Louis should be reflected
in the yields of crops that are highly dependent on
rainfall in summer. Of course, effects will be complex
because of added hail and wind, shifting agricultural
technologies, and the unknown influences of air and
water pollution. Corn and soybeans are the principal
crops of the region and are very responsive to rainfall
in July and August (Changnon and Neill, 1967).
The regional corn yield values of pre- and post-
anomaly periods are shown in Fig. 3. 1 Results of sub-
tracting the differences in the dry, normal and wet
seasons furnish a measure of the early-to-late
"change," if any existed. The change in the corn
yields of the dry seasons represented a 4.9 bu/acre
'Crop yield statistics are presented in bushels per acre to conform
with widespread use and understanding. Conversion to SI units can
be done as 1 bushel = 35.328 liters, and an acre is 4047 m2 •
1961-1976
PERIOD
CONTROL
25. I 57.7
DRY SEASONS
Dl FFERENCE = 2. 9 bu/acre DIFFERENCE = 7. 8 bu/acre
GAIN = 4. 9 bu/acre

8.3 Agricultural impacts

Information available on the agricultural impacts re-
lated to the St. Louis urban weather anomalies relates
to 1) possible shifts in the yields of the primary crops, 76.7
corn and soybeans, and 2) changes in agricultural
practices in the effect area. Measures of the local al- .-'----'----' NEAR-NORMAL
SEASONS
terations of crop yields were derived through a double DIFFERENCE = 3.8 bu/acre DIFFERENCE = 6.1 bu/acre
"target-control" analysis wherein the yields in the ef- GAitl = 2.3 bu/acre

fect (target) area were compared with those in a sur-

rounding (control) area for the pre-anomaly period
(considered that before 1950) and post-anomaly (con-
sidered to be well established after 1960) period
(Changnon et al., 1977a).
The urban-increased temperatures and cloudiness
appear to have no notable effect on agricultural prac-
tices or local yields. The length of the growing season WET SEASONS
in the rural areas has not been altered, and the primary Dl FFERENCE = 4. 3 bu/ acre Dl FFERENCE = 4. 3 bu/ acre
NO CHANGE
urban-produced weather change relating to agriculture
is that due to precipitation. The effects of more rain- FIG. 3. Average corn yields in the target (urban-related rain
changes) and control (no rain changes) counties in wet, near-normal
fall, dirtier rain, and more hail and wind damage were and dry summers for 193~5 and 1961-76 periods. After Changnon
investigated and found to be identifiable. et a/. (1977a).
METROMEX: A REVIEW AND SUMMARY
gain or increase in the effect area. Near normal rainfall
seasons indicated a gain of 2.3 bu/acre. In wet seasons
in the no-effect (early) period the difference was 4.3
bu/acre, the same as in the effect (late) period. Thus,
the urban-related rainfall increase in "wet"
July-August seasons produced no apparent change in
yields. The average annual change in the recent period
amounted to 2.6 bu/acre. The late period average yield
of the anomaly area was 76 bu/acre; thus, the urban-
produced rain increase represented 3.4% of the total
yield. In dry summers, the increase was 7.5% of the
total yield. The annual corn yields in the early period
(effect and control) were compared with those in the
late period, and a two-way analysis of variance showed
the recent differences were significant at the 5% level
(Changnon et al., 1977a).
A similar analysis for soybean yields revealed the
greatest gain in the dry summers (1.7 bu/acre) with an
average of 1.3 bu/acre for all summers. This is 4% of
the effect area average yield (32 bu/acre) for 1961-76.
These yield shifts related to the anomaly were an in-
tegration of all weather influences on crop production
including more rain, more hail, higher wind gusts and
dirtier rain.
Hail insurance loss values established the presence
of a locally high loss area in portions of the anomaly
where loss costs were 100% greater than elsewhere in
the area. The added crop loss in the anomaly area was
0.6 bu/acre in corn yields (1% of the total yield), and
0.2 bu/acre in soybean yields (1% of the total yield)
for the 1948-75 period.
The agricultural effects of additional pollution scav-
enged and deposited by the added summer rainfall in
the effect area could not be directly ascertained. Se-
monin (1976) has shown that, on the average, the rain
in portions of the effect area is often relatively acid
(pH < 4.5), but the pH of the rainfall is quite variable
within the area (see Fig. 18, Chapter 5). A recent soil
investigation (Changnon et al., 1979) revealed that
soils in most of the anomalous rain area were not se-
riously polluted with Cd, Zn and Pb, except within 10
km of the city. The low available amounts of Zn, Pb
and Cd, while offering no threat to crop production,
show that atmospheric deposition has been the main
source of these metals found in soils from 10 to 40 km
east of St. Louis.
The results of the yield analyses are summarized in
Table 3. These show that yield increases in corn and
soybeans, without the associated hail-wind losses,
would be 3.2 bu/acre for corn and 1.5 bu/acre for soy-
beans. These yield shifts (without the hail-wind losses)
are as great as those predicted using weather-yield re-
lationships for southern Illinois and the average urban
rain increase (Changnon and Neill, 1967). These pre-
dicted shifts appear in Table 3. Hence, no appreciable
161
TABLE 3. Effects of summer precipitation anomalies on crop
yields in effect area.
Yield (bushels/acre)
Corn Soybeans
Net yield shifts, all summers +2.6 +1.3
Hail-wind losses -0.6 -0.2
Yield shifts without hail-wind
losses 3.2 1.5
Predicted shifts based on
average rain changes input
into rain-crop yield equations
(Changnon and Neill, 1967;
Changnon, 1968) +2.0to +3.0 +l.Oto +2.0
effect due to altered (poorer) rain quality or the loss
of soil is apparent. The gains from the summer rainfall
increases have essentially overwhelmed the losses due
to added hail and wind and any other factors. The net
yield increases, as shown in Table 3, are 2.6 bu/acre
for corn and 1.3 bu/acre for soybeans.
Coupling these net yield increases with average farm
acreages in the anomaly area indicates that the average
impact (benefit) of the altered weather on the typical
farm has been 146 more bushels of corn per year and
90 more bushels of soybeans. Extending these values
to the two-county effect (anomaly) area (2642 farms)
produced a regional annual increase of 385 700 bushels
of corn and 237 000 bushels of soybeans (Changnon
et al., 1977a). The 10% average July-August rain in-
crease over the two-county area produced a 3% (corn)
to 4% (soybean) increase in the major crop yields.
Translation of these values to monetary gains de-
pends on price assumptions. In recent years prices
have typically been $2.0 per bushel for corn and $6.0
per bushel for soybeans. Use of these values and the
average yield gains per farm (for 1961-76) indicates a
net monetary gain per farm of $832 ($292 for corn and
$450 for soybeans). The gains for the anomaly area
are $771 000 for corn and $1 422 000 for soybeans.
Thus, the average area gain at these prices has been
about $2.2 million per year.
The yield gains are greatest in dry summers (7 .5%
for corn and 6.0% for soybeans) when the increases
are of appreciable value to the farmers. In the wet
summers no corn yield effect is realized and that to
soybeans (1 bu/acre, or 3%) is slight. This greater yield
increase in dry and near normal summers, with little
or no increase in wet summers, has resulted in a more
stable farm income, noted as a tertiary impact (Fig.
1). Thus, there has been both greater income and more
income stability from this locally modified weather.
Studies of public attitudes indicate that some area
farmers are aware of the increased rain east of St.
Louis (Farhar, 1979).
162 METEOROLOGICAL MONOGRAPHS

8.3.2 EFFECTS ON AGRICULTURAL PRACTICES before rates are changed. Furthermore, crop-hail in-
surance companies typically do not yet set different
The effects on crop production due to urban weather rates for relatively small areas in the Midwest. Since
anomalies have further, third-order impacts on agri- rates in the effect area have not gone up commensu-
cultural practices and related activities. The yield rate with the higher loss, others are subsidizing the
analysis showed that the owner (and tenant) of an av- costs of the greater loss in this area.
erage farm gained most ($540 or 65% of the total gain) Another agricultural response, at least institution-
from increases in soybean yields. Corn yield benefits, ally, to the increased erosion from higher rain rates
although bigger in total bushels, were not nearly as has occurred in the Soil Conservation Service. Begin-
important largely because of the higher prices for ning in 1973, local SCS county advisors altered the
beans. Interestingly, crop planting strategies changed rain factors within the design criteria they use to ad-
in the anomaly area in line with this outcome. Chang- vise local farmers for conservation methods (weirs,
non et al. (1977a) found a relative gain of 6% in the grass-waterways, terracing, etc).
area planted with soybeans in the effect area, as com-
8.4 Impacts on business and industry
pared to those in the control area. Corn area showed
a relative decrease. This relative shift to more soybean 8.4.1 AGRIBUSINESS
acreage and to less corn acreage in the effect area is
economically consistent with effects of more rain. The effects of rain-increased yields and the resulting
These relative shifts in acreage in the effect area sug- impacts (more farm income, higher farmland values,
gest that at least some local farmers were aware that changes in soybean and corn acreage, etc.) on local
the area had, for some reason, more rainfall and a agribusiness were not studied. However, some likely
greater potential for soybean yield gains. effects were identified (Changnon et al., 1977a). Rel-
Increased benefits, higher incomes and more stable atively minor (slight) impacts that should exist in the
farm incomes should produce in recent years (iflocally effect area east of St. Louis, compared to the no effect
recognized) a relative increase in the value of the ag- area, include 1) increased profitability to farm equip-
ricultural lands in the effect area (Changnon, 1979). ment firms; 2) increased sales of fertilizers, herbicides
The early and late land values for the effect and con- and pesticides; 3) increases in the number and favor-
trol areas showed a 3.6% relatively greater increase in ability of loans; and 4) additional crop storage space
agricultural land values in the effect area. This indi- and transportation usage to handle the added produc-
cates 1) that some local farmers were aware of the tion.
greater agricultural benefits resulting from more rain The greater number and intensity of hailstorms, and
(but not necessarily the cause, Farhar, 1979), and 2) the resulting higher crop and property losses, impact
that as a result, they were willing to pay more for this slightly on the insurance industry. To date, the private
land. This is a tertiary impact (Fig. 1) but one that insurance companies have not increased their rates in
further relates to increased taxable income to the gov- the effect area. However, losses there are greater and
ernment and sales by agribusinesses. their profitability in the two-county area is lower. As
The added hail (a direct impact), which produces more farmers seek more insurance coverage, a trend
losses to crops and property (secondary impacts), reflected in the crop liability, the relative amount of
could lead to increased purchases of insurance cov- loss to the companies will increase due to the locally
erage (liability) and altered insurance rates in the effect concentrated nature of hailstorm damage. This will not
area. After adjusting the liability values of the effect result in any major problem, because of the statewide
and control areas for differences in areal extent, the or larger base of selling by most companies, but it may
effect area value was higher by $63 per square mile. lead to rate adjustments in the effect area.
This increase, expressed as a percent of the control
value, shows the liability purchased in the effect area 8.4.2 PLANNED WEATHER MODIFICATION
was 12% higher than that in the control area. This INDUSTRY
further indicates that some of the local farmers in the
effect area perceived their greater hail loss. The rela- The results of METROMEX, including the opera-
tively greater income from the higher yields also al- tional experience (Cataneo, 1974), have several im-
lows for more expenditures including more insurance pacts on the weather modification industry (Semonin
coverage. and Changnon, 1975). The field operations and the
The added crop-hail loss has not been reflected in evaluation procedures of METROMEX have provided
increased insurance rates in the effect area. As shown valuable lessons applicable to the design of a planned
in a study of hail suppression (Changnon et al., 1977b), weather modification experiment in Illinois. Experi-
a sizeable and prolonged shift in hail loss is needed ence in conducting the tracer releases into clouds in-
METROMEX: A REVIEW AND SUMMARY
dicated that the cloud-base seeding technique could be
conducted (Semonin, 1972). These tracer operations
have also shown that the real-time display of quanti-
tative radar data is necessary if cloud-base aircraft
seeding is to be used.
The METROMEX synoptic-rain results are encour-
aging for planned weather modification, regardless of
the urban mechanisms that cause modification. The
fact that squall lines exhibit the potential for rainfall
enhancement is particularly important since they are
the major rain producers of the summer. On the other
hand, the results do suggest that increases in rainfall
from squall lines will be accompanied by increases in
hail.
Urban effects also led to local showers and convec-
tive storms during regional dry periods. These kept
the immediate urban and rural area from being as
drought prone as other nearby areas. This suggests
there is some hope to partially alleviate some drought
conditions, and it illustrates those weather conditions
when rain presumably could be purposely induced
during dry summer periods (Huff and Semonin, 1974).
8.4.3 IMPACTS ON UTILITIES
Water supply systems in the effect area are im-
pacted by the anomaly, as indicated in the water re-
sources impacts. There are benefits from the added
water to the surface storage sources and groundwater
sources. The atmospheric related increased rain pol-
lution is slight and of little consequence to local water
utilities.
One of the impacts to the power industry and their
consumers in the effect area is an increase in power
outages due to lightning strikes (Changnon, 1978a).
The pattern of outages, based on township frequen-
cies, is shown in Fig. 4. Also shown is the thunder-
day pattern established from the METROMEX data.
The outage pattern confirms the validity of the MET-
ROMEX thunderstorm and lightning anomaly, and re-
veals a concentration of outages in the suburban area
east of St. Louis. Outages are 5-10 times greater there
than in the eastern control counties (Bond and Wash-
ington). These outages are both costly and inconven-
ient to the power company, the public, business and
industry in the effect area.
8.4.4 OTHER IMPACTS
In general, business and industrial structures in the
area suffer from the urban-generated weather anom-
alies. The increased storminess (hail, wind and light-
ning) produces losses. One such case involved the col-
lapse of an industrial roof during an urban-affected
severe storm in 1973 (Changnon, 1977a). The accu-
mulation of heavy rain on the flat roof because down-
163
!SOLINES BASED ON
10 - - LIGHTN!NG-CAUSEO
POWER OUTAGES
!SOCERAUN!CS Of
18 - - - AVERAGE THUNDERSTORM
DAYS
I
I
I
All 3 or less
?" tages
A112or l ess
outages
FIG. 4. Patterns of lightning-caused summer power outages in
1972-75 and average summer thunderstorm days, 1973-75. After
Changnon et a/. (l977a).
spouts were clogged with hail resulted in too heavy a
load. Such damages are likely covered by insurance
and thus result in either higher premiums for buyers
or losses to the insurance companies. Clearly, storm-
related losses produce work delays and other second-
ary but unmeasured losses.
Another structural loss is produced by rainbome
pollutants. Acid rain has been noted to damage paint
and stone-type structures (Tobin, 1976). However, no
direct measures of such impacts on buildings have
been made in the St. Louis area. Semonin (1976) pre-
sents rain quality data which shows that the industrial
area of East St. Louis-Granite City experiences , on
the average, quite acid rainfall.
8.5 Ecological effects
8.5.1 PLANTS AND TREES
The possibility exists that the localized urban-in-
dustrial alterations in rain and related conditions affect
some flora. Interest in this possibility has resulted in
studies of rain effects on the growth of trees. Harmon
and Elton (1971) and Ashby and Fritts (1972) initiated
ecologically oriented studies based on analyses of tree
rings in and around the La Porte (Indiana) area down-
wind of Chicago where Changnon (1968, 1970) found
an urban-related rain increase. Harmon and Elton
showed a weather-related anomaly in tree rings in the
164
area that they ascribed to a "more favorable rainfall"
climate resulting from a combination of urban effects
on rainfall and of lake effects on weather. Ashby and
Fritts concluded that trees in the area showed increas-
ing effects of man-made pollution on growth. Their
analysis of rainfall could only partially support the oc-
currence of a 30% precipitation change because it was
too insignificant to be detectable statistically. Their
results showed that the total precipitation variable
could explain only about 25% of the total variability
in tree ring size. Similar studies were not pursued in
the St. Louis area because the effect of the rain change
on trees is clearly minor, estimated to be 5%.
Air pollution injury to trees and other plants is a
problem in the St. Louis area (Lanphear, 1970). How
much of this is attributable to increased pollution de-
position by the added rainfall is not clear. Half the
deposition of pollutants occurs in rain and if the urban
rain anomaly of + 20% is roughly linearly related to
the amount of deposition, calculations would suggest
that ~ 10% ofthe total summer pollution deposition on
vegetation is due to the rain anomaly.
In the early part of the 20th century, air pollution
in the St. Louis area was so severe that many ever-
greens and certain other plants did not survive. This
damage was associated with so2 from smoke in the
burning of soft coal by many local industries. Local
regulations enacted in 1939-40 reduced the level of
S02 by more than 75% by 1950. Lanphear's studies in
the 1960's showed so2 damage to sugar maples and
ginkgo trees, primarily in June and July, and it was
not associated with high temperatures nor drought
conditions. Although damage was noted in the city and
St. Louis County, it was usually greatest within 3.2 to
4.8 krn of sources.
Hurne (1968) noted pollution damage to trees
(Chinese elms, locusts, sycamores), shrubs (lilacs,
mock orange), and vines in the East St. Louis area.
Plant vigor was notably reduced and leaves were vis-
ibly damaged. The general lack of flora in this indus-
trial area is considered a result of pollution damage.
The extent of this type of damage beyond the metro-
politan area has not been documented, and it is hard
to assess what impact the urban rain anomaly has had
on natural vegetation. Clearly, more rain should en-
courage growth, as found in the crop yields. Much of
the area relatively close ( 1.6-4.8 krn) to industrial-
vehicular sources, and hence to dry deposition, re-
ceives much of the observed plant damage and soil
pollution.
8.5.2 SoiLs
The erosion of top soils is likely one of the most
important impacts of the urban rain anomaly. Calcu-
METEOROLOGICAL MONOGRAPHS
lations indicate a 34% increase in soil loss (2.8 tons
per acre per year) related to the local increase in high
rain rates.
The possible effect of added rainborne pollutants on
soil quality has been measured east of St. Louis
(Changnon et al., 1979). This study indicated that soil
pollution by Cd, Zn, and Pb has been increased from
Granite City eastward to 40 krn. However, only a
small part of this is due to the rain anomaly. The rain,
surface water and groundwater quality analyses (Huff,
1975b; Schicht, 1977) indicate that expected amounts
of zinc, nitrates and potassium in groundwater in the
bottornlands close to St. Louis do not appear because
of their attenuation (capture) on soil particles. The cal-
culated zinc cycle in the northern floodplain area is
interesting (Changnon et al., 1977a). The atmospheric
deposition east of St. Louis is 0.19 rng L - 1 with 0.08
rng L - 1 appearing in local strearnwater and 0.01 rng
L - 1 in the groundwater. Thus 0.10 rng L. - 1 or 53% is
left in the soil. Changnon eta/. ( 1979) found the typical
zinc content in a 1 crn 2 soil column of 45 ern depth to
be 210 J-tg and calculated it had taken about 55 years
of deposition to accumulate to this value. Since the
rain anomaly in the area accounts for ~ 5% of the total
annual deposition (10% of the 50% of the total depo-
sition from rainfall), the rain anomaly probably adds
only 0.01 rng L - 1 • However, the anomaly induced
added zinc deposition is equal to the amount found in
the groundwater.
The effect of the added soil pollution on the growth
and production of plants and crops was not measured,
although the crop yield analyses suggested no appar-
ent effects. The values of Zn, Cd and Pb found beyond
10 krn of St. Louis are well within natural values found
in some soils and appear to offer no threat to crop
production or contamination of the food chain. In fact,
certain pollutants (nitrates) in soil and a more acid soil
can be helpful to certain crops. Acid soils are benefi-
cial to certain plants (horseradish) grown in the flood-
plain. If soil pollution due to the anomaly is helping
or hurting plants and crops, it is likely a very minor
effect.
8.5.3 ANIMALS
No direct measurements of possible effects of the
urban rain anomaly on wild or domestic animals have
been made. Havera (1973) studied the effects of var-
ious weather variables on the population of rabbits in
the area. He concluded that rainfall in summer months
had no detectable relation to the number of rabbits in
the Illinois areas east of St. Louis. The occasional
incidence of very large hailstones, potentially related
to urban influences, could produce death and damage
to domestic animals (chickens, hogs, etc.) and to small
METROMEX: A REVIEW AND SUMMARY
game animals. Basically, no impact on animals is be-
lieved to exist (Changnon et al., 1977a).
8.6 Impacts on human health and activities
8.6.1 HEALTH AND SAFETY
It has long been recognized that excessively high
temperatures affect health and occasionally result in
death to the elderly and to those with certain illnesses
(heart and respiratory diseases). St. Louis produces
a heat island that increases the high daytime summer
temperatures (Jones and Schickedanz, 1974), and
health problems may result (Landsberg, 1970). How-
ever, the urban additions of rainfall and severe weather
in and beyond St. Louis are not known to produce any
direct effects on health, other than to be helpful in
reducing high temperatures. The added cloudiness
east of St. Louis is beneficial on some days for reduc-
ing maximum temperatures by 1 or 2°C.
A 10-year study of health impacts of air pollution at
St. Louis began in 1974 during METROMEX. The
study under the direction of Harvard University sci-
entists, includes five other towns and cities with lesser
pollution than at St. Louis. The study measures res-
pirable air pollutants (S0 2 , N0 2 , MRP and particulate
sulfates) in and outside a group of homes in south St.
Louis. Gauging the health of adults and children is
done by annual sampling involving questionnaires (to
document respiratory symptoms) and by performing
pulmonary function tests. Preliminary results (EST,
1977) indicate more respiratory symptoms and changes
in pulmonary functions in St. Louis than in other sam-
pled (less polluted) locales. Influences of cigarette
smoking were found to overwhelm any air pollution
effect. A major long-run impact of these results will
be their effect on setting of future air pollution stan-
dards.
The rain anomaly has a sizeable effect on transpor-
tation, causing either delays or accidents, which can
produce injuries and loss of life. For example, down-
drafts from thunderstorms can and do produce acci-
dents to aircraft, particularly to large jets on landing.
A recent assessment of impacts of inadvertent weather
modification (CEM, 1977) pointed to this problem.
Since results show a 40% increase in thunderstorms
and their durations over and east of St. Louis, the
anomaly increases the thunderstorm danger to com-
mercial aircraft operating in the St. Louis area. The
crash of a commercial airliner on its landing approach
at St. Louis in 1973 appeared to occur in the gust front
generated by downdrafts of an urban-affected thun-
derstorm (Schickedanz and Gatz, 1975).
The urban-related increase in heavy rain events in
a densely populated urban and suburban region re-
duces visibility on highways and arteries and de-
165
creases traction. These factors obviously help in caus-
ing vehicular accidents, injuries and occasional deaths.
A study of afternoon (1600-2100 LST) automobile ac-
cidents in seven suburban cities east of St. Louis
(Sherretz and Farhar, 1978) showed that the mean ac-
cident rate per mile (1.6 km) increased linearly with
rainfall amount. The mean (per city) 5 h accident rate
(and associated rainfall amount) was 0.97 in no rain
conditions; 2.03 in afternoon rains of 0.3 to 25.0 mm,
and 2.60 with rains ;;;.25.0 mm. As noted previously,
urban influences increased the number of rain events
of ;;;.25.0 mm by 30 to 100%, depending on location
within the anomaly area. Thus, the precipitation
changes have resulted in an increase in the number of
accidents, calculated as being 20% greater due to the
urban-induced effects. This is of major significance,
both economically and to human health and life.
8.6.2 EFFECTS ON HUMAN ACTIVITIES AND
RELATED ECONOMIC IMPACTS
Reductions in visibility were studied in METRO-
MEX, and these reductions affect human activity in
a variety of ways. Areas of visibility restriction
amounting to 50% of prevailing regional visibilities
were found downwind of the metropolitan area (Komp
and Auer, 1978). During a weekend when lighter ve-
hicular and industrial activity occurs, visibilities within
the anomaly were reduced by only 20%. Downwind
of the Alton-Wood River industrial complex a reduc-
tion by 40% of regional visibility was noted. The con-
figuration of the visibility anomaly was found to be
dependent on the orientation of the winds (see Figs.
5 and 6). Winds from northerly or southerly directions
(Fig. 6) resulted in a reduction by 60% of regional
visibility. Gradients of visibility showed values rang-
ing from 0.1 to 1.0 visibility mile per land mile ( 1.6
km). Such major reductions in visibility affect many
processes (including plant photosynthesis), human ac-
tivities, and the aesthetics of the area. Major among
these are influences on surface and air transportation
wherein 40 to 60% reductions endanger and slow ve-
hicular and air traffic in this sizeable urban pollution
plume. The UC visibility studies (see Chapter 6) also
discerned processes of regional haze development and
the role of clouds in the transformation of S02 , Aitken
particles and CCN into long lasting droplets and ad-
vect over great distances.
Tobin (1976) made an extensive study of how
weather changes affect urban areas. He showed that
heavier rain and poorer visibility resulted in increased
crime. This, in turn, results in more police activity and
added costs in law enforcement. Sassone (1976) points
to a positive linear relationship in urban areas between
166 METEOROLOGICAL MONOGRAPHS

-,
INDUSTRIAL AREA

D 0 hr 1 hr
VISIBILITY REDUCTIONS
UP TO 40%
'-----------------~

2 hr
I
I
J

3 hr 4 hr 5 hr

...... ,
I
I
I
'1ETROPOL !TAN VISIBILITY REDUCTIONS I
AREA UP TO 50% I
I
I
I
I
'--------------------------/
MEAN TRANSPORT WIND ..

FIG. 5. Conceptual model of downwind visibility anomalies when transport wind is west-
erly. After Komp and Auer (1978).

rain amount and 1) costs for education, and 2) costs
for sanitation and sewage treatment. Hence, taxes for
law enforcement, education and water resource sys-
tems have increased in the affected area as a result of
the rain anomaly.
More rain that is polluted increases damage to paint
and stone on houses. This has led to higher repair
costs, and in general, air pollution has reduced median
property values (Tobin, 1976). As a result, income
from property taxes decreased. This is likely confined
largely to the metropolitan area.
The sum of these likely (but largely unquantified)
impacts (higher personal taxes, lower property values
and decreased tax income to cities) from the bad
weather anomaly may have helped encourage emigra-
tion of the more wealthy population to suburbs beyond
the effect area. This, in turn, resulted in relatively
more low income residents in the urban center and
then less taxable income for the affected urban gov-
ernments. In a sense, the urban-produced rain anom-
aly helped produce a series of events that may have
played a small part in the urban decay at St. Louis.
Hoch (1976) revealed that not all social impacts of
increased rainfall are adverse. His findings show that
a 20% rain increase, such as that found in Madison
and St. Clair Counties (but not in St. Louis), will lead
to a 2% increase in average wages for a typical area
in the central United States. Crocker (1976) showed
that altered climate, including more rain, would alter
the local demand for market purchased goods (e.g.,
more rain gear). This could affect the stocks and in-
come of commercial firms in the effect area. Farhar
(1979) has studied summer recreation in the effect area
and concludes that the added rainfall has a slight effect
on the type and timing (delays) of planned activities.
In essence, these impacts (some without direct mea-
surements in the St. Louis area) suggest that living in
the rain effect area is more costly (higher taxes for
several reasons) and that they simultaneously increase
income (Changnon et al., l977a). The higher costs

0 hr 1 hr 2 hr 3 hr
~------------~----------~~-----------4----------~
0 hr 1 hr 2 hr 3 hr 4 hr

INDUSTRIAL AREA

D METRO POL !TAN

AREA

MEAN TRANSPORT WIND

FIG. 6. Conceptual model of downwind visibility anomaly when transport wind is parallel
(north or south) to orientation of aerosol sources. After Komp and Auer (1978).
METROMEX: A REVIEW AND SUMMARY 167

coupled to the rain pollution damages form a sequence weather modification (Changnon, 1977b). These in-
of events that likely help cause or reinforce existing clude information as to seeding technologies, seeding
1) emigration from the city and suburban weather ef- methodologies, modification potential in various
fect areas, and 2) the development of slums, both re- weather and climatic conditions and operational as-
sulting from a host of other socio-economic reasons. pects.
A major question is whether the amount of social im- The rainfall in the area downwind of the St. Louis-
pact of the weather anomaly is significant. Tobin related rain effect area (beyond 50 km or 30 mi east
(1976) and Farhar (1979) indicate that few urban or of St. Louis) approximated that measured west of St.
nearby affected suburban rural dwellers realize that Louis (Changnon et a/., 1978). This suggests that in-
by choosing to live in or near the city, they have cho- creases in rainfall produced through intentional cloud-
sen a climate different from the rural one. Thus, the seeding over localized areas in the Midwest may not
social and real costs resulting from the urban anomaly, produce a measurable increase or decrease in rainfall
which help lead some people to relocate (urban to ru- beyond the major effect (seeded) area (Changnon,
ral) may be largely unknown or unrecognized by most 1976b). The simultaneous increase in rainfall rates,
local dwellers whose responses are but third order im- hailfalls and surface gusts found east of St. Louis sug-
pacts to recognized second order impacts (see Fig. 1). gests that modification to change one of these condi-
tions (rain) may change them all (Changnon, 1977b).
8.7 Impacts on atmospheric sciences Braham (1974) has described several urban-related
changes in cloud droplet sizes, illustrating how certain
8. 7.1 FORECASTING OF WEATHER cumuliform clouds could be made less stable and more
METROMEX results provide a better understand- apt to rain. The METROMEX results of Braham
ing of precipitation conditions and thus will lead to (1976) showed very high CCN concentrations now
better forecasting of precipitation conditions around routinely measured over central and eastern United
St. Louis and other similar urban areas (Changnon, States, presumably as a result of anthropogenic nuclei
1977b ). Any improvements in rainfall forecasts in production. These measurements suggest caution in
areas where 70% of the nation's population reside choosing the nucleus spectrum used in studies of con-
have clear advantages in operating urban water man- tinental clouds. Changnon (1976a) studied radar echo
agement systems and in planning personal and indus- behavior as part of METROMEX, and found that
trial activities. The results also are relevant to air pol- echoes that resulted from merged echoes grew faster
lution forecasting, and have utility in forecasting of (50%), became taller (52%), and lasted longer (122%)
cloud cover for urban airport operations. than non-merged echoes. The study of urban-effects
Analyses of the synoptic weather conditions when on tallest cumulus echoes (Braham and Wilson, 1978)
urban effects on rain develop have pointed to the fact suggests three mechanisms: 1) urban enhanced up-
that under certain conditions, such as squall line and drafts (by thermal forcing and convergence); 2) urban
cold frontal passages, urban effects often act to en- inducement of more clouds that progressively moisten
hance the rainfall (Vogel, 1974). Of particular rele- the lower troposphere; and 3) seeding of warm clouds
vance to people and their activities will be the im- by anthropogenic ice nuclei. Warm clouds are those
proved capability to forecast those conditions when that form precipitation in a cloud volume with tem-
urban effects combine with nature to produce 25 mm peratures above freezing. These concepts and the
or greater rainfalls in and/or beyond the city (Huff, merging results collectively indicate that rural clouds
1976a). Better forecasts of these conditions would be in Illinois may be suitable for rain enhancement
of consideration to local transportation systems and through dynamic seeding.
the operations of urban water management systems. METROMEX supported research led to improve-
METROMEX results on wind and other atmospheric ments and development of various weather instru-
conditions describing the atmospheric stability indi- ments. A remote recording, audio thunder detector
cate the area and time where urban-induced clouds was developed (Changnon et al., 1977a) and several
will form (usually before natural clouds) and those improvements in aircraft-mounted sensors of the
conditions (times and days) when urban effects will Knollenberg type were made. METROMEX results
initiate local showers and storms (Vogel, 1975; Chang- reveal that single supersaturation CCN counters should
non and Semonin, 1975; Ackerman et al., 1978). be used with great care if meteorologically significant
data are to be obtained. Winter results showed that a
thermal diffusion device (operating at a supersatura-
8.7.2 PLANNED WEATHER MODIFICATION AND
tion of 0.7%) would have overestimated cloud drop
CLOUD PHYSICS
concentrations 95% of the time with an average error
METROMEX has generated a host of results relat- of 58%. Other University of Chicago results (Chapters
ing to both experimental and operational aspects of 4, 5, and 6) were important in adding knowledge 1) to
168
the theory of transient supersaturations in CCN
counters leading to an understanding of how such de-
vices must be used; 2) of how to construct a constant
flow IFN chamber; and 3) about how haze develops
and expands over entire regions (Braham, 1977).
The METROMEX results also suggest that large
urban-generated condensation nuclei have little effect
on the coalescence process. This suggests the hygro-
scopic modification process may not be important in
the Midwest. Changnon et al. (1976) have shown that
the enhancement of the rainfall in the St. Louis area
often occurs as a result of more mergers of the greater
number of convective cells generated over the city
area. The high-cumulus results (Chapter 5) also indi-
cate that thermodynamic enhancement by the city
leads to more rain. These two results suggest that dy-
namic (heavy) seeding to change the ice phase, so as
to increase cloud growth and to secure merging, is a
feasible way to enhance rainfall in the Midwest.
These and many other results (Auer and Dirks,
1974) illustrate the large variety of useful results and
information generated by METROMEX that help in-
crease basic knowledge in the field of cloud physics.
Changnon et al. (1976) and Boatman and Auer (1974)
have developed hypotheses of how urban conditions
affect both the microphysics and dynamics of clouds.
Braham and Squires (1974) in an inventory of cloud
physics, described the value of METROMEX studies
to the field of cloud physics. The first direct airborne
measurements of ultragiant particles occurred as part
of METROMEX (Johnson, 1978).
Other important findings relate to thermodynamic
effects on the atmosphere. For example, observations
of aircraft-dispensed chaff, and as detected by dual-
Doppler radar systems in METROMEX, indicated
that under certain conditions the horizontal roll mode
of convection exists over an urban area, and it is
locked to thermal features generated by the city. The
observations within METROMEX are the first such in
connection with strong forcing from an urban heat is-
land. METROMEX measurements were also the first
to show that the daytime mixing heights are domed
upward over a city like St. Louis, and the results ver-
ified earlier computer modeling of urban effects on the
missing layer depth.
8.7.3 NETWORK DESIGN AND OPERATIONS
Changnon (1975) used METROMEX results to re-
view all of the various factors that go into the design
and operations of a successful mesoscale network, in-
cluding its data processing. The major network mea-
surements defined by the METROMEX goals as being
essential included the anomaly-related network com-
ponents, the networks to measure related surface con-
METEOROLOGICAL MONOGRAPHS
ditions, and certain other instrumentation relating to
the measurements of the low-level atmosphere.
Mesoscale networks like those used in METRO-
MEX often involve complicated operations and hence
they require quality field staff and quality instrumen-
tation. Representative data of high quality can only be
obtained through careful attention to comparable sit-
ing of instrumentation, good servicing and frequent
calibrations, and finally to careful editing and thorough
checking of the data. For example, it was found that
the temperature lapse rate may have definite and
sometimes adverse effects on the ascent rates of
spherical meteorological pilot balloons. An alternative
to the common usage of assumed ascent rates was
proposed for the common environmental lapse rates
(Auer, 1978b). The good liaison established between
METROMEX aircraft crews and local controllers of
the Federal Aviation Agency helped prove that me-
teorological aircraft operations can be conducted in
busy air space if proper liaison is maintained.
8.7.4 INFORMATION FOR NUMERICAL MODELING
Computer modeling using one-dimensional and two-
dimensional cloud models (Ochs, 1974; Murray et al.,
1978; Johnson, 1978; Fitzgerald, 1974; Ochs and Se-
monin, 1976) and mesoscale models (Hane, 1978; and
Dirks, 1974) was extensively employed in METRO-
MEX. These models have been primarily used to help
understand the physical relationships between the ur-
ban surface and effluents and the atmospheric behav-
ior. They have pointed to the importance of surface
temperature anomalies in affecting wind fields and in-
itiating clouds, and the possible role of urban conden-
sation nuclei in rain formation.
One resulting impact of METROMEX has been to
help refine and develop numerical models suitable for
interpreting midwestern convection and for predicting
the occurrence and magnitude of urban rain anomalies
at other cities. The extensive field data generated by
the METROMEX network and meteorological aircraft
have also had an impact on modeling research being
performed by other scientists (Silverman and Nelson,
1975; Chen, 1976).
8.7.5 SETTING RESEARCH PRIORITIES
Some extremely valuable aspects of METROMEX
have been 1) to dimensionalize, for the first time, a
major case of inadvertent weather modification, 2) to
help delineate further research needed in inadvertent
weather modification, and 3) to generate interest
among the scientific community in inadvertent weather
and climate modification. Changnon (1973a) defined
the few knowns and many scientific unknowns of in-
advertent weather modification. This led to an assess-
METROMEX: A REVIEW AND SUMMARY 169

ment of research needs (Changnon, 1974). A critical 50~----,-----,-----,-----~----·

finding was the extreme importance of identifying

those conditions on land use arrays such as a mega-
lopolis, that could lead from local weather changes to
regional climatic changes and on to hemispheric and
global changes.
The impact of METROMEX on research efforts and
their priorities has been reflected in several places. An
ERDA-sponsored energy workshop in 1975 (ERDA, URBAN POPULATION, mill ions ( 1970 data)
1975) focused on many METROMEX results to de- FIG. 7. Relationship between urban area population and maximum
velop research recommendations for ERDA. Two percentage increase in thunder days found in or near the urban area.
After Changnon et al. (1977a).
workshops sponsored by NSF have concerned inad-
vertent weather modification and both have focused
heavily on METROMEX results to define future re- predictive equation shown in Fig. 7. It is interesting
search priorities (Blanchard, 1975; CEM, 1977), and that the St. Louis rain anomaly and its ensuing impacts
a 1975 conference on the Urban Physical Environment on runoff, water quality and agriculture were detected
sponsored by USDA focused on METROMEX re- to begin in the 1940--55 period when the contiguous
sults. A national assessment of weather modification metropolitan area finally exceeded 1 million in popu-
(WMAB, 1978) drew heavily on the METROMEX re- lation and an area of 260 km2 •
sults, and concluded that extensive studies of inad- The evidence concerning the restricted visibility
vertent modification should be an integral part of an downwind of St. Louis from 1975 and 1976 case stud-
expanded national weather modification program. Re- ies has been synthesized into two conceptual models
search priorities set forth have included needs for 1) (Figs. 5 and 6) that may have application to other met-
further urban studies in other areas, 2) investigations ropolitan areas (Komp and Auer, 1978; and Shea and
of effects of major heat releases (i.e., power parks), Auer, 1978). In these models the magnitude of the
3) initiation of research on influences of regional land anomaly is expressed as a reduction equaling a certain
use changes, and 4) investigation of the social and en- percentage of prevailing regional values; e.g., with a
vironmental impacts. 16 km prevailing visibility the magnitude of the reduc-
tions downwind from an industrial area is 40% or 6.4
8. 7.6 PREDICTION OF URBAN ANOMALIES km, indicating a 9.6 km visual range. These models
also describe the orientation, the areal extent and time
A major goal of METROMEX was to develop suf- of initiation of the visibility restrictions with conjec-
ficient understanding of the causes of urban precipi- tures to the downwind limit of the visibility reductions.
tation modification, in conjunction with cloud and me- The University of Chicago haze development model
soscale modeling, to allow prediction of anomalies at (see Chapter 6) reveals how haze and ozone develop
other cities with a minimum of measurements. A con- over larger areas (entire states).
ceptual model involving use of synoptic weather con-
ditions and radar echo characteristics has been pro- 8.7.7 UNDERSTANDING CLIMATE CHANGE
posed for testing at Chicago, as part of the Chicago
Area Program (Changnon and Semonin, 1978). As yet, A major concern of the scientific community in re-
understanding of the St. Louis anomaly and model cent years has been to understand climatic changes
development are not adequate for quality predictions. and the possible role man has or could have in chang-
However, it has been important to learn from MET- ing climate (MIT, 1970; Landsberg, 1970).
ROMEX (Changnon et al., 1976) that the 1971-75 rain METROMEX has improved knowledge of this is-
findings agreed well with the earlier climatic studies sue. METROMEX findings indicate that the spatial
of the St. Louis (then potential) anomaly (Huff and extent of an urban area's alterations of climate is lim-
Changnon, 1972). This helps verify the reality of the ited. Many of the changes are truly local and exist
earlier climatic findings about several other urban cen- largely within the urban complex and a few hundred
ters in humid climates (Huff and Changnon, 1973). The feet above it. A few other changes, particularly those
fact that the magnitude of the urban thunder anomalies of visibility, clouds and rainfall, extend out to 50 km
found by climatic analyses at several cities was lin- downwind and thus produce mesoscale climatic dif-
early related to population (see Fig. 7), when coupled ferences. Urban aerosol and odors were found to ex-
with the knowledge that the St. Louis thunder anom- tend downwind as far as 100 km in certain circum-
aly was conclusively established in METROMEX, stances.
gives greater credence to the climatic findings and the However, the ability of METROMEX results to
170
specify how the complexities of an urban-industrial
area affect the weather to induce a climatic change
has been critically important both in establishing proof
of man-made climate changes well beyond the city it-
self, and in understanding how the change occurs. A
key issue within the field of man-made climatic change
concerns the areal scale of the changes. Hence, a sig-
nificant question relating to urban effects on climate
concerns the potential combined effects from mega-
lopolises. Could they be additive and trigger climatic
changes on the macro scale?
8.8 Institutional impacts
The institutional impacts have been presented last
since they are all tertiary impacts (and responses) oc-
curring as a result of the direct (weather) impacts and
often integrating the secondary impacts (Fig. 1). The
origins and explanations for many of these institutional
impacts are found in the prior sections of this text.
8.8.1 URBAN AND REGIONAL PLANNING
Two major questions related to urban and regional
planning arise from the METROMEX findings (Chang-
non, 1978). First, what land use decisions can be made
to reduce the urban influences on precipitation, as-
suming they are deemed largely detrimental? Second,
what land use decisions can be made in the weather
anomaly area to minimize adverse effects and to op-
timize benefits?
Let us consider first the strategy of reducing or elim-
inating the effect, at least in future large urban areas.
Since the basic cause of the precipitation changes re-
lates to overall city size, control (through land use
management) seems appropriate to consider only for
future cities. Remedial action for large existing cities
does not seem reasonable.
A key to answering this question is the identification
of the urban factors that influence precipitation. The
METROMEX studies aimed at understanding the
causes of precipitation changes at St. Louis, and those
described herein have indicated the changes result
largely from urban-produced perturbations (thermo-
dynamic and kinematic) extending from the surface
into the boundary layer. Hence, those surface factors,
or man-made alterations, that induce major thermo-
dynamic perturbations into the boundary layer are
most relevant.
Changes in land use to serve urban needs, and that
influence surface weather, include changes in the char-
acter of surface materials, land-scale shape and struc-
ture, heat sources and retention, and evaportranspir-
ation.
The land use typing of St. Louis by Auer (1978a),
which is described as a meteorologically-oriented clas-
METEOROLOGICAL MONOGRAPHS
sification, was compared with (i) detailed surface
weather patterns (temperature, humidity and winds),
and (ii) boundary layer (surface to cloud base) patterns
of the same weather conditions. The surface pertur-
bations in temperature, winds and humidity included
all industrial, commercial and residential areas of met-
ropolitan St. Louis, but they were greatest in the city
core, the adjacent commercial, industrial and compact
residential areas.
The speed and direction of boundary layer winds
(those up to cloud bases) were influenced over the
entire metropolitan area, and as a result, they were
also altered in the rural areas upwind and downwind
of the city (Ackerman et al., 1978). Auer (1978a)
shows how the temperature and moisture changes in
the boundary layer also typically extend over most of
the metropolitan area but with the greatest departures
(> 1.5 g kg- 1 in specific humidity) over the city core.
Auer concludes that "no single land use type may
itself be responsible for a particular thermodynamic,
kinetic or radiative anomaly." However, taken in to-
tality, the METROMEX results show that all urban
land use types are coincident with observed surface
weather anomalies, with their magnitude increasing
with decreasing vegetative areas. Areas with less than
35% vegetative cover (identified by Auer for St. Louis
as including two classes of industrial, one class of
commercial, and two classes of compact residential)
which cover at least 100 km2 appear to be a key for
causing perturbations in the surface and boundary lay-
ers sufficient to lead to identifiable modification of pre-
cipitation. Areas with low vegetative cover ( <35%)
that extend over many kilometers do affect weather
in the boundary layer but alone do not appear to per-
turbate the cloud and rain processes.
However, two exceptions to these two important
findings exist. Residential areas with 50% or more veg-
etative cover help influence the amount and placement
of urban influences if these residential areas are ad-
jacent to the built up, low ( <35%) vegetation areas.
Second, relatively small (10-30 km 2 ) low vegetation
areas incorporating industrial processes with strong
concentrated releases of heat (7 x 1011 cal h- 1 at St.
Louis) and moisture (3 x 108 gm h- 1) can produce
localized effects on clouds and precipitation.
The St. Louis findings, particularly on the signifi-
cant influence of the city center where vegetative
cover is less than 35%, indicate that dispersion of in-
dustrial, commercial and compact residential areas
into small ( <20 km 2 ) zones would be a key to elimi-
nating measureable effects on precipitation processes,
if indeed these are ultimately defined as major disbe-
nefits.
The character of the downwind (east) area of a city
is an important consideration in defining the degree of
METROMEX: A REVIEW AND SUMMARY
benefit or injury from urban-generated precipitation
anomalies. For example, the anomaly area east of St.
Louis receives an inordinate number of disbenefits as
a result of the poorly drained bottomlands. Dispersion
of individual industries that release large volumes of
heat and/or moisture (power plants, refineries and
steel plants) also is desirable as a design strategy to
minimize precipitation effects in future urban and re-
gional development.
The other major planning question relates to what
types of land use decisions can be made within the
anomaly area to minimize the adverse effects and op-
timize the benefits (Changnon, 1979). Inherent in this
decision process is awareness of the disbenefits and
benefits.
The weather changes producing major problems are
esstentially those relating to increased storminess-
more rainstorms, higher rain rates, more lightning,
higher wind gusts and more hail. These impact nega-
tively on everyone-the urban, suburban and rural
resident-and affect most activities, including indus-
try, farming, transportation and recreation. A variety
of design and social adjustments can be considered
including more insurance coverage, better soil con-
servation practices, larger drainage systems, stronger
structures, better grounding of structures and standby
power generators for lightning strikes, greater public
awareness and forecasting, etc. The net effects are
higher costs, inconvenience and anxiety, although cer-
tain interests such as insurance firms and repair firms
ultimately benefit.
The weather changes producing general benefits at
St. Louis relate to the increase in total summer rain-
fall. This is clearly beneficial to those growing soy-
beans and corn and related profit-making agribusi-
nesses. In general, more rain is also beneficial to water
resource systems (improved quality through more di-
lution, more runoff and available water in surface and
groundwater resources).
Other weather changes result in mixed effects. The
increased local deposition of pollutants by rain (which
collects the pollutants) cleanses the air and, as such,
is a general benefit over and beyond the city. How-
ever, the pollutants, including acid rainfall, that are
deposited ultimately hurt soil quality, structures,
stream and groundwater quality, and natural vegeta-
tion. Alternatives to be considered in design include
use of structural materials not harmed by acids, and
use of vegetation varieties less susceptible to pollutant
damage.
Consideration of these anomaly-related disbenefits
and benefits is an issue in regional planning. In gen-
eral, the major benefits accrue to farming and agricul-
ture, indicating that agricultural land use would be a
wise choice for areas east of large cities (and within
171
50 km). However, farmers in such impacted areas will
have to adopt certain practices (more hail and wind
insurance, better soil conservation practices, more soil
treatment to offset pollutants in rain, etc.). The greater
rainfall, coupled with the need for systems to minimize
flooding, suggest that the fan-shaped anomaly area
east of cities would be an excellent place to site water
supply reservoirs.
It appears, based on available results and informa-
tion, that suburban growth, industrial development,
major transportation facilities, and major recreational
facilities, should not be planned for anomaly areas.
For example, the added storminess and decreased vis-
ibility will delay or endanger air traffic to major air-
ports and produce more vehicular traffic accidents. A
major planning recommendation with any land use
east (northeast, east and southeast) of major metro-
politan areas is to design and construct bigger, more
adequate drainage facilities. Corrected (higher) rain
frequency values should be employed in the design of
all water resource facilities, including storm and sewer
drainage systems, storage facilities, sediment removal
systems and water treatment installations. Present de-
sign values based on older data without urban influ-
ences are too low.
Questions that relate to a comprehensive consider-
ation of all economic costs and benefits of the many
impacts cannot be adequately answered because a
complete economic assessment has not been done.
One could claim that the increased tax benefits from
more suburban development (in the anomaly area)
would sufficiently exceed the known $2.2 million av-
erage annual income increase from higher crop yields
to provide enough surplus funds to offset the many
costs for adjusting to the added rainfall problems in
these suburbs. At this time, the costs related to ad-
ditional drainage and water treatment facilities that
can be traced to the anomalies is about $100 million
at St. Louis, indicating costs will greatly outweigh the
agricultural benefits over a reasonable time horizon.
8.8.2 LOCAL AND REGIONAL INSTITUTIONS
The shifts in cloudiness, rainfall and severe storms,
and the host of secondary impacts that occur within
the two-county area east of St. Louis collectively, di-
rectly or indirectly, affect most local governmental in-
stitutions. It is doubtful that many of these institutions
are aware of the scale of the weather changes and the
impacts that affect them. Release of information about
METROMEX has been extensive (RTI, 1976), but the
extent and severity of the impacts have not reached
many who would make decisions. For example, it is
not easy to discern the added heavy rain from that
which nature provides.
172
There is also a lack of use of the METROMEX re-
sults by many who have been well informed about the
impacts. For example, considerable efforts, beginning
in 1973, have been made by lllinois State Water Sur-
vey scientists to inform water resource engineers in
the St. Louis area about the sizeable changes in heavy
rainfall, sufficient to require a major shift in the design
of hydrologic structures. Yet to date, very little ap-
plication of the new values has been made as most
individual government institutions, their engineers, or
consulting firms still design around the old and inad-
equate (and less costly) rainfall-runoff values. Adop-
tion of the newer and correct design values will take
time.
The areal scale of the water problems in the flood-
plain, particularly the floodwater issue including basin
sedimentation, has defied well-organized local solu-
tions. Nearly 200 governmental agencies (city, town-
ship, county, regional, state and federal) have varying
jurisdictions in the two-county affected area. The in-
ability of these agencies to function coherently either
in developing regional water resource (quantity and
quality) plans, or in developing adequate management,
operations and maintenance of regional water resource
systems is apparent. The general lack of adoption of
these correct design values by engineers as yet reflects
the lack of a centralized authority for controlling and
implementing water resource design values. However,
a new suburb (Fairview Heights) in the floodplain area
has used the METROMEX produced heavier rain de-
sign criteria in their design and construction of city
storm drains and sewer systems.
Major findings of METROMEX are the heavy rain
increases and their implications for the water resource
systems in the anomaly area. These findings, however,
are not being responded to largely because there are
no regulatory agencies with anomaly scale authority,
and because of the multitude of uncoordinated, only
partially responsible local governmental entities (Davis,
1979). This outcome may not be surprising for several
reasons. Sassone (1976), after a study of 196 cities
concluded that local climate effects are not significant
in explaining total municipal costs.
A recent cloud burst on 14 July 1978 (a 3 h rainstorm
that produced up to 3 inches of rain in St. Louis) is an
excellent example of the direct impacts (the losses),
the public reaction, and the ensuing responses of local
officials to such events. This heavy short-duration
rainstorm caused a local river (Des Peres) to experi-
ence a flash flood affecting more than 100 homes and
apartments and causing $2 million in losses. The flood
victims blamed local officials of the city and the Met-
ropolitan Sanitary District for installing inadequate
planning and structures for handling storm water.
METEOROLOGICAL MONOGRAPHS
Local officials in tum largely blamed the problem
on the improper housing developments in local flood-
plains. The Executive Director of MSD said, "severe
thunderstorms cause as much or more damage in the
St. Louis area than a major flood (of the Mississippi
and Missouri Rivers)." The local officials tried to
counter public pressure in three ways. One response
was to suggest the purchase of floodplain insurance;
a solution to future problems. Governor Teasdale of
Missouri declared the flood damaged urban area a dis-
aster area, qualifying residents for emergency (low in-
terest) loans from the Federal Small Business Admin-
istration. However, these are not available to renters,
and since so many flood damages were to renters of
apartments, this was not a major solution. The third
solution offered was to ask the federal government
(Corps of Engineers) to build the water control struc-
ture needed. Corps officials estimated the cost to be
$1 billion, requiring 20-30 years to complete. No local
official blamed urban effects for intensifying the storm,
but had they known of that possibility, it seems a likely
explanation incorporated.
Tobin ( 1976) adds further insight to this problem of
general lack of governmental action and responses to
weather-related problems. He noted two factors in his
study of the urban-related rain anomaly at La Porte,
Indiana, a situation similar to St. Louis and its rain
anomaly. First was the areal separation (40 km or 25
mi) between the La Porte rainfall anomaly and the
causative factor, Chicago. The second was the pleth-
ora of governmental units in the area, producing a
fragmentation and absence of political responsibility.
These two factors essentially make the "producer" of
the effect (the city) free from responsibility and reg-
ulation. As Tobin states, "Those responsible are not
those bearing the cost."
This same situation is apparent in the area of the
anomaly-produced increase in local flooding in and
east of St. Louis. Those who should help pay for the
added costs for management, operation and mainte-
nance of flood control systems, plus new expensive
hydrologic facilities, are not paying. Under the exist-
ing taxing structure, those in the problem area (the
floodplain) pay the costs for the additional water and
related problems like sediment removal. Those in the
upland rural country where much of the anomaly's
heavy rain occurs (and where the large volume of run-
off and sediment is generated) do not pay the costs of
the water management problems of the floodplain.
Payments for the added costs could be argued to rest
with the metropolitan area which is producing the
added heavy rainfall.
No action describes the attitude oflocal government
to date toward inadvertent weather modification
METROMEX: A REVIEW AND SUMMARY
(Davis, 1979). This results largely because few urban
area residents realize that they live in a different and
potentially worse climate. When the undesirable ur-
ban-induced weather changes intensify, and when
more citizens in the urban-effect area become aware
of them, these factors will likely lead to demands for
public action. This could change the past patterns
wherein public officials have not, at least consciously,
allocated funds to handle the consequences of the ur-
ban-modified climate.
Impacts on human activities revealed how the lo-
cally-altered precipitation would affect taxes. A 20%
increase in rainfall in an area would increase wages
about 2%. In the rural effect area agricultural benefits
do increase income. Value of agricultural lands has
also increased and the net effect of this has been to
increase tax income for local, state and federal gov-
ernments. Sassone (1976) concluded that rain modifi-
cation will affect long-term social costs, with added
rain producing higher costs for local education, storm
systems and sewage treatment. These public sector
costs will necessitate higher taxes and in essence may
negate the added income from higher taxes as a result
from the anomaly.
In the urban and rural effect areas in the floodplain,
added taxable income will occur for the same reasons,
but the total will be less because of the relatively
smaller amount of agricultural lands and resulting in-
come benefits. The relatively greater water manage-
ment costs in the floodplain bring much greater costs
and higher taxes to those living there. These higher
costs and taxes, borne heavily by the east side urban
areas, may help cause emigration of the wealthier cit-
izens. This in turn leads to a lower community income
and a lower taxable urban base, ultimately followed
by poorer urban service and less fiscal capability to
deal with water resource problems. If available, more
federal funds must be used to subsidize local urban
needs; thus, the nation is helping to partially pay for
the urban water resources problems created by the
weather anomaly.
A qualitative assessment of the urban anomaly's im-
pact on institutions in the effect area suggests 1) net
benefits to many institutions in the rural areas and
communities in the uplands, and 2) net disbenefits to
most governmental institutions in the floodplain area.
As the regional water problems and their costs, par-
ticularly those caused by the anomaly, become known
to the local public, action relating 1) to more equitable
tax support of costs, and 2) to proper design, conser-
vation and operational practices for water systems will
likely develop. Then public officials and institutions
would react. One possible future outcome includes
giving the local regional planning commission or other
173
entity authority to regulate or develop a regional
(anomaly scale) water resource institution (district)
and with the commission to perform these tasks and
to raise taxes needed on a more equitable basis.
8.8.3 FEDERAL AGENCIES
One set of relevant impacts of the METROMEX
results includes those to federal agencies that must
encourage energy development (DOE) and set con-
trols (EPA), based on knowledge of environmental im-
pacts related to energy production and other human
alterations of the land surface. The St. Louis studies
have shown that urban-produced alterations of clouds
and rain are largely the result of the entire urban ag-
glomeration, not just one easily controllable human or
industrial activity.
The qualitative results about sharply reduced visi-
bility over and well beyond St. Louis can be utilized
by various local, state and federal agencies. At the
federal level, aviation officials should be cognizant of
prevailing visibility restrictions in the vicinity of large
metropolitan areas. Information regarding the visibil-
ity anomaly could serve as input in establishing air
traffic patterns and new airport siting proposals (Komp
and Auer, 1978). Knowledge of the locally increased
thunderstorm occurrences with their attendent in-
creases in hailfalls and downdrafts (gust fronts) which
affect private and especially commercial jet aircraft,
will be valuable to the FAA.
Plumes of airborne particulate matter, condensable
gases and visibility reductions present yet another ex-
ample of a mesoscale transport event downwind from
a large metropolitan area. Such events must be rec-
ognized as problems and be the concern of state air
pollution agencies. METROMEX measurements in-
dicate that the concentration of particles between 0.05
and 55 JLm in diameter provide an average mass load-
ing of 44 and 86 J.tg m- 3 upwind and downwind, re-
spectively, of St. Louis (Johnson, 1978). There are to
be compared with present clean air standards of 75 J.tg
m- 3 • State and local planners within an expanding
metropolitan area must also consider the aesthetic,
societal and economic impact on communal air qual-
ity. The prevention or control of undesirable inad-
vertent weather and climate modification would allow
the urban socio-economic lifestyle to become more
desirable.
METROMEX results helped show that the fossil-
fuel power plants around St. Louis are among the ma-
jor contributors to the local so2 concentrations in the
air. Furthermore, METROMEX results have shown
that the conversion of S0 2 to sulfates is aided by the
cloud processes; thus conversion to sulfate and the
acid rain problem are partially aided by the increase
174
in urban clouds. A study of aerosol evolution in the
plumes of St. Louis power plants strongly suggests
active aerosol growth mechanisms in the plumes,
without regard to chemical identities of the aerosols.
This volumetric conversion occurs during a 3-4 h pe-
riod.
Although METROMEX has not been aimed at the
control-optimization issue of air pollution, the project
has defined several pollution aspects that relate to fos-
sil-fuel power plants. These include 1) how plants
should be spatially distributed, and 2) how they should
be allowed to release effluents so as to minimize ef-
fects on clouds, rainfall quality and air quality. The
increases in convective clouds and rainfall initiation
related to a large refinery complex near St. Louis sug-
gest that large industrial complexes and future power
parks with concentrated releases of waste heat, mois-
ture and aerosols to the atmosphere in a humid con-
tinental climate could produce frequent convective
clouds and increased precipitation.
These issues of changes in weather, air quality and
climate over areas considerably downwind of indus-
trial centers and large cities point to another important
institutional concern, international relations. The State
Department and Department of Commerce (NOAA)
have to be concerned with portent for exportation of
altered weather from the United States to other na-
tions.
As noted in the previous section on taxing of local
institutions, the anomaly in agricultural areas gener-
ates higher personal incomes and thus more income
tax to the IRS for the federal government to use. How-
ever, it appears that the anomaly related water re-
source problems in the floodplain area contribute to
costly 1) planning, operational and development pro-
grams for the federal government (Corps of Engi-
neers); 2) urban flooding damages and resulting ren-
ovation (HUD and FHA); and 3) water quality moni-
toring and treatment development (EPA). The net
economic effect is likely one of disbenefits, more costs
than income from the anomaly, to the federal govern-
ment.
Other impacts on federal agencies include the local
adjustment by the Soil Conservation Service of the
USDA to the new heavy rain values in their local de-
sign criteria for terraces and other control structures.
METROMEX results on conditions for rain initiation
and intensifications have been presented to the St.
Louis area weather forecasters of the NWS. These
should be incorporated in the local forecast criteria.
Increased storm losses (hail, rain and wind) to crops
will also impact on the local loss payments of the
FCIC.
Other impacts include the METROMEX generated
findings relating to research needs and priorities for
METEOROLOGICAL MONOGRAPHS
inadvertent weather modification. These have affected
the orientation of research programs supported by
NSF and DOE.
8.9 Review of impacts and implications
Changnon et al. (1977a) delineated 58 impacts re-
sulting from the urban-altered summer clouds, rainfall
and storminess. The impacts were rated as having a
general benefit, disbenefit, or neutral (some benefit
and some disbenefit) impact to the major impacted
group. Second, the percent change, if known, was
listed. Tables 4 and 5 illustrate several key impacts.
Table 6 presents the distribution of all 58 impacts. The
minor impacts show an even distribution (11 vs 11) in
the benefit and disbenefit categories. However, 17 of
the 24 major impacts (71%) are listed as a local dis-
benefit. The total for the disbeneficial impacts is 34
(or 59%) of all the local impacts. This suggests a net
disbenefit to the area resulting from the urban-induced
precipitation anomaly, and the preponderance of the
major disbenefits (12 of 17) relate to the heavier rains
(see Table 5). The scientific knowledge and technical
information generated by METROMEX, in both basic
and applied scientific areas, is seen to represent a ma-
jor benefit.
The study of the impacts suggested a considerable
regional difference within the effect area of -4000
TABLE 4. Impacts of more total summer rainfall and pollutants.
After Changnon et al. (l977a).
Impacts Effect* Change**
Increased summer runoff B +11%
Increased storage in lakes and ponds B +10%
Increased groundwater supplies B +5%
Dilution of water pollutants B Unk
Altered low flow levels and water
quality regulations exceeded D +97%
More pollutants in streams D +l to +200t
More deposition of pollutants D +15%
More pollutant damage to buildings D Unk
Increased groundwater pollution D +10%
Increases in com yields B +3%
Increases in com yields in dry years B +7%
Increases in soybean yields B +4%
Increases in soybean yields in dry
years B +6%
Increased pollution effect on crop
yields N Unk
Greater farm income and more stable
income B Unk
More acreage in soybeans N +6%
Less acreage in com N -16%
Higher value of agricultural lands B +3%
Increased tax income to government
rural areas B +5%
*Effect classed as a benefit (B), disbenefit (D), and neutral (N),
good and bad.
**Percent change, + or -, or unknown (Unk), based on the most
common, area-wide value if several were available.
tRange shown reflects differences found with varying pollutants,
including I% for potassium, 100% for zinc and 200% for sulfates.
METROMEX: A REVIEW AND SUMMARY 175

TABLE 5. Impacts of higher rain rates and more heavy rainstorms tially minimized by changing (dispersing) arrays of fu-
due to St. Louis effects on the climate. After Changnon et al. ture large industrial complexes.
(1977a).
At this time little, if anything, is being done con-
Impacts Effect* Change** sciously by local or regional public officials to manage
urban weather anomalies or to adjust to their impacts.
More bypasses of sewage treatment
plants D +40% They have many more pressing problems such as
More fluctuations in groundwater crime and transportation. Inadvertent urban-produced
levels D +100%
More frequent urban flooding D +100% climate modification will likely not be addressed by
Increased automobile accidents D +20% public officials until the climate modification and its
Increased soil erosion in uplands D +34% impacts receive wider public awareness, the public
Increased sedimentation of streams D + 10 to +30t
Sedimentation in floodplain facilities D Unk becomes sufficiently concerned to believe action is
Added operations and management of needed, and until the public perceives that the matter
floodplain water facilities D >100%
Added drainage systems D Unk is appropriate for action by some governmental agency.
Altered design for hydrologic A key question concerns whether there will be even
structures D Unk greater modification of climate by urban areas in the
Added soil losses on crop production D Unk
Delays and danger to air future. The answer appears to be yes for several rea-
transportation D Unk sons. First, the national population shift and growth
Decreased visibility and increased
crime D +30% is still into urban centers, and all available evidence
Inequitable local taxing for water shows that the amount of climate change is related to
management facilities D Unk
Higher costs to government agencies
urban size. It appears that a megalopolis such as the
for water management, treatment Boston-to-Washington corridor or the Chicago-to-
and planning D Unk Boston corridor has brought or may bring climatic
*Effect classed as a benefit (B), disbenefit (D), and neutral (N), changes over large areas. It also seems likely that fos-
good and bad. sil-fueled power plants will increase over the next 20
**Percent change, + or-, or unknown (Unk), based on the most
common, area-wide value if several were available.
years in and around cities and that automobile trans-
tRange reflects differences found in various streams. portation will be sustained if not expanded. Another
national activity that will induce more weather effects
is the continuing thrust for economic and industrial
km 2 • This difference related to the major physio-
expansion and use of fossil fuels. A study of the effect
graphic and land use differences. In the urban-subur-
of Chicago on precipitation is now under way to assess
ban floodplain area, 15 of 17 major impacts were dis-
the transferability of the St. Louis causation factors
benefits. In the rural (agricultural) uplands, 7 of the 17
to a larger city situated near a large water body
major impacts were benefits. Clearly, the urban-in-
(Changnon and Semonin, 1978).
duced rain anomaly is mainly disbeneficial in the
floodplain but has mixed impacts in the uplands. REFERENCES
The city, as a whole, leads to most of the rain Ackerman, B., S. A. Changnon, Jr., G. Dzurisin, D. F. Gatz, R. C.
changes discerned, indicating there is no practical way Grosh, S.D. Hilberg, F. A. Huff, J. W. ManselJ, H. T. Ochs,
to control the climate shifts for existing large cities. M. E. Peden, P. T. Schickedanz, R. G. Semonin and J. L.
The precipitation changes are not due primarily to one Vogel, 1978: Summary of METROMEX, Vol. 2, Causes of
isolatable and easily controllable factor such as a par- Precipitation Anomalies, Illinois State Water Survey BulJ. 63,
Urbana, 395 pp.
ticular compound in the effluent from automobile ex- Agricultural Research Service (ARS), 1972: Rainfall-erosion losses
hausts or from a specific industry. METROMEX re- from cropland east of the Rocky Mountains. Agricultural Hand-
vealed that a large refinery complex produces sizeable book 282, U.S. Dept. of Agriculture, Washington, DC, 47 pp.
but very local increases in convective clouds and in Ashby, W. C., and H. C. Fritts, 1972: Tree growth, air pollution,
the initiation of rainfall. These effects could be poten- and climate near La Porte, Indiana. Bull. Amer. Meteor. Soc.,
53, 246-251.
Auer, A. H., 1977a: Impact of waste heat/humidity dissipation on
TABLE 6. Impacts from changed precipitation and storminess in convective clouds. Paper, Workshop on Atmospheric Effects
the St. Louis area. of Nuclear Energy Centers. Argonne National Laboratory, 13
pp.
Major Minor Unknown - - , 1977b: The evolution of visibility deterioration and aerosol
Effect impacts impacts influence Totals spectra in the St. Louis urban plume. J. Air Pollut. Control
Assoc., 21, ll88-ll9l.
Benefit 7 11 2 20 - - , 1978a: Correlation of land use and cover with meteorological
Dis benefit 17 II 6 34
Neutral 0 4 0 4 anomalies. J. Appl. Meteor., 17, 636-643.
- - , 1978b: Modification of convective cloud activity by an urban
Totals 24 26 8 58 area. Final Report, ENV 73-07881, University of Wyoming,
Laramie, 25 pp.
176
- - , and R. A. Dirks, 1974: Contributions to an urban meteoro-
logical study: METROMEX. Bull. Amer. Meteor. Soc., 55,
106-110.
Blanchard, D., 1975: Final Report on Second Inadvertent Weather
Modification Workshop. Institute of Man and Science, State
University of New York, Albany, 81 pp.
Boatman, J. F., and A. H. Auer, 1974: Inadvertent thunderstorm
modification by an urban area. Preprints 4th Conf. Weather
Modification, Ft. Lauderdale, Amer. Meteor. Soc., 366-373.
Braham, R. R., 1974: Cloud physics of urban weather modifica-
tion-a preliminary report. Bull. Amer. Meteor. Soc., 55,
100-106.
- - , 1976: CCN spectra in C-k space. J. Atmos. Sci., 23, 343-345.
- - , 1977: Overview of urban climate. Proc. Conf Metropolitan
Physical Environment, U.S. Department of Agriculture, Upper
Darby, PA, 3-17.
--,and P. Squires, 1974: Cloud physics-1974. Bull. Amer. Me-
teor. Soc., 55, 543-586.
- - , and D. Wilson, 1978: Effects of St. Louis on convective cloud
heights. J. Appl. Meteor., 17, 587-592.
Cataneo, R., 1974: Operational aspects of project METROMEX-
an inadvertent weather modification study. Preprints 4th Conf.
Weather Modification, Ft. Lauderdale, Amer. Meteor. Soc.,
379-400.
Center for the Environment and Man (CEM), 1977: Report of the
Workshop on Inadvertent Weather Modification. Hartford, CT,
118 pp.
Changnon, S. A., 1968: The La Porte anomaly-fact or fiction?
Bull. Amer. Meteor. Soc., 49, 4-11.
- - , 1970: Reply. Bull. Amer. Meteor. Soc., 51, 337-342.
- - , 1972: Urban effects on thunderstorm and hailstorm frequen-
cies. Preprints Conf Urban Environment, Philadelphia, Amer.
Meteor. Soc., 177-184.
- - , 1973a: Atmospheric alterations from man-made biospheric
changes. Modifying the Weather, W. D. Sewell, Ed., Western
Geographical Service, Vol. 9, 135-184.
- - , 1973b: Inadvertent weather and precipitation modification by
urbanization. J. Irrig. Drainage Div., ASCE, 99(IRI), 27-41.
- - , 1974: A review of inadvertent mesoscale and climate modifi-
cation and assessment of research needs. Preprints 4th Conf.
Weather Modification, Ft. Lauderdale, Amer. Meteor. Soc.,
347-352.
- - , 1975: Operations of mesoscale networks: As illustrated by
METROMEX. Bull. Amer. Meteor. Soc., 56, 971-977.
- - , 1976a: Effects of urban areas and echo merging on radar echo
behavior. J. Appl. Meteor., 15, 561-570.
- - , 1976b: Inadvertent weather modification. Water Res. Bull.,
AWRA, 12,95-718.
- - , 1977: Impacts of urban-modified precipitation on man's activ-
ities. J. Wea. Mod., 9, 8-18.
- - , 1978a: Heavy falls of hail and rain leading to roof collapse. J.
Structural Div. ASCE, ST1, 198-200.
- - , 1978b: Urban effects on severe local storms at St. Louis. J.
Appl. Meteor., 17, 578-586.
- - , 1979: What to do about urban-generated weather and climate
changes. J. Amer. Inst. Planners, January, 36-48.
- - , and J. C. Neill, 1967: Areal variations in corn-weather rela-
tions in Illinois. Trans. Ill. Acad. Sci., 60, 221-230.
- - , and R. G. Semonin, 1975: Studies of selected precipitation
cases from METROMEX. Illinois State Water Survey, Report
of Investigation, No. 81, Urbana, 318 pp.
--and--, 1978: The Chicago area program. Bull. Amer. Me-
teor. Soc., 59, 153-160.
- - , R. G. Semonin, and F. A. Huff, 1976: A hypothesis for urban
rainfall anomalies. J. Appl. Meteor., 15, 544-560.
--,F. A. Huff, P. T. Schickedanz and J. L. Vogel, 1977a: Sum-
METEOROLOGICAL MONOGRAPHS
maryofMETROMEX, Vol. 1,Anomaliesand Impacts, lllinois
State Water Survey Bull. 62, Urbana, 260 pp.
- - , D. F. Gatz, J. Bertness, S. T. Sonka, J. Bartlett and J. J.
Hassett, 1979: Studies of impacts of urban-related weather and
climate changes at Chicago and St. Louis. lllinois State Water
Survey, Urbana, 112 pp.
- - , R. J. Davis, B. C. Farhar, J. E. Haas, J. L. Ivens, M. V.
Jones, D. A. Klein, D. Mann, G. M. Morgan, S. T. Sonka, E.
R. Swanson, C. R. Taylor and J. Van Blokland, 1977b: Hail
suppression: Impacts and issues, Final Report ERP75-09980,
Illinois State Water Survey, Urbana, 427 pp.
Chen, C. S., 1976: The effects of pressure perturbation on precip-
itating clouds. Preprints Int. Conf. Cloud Physics, Boulder,
Amer. Meteor. Soc., 374-381.
Crocker, T. D., 1976: Climate and household budget expenditures.
The Urban Costs of Climate Modification, T. A. Ferrar, Ed.,
Wiley, 157-170.
Davis, R. J., 1979: The Legal Implications of Inadvertent Weather
Modification. University of Arizona, 101 pp.
Dettwiller, J. and S. A. Changnon, 1976: Possible urban effects on
maximum daily rainfall at Paris, St. Louis and Chicago. J. Appl.
Meteor., 15, 517-519.
Environ. Sci. Tech., 1977: Air pollutants and health: An epidemi-
ologic approach, 11, 648-650.
ERDA, Division of Biomedical and Environmental Research, 1975:
Chemist-Meteorologist Workshop-/975. U.S. Energy Res.
and Dev. Admin., Washington, DC, 97 pp.
Farhar, B. C., 1979: The influence of the St. Louis rain anomaly on
human activities. IBS Report, University of Colorado, Boulder,
211 pp.
Fitzgerald, J. W., 1974: Effect of aerosol composition on cloud
droplet size distribution: A numerical study. J. Atmos. Sci., 31,
1357-1358.
Hane, C., 1978: Scavenging of pollutants by thunderstorm rainfall:
Numerical simulation. J. Appl. Meteor., 17, 699-710.
Harmon, J. R., and W. M. Elton, 1971: The La Porte precipitation
anomaly. Ann. Assoc. Amer. Geogr., 61, 468-480.
Havera, S. P., 1973: The relationship of Illinois weather and agri-
culture to the eastern cottontail rabbit. Tech. Rep. 4, Contract
14-06-D-7197, Illinois State Water Survey, Urbana, 92 pp.
Helfand, L., 1975: Personal communication.
Hoch, I., 1976: Climate wages and urban scale. The Urban Costs
of Climate Modification, T. A. Ferrar, Ed., Wiley, 171-216.
Huff, F. A., 1975a: Urban effects on the distribution of heavy con-
vective rainfall. Water Resour. Res., 11, 889-896.
- - , 1975b: The effect of precipitation scavenging of airborne and
surface pollutants on surface and groundwater quality in urban
areas. Final Report, Part 2, NSF GK-38329, Illinois State Water
Survey, Urbana, 43 pp.
--,and S. A. Changnon, 1972: Climatological assessment of urban
effects on precipitation at St. Louis. J. Appl. Meteor., 11,
823-842.
- - , and - - , 1973: Precipitation modification by major urban
areas. Bull. Amer. Meteor. Soc., 54, 1220--1232.
- - , and R. G. Semonin, 1974: Potential of precipitation modifi-
cation in moderate to severe droughts. Preprints 4th Conf
Weather Modification, Ft. Lauderdale, Amer. Meteor. Soc.,
490-495.
Hume, E. P., 1968: Apparent air pollution injury to ornamental
plants in the East St. Louis area. Bull. Arbor. Rotan. Garden.,
2, 54-55.
Johnson, D. B., 1978: Giant nuclei and warm rain initiation. Pre-
prints Conf. Cloud Physics and Atmospheric Electricity. Issa-
quah, Amer. Meteor. Soc., 31-34.
Jones, D. M.A., and P. T. Schickedanz, 1974: Surface temperature
moisture and wind studies. Interim Report of METROMEX
METROMEX: A REVIEW AND SUMMARY
Studies: 1972-1973, F. A. Huff, Ed., NSF GI-38317, Illinois
State Water Survey, Urbana, 98-130.
Komp, M. J., and A. H. Auer, 1978: Visibility reduction and ac-
companying aerosol evolution downwind of St. Louis. J. Appl.
Meteor., 17, 1357-1367.
Landsberg, H. E., 1970: Man-made climatic changes. Science, 170,
1265-1274.
Lanphear, F. 0., 1970: Air pollution injury to plants in St. Louis.
Dept. of Horticulture, Purdue University, Lafayette, 13 pp.
Lawson, R. R., and A. H. Auer, 1976: Urban-induced influences on
convective cloud activity. J. Rech. Atmos., 10, 143-156.
Lowry, W. P., 1972: Urban effects on the atmosphere-who in the
world cares? Preprints Conf. Urban Environment, Philadel-
phia, Amer. Meteor. Soc., 135-138.
Massachusetts Institute of Technology, 1970: Man's Impact on the
Global Environment. Report on the Study of Critical Environ-
mental Problems, Cambridge, 319 pp.
Murray, F. W., L. R. Koenig and P. Tag, 1978: Numerical simu-
lation of an industrial cumulus and comparison with observa-
tions. J. Appl. Meteor., 17, 655-668.
Ochs, H. T., 1974: Cloud modeling in METROMEX. Preprints 4th
Conf. Weather Modification, Ft. Lauderdale, Amer. Meteor.
Soc., 396-400.
--,and R. G. Semonin, 1976: Microphysical computations in ur-
ban and rural clouds. Preprints Int. Conf. Cloud Physics, Boul-
der, Amer. Meteor. Soc., 22-26.
Research Triangle Institute, 1976: METROMEX, Case Study No.
37, RANN utilization experience. Contract C76-17165, Re-
search Triangle Park, NC, 27 pp.
Sassone, P. G., 1976: Climate modification and some public sector
considerations. The Urban Costs of Climate Modification, T.
A. Ferrar, Ed., Wiley, 217-238.
Schicht, R. J., 1965: Groundwater development in the East St.
Louis Area, Illinois. Report of Investigation 51, Illinois State
Water Survey, Urbana, 58 pp.
- - , 1977: The Effect of Precipitation Scavenging of Airborne and
Surface Pollutants on Surface and Groundwater Quality in Ur-
ban Areas, Groundwater Studies. Part 2, Final Report NSF
GL-38329, Illinois State Water Survey, Urbana, 58 pp.
--,and E. G. Jones, 1962: Ground-water levels and pumpage in
East St. Louis Area, Illinois, 1890-1961. Illinois State Water
Survey, Report oflnvestigation 44, Urbana, 40 pp.
177
Schickedanz, P. T., and D. F. Gatz, 1975: Severe storms of23 July
1973. Studies of selected precipitation cases from METRO-
MEX, S. A. Changnon and R. G. Semonin, Ed., illinois State
Water Survey, Report oflnvestigation 81, Urbana, 90-126.
Semonin, R. G., 1972: Tracer chemical experiments in Midwest
convective clouds. Preprints 3rd Conf. Weather Modification,
Rapid City, Amer. Meteor. Soc., 83-87.
- - , 1976: The variability of pH in convective storms. Water Air
Soil Pollut., 6, 395-406.
- - , and S. A. Changnon, 1974: METROMEX: Summary of
1971-1972 results. Bull. Amer. Meteor. Soc., 55, 95-99.
--and--, 1975: METROMEX: Lessons for precipitation en-
hancement in the Midwest. J. Wea. Mod., 7, 77-86.
Shea, D. M., and A. H. Auer, 1978: Thermodynamic properties and
aerosol patterns in the plume downwind of St. Louis. J. Appl.
Meteor., 17, 689-698.
Sherretz, L. A., and B. C. Farhar, 1978: An analysis of the rela-
tionship between rainfall and the occurrence of traffic acci-
dents. J. Appl. Meteor., 17, 711-715.
Silverman, B. A., and L. Nelson, 1975: Some Model Results for
Natural and Seeded Hail-Bearing Clouds. National Hail Re-
search Experiment Symposiurn!Workshop on Hail, Preprint
Vol. 1, NCAR, 15 pp.
Southwestern Illinois Metropolitan and Regional Planning Commis-
sion (SIMRPC), 1975: Plan for major drainage: The American
bottoms and hillside drainage area planning basin. Collinsville,
IL, 139 pp.
Tobin, R. J., 1976: Climatic changes, local government and political
problems. The Urban Costs of Climate Modification, T. A.
Ferrar, Ed., Wiley, 239-258.
Vogel, J. L., 1973: Analyses of possible urban effect thunderstorms.
Preprints 5th Conf. Weather Forecasting and Analysis, San
Diego, Amer. Meteor. Soc., 189-191.
- - , 1974: Synoptic analyses. Interim Report of METROMEX
Studies: 1971-1973, F. A. Huff, Ed., NSF GI-38317, Illinois
State Water Survey, Urbana, 6-16.
Weather Modification Advisory Board, 1978: The Management of
Weather Resources. U.S. Dept. of Commerce, Washington,
DC, 229 pp.
Wilmarth, A., 1977: Personal communication. Southwestern Illinois
Metropolitan and Regional Planning Commission, Collinsville,
IL.
 APPENDIX
Management and Organization
METROMEX was designed to be "open-ended" in
that any research groups who wished to become in-
volved, in either the atmospheric sciences phases or
in the allied projects (which would benefit from, or
utilize the project's weather data), were welcomed and
became involved along the lines of independent effort
and funding. In addition to original planning institu-
tions (Argonne National Laboratory, Illinois State
Water Survey, University of Chicago and University
of Wyoming) research groups from the Battelle Pacific
Northwest Laboratories (PNL), the Stanford Re-
search Institute (SRI), and the University of California
at San Diego (UCSD), each brought equipment and
staff to St. Louis during the first season, 1971. The
PNL and SRI groups continued their field efforts from
1972 through 1975; the University of Missouri at Rolla
(UMR)joined in 1973; NOAA scientists joined in 1975;
and social scientists of the University of Arizona (UA)
and the Institute of Behavioral Sciences at the Uni-
versity of Colorado (IBS) began their research in 1977.
Basically, their projects were established at St. Louis
to take advantage of the data emanating from the
METROMEX field project. The major field phase oc-
curred between June 1971 and August 1975 when there
were five summer field programs. In all five summers,
UC, ISWS, UW, PNL and SRI participated in field
activities and essentially served as the lead research
groups. A calendar of summer operations between
1971 and 1976 of the METROMEX participants is
shown in Table 1. The UC carried out limited field
operations during January-March 1972 and
March-April 1973, and the ISWS operated the rain-
gage network on a reduced basis during the periods
September-May 1971-75.
Another major field project for the St. Louis area,
labeled as the Regional Air Pollution Study (RAPS) by
the Environmental Protection Agency, was initiated
in 1973. RAPS focused on the effects of the weather
on air pollution. Field data were gathered in 1974-76
to develop a data bank for testing air pollution models
and developing control strategies. Certain RAPS data
have been used by METROMEX scientists, and MET-
179
ROMEX data and equipment were used by RAPS sci-
entists. In fact, some projects of the RAPS program
were conducted by METROMEX participants (UW,
PNL, ISWS) in addition to their METROMEX fo-
cused projects.
Continuity in planning and operations of METRO-
MEX was handled by the principal investigators and
scientists from each group who were organized into a
project panel. A program coordinator, chosen each
year by this panel, acted as a focal point for program
communications, both scientific and public, and for
data exchange.
Project sponsors included the State of Illinois; the
Department of Energy (including its predecessors,
AEC and ERDA); the National Science Foundation
(including the Atmospheric Sciences Division, Engi-
neering Division and Research Applied to National
Needs Program); the U.S. Environmental Protection
Agency; U.S. Public Health Service; the National
Oceanic and Atmospheric Administration; and the
U.S. Air Force. The estimated cost for METROMEX
for the 1971-78 period was about $10.5 million.
The fact that METROMEX has been successfully
conducted without strong centralized governmental
control proves that "big science" programs can be
planned and completed by scientific groups. It can
serve as a model wherein all of those involved are
there entirely by their own institutional interest for
scientific research. In the final analysis, METROMEX
was a successful program because the scientists in-
volved were willing to cooperate in operations, to ex-
change data and to derive results, often collectively,
in a reasonable time.
METROMEX Components
Argonne National Laboratory (ANL). The ANL
effort in 1971 was twofold: 1) a study of precipitation
scavenging under the direction of D. F. Gatz, and 2)
an investigation of the effect of an urban complex on
the winds in the boundary layer, under the direction
180
TABLE I. Calendar of summer field operations for METROMEX participants.
1971 1972 1973
Jun Jul Aug Jun Jul Aug Jun Jul
ANL X X X X
uc X X X X X X X X
ISWS X X X X X X X X
uw X X
PNL X X X X
SRI X X
NOAA
UMR
UCSD X
DRI
IBS
VA
of B. Ackerman. These activities were ended in 1972
as both investigators became employees of the Illinois
State Water Survey where their research in METRO-
MEX was similar to that at Argonne. Subsequent Ar-
gonne efforts in the 1972-73 period concerned study
of the structure of the lower atmosphere by E. Miller.
Their work was sponsored by the Atomic Energy
Commission. The Air Weather Service provided sup-
port equipment and personnel for the 1971-72 pilot
balloon observations, and the National Center for
Atmospheric Research provided the airplane support
in 1971 for the Argonne efforts.
University of Arizona (VA). This impact-related
research addressed the legal implications of inadver-
tent weather modification. It was performed by R. J.
Davis. The one-year project had two goals: I) to as-
certain the legal constraints upon possible public pol-
icy options in light of the METROMEX findings and
urban-related weather changes elsewhere; and 2) to
determine how the law might have an impact upon
inadvertent weather modification. This research was
supported by the National Science Foundation.
Battelle Pacific Northwest Laboratories (PNL).
Two groups from these Laboratories participated dur-
ing 1972-78 in METROMEX: the Atmospheric Sci-
ences Department involving A. J. Alkezweeny, J. M.
Hales and M. T. Dana, and the Radiological Sciences
Department involving N. A. Wogman and J. A.
Young. The major objectives included 1) determina-
tion of the physical and chemical changes of pollutants
in the St. Louis area, a source-to-sink evolutionary
type study; 2) study of the scavenging of pollutants
and atmospheric tracers by convective rainfall; and 3)
study of urban aerosol size distributions. This research
was supported by the Department of Energy (previ-
ously AEC and then ERDA) and by the Environmental
Protection Agency.
University of California at San Diego
(UCSD). This two-year project was carried out in
conjunction with Scripps Institute of Oceanography,
METEOROLOGICAL MONOGRAPHS
1974 1975 1976 1977
Aug Jun Jul Aug Jun Jul Aug Aug May Jun
X
X X X X X
X X X X X X X
X X X X
X X X
X X X
X X
X
X
X X
X
and was under the direction of S. Q. Duntley, R. W.
Johnson and J. I. Gordon. The principal project task
was to take daytime atmospheric optical measure-
ments in the St. Louis area and southern Illinois and
from these measurements to determine optical prop-
erties for various downward-inclined paths of sight.
This research was supported by the U.S. Air Force.
University of Chicago (UC). The Cloud Physics
Laboratory, under the direction of R. R. Braham, Jr.,
had four major objectives: 1) establish cause-and-ef-
fect relationships and levels of response of urban
clouds to inadvertent modification; 2) determine source
strengths and loss functions for cloud active nuclei in
urban atmospheres; 3) map regions of urban-altered
cloud structures; and 4) carry out descriptive studies
of urban aerosols, the urban heat island and other fea-
tures of the urban boundary layer. The operational
period was June-August 1971-72, July-August during
1973-75, and in two winter periods (January-March
1972 and March-April 1973). This research work was
sponsored by the National Science Foundation.
Denver Research Institute (DRI). This one-year
project, in 1976, in collaboration with UC, focused on
the measurement of local CCN patterns using special
equipment on the UC aircraft. This was under the di-
rection of D. Saxena. The effort was NSF sponsored.
Illinois State Water Survey (ISWS). As with the
other groups, the Water Survey's plans and interests
in METROMEX evolved from past principal research
areas and available instrumentation. The principal in-
vestigators were S. A. Changnon, Jr., F. A. Huff and
R. G. Semonin. The Water Survey's historical interest
in precipitation studies, climatology, instrumentation
and cloud physics led to six major objectives.
The major METROMEX objectives of the Survey
throughout 1971-78 included I) the study of severe
local weather phenomena (heavy rainstorms, thunder-
storms and hailstorms) in summer so as to describe
the temporal-spatial relationships of these events in
the St. Louis urban area with special reference to their
METROMEX: A REVIEW AND SUMMARY
relationships under varying synoptic weather condi-
tions; 2) the study of raingage and radar data to assess
the magnitude and location of the urban-related pre-
cipitation changes; 3) an investigation of urban-in-
duced changes in winds in the planetary boundary
layer; 4) studies of surface weather (temperature,
winds and humidity patterns); and 5) an atmospheric
chemistry project including use of unique chemical
tracers to ascertain sources, scavenging and deposi-
tion of pollutants by precipitation; and 6) the local
impacts of the urban precipitation changes. The Water
Survey's principal overall interest was to study, un-
derstand and evaluate the urban-induced rainfall in-
creases with respect to the water resources (both qual-
ity and quantity) of Illinois.
Funding was received from the State of Illinois,
Department of Energy (AEC and ERDA), National
Science Foundation and Environmental Protection
Agency. Most measurements were made throughout
the summers of 1971-75, and the precipitation mea-
surements were continuously collected from June 1971
through August 1975. Support in equipment and per-
sonnel was furnished by NCAR and the Air Weather
Service.
Institute of Behavioral Sciences at University of
Colorado (IBS). This impact-focused research proj-
ect considered certain societal ramifications resulting
from inadvertent precipitation modification in the St.
Louis metropolitan area. This one-year project was
performed by B. C. Farhar and sought to measure
public and institutional responses to the anomalies and
related effects on human activities. This research was
supported by the National Science Foundation.
University of Missouri at Rolla ( UMR). This one-
year (1972-73) research effort focused on a study of
aerosols in the lower troposphere. It was performed
by J. F. Stampfer, Jr. The objectives were 1) to de-
termine the effect of synoptic conditions on the ver-
tical and horizontal dispersion of the aerosol, 2) to
correlate ground based lidar returns with in situ aero-
sol measurements, and 3) to calculate vertical aerosol
exchange coefficients. This research was supported by
the National Science Foundation.
Natimial Oceanic and Atmospheric Administration
(NOAA-WPL). The Wave Propagation Laboratory
participated in METROMEX during 1975 to study
wind patterns using high-resolution Doppler radars
over the St. Louis urban and rural areas. The research
effort was directed by R. A. Kropfli. The research was
aimed at four goals: 1) observing the rural-urban dif-
ference in the wind field within the mixing layer under
a variety of conditions; 2) observing the wind field
within convective storms as they passed through the
urban area; 3) applying results in an evaluation of ex-
181
isting numerical models of the urban mixing layer and
convective storms; and 4) providing other METRO-
MEX participants with measurements of the three-di-
mensional urban wind field. This research was sup-
ported by the National Science Foundation (RANN
Program) and NOAA.
Stanford Research Institute (SRI). The primary
goals of this project directed by E. Uthe were to pro-
vide objective documentation of urban influence on 1)
mixing depth and other boundary layer structure, 2)
cloud-base heights and formation frequency, and 3)
precipitation elements in the St. Louis urban/rural
area. An additional goal was to quantitatively relate
time and space variations of these observed atmo-
spheric structures to other meteorological measure-
ments, and thus to identify dynamic, radiative and
physical processes that may contribute to the precip-
itation anomaly. The project extended from 1971 to
1976 and was funded by the National Science Foun-
dation.
University of Wyoming (UW). The Wyoming re-
search group, under the leadership of A. H. Auer, Jr.,
and R. A. Dirks, had six principal objectives: 1) ob-
serve and analyze the kinematics of the general me-
soscale motion; 2) investigate the temporal and spatial
variations of the static meteorological parameters
(e.g., temperature and humidity) in the urban environ-
ment and relate the observed air mass modification to
thermodynamic energy, humidity and aerosol (Aitken
Nuclei) budgets in the urban atmosphere; 3) observe
and analyze the character and modification of the life
cycles of growing convective clouds due to mesoscale
influences of the urban area; 4) observe and analyze
the character and extent of the downwind plume (heat,
humidity, aerosols) emanating from the urban area; 5)
observe and analyze daily variations of solar and in-
frared radiation in the urban and rural atmospheres
and relate these to the energy budgets of the urban
and rural areas; and 6) observe and analyze the flow
structure and turbulence in a planetary boundary layer
in urban and rural atmospheres to relate horizontal
and vertical motions to energy sources, such as sur-
face heating, and to ascertain the fluxes of heat and
water vapor within and out of the urban plume.
Within this framework, the goal of the University
of Wyoming's activities within METROMEX from
1971 through 1978 was to make comparative obser-
vations of urban and non-urban environments to de-
termine processes responsible for differences in
boundary layer signature and wind field, traceable to
urban-induced changes in heat, humidity and aerosols.
The effort was funded by the National Science Foun-
dation and the Environmental Protection Agency.
 METEOROLOGICAL MONOGRAPHS
Vol. 1, No. 1. Wartime Developments in Applied Climatology by
W. C. Jacobs. August, 1947. Out of print.
No. 2. The Observations and Photochemistry of Atmo-
spheric Ozone and Their Meteorological Signifi-
cance by R. A. Craig. September, 1950. Out of
print.
No. 3. On the Rainfall of Hawaii: A group of contribu-
tions by L. B. Leopold; H. Landsberg; C. K. Stidd
and L. B. Leopold; T. C. Yeh, C. C. Wallen and
J. E. Carson; and T. C. Yeh, J. E. Carson and J.
J. Marciano. June 1951. Out of print.
No. 4. On Atmospheric Pollution: A group of contribu-
tions by J. H. Carter, C. A. Gosline, E. W. Hew-
son and H. Landsberg; M. L. Barad; G. W. Brier;
W. C. L. Hemeon; P. H. Lowry; P. H. Lowry, D.
A. Mazzarella and M. E. Smith; H. F. Poppen-
diek; H. Rouse; R. H. Sherlock; and M. E.
Smith. November, 1951. ISBN 0-933876-00-9
No. 5. Forecasting in Middle Latitudes by H. Riehl; and
J. Badner; J. E. Hovde; N. E. LaSeur; L. L.
Means; W. C. Palmer; M. J. Schroeder; L. W.
Snellman; and others. June, 1952. Out of print.
Vol. 2, No. 6. Thirty-Day Forecasting: A review of a ten-year
experiment by Jerome Namias. July, 1953.
ISBN 0-933876-01-7
No. 7. The Jet Stream by H. Riehl, M. A. Alaka, C. L.
Jordan and R. J. Renard. August, 1954. ISBN
0-933876-02-5
No. 8. Recent Studies in Bioclimatology. Edited by F.
Sargent, II, and R. G. Stone. October, 1974.
ISBN 0-933876-03-3
No. 9. Industrial Operations under Extremes of Weather.
Edited by J. A. Russell. May, 1957. ISBN
0-933876-04-1
No. 10. Interaction of Sea and Atmosphere: A group
of contributions by A. C. Redfield and A. R.
Miller; G. W. Groves; D. L. Harris; R. 0. Reid;
and W. Marks and J. Chase. June, 1957. ISBN
0-933876-05-X
No. 11. Cloud and Weather Modification: A group offield
experiments by S. Petterssen; J. Spar; F. Hall; R.
R. Braham, Jr., L. J. Battan and H. R. Byers; and
H. J. aufm Kampe, J. J. Kelly and H. K. Weick-
mann. July, 1957. ISBN 0-933876-06-8
Vol. 3. Meteorological Research Reviews: Summaries of progress
from 1951 to 1955. July, 1957. (Numers 12-20, Volume
3, are bound as one volume.) ISBN 0-933876-07-6
No. 12. Review of Climatology, 1951-1955 by H. E.
Landsberg.
No. 13. Meteorological Instruments by J. C. Bellamy.
No. 14. Radiometeorology by J. S. Marshall and W. E.
Gordon.
No. 15. Weather Observations, Analysis, and Forecasting
by S. Petterssen.
No. 16. Applied Meteorology by T. F. Malone.
No. 17. Physics of the Upper Atmosphere by E. 0. Hul-
burt.
No. 18. Physics of Clouds by H. J. aufm Kampe and H.
K. Weickmann.
No. 19. Physics of Precipitation by H. K. Weickmann.
No. 20. Atmospheric Electricity by H. J. aufm Kampe.
Vol. 4, No. 21. Studies of Thermal Convection in a Rotating Cyl-
inder with some Implications for Large-Scale
Atmospheric Motions by Dave Fultz, Robert R.
Long, George V. Owens, Walter Bohan, Robert
Kaylor and Joyce Wei!. December, 1959. ISBN
0-933876-09-2
No. 22. Topics in Engineering Meteorology by J. M.
Biggs, G. S. Vincent; R. K. Blackadar; H. E. Cra-
mer; E. P. Segner; E. Cohen; C. C. Bates and M.
A. Kohler; S. A. Changnon, F. A. Huff and R. G.
Semonin; M. K. Thomas; R. W. Gerdel; A. H.
Murphy; R. A. Boyd; I. Solomon and W. C.
Spreen; 0. L. Stokstad; and F. E. Legg. May,
1960. ISBN 0-933876-10-6
No. 23. Atmospheric Radiation Tables by Walter M. El-
sasser with Margaret F. Culbertson. August,
1960. ISBN 0-933876-11-4
No. 24. Fluctuations in the Atmospheric Inertia: 1873-1950
by El Sayed Mohammed Hassan. May,
1961. ISBN 0-933876-12-2
No. 25. Statistical Prediction by Discriminant Analysis by
Robert G. Miller. October, 1962. ISBN
0-933876-13-0
No. 26. The Dynamical Prediction of Wind Tides on Lake
Erie by George W. Platzman. September,
1963. ISBN 0-933876-15-7
Vol. 5, No. 27. Severe Local Storms. Edited by David
Atlas. September, 1963. ISBN 0-933876-17-3
Vol. 6, No. 28. Agricultural Meteorology. Edited by Paul E. Wag-
goner. July, 1965. Clothbound. ISBN
0-933876-18-1. Paperbound. ISBN 0-933876-19- X
Vol. 7, No. 29. Scattered Radiation in the Ozone Absorption
Bands at Selected Levels of a Terrestrial, Ray-
leigh Atmosphere by J. V. Dave and P.M. Furu-
kawa. January, 1966. Clothbound. ISBN
0-933876-21-1. Paperbound. ISBN 0-933876-22-X
Vol. 8, No. 30. Causes of Climatic Change, Edited by J. Murray
Mitchell, Jr. February, 1968. Clothbound.
ISBN 0-933876-28-9
Vol. 9, No. 31. Meteorological Investigations of the Upper At-
mosphere. Edited by R. S. Quiroz. April,
1968. Clothbound. ISBN 0-933876-29-7
Vol. 10, No. 32. On the Distribution and Continuity of Water Sub-
stance in Atmospheric Circulations by Edwin
Kessler. November, 1969. Clothbound. ISBN
0-933876-30-0
Vol. 11, No. 33. Meteorological Observations and Instrumenta-
tion. Edited by S. Teweles and J. Gi-
raytys. October, 1970. Clothbound. ISBN
0-933876-31-9
Vol. 12, No. 34. Long-Period Global Variations of Incoming Solar
Radiation by Anandu D. Vernekar. February,
1972. Clothbound. ISBN 0-933876-37-8
Vol. 13, No. 35. Meteorology of the Southern Hemisphere. Edited
by Chester W. Newton. November, 1972.
Clothbound. ISBN 0-933876-38-6
Vol. 14, No. 36. Alberta Hailstorms by A. J. Chisholm and Mar-
ianne English. November, 1973 Cloth-
bound. ISBN 0-933876-39-4
Vol. 15, No. 37. The Dynamic Meteorology of the Stratosphere
and Mesophere by James R. Holton. August,
1975. Clothbound. 6W' by 9" book. ISBN
0-933876-41-6
Vol. 16, No. 38. Hail: A Review of Hail Science and Hail Suppres-
sion. Edited by G. Brant Foote and Charles A.
Knight. December, 1977. Clothbound. ISBN
0-933876-46-7
Vol. 17, No. 39. Solar Radiation and Clouds by Ronald M. Welch.
Stephen K. Cox and John M. Davis. May,
1980. Clothbound. ISBN 0-933876-49-l
Vol. 18, No. 40. METROMEX: A Review and Summary by Stan-
ley A. Changnon, Jr., Richard G. Semonin, August
H. Auer, Roscoe R. Braham, Jr., and Jer-
emy M. Hales; Stanley A. Changnon, Jr.,
Editor. October, 1981. Clothbound. ISBN 0-
933876-52-1