Beruflich Dokumente
Kultur Dokumente
Including
IUGG Volumes
Maurice Ewing Volumes
Mineral Physics
GEOPHYSICAL MONOGRAPH SERIES
A. Berger
R. E. Dickinson
John W. Kidson
Editors
• American
Geophysical
Union
International
Union
ofGeodesy
and
Geophysical
Monograph/IUGG
Series
Library of CongressCataloging-in-PublicationData
Understandingclimatechange.
(Geophysicalmonograph;52/IUGG series;7)
1. Climaticchanges•Congresses.
I. Berger,A. (Andr(•).
II. Dickinson, Robert E. (Robert Earl).
III. Kidson, John W. IV. Series.
QC981.8.C5U48 1989 551.6 89-6746
ISBN 0-87590-457-2
Preface
A. Berger,R. E. Dickinson
andJohnW. Kidson ix
one involvingall branchesof geophysics and others, of the importantwork done throughoutthe Interna-
like chemistryand biology, in particular. It remains tional Union of Geodesyand Geophysics.
thus one of the most relevant examplesfor global
changesstudieswithin an InternationalGeosphere- A. Berger is from the Universit(• Catholique de
BiosphereProgram.As is customary,these papers Louvain, the Institute of Astronomyand Geophysics
were all reviewedby the authors'peersin order to G. Lemaitre in Louvain-la-Neuve, Belgium; R. E.
ensuretheir strictadherenceto the high standardsof Dickinsonis from National Center for Atmospheric
internationalresearchpublishing.The IUGG editors Researchin Boulder, Colorado and John W. Kidson is
expresstheir appreciationto all the scientistswho from New Zealand MeteorologicalService,Welling-
gavetime and energyas refereesfor this expression ton, New
Section I
WORLD CLIMATE
RESEARCH
Geophysical Monograph Series Understanding Climate Change Vol. 52
Gordon A. McBean
S S
T T
R R
E E
A A
M M
i II-- STREAM
2
STRE
AM :5
Fig. 1. Schematic box diagram of the earth sciences showing how the WCRP's three stream
strategy includes study of increasingly larger fractions of the earth. The main WCRP
activities are indicated on the boxes. See the text for further explanation.
MCBEAN 5
advances in computing power and analytical transports of mass, heat and ice through Fram
techniques are making it possible to consider Strait will provide a constraint on Arctic Ocean
comprehensive global-ocean models. The World models. The air-ice-sea component of the
Ocean Circulation Experiment (WMO, 1986) has two Greenland Sea Progrmmme will provide data for
basic goals: development and validation of sea ice models.
I To develop models useful for predicting Ways of measuring the vertical oceanic heat flux
climate change and to collect the data are being explored.
necessary to test them; and Although Arctic sea ice variations are
II To determine the representativeness of reasonably well documented by existing
specific WOCEdata sets for the long-term observations, the situation in the Antarctic is
behaviour of the ocean, and to find methods much more difficult. In cooperation with SCOR
for determining long-term changes in the ocean and SCAR, an Antarctic sea ice project is being
circulation. developed to coordinate ship-based hydrographic
The WOCEscientific plan identifies three main surveys, deploy suitably instrumented data buoys
within the Antarctic sea ice zone and examine the
thrusts or 'Core Projects":
(1) The Global Description - essentially to formation of deep ocean and bottom water. These
raise the level of description of the global activities will constitute a part of or will be
oceans to that presently available for the closely co-ordinated with the WOCESouthern Ocean
North Atlantic, with the addition of the Project.
measurement of sea level and wind stress by The optical properties of clouds and their
satellite-borne systems. effect on radiative fluxes are very important to
(2) The Southern Ocean Experiment - concerned the heat balance of the polar regions. The ISCCP
with the Antarctic Circumpolar Current and (see (ii) above) has a special project on
its interaction with the oceans to the developing algorithms for detection of polar
north. clouds from space. Improvements in modelling the
albedo of sea ice and snow are also essential.
(3) The Gyre Dynamics Experiment - designed to
support the development of eddy-resolving The important hydrological and snow-albedo
oceanic circulation models, with priority feedback processes at high latitudes are to be
included in a boreal forest HAPEX of the Land-
given to studying small-scale dynamical
processes that may need to be parameterized Surface Processes and Climate Progrmmme(iii).
in global ocean models.
The intensive WOCE observations will be Global Energ•y and Water Cycle Experiment - GEWEX
concentrated within a five-year period in the
early 1990s to coincide with the time- frame of The distribution, duration and amounts of
the altimetry and scatterometer satellites. Some precipitation constitute the most significant
manifestations of climate for human activities
ship-borne surveys are already underway.
andare sensitive indicators of climate change.
Polar Regions Programme Alterations of precipitation by a CO -induced
climate change may have more socioeconomic impact
While the tropical regions are the main than a temperature rise. However, at present,
recipients of solar radiation, the polar regions the representation of the water cycle in global
are the main sink of radiative energy. To fully weather and climate models is still too crude to
understand the global climate we must examine allow useful projections of future conditions.
both. Variations in the extent and thickness of Looking ahead to the next decade, it is
sea ice in the polar regions (Untersteiner, 1984) expected that there will be increased computer
have an impact on both interannual and longer capability to deal with the finer-mesh models
time-scale variability. The uptake of that will be required to deal adequately with
atmospheric carbon dioxide into the oceans occurs clouds and precipitation. The WCRP's and others'
principally at high latitudes and so does the progrmmmes on clouds and radiation, numerical
formation of deep ocean water. The Sea Ice and experimentation, land- surface processes and other
Climate Progrmmme is the main WCRPactivity in areas will have greatly improved our
the polar regions but other studies are also understanding of the processes of the global
relevant and hence included here. Towards the water cycle.
objective of developing fully interactive There is now growing concern about major
atmosphere-ice-ocean models, sea ice models will global changes. Water and energy cycles are
be tested with time-dependent atmospheric forcing critical and are the basis on which all other
fields (from observations) and specified oceanic global cycles must rely.
heat fluxes for their ability to reproduce the In the period 1995 and beyond, it is expected
time-dependent distribution of sea ice over that new space platforms will offer greatly
periods of several years. The performance of increased opportunities for new remote sensing
atmospheric models in polar regions is being instruments and earth observation. They will
evaluated. The WCRP cooperates with the Arctic make possible the deployment of much larger and
Ocean Sciences Board in its study of the higher-power instruments, such as active remote-
Greenland Sea. The determination of the sounding radars and lidars.
Geophysical Monograph Series Understanding Climate Change Vol. 52
MCBEAN 7
TOGA , I
I WOCE !
ERS-I,2
NSCAT
TOPEX/POSEIDON
ISCCP - --
G
RADIATION PROGRAMME-•- E
W
SEA ICE E
X
PLSPC/ISLSCP
WINDSAT
',[•1 $ c RAINSAT
M F 0 H
0 I V I ß
B F I N
I E E A
L T
H
Y
AIgorithm development
Model development,
NUMERICAL EXPERIMENTATION
Global Observations
Fig. 2. Time schedule of major WCRP activities. See text for further explanation
The Joint Scientific Committee is now solid) and energy within the global atmosphere
formulating a strategy for a Global Energy and and on the underlying surface, which may occur
Water Cycle Experiment, GEWEX, which would start naturally or through the influence of man's
in the period 1995+ and run about five years. activity.
The proposed objectives for GEWEXare:
I To describe and understand the transport of Greenhouse Gases Programme
water (vapour, liquid and solid) and energy in
the global atmosphere and at the underlying It is evident that changing concentrations of
surface; and greenhouse gases (e.g., water vapour, carbon
II To develop methods of predicting changes in dioxide, ozone, methane, chloro-fluoro-carbons)
the distribution of water (vapour, liquid and will have a major impact on the earth's climate.
Geophysical Monograph Series Understanding Climate Change Vol. 52
The WCRPis already developing models of of IOC/SCOR and their working groups and
radiative fluxes and their effect on the Prof. P. Morel, Director/WCRP have all made
atmosphere. A new major programme on the water special contributions.
cycle is being envisaged. What is still needed
is the development of comprehensive global models References
of the dynamical, physical, chemical and
biochemical interactions which control the large- Andre, J.C., J.P. Goutorbe and A. Perrier, 1986:
scale distribution of greenhouse gases and HAPEX-MOBILHY, A hydrologic atmospheric pilot
related chemical species. A specific programme experiment for the study of water budget and
towards this objective is under development. evaporation flux at the climatic scale. Bull,
Amer, Meteor, Soc., 67, 138-144.
Summary Houghton, J.T., and P. Morel, 1984: The World
Climate Research Programme. In The Global
The World Climate Research Programme is a Climate (J.T. Houghton, ed.), Cambridge Univ.
multi-component, endcavour to address the large- Press, 1-11.
scale variability of the global climate system. Mintz, Y,. 1984: The sensitivity of numerically
Several activities are underway (Figure 2)(e.g., simulated climates to land-surface boundary
TOGA, HAPEX-MOBILHY, ISCCP) and several more are conditions. In The Global Climate (J.T.
soon to start (e.g., WOCE). The WCRPis looking Houghton, ed.), Cambridge Univ. Press, 79-105.
ahead to a global water cycle experiment to start Untersteiner, N., 1984: The cryosphere. In The
before the end of the century. The WCRPcalls Global Climate (J.T. Houghton, ed.), Cambridge
upon scientists from many disciplines and Univ. Press, 121-140.
nationalities to work together towards WMO, 1984: Scientific Plan for the World Climate
understanding and prediction of the global Research Programme. WCRP-2, September. WMO/ICSU
climate. Only with their assistance can the (available from WMO, Geneva).
programme be a success. WMO, 1985: Scientific Plan for the Tropical Ocean
and Global Atmosphere Programme. WCRP-3,
September. WMO/ICSU (available from WMO,
Acknowledmments. _
The WCRP has been Geneva).
developed through the efforts of many scientists. WMO, 1986: Scientific Plan for the World Ocean
The members of the Joint Scientific Committee, Circulation Experiment. WCRP-6, July, WMO/ICSU
the Committee on Climate Changes and the Oceans (available from WMO, Geneva).
Section II
PALEOCLIMATES AND
Geophysical Monograph Series Understanding Climate Change Vol. 52
•C. Lorius, •J-M. Barnola,•M. Legrand, •J. R. Petit, •D. Raynaud, •C. Ritz,
2N. Barkov, 2Y. S. Korotkevich, •V. N. Petrov, 3C. Genthon, 3j. Jouzel,
4v. M. Kotlyakov, SF.Yiou, and SG. Raisbeck.
Abstract. Various records obtained from the greater exposure of continental shelves. On the
Vostok (East Antarctica) ice core allow recons- other hand there is no indication of a long term
truction of temperature, accumulation (precipita- relationship between volcanism and climate.
tion), aerosol loading and atmospheric CO concen-
tration histories over the last climatic cycle Introduction
(160 000 years). The results agree with those pre-
viously obtained from two other deep Antarctic ice Data on environmental and climatic changes
cores going back to the Last Glacial Maximum. having affected our planet in the past can be
The Vostok isotope-based temperature and CO obtained from various marine and terrestrial sedi-
records show a large 100 ky signal with changes ments and from ice cores. Although ice-core stu-
of the order of 10øC and 70 ppmv respectively. dies have some inherent limitations (small number
They are closely associated and show periodicities of available records, relative shortness of the
characteristic of the earth orbital parameters. period covered, difficulty of accurate dating},
These features suggest a fundamental link between they also offer certain advantages and unique pos-
the climate system and the carbon cycle and point sibilities. In particular, they can provide high
out the possible role of CO , in addition to inso- resolution records and access to the most impor-
lation inputs, in accountin•for the observed
tem- rant climatic parameters
(temperature,precipita-
perature history. tion, relative humidity and wind strength) as well
The accumulation (precipitation} record appears as the past atmospheric composition including
to be governed by temperature with values during trace gases and aerosols of various origins.
the coldest stages reduced to about 50 % of the This paper focuses mainly on the isotope,
current rate. Ice deposited during these coldest aerosol and CO data from the Vostok ice core over
stages is also characterized by high concentra- the last 160 k• and their climatic interpreta-
tions of marine and terrestrial aerosols ; these tion. For the Last Glacial Maximum, these results
peaks likely reflect strengthened sources and essentially confirm those already obtained from
meridional transport during full glacial condi- the two other Antarctic deep ice cores (Byrd and
tions, linked to higher wind speeds, more exten- DomeC} as shortly discussed for each of these pa-
sive arid areas on surrounding continents and the rameters. Over the full glacial-interglacial
cycle, the Vostok isotope-based temperature
history is discussed in terms of spectral charac-
• teristics and in relation to CO , orbital forcing,
Laboratoire
de l'Environnement,
de Glaciologie
B.P.
96
et de G•ophysique
38•02 St Martin
precipitationandaerosolloading.
d'H&res
2
cedex (France). The Vostok Ice Core
Arctic and Antarctic Research Institute,
BeriDgaStreet 38, 199236Leningrad
(USSR). The Soviet Antarctic station of Vostok is
3Laboratoire
deG&ochimie
Isotopique-LODYC
(UA located
in EastAntarctica
(78ø28'S
and106ø•8'E)
CNRS1206) CEA/IRDI/DESICP/DPC
91191Giœsur at an elevationof 3•90m, with a mean
annualtem-
Yvettecedex(France). peratureof - 55ø C anda current snowaccumula-
Institute of Geography,
Academy
of Scienceof tion of about2.3 g cm
-2 yr-1 Aninitial dril- .
LORIUS ET AL. 15
8•s0 record (fig lb) of Martinson et al. [1987] both may be affected by atmospheric
circulation
which is thought to essentially represent global changes, the overall consistency between the two
continental ice volume changes. The two profiles methods supports the estimate of past accumulation
correspond very closely down to 110 ky BP. However changes and also indirectly supports the deuterium
such consistency is no longer observed for the temperature interpretation.
earlier part of the record (before 110 ky BP} as
there is a major difference in the duration of the CO and Climate
last interglacial estimated respectively from the 2
ice and marine sediments. This likely results from Atmospheric CO data obtained recently both
relative uncertainties in both chronologiesand from Greenlandand2Antarcticice cores have shown
does not affect the above conclusion on the repre- in particular that concentrations during the Last
sentativeness of the Vostok record. Glacial Maximum were lower than Holocene values by
Aside from the large ~ 100 ky signal, visual 25-30 % and that the last deglaciation was charac-
inspection of the temperature curve (fig. lc) also terized by a concentration increase from about
clearly shows a ~ 40 ky oscillation with four well 190-200 to 270-280 ppmv [De!mas et al., 1980 ;
marked temperature minima roughly in phase with Neftel et al., 1982].
the past total insolation at the Vostok latitude The Vostok ice core has made it possible to
(governed by the obliquity cycle with a period of extend the record of past atmospheric CO right
41 ky). Thereare also similarities betweenthis throughthe glacial-interglacial cycle [Ba•nola et
Vostok record and the 65øN July insolation changes al., 1987]. Measurements performed at 66 different
which play a key role in the Milankovitch theory depth levels are separated by time intervals
of the ice ages (this July 65 ø N insolation is ranging from about 2 to 4.5 ky. Due to the gradual
largely influenced by precessional changes with enclosure of atmospheric air in ice, air extracted
periodicities of 19 and 23 ky}. Spectral analysis from the core is younger than the age of the snow
[Jouzel et al., 1987] confirms these visual featu- deposit and this difference (ranging from 2.5 to
res showing that aside from the ~ 100 ky glacial- 4.3 ky depending upon climatic conditions) has
interglacial oscillation, the Vostok temperature been taken into account in establishing the CO
record is dominated
by a strong ~ 40 ky signal. It time scale. The best estimatesof the CO concen5
also suggests a lesser influence by a component trations are plotted (fig. 2a) togethe• with the
slightly greater than 20 ky. The two frequency associated uncertainty bands. The CO concentra-
bands can be associated with the obliquity and tion exhibits twovery large change•betweentwo
precession cycles respectively, thus supporting levels centred around 190-200 and 260-280 ppmv
the role of astronomical forcing in determining with the low and high values associated with full
the late Pleistocene climate [Berger, in press], glacial and interglacial conditions respectively.
already convincingly demonstrated on the basis of The low values previously recorded in other ice
deep sea core records [Hays et al., 1976]. cores [Delmas et al., 1980 ; Neftel et al., 1982]
Beyond this temperature record, Vostok data are fully confirmed. The high level is comparable
allow evaluation of past accumulation changes. As with the so-called "pre-industrial" CO2 concentra-
noted above knowledge of this parameter is requi- tion which prevailed about 200 years ugo.
red for dating ; also, in a more global perspec- Indeed there ms a remarkable correlation (r =
tive growth and retreat of ice sheets largely ß 79) between the Vostok CO and temperature
depend on accumulation
changesoccurringover the records (figures 2a and2b}. Althoughsome diffe-
glacial-interglacial time scale. One approach (the fences are observed between the two records, such
one used for dating} assumes [Robin, 1977] that as the absence of a low CO value associated with
the precipitation rate is governedby the amount the rather cold stage •round 110 ky BP and the
of water vapor circulating above the inversion existence of a CO2 -temperature lag when going from
layer, itself controlled by the saturation vapor warm to cold periods, these results nevertheless
pressure, i.e. by the temperature [Lorius et al., provide the first direct evidence of a close asso-
1985). From the Vostok temperature data we deduce ciation between atmospheric CO and climatic (tem-
that snowaccumulationwas quite similar during perature) changeson a glacial-interglacial time
the Holocene and the previous Interglacial periods scale.
and reduced to about 50 % of the modern value Although there are too few CO data to study
during the coldest periods. Another approach the phaserelation with the isotOpe-temperature
et al.f0 1981
[Raisbeck ; Yiou et al ., 1985] is record , such a close association suggests that CO
based on the Be profile (fig. ld). At Vostok, mayhave, throughits radiative effect and th•
concentrations of this cosmogenic isotope were associated feedbacks, participated in the glacial-
quite similar during interglacials. The glacial- interglacial change. Indeed a simple linear mul-
interglacial concentration changes have been tivariate analysis suggests that CO changes may
interpreted as reflecting lower precipitation have accountedfor morethan 50 % of2 the Vostok
rates during cold periods. Indeed, apart from two temperature variability, the remaining part being
beryllium peaks [Raisbeck et al., 1987], there is associated with orbital forcing [Genthon et al.,
& goodcorrelation betweenthe •recipitation rate 1987]. Qualitatively these results agreewith the
estimatedassuming a constant•VBedeposition flux findings of Broccoli andManabe[1987] who found
and that derived from temperature change. Although that the low CO concentration was the main cause
Geophysical Monograph Series Understanding Climate Change Vol. 52
LORIUS ET AL. 15
ting from strengthened sources and meridional Acknowledgments. This work was supported in
transport linked to higher wind speeds (likely France by several institutions (Terres Australes
induced by higher temperature gradients with lati- et Antarctiques Fran•aises, CNRS-Institut National
rude}, more extensive arid areas over the surroun- des Sciences de l'Univers and C.E.A} and in the
ding continents and the greater exposure of conti- USSR by the Arctic and Antarctic Research Insti-
nental shelves due to a lower sea level [Petit et tute. It is based on a drilling programme
al., 1981]. More generally marine and terrestrial performed by Soviet Antarctic Expeditions. We are
aerosol concentrations measured in ice are also grateful to the NSF-Division of Polar
strongly dependent upon climatic conditions of Programs who provided logistic support.
global {source strength and atmospheric transport
efficiency}, regional {sea ice extent) and local References
{rate of snow accumulation) concern. Changes in
continental and marine inputs and in all major De Angelis, M. , N.I. Barkov and V.N. Petrov,
ions is now documented over the full glacial- Aerosol concentrations over the Last Climatic
interglacial cycle [De Angelis et al., 1987 ß Cycle (160 ky) from an Antarctic ice core.
Legrand et al.' 1988]. As an illustration, the Nature, 325, 318-321, 1987.
Vostok aluminium record, an indicator of continen- Barnola J.M., D. Raynaud, Y. S. Korotkevich and
tal aerosol, is given in figure 2c as evaluated by C. Lorius, Vostok ice core : a 160,000 year
Legrand et al., in press. This record is characte- record of atmospheric CO , Nature, 329, 408-414,
rized by sharp increasesduring cold stages B, D 1987. 2
and H and also by a slight increasing trend in the Berger A., Milankovitch theory and climate, Review
A1 background over the last glacial period which of Geophysics, in press.
may be explained by the emergence of large parts Broccoli, A.J. and S. Manabe, The influence of
of the continental shelves due to sea level continental ice, atmospheric CO2 and land
lowering [De Angelis et al., 1987] and also con- albedo on the climate of the Last Glacial
nected to the snow accumulation trend. As for the Maximum,Climate Dynamics, •, 87-99, 1987.
LGM, the sharp increase of continental input at Charlson, R.J., J.E.Lovelock, M. 0. Andrae and
the end of the previous ice age and during the S.G. Warren, Oceanic phytoplankton, atmospheric
cold stage D may be explained by expansion of arid sulfur cloud albedo and climate, Nature, 326,
areas, higher emissions due to higher wind speed 655-661, 1987.
over continents and more efficient atmospheric Cragin, J.H., M. M. Herron, C. C. Jr. Langway,
transport. There is not yet a clear explanation of G. Klouda, Interhemispheric comparison of chan-
the absence of large concentration values during ges in the composition of atmospheric precipita-
stage F, but this suggests that global climatic tion during the late cenozoic era. in Dunbar
conditions may have been different from full Maxwell J. (ed.), Polar oceans, Proceedings of
glacial during this stage at the beginning of the the polar oceans conference, Montreal, 1974,
last glacial period. Calgary, Arctic Institute of North America, 617-
The marine aerosol record (as illustrated by 63•, 1977.
the Na marine concentration profile in fig. 2d) Delmas, R.J., J. M. Ascencio and M. Legrand, Polar
shows also higher values during cold climatic con- ice evidence that atmospheric CO , 20,000 yr BP
ditions superimposed
on a well marked increasing was 50 % of the present, Nature• 284, 155-157,
trend for the concentrations during the last 1980.
glacial period. In this case we observe a moderate Genthon C., J.M. Barnola, D. Raynaud, C. Lorius,
increase of marine deposits during stage F, possi- J. Jouzel, N.I. Barkov, Y. S. Korotkevich and
bly related to climatic changes around Antarctica V.M. Kotlyakov, Vostok ice core : the climate
as depicted in the Vostok temperature record. In response to CO and orbital forcing changes
comparison,the S0,-- profile (fig. 2e) exhibits over the last •limatic cycle (160,000 years),
rather weakvariations (up to a factor 3). Diffe- Nature, 329, 414-418, 1987.
rent sources contribute to the observed inputs ; Hays, J.D., J. Imbrie and N.J. Shackleton, Varia-
continental dust and sea salt inputs are higher tions in the Earth orbit : pacemaker of the ice
during cold climatic stages but the acidic-gas ages, Science, 194, 1121-1132, 1976.
(H_S0j) derived contribution remains relatively Imbrie J and J Z Imbrie Modelling the climatic
stable over the whole climatic cycle, indicating response to orbital variations, Science, 207,
the absence of a long-term relationship between 243-253, 1980.
volcanism and climate. More generally a comprehen- Johnsen, S. J., W. Dansgaard, H.B. Clausen and
sive study of all major ions present in the ice C.C. Langway, Oxygen isotope profiles through
points out significant changes in the aerosol the Antarctic and Greenland ice sheets, Nature,
loading associated with climate which may have 2•5, 429-434, 1972.
some impact on the radiation balance of the earth Jouzel,J. and L. Merlivat, Deuterium and oxygen 18
atmosphere system. The mineral acid contribution in precipitation. Modelling of the isotopic
represents a large part (over 50 %) of ice impuri- effect during snow formation, J. Geophys. Res.,
ties deposited during interglacials. For glacial 89, 117•9-11757, 198•.
ice the inputs of marine and terrestrial salts Jouzel C., C. Lorius, J. R. Petit, C. Genthon,
become preponderant, representing up to 75 % of N.I. Barkov, V.M. Kotlyakov and V. N. Petroy,
total soluble impurities [Legrand et al., 1988]. Vostok ice core : a continuous isotope tempera-
Geophysical Monograph Series Understanding Climate Change Vol. 52
ture record over the last climatic cycle sol content from East Antarctic ice core samples
(160,000 years), Nature, 329, 403-408, 1987. and past wind strength, Nature, 293, 391-394,
Legrand M., C. Lorius, N.I. Barkov, V. N. Petrov, 1981.
Vostok (Antarctica) ice core ß atmospheric Petit, J. R., P. Duval and C. Lorius, Long-term
chemistry changes over the last climatic cycle climatic changes indicated by crystal growth in
(160,000 yr), Atmospheric Environment, 22, 2, polar ice, Nature, 326, 62-64, 1987.
317-331, 1988. Raisbeck, G.M., F. Yiou, M. Fruneau, J.M.
Le Treut, H. and M. Ghil, Orbital forcing, cli- Loiseaux, M.
•0
Lieuvin, J.C. Ravel and C. Lorius
matic interactions and glaciation cycles, J. Cosmogenic Be concentrations in Antarctic ice
Geophys. Res., 88, 5167-5190, 1983. during the past 30 000 years, Nature, 292, 825-
Lorius, C. and L. Merlivat, Distribution of mean 826, 1981.
surface stable isotope values in East Antarctica Raisbeck G.M.,= F. Yiou, D. Bourles, C. Lorius,
observed changeswith depth in a coastal area. J. Jouzel, andN. I. Barko•d Evidence for two
In ß Isotopes and Impurities in Snow and Ice. intervals of Enhanced Be Deposition in
Proc. Grenoble Symp. August-September 1975, IAHS Antarctic Ice during the Last Glacial period,
Nø 118, 127-137, 1977. Nature, •26, 273-277 ,1987.
Lorius, C., L. Merlivat, J. Jouzel and M. Pourchet Rasool, S.I. and S. H. Schneider, Atmospheric
A 30,000 yr isotope climatic record from carbon dioxide and aerosols : effects of large
Antarctic ice, Nature, 280, 644-48, 1979. increases on global climate, Science, 173, 138-
Lorius, C., J. Jouzel, C. Ritz, L. Merlivat, 141, 1971.
N.I. Barkov, Y.S. Korotkevich and V.M. Kotlyakov Robin, G. de Q., Ice cores and climatic changes,
A 150, 000 year climaticrecordfromAntarctic Land.,B 280, 143-•6 • ,
Phil.Trans.R.Soc. •977.
ice, Nature, •16, 591-596, 1985. Sa[[•[ • C--•r•o••6xide andthe 81 0 record
Martinson, D. G., N. G. Pisias, J. D. Hays, J. of late-quaternary climatic change : a global
Imbrie, T. C. Moore and N.J. Shackleton, Age model, Climate Dynamics, [, 77-85, 1987.
dating and the orbital theory of the ice ages ß Sundquist, E., Ice core links CO to climate,
development
of a high resolution 0 to 300 000 y. Nature, •29, 389-390, 1987. •
chronostratigraphy, Quaternary Research, 27, 1, Thompson, L. and E. Mosley-Thompson, Microparticle
1-30, 1987. concentration variations linked with climatic
Neftel, A., H. 0eschger, J. Schwander, B. Stauffer change : evidence for polar ice cores, Science,
and R. Zumbrunn, Ice core sample measurements 212, 4496, 812-815, 1981.
give atmospheric CO content during the past Wang, W.C., D. J. Wuebbles, V.M. Washington, R. G.
40,000 years, Nature,2295, 220-223, 1982. Isaac and G. Mollnat, Trace gases and other po-
Palais, J.M. and M. Legrand, Soluble impurities tential perturbations to global climate, Reviews
in the Byrd station ice core, Antarctica ß of Geophysics, 24, 110-140, 1986.
their origin and sources, J. Geophys. Res.,90, Yiou, F., G. Raisbeck, D. Bourles, C. Lorius and
1143-1154,1985. Barkov N.I., •øBe at VostokAntarctica during
Petit, J.R., M. Briat and A. Royer, Ice age aero- the last climatic cycle, Nature, 316,616-617,1985.
Geophysical Monograph Series Understanding Climate Change Vol. 52
Michael H. Kuhn
Southern hemisphere
Mountain glaciers 0.003 0.01 0.01 0.3
Antarctica including 13.6 30.1 27.1 2.2
ice shelves
R = H + LE + S + M (1) 8
0
where the individual terms are most con-
1973 75 77 79 81 83 1985
venientl¾ taken as energy flux densities
(W m-2). Figure 1. Time series of Northern Hemi-
Table 2 presents examplesfor melting sphere snowcover area, in 106 km2. The
snow and for Antarctic mid-winter condi- dots mark the November snow cover for
tions. each year. After Ropelewski (1986).
Geophysical Monograph Series Understanding Climate Change Vol. 52
KUHN 19
R = H + LE + S + M
Albedo
80-
27.3.1978
Increasing grain size, liquid water -
KUHN 21
Snow
-6
50 0.007 0.07- 10 0.8
100 0.03 0.15 1.2
300 0.27 0.45 2.1
Ice
Sand
Rock
B = P + D + A + E + M + C (3)
90 '70
• I•
where P, D, A
(accumulation)
(ablation).
and are
C, usually
E, M positive
negative
KUHN 2•
higher frequency between less long-lived than 4 K, about 0.4 K of this being due
stable states of the climate system. to surface albedo changes.
An obvious requirement for such sharp As was mentioned, planetary albedo is
or rapid transitions is the existence of higher than surface albedo over most of
feedbacks involving snow and ice, and it the globe but lower (about 0.75) than
is helped by the potential of ice to suf- surface albedo (0.84) over the dry snow
fer catastrophic, dynamic changes. facies of the Antarctic Plateau. This
masking of surface contrasts by atmo-
Feedbacks and Instabilities spheric backscattering and absorption
decreases the efficiency of the albedo
The Albedo-Temperature Feedback feedback. Shine et al. (1984), among
others, have pointed out that cloudiness
We have mentioned so far two is higher over snow free than over snow
feedbacks, one involving surface layer covered surfaces, a fact that further
instability and one involving orographic contributes to reduce the surface albedo
forcing of planetary waves. The one most feedback.
often quoted in connection with snow and
ice is the albedo-temperature feedback. Topographic Feedbacks Promotinq Ice
With rising global temperature, planetary Sheet Development
albedo is diminished by the melting of
ice thus adding to the original forcing. It would be more precise to speak of
This feedback was the central theme of an energy budget-albedo feedback rather
the classical papers of Sellers (1969) than of a temperature effect. However,
and Budyko (1969). both temperature and energy balance dis-
Hansen et al. (1984) have modelled the play a strong negative correlation with
effects of doubling CO2 or increasing the surface elevation, both dT/dz and db/dz
solar constant by 2 % and found that an are negative as is change of T and b with
original temperature change of about 1.2 latitude • . It was long known to alpine
K (simila• in both cases) was amplified glaciologists that a glacier with a flat
by various feedbacks to a total of more part at the altitude of its equilibrium
Geophysical Monograph Series Understanding Climate Change Vol. 52
dB d(bS) dz _ dS db dz •
d• = dz dT- (b(z) • + S(z) •) d-• •-
A
(4) •
- 280
S being the glacier surface area up to •
altitude z. A flat part means a large va- ß
lue of dS/dz and the value of db/dz is E
characteristic
1984), while
of a
dT/dz
given climate
is roughly
(Kuhn •
constant.
This shows that dS/dz determines the sen- 270
sitivity of glacier mass balance to cli-
matic changes on a local scale. Consi- 1
dering the global scale one finds 1320 1340 13'60 13'80 14'00
d:B
= dB d__z
• (5) Solar
constant
(W/m
2)
dT dz de dT Figure 7. Mean annual Northern Hemisphere
surface air temperature plotted as a func-
For a given meridional temperature gra- tion of solar constant. (A) with realistic
dient dT/d½ the continental slope dz/d½ topography, (B) topography has been elimi-
is thus the parameter that determines nated. From Birchfield and Wertman, 1983.
mass balance sensitivity to a change in
global temperature. Birchfield and Wert-
man (1983) noted that "The Himalaya- librium line, since occasional, extreme
Alpine Belt, the Tibet Plateau, and the summers may waste the ice accumulated
Colorado Plateau appear in the zonally under nearly average conditions.
averaged elevation of the continents as a The Tibetan Plateau is in a similar
single plateau at an altitude of more situation with respect to present equi-
than 1600 m centered between 30 ø and librium lines and may well have experien-
40øN". Any climatic forcing that moves ced a rapid glaciation in the quaternary.
the snow line (equilibrium line) south- Kuhle (1986) believes to have found mor-
ward will thus be enhanced when a further phological evidence for a substantial
southward shift means an increase of sur- pleistocene ice cover in spite of the low
face above sea level. This effect is il- latitude position of the plateau, his as-
lustrated in Figure 7 which shows Nort- sociated climatic speculations, however,
hem Hemisphere temperatures following a have not been confirmed.
change in solar constant for sea level
and for a realistic topography. Reversible and Irreversible Chanqes
Koerner (1980) has given an interes-
ting contribution to the problem of the Annual snow cover extent, equilibrium
initiation of the Labrador-Ungava Ice line lowering and other processes invol-
Sheet. In defeat of the earlier hypothe- viQg snow masses of the order of 103 kg
sis that this ice sheet had been initi- m-2 have an almost instantaneous adjust-
ated by the buildup and subsequent spread ment and are reversible on a time scale
of mountain glaciers, Koerner points at of years. A mass balance disturbance is
the situation of the Baffin Island Plate- transmitted downglacier at a speed se-
au just below today's equilibrium line. veral times that of mean mass flow so
In the terms used above this plateau me- that most ice bodies are continuously in
ans high climatic sensitivity due to high a transient state, lagging behind smaller
values of dS/dz. climatic changes.
Barry et al. (1975) found Little Ice A developing ice sheet soon reaches a
Age conditions in Labrador approaching size large enough to modify atmospheric
ice sheet initiation and stated that temperatures or circulation. For example,
(today) "the climatic changes required to Braithwaite (1977) found surface air
initiate the necessary snow line lowering temperatures over Ellesmere Island gla-
may involve only a minor summer cooling". ciers 1 to 2 degrees lower than over ad-
Koerner (1980) stresses the point that a jacent bare ground. This cooling effect
reduction of annual mass balance variance is magnified when an ice sheet builds up
is just as important for ice sheet ini- to several hundred meters thickness at-
tiation as a lowering of the mean equi- raining a surface climate significantly
Geophysical Monograph Series Understanding Climate Change Vol. 52
KUHN 25
different from that governing its initial sea level changes or by thinning or
growth. When sea level temperature or sea thickening of the ice shelf. When ice at
level energy balance have returned to the old grounding line becomes afloat due
their original, no-snow values, the top to either sea level rise or ice sheet
of the new ice sheet is still in a healt- thinning the grounding line retreats. Ma-
hy glacial condition. ny marine ice sheets are terminating on
This, by definition, is an unstable sills, bedrock sloping downward inland
reaction to climatic forcing. It is not due to a combination of glacial erosion
entirely irreversible, however, and on a and isostatic depression (Fastook 1984).
large enough time scale the ice sheet Under such conditions the grounding line
surface temperature will follow a hy- retreats into deeper water, greatly am-
steresis loop around its original point plifying the response to the climatic
when forced by a negative and subsequent signal and possibly becoming unstable as
positive sea level temperature excursion. was recently the case with Columbia Gla-
cier in Alaska (Meier et al. 1985).
Surges, a Periodic Instability With long distance grounding line re-
treat the ice sheet surface becomes
Glacier surges are a dynamical insta- steeper and ice outflow is accelerated,
bility that is due to internal causes, the ice sheet becomes thinner and requi-
not triggered by climatic forcing. Only a res further grounding line retreat. Nume-
minor fraction of glaciers are capable of rous speculations have been made about a
surging and few have been observed, one possible catastrophic drainage or collap-
outstanding example being the 1982/83 se of the West Anatarctic Ice Sheet. Af-
surge of Variegated Glacier (Kamb et al. ter detailed consideration of all dyn-
1985). When basal water pressure comes amical processes involved Bentley (1984)
near ice overburden pressure, basal sli- comes to the conclusion that, if it
ding greatly accelerates producing ice drains, it will take at least 500 years
velocities of up to 50 m per day. Under for doing so.
usual conditions, higher water pressure In a classical paper Wilson (1964)
enlarges the subglacial tunnel system discusses global cooling due to the al-
which controls water outflow. Only at bedo increase by ice bergs spreading over
high enough speeds frictional melting the Southern Ocean following an Antarctic
enables a subglacial cavity network surge.
without major tunnels to be established
that in turn promotes further pressure to Snow and Ice in the Hydrological Cycle
build up. The conditions suitable for
this second mode of subglacial water flow Apart from the various ways in which
seem to be fulfilled cyclically with pe- ice and snow cover modifies the climate
riods of several decades. West Greenland via the surface energy budget, some cli-
outlet glaciers are believed to be perma- matically relevant processes involve mass
nently in the surging mode. transport or mass storage in the
Since surges are not climate-related cryosphere. These range from the daily
their moraines should not be used indis- freeze and thaw cycle to glacial sea le-
criminantly in paleoclimatic reconstruc- vel changes.
tions. The role of surges in the climate
system lies in their potential of drai- The Seasonal Cycle
ning large ice basins in extraordinarily
short time. Figure 1 shows that an appreciable
amount of water is seasonally stored on
The Groundinq Line Instability and the the land surface. Compared to no-snow
Possible Drainaqe of West Antarctica conditions on the same soil this means
delayed runoff and prolonged evaporation
A survey of climate-related dynamical at the potential rate not only from the
behavior of terrestrial and marine ice snow but also from the soil moisture that
sheets was given by Bentley (1984). He is charged by melt water. The hydrologic
shows that changes in the surface mass effect is not restricted to the local
balance (precipitation or melting) are scale: many semi-arid areas receive an
not nearly as effective in accelerating important fraction of their water balance
ice flow as are shifts in the grounding from ice and snow melt in distant moun-
line position. At the grounding line the tain. This is particularly important in
hydrostatic pressure of the ice load summer-dry regions with sufficient winter
equals that of the sea water, an equi- precipitation that would otherwise drain
librium that may be disturbed either by unused.
Geophysical Monograph Series Understanding Climate Change Vol. 52
hA J F M A M J J A S O N D
mm
4o0 83 %
1500
glacierized
200
2640m
L-'t___
200 44%
Om
1905m
• o - 1o o 1o 2oøc
0
10 I '0
I 925m •c •• sensitivity of this effect to climatic
0 '
! i i •:.• change
•o
is highest
area with altitude
wherethe change
is most rapid
of
and whe-
Fig. 8. Monthly
runoffexpressed
in mm re temperature changesmost slowly with
(kgm-2 mo
-1)for fouralpinegauging
stations
that altitude.
are successively lower and drain less glacierized
catchment
areas in the basinof 0tztaler Ache. Sea Level Chanqes
KUHN 27
polar regions are all insignificant com- feedback effects at the beginning of an
pared to glacier melt or whether they ice age, Quaternary Research, 13, 153-
simply have cancelled in this century is 159, 1980.
still a matter of stimulating discus- Kuhle, M., Die Vergletscherung Tibets und
sions. die Entstehung von Eiszeiten, Spektrum
der Wissenschaften, 1986 (9), 42-54,
References 1986.
Kuhn, M., Climate and glaciers. Procee-
Barnett, T. P., L. D'•menil, U. Schlese, dings of the Canberra Symposium on sea
and E. Roeckner, The effect of Eurasian level, ice and climatic change, Inter-
snow cover on global climate, Science, national Association of Hydrological
239, 504-507, 1988. Sciences Publication, 131, 3-20, 1981.
Barry, R. G., J. T. Andrews, and M. A. Kuhn, M., Mass budget imbalances as cri-
Mahaffy, Continental ice sheets: condi- terion for a climatic classification of
tions for growth, Science, 190, 979- glaciers, Geografiska Annaler, 66 A
981, 1975. (3), 229-238, 1984.
Bentley, C. R., Some aspects of the cryo- Kuhn, M., The bidirectional reflectance
sphere and its role in climatic change, of polar and alpine snow surfaces, An-
AGU Geophysical Monograph, 29, Maurice nals of Glacioloqy, •, 164-167, 1985.
Ewing Volume •, 207-220, 1984. Kuhn, M., A. J. Riordan, and I. A.
Birchfield, G., and J. Wertman, Topogra Wagner, The climate of Plateau Station,
phy, albedo-temperature feedback, and In The Climate of the Arctic, (G.
climate sensitivity, Science, 219, 284- Weller, and S. A. Bowling, eds.). AMS,
285, 1983. Fairbanks, 255-267, 1975.
Braithwaite, R. J., Air temperature and Kuhn, M., L. S. Kundla, and L. A.
glacier ablation, a parametric appro- Stroschein, The radiation budget at
ach, Ph.D. Thesis, McGill University, Plateau Station 1966/67, AGU, Antarctic
Montreal, 146 pp., 1977. Research Series, 25, 41-73, 1977.
Budyko, M. I., The effect of solar radia- Kuhn, M., Possible future contributions
tion variations on the climate of the to sea level change from small gla-
Earth, Tellus, XXI (5), 611-619, 1969. ciers. Paper given at the International
Dewey, K. F., The relationship between Workshop on the Effects of Climatic
snow cover and atmospheric thermal and Change on Sea Level, Severe Tropical
circulation anomalies, Snow Watch 1985, Storms and their Associated Impacts.
Glacioloqical Data, Report GD-18, World Norwich, UK, 1-4 Sep. 1987 (in press).
Data Center A for Glaciology, Boulder, Lindeman, M., and J. Oerlemans, Northern
37-53, 1986. Hemisphere ice sheets and planetary
Fastook, J. L., West Antarctica, the sea- waves- a strong feedback mechanism,
level controlled marine instability: Journal of Climatoloqy, •, 109-117,
past and future, AGU Geophysical Mono- 1987.
graph, 29, Maurice Ewinq Volume •, 275- Liljequist, G. H., Energy Exchange of an
287, 1984. Antarctic snow-field, Norweqian-
Gornitz, V., Mean sea level changes in British-Swedish Antarctic Expedition,
the recent past. Paper given at the In- 1949-52, Scientific Results, Vol. II,
ternational Workshop on the Effects of Part 1 D, Norsk Polarinstitutt, Oslo,
Climatic Change on Sea Level, Severe 1957.
Tropical Storms and their Associated Male, D. H., and D. M. Gray, Snowcover
Impacts. Norwich, UK, 1-4 Sep. 1987 (in ablation and runoff, In Handbook of
press). Snow (D. M. Gray, and D. H. Male,
Hansen, J., A. Lacis, D. Rind, G. Russel, eds.), Pergamon Press, 360-436, 1981.
P. Stone, I. Fung, R.Ruedy, and J. Meier, M. F., Contribution of small gla-
Lerner, Climate sensitivity: Analysis ciers to global sea level rise.
of feedback mechanisms, AGU Geophysical Science, 226, 1418-1421, 1984.
Monograph, 29, Maurice Ewinq Volume •, Meier, M. F., L. A. Rasmussen, and D. S.
130-163, 1984. Miller, Columbia Glacier in 1984:
Kamb, B., C. F. Raymond, W. D. Harrison, Disintegration underway. U.S. Geolo-
H. Engelhardt, K. A. Echelmeyer, N. gical Survey Open-File Report, 85-81,
Humphrey, M. M. Brugman, and T. 17pp, 1985.
Pfeffer, Glacier surge mechanism: 1982- Moser, H., H. Escher-Vetter,
H. Oerter,
1983 surge of Variegated Glacier, O. Reinwarth, D.
AbfluBZunke,
in und
Alaska, Science, 227, 469-479, 1985. von Gletschern, GSF-Bericht 41/86 (ISSN
Koerner, R. M., Instantaneous glacieriza- 0721-1694), Gesellschaft f•r Strahlen-
tion, the rate of albedo change, and und Umweltforschung M•nchen, 1986.
Geophysical Monograph Series Understanding Climate Change Vol. 52
VOLCANOES AND
Geophysical Monograph Series Understanding Climate Change Vol. 52
Abstract.Estimatesof volcanicvolatilecomposition
and earlywork in thisfield attributedtheclimaticeffectsof
massreleaseof sulfur, chlorine and fluorine to the volcaniceruptionsto dustveilsof volcanicash•..amb, 1970;
atmosphere fromtwelvelargeRecentandQuaternary 1971], more recent studieshave shownthat acidic volcanic
volcaniceruptions hasbeenmadeon basisof pre-eruption aerosols aremoreimportantin producinga climaticimpact,
volatilecomposition, asdetermined by electronmicroprobe bexause of theirlongeratmospheric residencetime [Pollacket
in glassinclusionstrappedin phenocrysts in tephra.These al., 1976;RampinoandSelf, 1982, 1984;Devineet al.,
estimates extendourknowledgeof atmospheric loadingby 1984;Rampinoet al., 1985].In fact,smallsulfur-rich
volcanicgasesto includeeventsmorethantwo ordersof eruptions(bothexplosiveandnon-explosive) mayproduce
magnitudelargerthanrecenteruptions observedwith modem similaror evengreateratmosp•c-c•c effects,as
methods. Resultsfor severaleventsagreewell withother meas• by thestratospheric opticalthickness and
independent estimates,basedonicecoresandatmospheric deviations in meanhemispheric temperature, aslarge,
studies.The resultsshow,thatyield of sulfur,chlorineand explosivesilicicsulfur-poor ash-producing eruptions[Serfet
fluorineto theaunosphere is notonlydependent on total al., 1981;Sigurdsson,1982;RampinoandSerf,1982, 1984;
eruptedmass,butlargelydetermined by thecomposition of Roseet al., 1983;Devineet al., 1984;Sigurdsson et al.,
theerupting magmas. Thusvolcanicvolatileyieldfrom 1985;Rampinoet al., 1985].It is nowapparent thatthe
high-silicaor rhyoliticexplosiveeruptions is oneor two potentialclimaticimpactof a volcaniceruptionis notonly
ordersof magnitude lowerthanduringeruptionof equal governed by theintensityor magnitude of theeruption,but
mass
of basaltic
ortrachytic
ma,g•as.Sulfuryieldduring probablymoreimporumtly by thechemicalcomposition of
individualeventsis up to 3x10lø g, suchasin thecaseof the themagma,i.e. theconcentration andtypeof volatile
basalticfissureeruptionsof Laki andEldgjain Iceland. comlxments, whicharedegassed duringeruptionandmay
However,in certaintrachyticeruptions theyieldof halogens generatea volcanicaerosol.
mayexceedsulfuroutput,suchasduringthegreat1815 The determination of pre-emptionvolatileconcentrationof
Tamboraeruptionin Indonesia,whenchlorineandfluorine themagmaandtheyieldof volcanicgases to theatmosphere
yieldtotheatmosphere
isestimated
as2x1014
and1.7x10
TM duringeruption is thereforeof majorinterest
in thestudyof
g, respectively.
Petrologicestimates of sulfuryieldcorrelate thepossiblerelationship betweenvolcanicaerosols and
closelywithnorthernhemisphere annualtemperature climate.Thepotentialof trappedglassinclusions in
anomalies observed followingtheeruptions, in agreement phenocrysts asrecorders of pre-emptionvolatilecontentof
with thefindingsof Devineet al [1984]on a smallerdata magmaswasfirstrecognizedby Anderson[ 1974],who
base. appliedthisme• in estimating thevolcanicvolatile
contributionto thesulfurandchlorinebudgetof theoceans.
Introduction The me• wasalsoappliedto the 1976Mt. St. Augustine
eruptionby Johnston[ 1980], who demonstrated
the
Volcanicactivityis oftencitedasoneof themajorcauses potentially
greatcontribution
of volcaniceruptions
to the
of climatecooling,ontimescales rangingfroma few years chlorinebudgetof theEarth'sstratosphere.
Thesestudies
to millionsof years[Lamb, 1970;KennettandThunell, pavedtheway for thepetrologicestimates
of volcanic
1975,1977;Porter,1981;Rampinoet al., 1985].While degassing
duringearlierhistoricandpre-historic
eruptions.
Petrologicestimatesof volatileemissionsfrom a number
of recenthistoricandseveralQuatenmryeruptions were
madeby Sigurdsson [1982], Devineet al.[1984], and
Sigurdsson et al. [1985] (Table 1). Devineeta!. [1984]
*Present
address:
Glacier
Research
Group,
University
of showedthat,in general,basalticmagmasreleaseaboutan
New Hampshire,Durham,N.H. 03824 orderof magnitude moresulfurthansiliciceruptions of
similarmagnitude.Theseauthorsalsofoundthatsurface
Copyright 1989 by temperature de• afterfourhistoricvolcaniceruptions
International Union of Geodesy and Geophysics (Laki 1783,Tambora1815,Krakatau1883,Agung1963)
and American Geophysical Union. werepositivelycorrelatedwith the estimatedmassof sulfur
Geophysical Monograph Series Understanding Climate Change Vol. 52
TABLE 3. Electron microprobe analyses of glass inclusions and matrix glasses from volcanic eruptions,
and estimates of sulfur, chlorine and fluorine yield to the atmosphere.
EruL• Volumo
(km3DRE): 9 2 1.4 0.7
Yield
(metrictons):
H2SO
4 8.4x 107 3.8x 105 0 3 x 104
HCI 3.9 x 106 6.6x 105 0 7.8x 105
HF 6.2 x 106 2.4x 106 1.7x 104 2.2x 105
Total Acids
(metric
tons): 9.4 x 107 3.4x 106 1.7x 104 1x 106
Yield (grams):
Sulfur 2.7 x 1013 1.2x 1011 0 9.7 x 109
1088+_98
ppmõ 27+Sppm 6+14ppm
(99.9%) (97.7%) (85.5%)
Chlorine 3.8 x 1012 6.4x 1011 0 7.8x1011
151+39ppmõ 140+_.65
ppm 470+_108ppm
(98.4%) (94.9%) (99.8%)
Fluorine 5.8 x 1012 2.2x1012 1.6x 1010 2x1011
230+_272ppmõ 478+309ppm 5+41ppm 126+_83ppm
(51.8%) (78.5%) (0%) (74.2%)
See
Appendix
forsample
identification.
Values
inparentheses
represent
thestandard
deviation
interms
ofleast
units
cited
forthe
valueto theirimmediateleft, thus47.12(144)indicates
a standard
deviationof 1.44wt. %. na. indicatesnotanalysed.* All iron
calculatedasFeO. tall analyses correctedfor sodiumlossby decaycurvemethodof NielsenandSigtmtsson [1981].õ Volatile
differencebetweenmeltinclusion meanandmauixglassmeanand•t confidencein parentheses.n.d.= notdetected.
Em_p• Volume
(km3DRE): 5.1 3 3 3
Yield (metrictons):
H2SO
4 2.9x 105 6.5 x 106 5.1x 105 8.9x 105
HCI 1.1x 106 0 2.9x 105 1.1x 106
HF 9.9x 105 0 0 6.6x 104
Total Acids
(metric
tons): 2.4x 106 6.5x 106 8 x 105 2x106
Yield(grams):
Sulfur 9.4 x 1010 2.1x 1012 1.7x 1011 2.9 x 1011
8+__8
ppm 306+_72
ppm 24_+36
ppm 42+_20ppm
(45.5%) (99.8%) (26.6%) (94.8%)
Chlorine 1.1x 1012 0 2.8 x 1011 1 x 1012
94+_41ppm 41+._69ppm 151+158ppm
(94.6%) (25.2%) (67.8%)
Fluorine 9.3 x 1011 0 0 6.2 x 1010
79_+41 ppm 9+_115ppm
(77.6%) (o%)
TABLE 3 cont. Electron microprobe analyses of glass inclusions and matrix glassesfrom volcanic
eruptions,and estimatesof sulfur, chlorine and fluorine yield to the atmosphere.
No. Anal. 11 12 10 8 11
F(ppm) 218(138) 413(86) 243(129) n.a. 264(92) n.a. n.a.
No. Anal. 5 8 6 11
Em•ned
Volume
0unøD•): 3 9 0.6 0.6 3.5
Yield (metrictons):
H2SO4 0 7.9x 106 1.5x 105 1.5x 106 0
HCI 2.9x 106 3.2x 106 3.8x 105 0 n.a.
HI: 0 6.6x 104 0 n.a. n.a.
Tolal Acids
(metric
tons): 2.9x 106 1.1x 107 5.3x 105 1.5x 106 0
Yield (grams):
Sulfur 0 5 x 1010 4.8x 1011 0
35+39ppm 336+_94ppm
(0%) (99.9%)
Chlorine 2.9x 1012 3.7x 1011 0 n.a.
413+_178ppm 254+146ppm
(97.4%) (90.6%)
Fluorine 0 n.•. n.•. n.•.
TABLE 3 cont. Electron microprobe analysesof glass inclusions and matrix glassesfrom volcanic
eruptions, and estimates of sulfur, chlorine and fluorine yield to the atmosphere.
CampanianPlinian TarawetsPlinian
25 26 27 28 29 30 31 32
Inclusions Matrix Inclusions Matrix Inclusions Matrix Inclusions Matrix
EruptedVolume
(km3DRE): 23 23 0.35 0.35
Yield (metrictons):
H2SO4 0 5.5x 107 5.4x 106 2.3x 106
HC1 0 0 8.7x 105 1.5x 106
HI: 0 0 4.6x 104 1.2x 106
Tolal Acids
(metrictons): 0 5.5x 107 6.4x 106 5 x 106
Yield (grams):
Sulfur 0 1.8x 1013 1.78x 1012 7.39x 1011
302+105ppm 1814+266ppm 754+116pp
(98.7%) (99.9%) (99.9%)
Chlorine 8.5x 1011 1.5x 1012
868+225ppm 1480•_71ppm
(99.6%) (98.2%)
Fluorine 4.3 x 1010 1.1x 1012
44+_71ppm 1157+370ppm
(42.7%) (95.9%)
conclusions
of Hildreth[ 1983] andsuggestthe zonationin theearly,middleandlatephasesof theeruption.Volatile
themagmachamber mayhavebeenthedominantfactor emission
fromtheearlyphase,oftheKatmaieruption
was
responsible
for thecompositional
variabilityin theKatmai dominatedby sulfur(6.45x10
v metrictonsH2SO4).The
deposits. eruptionappearsto havebecomeprogressively
enrichedin
We have determinedthe sulfur, chlorine and fluorine
contentof glassinclusionsin plagioclaseandpyroxene halogens
(mainly
C1),
e.specially
inlatestages,
whenlx106
phenocrysts andof matrixglasses in threesamples from (pyroxene)
to2.92x10
ø (plagioclase)
metrictonsof HC1
early,middleandlatestages of tephrafall deposition wereemittedto theatmosphere. Lyonset al. [in prep.]have
(columns11 to 19, table3). We havedividedthe volumeof identifiedtheKatmaieruptionin an ice corefrom Dye 3,
theentire
plinian
falldeposit
(9km3D.R.E.)
equally
among Greenlandon thebasisof peaksin excesschlorideand
thethreephasesof theeruptionin orderto estimatevolatile sulfate.They alsofind a patternof increasedsulfate
mass.Thereforethereportedtotalvolatilesreleasedfromthe followedby chlorideduringthecourseof theKatmai
Katmaieruption(column19) is the sumof thesamplesfrom eruption.
Geophysical Monograph Series Understanding Climate Change Vol. 52
TABLE 3 cont. Electron microprobe analyses of glass inclusions and matrix glasses from volcanic
eruptions, and estimates of sulfur, chlorine and fluorine yield to the atmosphere.
Eru_lT,
ed Volume
(km3DRE): 1 0.25 10
Yield (metrictons):
H2SO4 2.3x 105 3.5x 103 0
HCI 8.3x 104 2.7x 105 1.2x 107
n.a. n.a. 0
Total Acids
(metrictons): 3.2x 105 2.7x 105 1.2x 107
Yield(grams):
Sulfur 7.6x 1010 1.2x 109 0
33+7ppm 2+12ppm
(99.9%) (0%)
Chlorine 8.1x 1010 2.6x 1011 1.2x 1013
35+34ppm 448+181
ppm 462+107
ppm
(0%) (96.2%) (99.9%)
Huorine n.a. n.a. 0
SeetheAppendix
forsample
identification.
Valuesinparentheses
represent
thestandard
deviation
intermsof leastunitscitedfor
thevalue
totheir
immediate
left,thus
47.12(144)
indicates
astandard
deviation
of1.44wt.%.n.a.indicates
notanalysed.
*All
ironcalculated
asFeO.tall analyses
corrected
forsodium
lossbydecaycurvemethod
of NielsenandSigurdsson
[1981].õ
Volatiledifference
betweenmeltinclusion
meanandmatrixglassmeanandpercentconfidence
in parentheses.
n.d.= notdetected.
Eruption Santa
Maria1902 Bezymianny
1956 Bezymianny
1956
Anal. No. 39 40 41 42 43 44
Glass Inclusions Matrix Inclusions Matrix Inclusions Matrix
EruptedVolume
0cm3DRE): 8.5 1.0 1.0
Yield (metrictons):
H2SO4 5.6x 106 3.7x 106
1.8x 105
HCI 3.7x 106 3.4x 106 n.a.
I-IF 0 0 0
Total Acids
(metric tons): 1.2x 107 7.06x 106 1.8x 105
Yield (grams):
Sulfur 1.84x 1012 1.2x 1012 5.7x 1010
90+_26ppm 467::L-45
ppm 22:!:12plmi
(99.9%) (99.9%) (79%)
Chlorine 3.59x 1012 3.28x 1012 n•.
176+__51
ppm 1263:!:232
ppm
(99.2%) (99.9%)
Fluorine o o o
plagioclase,
hornblende, orthopyroxene,
clinopyroxene and H2SO
4(5.63x106)
and
39%(3.68x106)
ofwhich
was
titanomagnetite[WilliamsandSelf, 1983;Rose,in press]. HCI. LegrandandDe• [ 1987]reporta smallw.ak of
Tracesof darkgray,scoriaceous high-A1basaltoccurbothas sulfin-icacidin theice at Dome C, Antarcticawhichthey
discrete
blebswithinthedaciticpumiceandasisolated attribute
totheeruptio•
ofSanta
Mariaandcalculate
global
fragments scattered
throughout thedeposit.
Mediumgray acid fallout of 2.2x10 t metric tons. No evidence of the Santa
mixedpumices arealsoa minorcomponent of thedeposit Mariaeruptionhasyetbeenrepon• in iceco•s from
andareevidencethattheeruptionmayhavebeentriggeredby Gre•nlmxl,buta smallpeakin theCreteicecoreat thistime
theinjection
of a hightempexaturemaficmagmaintoan intervalmay•spond to theevent[Hammeret al. 1980].
alreadyvapor-saturateddaciticmagmabody[Sparkset al.,
1977;Williams, 1979; Rose,in press].
Ourresultsof themajorandtraceelementcompositions TaraweraEruption,1886 A D.
of glassinclusions
andmatrixglassfromthe1902Santa
Mariaeruptionaregivenin columns 39 and40, respectively, The largestandmostdeslxuctive volcaniceruptionin New
ofTable3.Weesl•ma•thattheSanta
Mariaeruption
emitled Zealandin historicaltimeswasthebasalticplinianfiss•
a total of 9.31x10 ø metric tons of acids, 61% of which was eruptionof Tarawerain 1886.Theactivitywasmainlyalong
Geophysical Monograph Series Understanding Climate Change Vol. 52
a4Innlong
fissure,
which
erupted
atotal
of0.7km3of thelackof outcropsof the 1835deposit[Williams,1952].
high-al•nina
basaltic
magma,
witha mass
eruption
rateof Original
estimatesforthevoll•me
ofthe1835eruption
1.Sx10ø kg/s[W•er et al., 1984].The pliniancolumn products
rangefrom150lqm • (D.R.E.)[Penck,
quoted
in
heightis estimated to havebeen28 km to 34 km, basedon Williams, 1952] to 50 km• (D.R.E.) [Reclus, 1891].
calculations of Walker et al. [1984] andCareyandSparks Williams
[195•]concluded
thatthetotalvolume
didnot
[1986],respectively. The powerfulnatureof theTarawera exceed10 km• (D.R.E.), althoughthisis only an orderof
eruptionis thoughtto bedueto theinteraction of thebasaltic magnitude estimate.
magmawithgroundwater froma nearbygeothermal field Ourresultsof themajorandtraceelementcompositions of
[Walkeret al., 1984].Violentphreaficexplosions associated glassinclusions andmatrixglassfrom the 1835Coseguina
withtheTaraweraeruptionsuggest thata significantfraction eruptionaregivenin columns37 and38, respectively, of
of thetotalvolafilesmayhavebeenderivedfromground Table3. AlthoughtheCoseguina eruptionmayhavebeen
water.Calculations of thegasexit velocityby Walkeret al. responsiblefor a largeemissionof fine ashto the
[ 1984] leadto an estimateof 1.5 to 3.0 % for the total atmosphere, our resultssuggestthattheaerosolcloudfrom
volatilecontentof theeruptedgas-clast mixture,including thiseruption
mayhavecontained
smalli}mounts
of sulfurbut
anygroundwater whichmayhaveinteracted withthe wascomposed
mostlyof HCI (1.23x10• metrictons).There
eruptingmagma. is no evidence from Cneenland ice cores of sulfuric acid
Petrologicstudyof glassinclusions in pyroxenes from deposition
from theCoseguina eruption.Legrandand
twotephrasamples fromtheeruption(early,TW-1 andlate, Delmas[ 1987]reporta smallsulfi•c acidpeakin theice at
TW-2 sta__geejecta)suggest a totalvolatileyieldof ab9ut DomeC, Antarctica,
atthetimeof theCo•guinaeruption
1.15x107
metric
tons
•asacids)
(early
ejecta-
6.5x10
ømetric andestimatea globalacidyield of 2.3x10• metrictons,only
tons;lateejecta-5x10u metrictons).Whilebothphases of a factorof 2 greaterthanour petrologicestimate.It shouldbe
theeruption appearto haveemittedsignificant quanfites
of noted,however,thatthereis a discrepancy in the
sulfur,chlorineandfluorine,theearlyejectawereenrichedin composition of the acids; our petrologicestimatesuggesting
sulfurrelativeto late-stageproducts, whichwer• richerin a volcanicvolatileemissioncomposed dominantlyof HC1,
halogens.The inclusions in theearlyeruptedtephracrw-1) whereasice corestudiessuggestan aerosolcomposed
containthehighestconcentration of sulfur(1888+594;n=5) entirelyof H2SO4.
of anynaturalglassthatwe haveanalyzedsofar.
Furthermore, according to majorelementanalyses, the Mount St. Helens Eruptions 1500 A1). and 1800 A1).
magmain theearlystageof theeruptioncontained about4.9
% totalvolatiles,mostof whichwerelost on eruption.The Sulfur emission from the 1980 Mount St. Helens
magmaeruptedin late-stage activitywasvirtuallydevoidof eruptionwaslow [Sigurdsson, 1982],asis typicalof
volatiles.
explosiveeruptionsof mostdacitemagnms[Devineet al.,
Becauseof its locationin mid-latituderegionof the 1984]. In orderto characterize
timherthevolatiledegassing
southernhemisphere, it is unlikelythattheTarawent of thisvolcano,we havestudieddacitetephrafall fromtwo
eruptionwouldberecorded in Cneenland icecoresor thatit previouseruptionsof Mount St. Helens.They arethe "T"
wouldhavean effecton northernhemisphere climate. tephrafall from theGoatRockseruptiveperiod(ca. 1800
LegrandandDelmas[ 1987],however,reporta peakof A.D.) andthe "W" tephrafrom theKalamaeruptiveperiod
sulfuric acid in an ice core from Dome C, Antarcticawhich [ca. 1530 A.D. Mullineaux and Crandell, 1981; Hoblitt et
theyattributeto theTaraweraeruption.
No estimate
of global al., 198ff, Mullineaux,1986].Thevolumeof eruptedtephra
acid fallout is feasible on basis of the ice core as the Tarawere
peakoverlapswiththatof the 1883Krakataueruption,but hasnotbeenaccurately
•termined,butweestimate
volumes
of about 0.25 and 1 km•(DRE) for the "T" and"W" tephras,
LegraMandl)elmas[ 1987]estimate
a H2SO4 deposition respectively,
on thebasisof publishedfield observations
fluxof3.7kg/kra
2fortheTarawera
eruption. [Crandell et al., 1975; Crandell and Mullineaux, 1978;
Mullineaux, 1986].
CoseguinaEruption,1835 A 1). Ourresults
indicate
thethe1800A.D.emotion
ofthe"T"
tephrahada negligiblesulfuryield (1.11Jx10:'
g S; basedon
The 1835eruptionof Coseguina volcanohasbeen 2_+12 ppmS inclusion-matrixglassdifference)butemitted
regardedasoneof thelargestandmostviolent volcanic substantial
quantifies
ofchlorine
(2.58x1011
gC1).
Ourdata
eruptions
occun/ngin historictimesin theAmericas. indicatethatthe 1530A.D. "W" eruptionmay haveemitted
Coseguinais locatedon the southernendof theGulf of asmuch
as7.6x1010
gSandnegligible
C1(inclusion-matrix
Fonsecain the northwestcomerof Nicaragua.The eruption glassdifferenceonly 35 +34 ppmC1).As with the"T"
waslocallyof greatsignificance[Williams,1952],with f'me eruption,we did notmeasurefluorinein thesamples of the
ashfall nearthe volcanoand as far as Jamaica(1300 km to
"W" tephra, however,basedon the analysesof the 1980
the east),anddarknesslastedlocally for 2 to 3 days. tephra,the 1530A.D. eruptiveeventmay have producedas
Composition of theejectais andesitic,
withpredominantly
(~ 66%) paleto darkbrownglass,containingminutegrains much
as2x1011
gF.
of augite,ironoxide,andabundant plagioclasemicrolites Theseresultsarevery similarto thoseof Sigtl.
rdssonet al.
[Wffi•s, 1952].Discretecrystalsof plagioclase, augite, [1985] whofoundS, C! andF yieldsof 2.5x10 10g,
hypersthene andmagnetite arealsocommonin the 1835 3.4x1010
gand5x1010
g,respectively,
forthe1980
Mount
ejecta,alongwith lithiefragments fromtheoriginalsummit St. Helenseruptiveevent.Theseresultsconrumourdataon
cone. Taupo,reportedbelow,andthefindingsof others[Devineet
Oneof themajordifficultiesin tryingto reconstruct
an al., 1984;Sigurdssonet al., 1985]thatdaciticmagmasare
accuratescenarioof theeruptionandtotaleruptedvolumeis generallysulfur-poor
andoftenhalogen-rich.
Geophysical Monograph Series Understanding Climate Change Vol. 52
farranging
effects,
andthatthe14Cageof131A.D.is discussedabove, we have found a similar enrichment of
probablya morereliableagefor theeruption.Nevertheless, halogensin matrix glassrelativeto glassinclusionsin
we do haveevidencethat significantamountsof HCI and HF samplesof the 1400 yr. B.P. Rabaulignimbrite.These
wereinjectedinto thestratosphere
(eruptioncolumnheight > resultssuggestthattheenrichmentof halogensin thematrix
50 km). Little is known aboutthe climatic effectsof the glassmay be relatedto eithera syn-eruptiveor
halogenacids,whichare not knownto form atmospheric post-deposifional
processes in the ignimbrite.
aerosols,
but theymay havehada majoreffecton the
chemistryof the stratosphere
(e.g.ozone depletion)at that Discussion
time [Crutzen,1974; Mankin andCoffey, 1984].
Evidencefor distantdeposition
fromtheTaupoeruption The resultsof petrologicestimatesof degassing
from the
hasbeenfound in Greenlandice cores. Hammer [ 1984] twelveeruptionspresentedin thispaperandthe seventeen
documenteda "dust"layerin theDye 3 corefrom South eventsstudiedby Devineet al. [ 1984] canbe comparedwith
Greenland, which is dated at about 174 to 175 A.D. The volcanic aerosol mass estimates based on ice core and aerosol
layeris composed of silicateparticlesof "highSi content" opticaldepthstudiesfor theseevents.Thesepetrologic
(precisecompositionunknown)andis not associated with studiesnow providethedatabaserequiredto re-evaluatethe
highacidity.In Antarctica,closerto thesourceof theTaupo potentialclimaticimpactof volcanicdegassing fromlarge
eruption,a layer at 104.2 to 104.5m depthin the Dome C ice eruptionsand a furthertestof the claim of Devine et al.
core(estimatedage .--2000 yr B.P.) wasfoundto have [ 1984] andRampinoandSelf [ 1984]of a directrelationship
elevatedlevelsof D.C. electricalconductivity, sulfate,and betweenmeannorthernhemispheresurfacetemperature
microparticles[Maccagnanet al., 1981;Benoistet al., 1982]. declineaftervolcaniceruptionsandthe massof sulfur
However,major elementanalysesof glassshardsfoundin releasedby the eruption.
thislayer(SiO2 = 66.54+1.51%)by deAngeliset a1.[1985]
Aerosol Estimates
do not supportthe conclusionthatthematerialis from the
Taupoeruption.Therefore,confirmationthattheTaupo
eruptionproducedeitheracidfalloutor ashdeposition in The massof totalacidsreleasedby theeruptionswhichwe
Antarcticamustawaitfutureicecorestudies. Ourpetrologic studied(thosefor whichaerosoldataalsoexist)areplotted
datasuggest,however,thattheeventmaybeonlybarely versusthetotalaerosolmassin figure 1, calculatedfrom
detectable
in polarice coreson thebasisof acidityandD.C. globalpeakopticaldepthestimates[Deirmendjian,1973;
conductivitymeasuremenu
alonebecauseof thelow volcanic Stothers,1984] anddirectmeasurements usingairbornelidar
volatileyield. andparticlecounters[McCormicket al., 1981;Kent, 1982;
McCormick,1982]. For mostof theeruptionsthe agreement
CampanianEruption, 35,000 yearsBJ•.
1012 ,
1:1 (Tarawera,KrakatauandTambora).The discrepancy
Agungeruptionis readilyaccountedfor by theobserved
two-thirds/one-thirddistribution
for the
of the 1963Agungaerosol
1012 1013 1014 101s
betweenthe Southernand NorthernHemispheres,
PetrologicEstimateof Total Acids(g) respectively[Castlemanet al., 1974;Delmaset al., 1985;
LegrandandDelmas,1987].The slightunderestimate of the
Fig. 2. Plotof totalacids(g) estimatedby petrologic depositionflux in AntarcitcafromtheSantaMaria eruptionis
methodversusglobalacidfallout(g) estimatedfrom likelydueto thenorthern latitudelocation(14ø N) of the
Greenland[Hammeret al., 1980] andAntarctic[Legrandand volcano.
Delmas, 1987] ice core studies.Key - sameasin Fig. 1
includingE- Eldgja,C- Coseguina.Linearregression Halogens
equation(y = 1.366e+8*x•).4111, R--0.92) calculatedfor
all datapoints.Line corresponding to 1:1relationship
also The resultsshowvery significantreleasesof HC1andHF
shownfor comparison. gasesto theatmospherein manyeruptions.Of thetwenty-six
Geophysical Monograph Series Understanding Climate Change Vol. 52
2OO
andchemicalmodelssuggestthatHC1andHF will not form
liquidaerosols
undernormalstratospheric
conditions
[Miller,
A 1983; Solomonand Garcia, 1984].
Studiesof theatmospherefollowingrecenteruptions
shows
highlysignificant
increase
of chlorine.Thusspectroscopic
observationsof the total columnamountof HC1, made
shortlyafterthe 1982E1Chichoneruption,showan increase
100
of the stratospheric
HC1burdenof 40 per centovera large
partof the globe,with an estimatedtotal increaseof about
4x1010
gHC1,
orabout
9percent
ofthetotal
global
stratosphericHC1burden[Mankin andCoffey, 1984]. The
hydrogenchloridemay in partbe derivedfrom releaseof
gaseous HC1duringbreakdownof haliteparticlesobserved
in the aerosolcloud from the volcano [Wood et al, 1985], as
o well asfromdirectdegassing of themagma.The atmospheric
10•2 10'•3 10'14 1015estimateof Cadle et al. [ 1977] of the massof HC1 released
bytheAgtmg 1963 eruption (1.2x1012 g)isvirtually
Petrologic
Estimateof TotalAcids(g) identical
toourpetrologic
estimate
(1.42x
1012g),whereas
the estimateof Stolarskiand Buffer [1978] for the Krakatau
20
1883eruptionis oneorderof magnitudelower thanthe
petrologicvalue.StolarskiandCicerone[1974] were the f'u'st
to considerthatchlorinefromvolcanoesmightperturbthe
stratosphericozonelayer.Basedon very conservative
estimatesof chlorinedegassing,the stratospheric
photochemistry modelof Stolarskiand Buffer[ 1978] for the
Krakatau1883eruptionestimatesabout7 per centozone
columndepletion.Outputof chlorinefrom the 1815 Tambora
eruptionis aboutthreeordersof magnitudehigheron basis
Agungß of petrologicdata,andmodelingof theimpacton theozone
layerof suchlargevolcanicchlorineemissions is urgently
needed.
Santa Maria
Tarawera
TemperatureData
0 !
40 50 80
Bezymianny
Heimaey 3.61x 106 0 0
SiO2 Magnin Surtsey 3.52x 106 1.44x 105 n.a.
Agung 3.16x 106 1.70x 106 8.89x 105
Rabaul 2.47x 106 0 n.a.
eruptionin 934 A.D. Eldgjatephrafall collectedby 15. K-192C, averagecomposition of melt inclusionstrapped
Thorvaldur Thordarson, 1.5 km SSW of Kambavam. by pyroxenephenocrysts in rhyolitictephra(latetephra,
Sampleno. 160883,sectionV. top7 cm of a 19 cm thicksection)fromlatephaseof 1912
2. ELD-1, averagecomposition of matrixglassin basaltic A.D. Katmaiplinian fall, Alaska,U.S.A. Sample
tephrafromEldgjaeruptionin 934 A.D. Eldgjatephra collectedby J. Riehle,9/85, 17 km NE of NovaruptaDome.
fall collectedby ThorvaldurThordarson,1.5 lcmSSW of 16. K-192C, averagecomposition of matrixglassin rhyolitic
Kambavam. Sampleno. 160883,sectionV. tephra (latetephra,top7 cm of a 19 cm thicksection)
3. O-1362,averagecomposition of meltinclusions trapped fromlatephaseof 1912A.D. Katmaiplinianfall, Alaska,
by pyroxenephenocrysts in rhyolitictephrafall from U.S.A. Samplecollectedby J. Riehle,9/85, 17 km NE
Oraefajokull,Icelanderuptionin 1362A.D. Collectedby of NovaruptaDome.
H. Sigurdsson, 29/7/83 at Fagurholsmyri, Iceland. 17. K-192C, averagecomposition of meltinclusions trapped
. O- 1362,averagecomposition of matrixglassin rhyolitic by plagioclasephenocrysts in rhyolitictephra(latetephra,
tephrafall from eruptionin Oraefajokull,Iceland1362 top7 cm of a 19 cm thicksection)fromlatephaseof 1912
A.D. Collectedby H. Sigurdsson, 29/7/83 at A.D. Katmaiplinian fall, Alaska,U.S.A. Sample
Fagurholsmyri,Iceland. collectedby J. Riehle,9/85, 17 km NE of Novarupta
5. TP-4, averagecomposition of melt inclusionstrappedby Dome.
pyroxenephenocrysts in rhyolitictephra,130 A.D. 18. K-192C, averagecompostion of matrixglassin rhyolitic
Hatepeplinianphaseof Taupoeruption,NorthIsland, tephra (late tephra,top 7 cm of a 19 cm thick section)
New Zealand. Collectedby C.J.N. Wilson, locality at from latephaseof 1912A.D. Katmaiplinianfall, Alaska,
N103/580133. U.S.A. Samplecollectedby J. Riehle,9/85, 17 km NE
, TP-4, averagecomposition of matrixglassin rhyolitic of NovaruptaDome.
tephra,130 A.D. Hatepeplinianphaseof Taupoeruption, 19. Total Katmai-Sumof thevolatilesreleasedin early,
North Island,New Zealand. Collectedby C.J.N. middleandlatephasetephra
Wilson, locality at N 103/580133. 20. Rab-1,averagecompositionof melt inclusionstrapped
7. TP-5, averagecomposition of meltinclusionstrappedby by plagioclasephenocrysts in dacitictephrafrom 1400 yr.
pyroxenephenocrysts in rhyolitictephra,130 A.D. B.P. Rabaulplinianfall deposit,Papua,New Guinea.
Rotongaiophreatoplinian phaseof Taupoeruption,North Samplescourtesyof Dr. P. Lowenstein.
Island,New Zealand.Collectedby C.J.N. Wilson, 21. Rab-1,averagecompositionof melt inclusionstrapped
localityat N103/580133. by pyroxenephenocrysts in dacitictephrafrom 1400 yr.
8. TP-5, averagecompositionof matrixglassin rhyolitic B.P. Rabaulplinianfall deposit,Papua,New Guinea.
tephra,130 A.D. Rotongaiophreatoplinian phaseof Samplescourtesyof Dr. P. Lowenstein.
Taupoerup6on,North Island,New Zealand.Collectedby 22. Rab-1, averagecompositionof matrix elassin dacitic
C.J.N. Wilson, locality at N103/580133. tephrafrom 1400yr. B.P. Rabaulplinianfall deposit,
9. TP-6, averagecompositionof melt inclusionstrappedby Papua,New Guinea.Samplescourtesyof Dr. P.
pyroxenephenocrysts in rhyolitictephra,130 A.D. Taupo Lowenstein.
plinianeruption,NorthIsland,New Zealand.Collected 23. Rab-3,averagecomposition of meltinclusionstrapped
by C.J.N. Wilson, locality at N103/549233. by plagioclase
phenocrysts in dacitictephrafrom 1400
10. TP-6, averagecompositionof matrixglassin rhyolitic yr. B.P. Rabaulpyroelasticflow deposit,Papua,New
tephra,130 A.D. Taupoplinianeruption,North Island, Guinea.Samplescourtesyof Dr. P. Lowenstein.
New Zealand.Collectedby C.J.N. Wilson, locality at 24. Rab-3,averagecomposition of matrixglassin dacitic
N103/549233. tephrafrom 1400 yr. B.P. Rabaulpyroelasticflow
11. K- 192A, averagecomposition of melt inclusionstrapped deposit,Papua,New Guinea.Samplescourtesyof Dr. P.
by pyroxenephenocrysts in rhyolitictephra(earlytephra, Lowenstein.
basal6 cm of a 19 cm thicksection)from earlyphaseof 25. IT-2053, averagecomposition of melt inclusionstrapped
1912 A.D. Katmai plinianashfall, Alaska,U.S.A. by pyroxenephenocrysts in trachytictephrafrom upper
Samplecollectedby J. Riehle,9/85, 17 km NE of half of plinianfall, Campanianeruption(.--35,000yrs.
NovaruptaDome. B.P.), PhlegreanFields,Italy. Samplescollectedby H.
12. K-192A, averagecomposition of matrixglassin rhyolitic Sigurdsson (5/84) in quarrynearLavorate,Italy.
tephra(earlytephra,basal6 cm of a 19 cm thick section) 26. IT-2053, averagecomposition of matrixglassin trachytic
from earlyphaseof 1912A.D. Katmaiplinianashfall, tephrafrom upperhalf of plinianfall, Campanianeruption
Alaska,U.S.A. Samplecollectedby J. Riehle,9/85, 17 (.--35,000 yrs.B.P.), PhlegreanFields,Italy. Samples
km NE of NovaruptaDome. collectedby H. Sigurdsson (5/84) in quarrynear
13. K-192B, averagecomposition of melt inclusionstrapped Lavorate,Italy.
by plagioclase phenocrystsin rhyolitictephra (middle 27. IT-2054, averagecomposition of melt inclusionstrapped
tephra,6 cm of a 19 cm thicksection)from middlephase by pyroxenephenocrysts in trachytictephrafrom lower
of 1912 A.D. Katmaiplinianashfall, Alaska,U.S.A. half of plinianfall, Campanianeruption(.--35,000yrs.
Samplecollectedby J. Riehle,9/85, 17 km NE of B.P.), PhlegreanFields,Italy. Samplescollectedby H.
NovaruptaDome. Sigurdsson (5/84) in quarrynearLavorate,Italy.
14. K-192B, averagecomposition of matrixglassin rhyolitic 28. IT-2054, averagecompositionof matrix glassin trachytic
tephra (middletephra,6 cm of a 19 cm thicksection) tephrafrom lowerhalf of plinianfall, Campanianeruption
from middlephaseof 1912A.D. Katmaiplinianfall, (.--35,000yrs. B.P.), PhlegreanFields,Italy. Samples
Alaska,U.S.A. Samplecollectedby J. Riehle, 9/85, 17 collectedby H. Sigurdsson (5/84) in quarrynear
km NE of NovaruptaDome. Lavorate,Italy.
Geophysical Monograph Series Understanding Climate Change Vol. 52
29. TW- 1, averagecompositionof melt inclusionstrapped 42. BZ-1956, averagecompositionof matrix glassfrom
by pyroxenephenocrysts in basaltictephrafromplinian andesitictephrafrom the March 30, 1956eruptionof
fall, 1886 A.D. Taraweraeruption,North Island,New Bezymiannyvolcano,U.S.S.R. Samplecollectedby Dr.
Zealand.Collectedby C.J.N. Wilson,localityat G.E. Bogoyavlenskaya at a distanceof about80 km north
N77/984913 from the volcano(10 km northfrom the settlementof
30. TW-1, averagecomposition of matrixglassin basaltic Klyuchi). At thissitethe tephralayer wasabout2 cm
tephrafromplinianfall, 1886A.D. Taraweraeruption, thick.
North Island,New Zealand.Collectedby C.J.N. Wilson, 43. BZ- 1956,averagecompositionof melt inclusions
localityat N77/984913 trappedin phenocrystsin andesitictephrafromtheMarch
31. TW-2, averagecomposition of meltinclusionstrapped 30, 1956 eruptionof Bezymiannyvolcano,U.S.S.R.
by pyroxenephenocrysts in basaltictephrafromplinian Samplecollected
by Dr. G.E. Bogoyavlenskaya
at a
fall, 1886A.D. Taraweraeruption,North Island,New distanceof about80 km north from the volcano(10 km
Zealand.Collectedby C.J.N.Wilson,localityat northfromthe settlement
of Klyuchi). At thissitethe
N77/973936,exposure1.5-2.0m belowtopof 1886 tephralayerwasabout2 cmthick.Samesampleas41/42
scoria fall. butaverageof high-silicainclusions.
32. TW-2, averagecomposition of matrixglassin basaltic 44. BZ-1956, averagecomposition of matrixglassfrom
tephrafromplinianfall, 1886A.D. Taraweraeruption, andesitic
tephrafromtheMarch30, 1956eruptionof
North Island,New Zealand.Collectedby C.J.N. Wilson, Bezymianny volcano,U.S.S.R. Samplecollected by Dr.
localityat N77/973936,exposure1.5-2.0m belowtopof G.E. Bogoyavlenskaya at a distanceof about80 km north
1886 scotia fall. from thevolcano(10 km northfrom the settlementof
33. MSH-201, averagecomposition
of melt inclusions Klyuchi). At thissitethetephralayerwasabout2 cm
trapped
byplagioclase
phenocrysts
in daciti6tephra, thick.Samesampleas41/42 butaverageof high-silica
Mount St. HelensKalamaeruptiveperiod-1530 A.D. matrix glass.
CW" tephra),Washington,U.S.A. Collectedin Bear
Meadow by H. Sigurdsson,8/14/84.
34. MSH-201, averagecomposition of matrixglassin dacitic Acknowledgements. We wish to thankD. Browningfor
tephra,MountSt. HelensKalamaeruptiveperiod-1530 hishelpin samplepreparationandmicroprobeanalysis.
A.D. CW" tephra),Washington,U.S.A. Collectedin S.N. Carey wrotethe computerprogramfor determiningthe
BearMeadowby H. Sigurdsson, 8/14/84. statisticalsignificanceof our resttim.We gratefully
35. MSH-202, averagecomposition of melt inclusions acknowledge thefollowingpeoplewho providedsamplesfor
trappedby plagioclase phenocrysts in dacitictephra, this study:ThorvaldurThordarson(Eldgja), C. J. N.
MountSt. HelensGoatRockseruptiveperiod-1800 Wilson (TaupoandTarawera),J. Riehle (Katmai), P.
A.D. CT" tephra),Washington,U.S.A. Collectedin Bear Lowenstein(Rabaul),S. Self (Coseguina),W.I. RoseJr.
Meadowby H. Sigurdsson, 8/14/84. (SantaMafia), andG.E. Bogoyavlenskaya (Bezymianny).
36. MSH-202, averagecomposition of matrixglassin dacitic We alsowishto thankS.N. CareyandW. Cornell for
tephra,Mount St. HelensGoatRockseruptiveperiod assistance with themicroprobeandfor helpfuldiscussions.
-1800 A.D. CT" tephra),Washington,U.S.A. Collected This work wasfundedby the NationalScienceFoundation
in BearMeadowby H. Sigurdsson, 8/14/84. grantEAR-8503104.
37. COS-5,averagecomposition of meltinclusions trapped
by phenocrysm in darkgrey-blackscoriafall from the
References
1835A.D. Coseguina, Nicaraguaeruption.Sample
collectedby SteveSelf on the E-SE flanksof thevolcano.
38. COS-5,averagecomposition of matrixglassin dark Anderson,A.T., Chlorine,sulfur,and water in magmasand
grey-blackscoriafall from the 1835A.D. Coseguina, oceans,Geol. Soc. Am. Bull., 85, 1485-1492, 1974.
Nicaraguaeruption.Samplecollectedby SteveSelf on Angell, J.K., The closerelationbetweenAntarctic
the E-SE flanks of the volcano. total-ozonedepletionandcoolingof the Antarcticlow
39. SM-1902, averagecomposition of meltinclusions stratosphere, Geophys.Res. Lett., 13, 1240-1243, 1986.
trappedin phenocrystsof pumiceousdacitictephrafrom Angell, J.K., andJ. Korshover,Surfacetemperature
the 1902A.D. eruption.Samplecollectedby Bill Roseon changesfollowingthe six majorvolcanicepisodes
therim of SieteOrejasvolcanoabout6 km NNW of the between1780 and 1980, Jour.Climate Appl. Meteorol.,
SantaMaria vent, Guatemala. 24, 937-951, 1985.
40. SM-1902, averagecomposition of matrixglassof Bailey, D.K., andR. Macdonald,Fluorineandchlorinein
pumiceous dacitictephrafrom the 1902A.D. eruption. peralkalineliquidsandtheneedfor magmageneration in
Samplecollectedby Bill Roseon therim of SieteOrejas an opensystem,Mineral. Mag., 40, 405-414, 1975.
volcanoabout6 km NNW of the SantaMaria vent, Barberi, F., F. Innocenti, L. Lirer, R. Munno, and T.
Guatemala. Pescatore,The Campanianignimbrite:A majorprehistoric
41. BZ- 1956,averagecomposition of meltinclusions eruptionin theNeapolitanarea(Italy), Bull. Volcanol.,
trappedin phenocrystsin andesitictephrafrom theMarch 41, 1-22, 1978.
30, 1956 eruptionof Bezymiannyvolcano,U.S.S.R. Benoist,J.P., J. Jouzel, C. Lorius, L. Merlivat, and M.
Samplecollectedby Dr. G.E. Bogoyavlenskaya at a Pourchet,Isotopeclimaticrecordoverthe last2.5 Ka
distanceof about80 km northfrom the volcano(10 km from Dome C, Antarctica, ice cores, Annals Glaciol,, 3,
northfnxn the settlement
of Klyuchi). At thissitethe 17-22, 1982.
tephralayerwasabout2 cm thick. Bull G.A. andD.G.James,Dustin the stratosphere
over
Geophysical Monograph Series Understanding Climate Change Vol. 52
refletde l'environment,
UniversiteScientifique
et Rose, W.I., Jr., R.E. Stoiber, and L.L. Malinconico,
Medicale de Grenoble, 1985. Eruptivegascompositions andfluxesof explosive
Legrand,M.R. and R.J. Delmas, The ionic balanceof volcanoes:budgetof S andC1emittedfromFuego
Antarcticsnow:A 10-yeardetailedrecord,Atmos. volcanoGuatemala,in Andesites,editedby R.S. Thorpe,
Environ,, 18, 1867-1874, 1984. pp. 669-676, JohnWiley, New York, 1982.
Maccagnan,M. J.M. Barnola,R. Delmas,andP. Duval, Rose, W.I., R.L. Wunderman, M.F. Hoffman, and L. Gale,
Staticelectricalconductivity
asanindicatorof thesulfate A volcanologist'sreviewof atmospheric
hazardsof
contentof polarice cores,Geophys.Res. Lett.,8, volcanicactivity,FuegoandMt. St. Helens,J. Volcanol.
970-972, 1981. Geotherm. Res., 17, 133-157, 1983.
Mankin, W.G. andM.T. Coffey,Incresesstratospheric Rose, W.I., Santa-Maria, Guatemala: Bimodal soda-rich
hydrogenchloridein theE1Chichoncloud.Science,226, calcalkalic stratovolcano,J. Volcanol. Geotherm. Res., in
170-172, 1984. press.
McCormick, M.P., Ground-basedand airborne Rosi,M., A. SbranaandC. Principe,The Phlegraeanfields:
measurements
of Mount St. Helensstratospheric Structuralevolution,volcanichistoryanderuptive
effluents, NASA Conf. Publ., 2240, 125-130, 1982. mechanisms, J. Volcanol. Geotherm. Res., 17, 273-288,
McCormick, M.P., G.S. Kent, G.K. Yue, and D.M. 1983.
Cunnold,SAGE measurements of the stratospheric Schatten,K.H., H.G. Mayr, I. Harris, and H.A. Taylor,
aerosoldispersionandloadingfrom the Sourfriere Jr., A zonallysymmetricmodelfor volcanicinfluence
volcano,NASA Tech. Pap., 1922, 1-20, 1981. uponatmosphericcirculation,Geophys.Res.Lett., 11,
Miller, E., Vapor-liquidequilibiraof water-hydrogen 303-306, 1984.
chloride solutionsbelow OøC. J. Chem. Ene. Data. 28. Self, S., M.R. Rampino,andJ.J.Barbera,The possible
363-367, 1983. effects
oflarge
19thand20th
century
volcanic
eruptions
Mullineaux,D.R., Summaryof pre-1980tephra-falldeposits on zonalandhemispheric
surfacetemperatures,
J.
eruptedfrom Mount St. Helens,Washingtonstate,USA, Volcanol. Geotherm.Res., 11, 41-60, 1981.
Bull, Volcanol,, 48, 17-26, 1986. Sigurdsson,H., Volcanicpollutionandclimate:The 1783
Mullineaux,D.R. andD.R. Crandell,The eruptivehistoryof Laki eruption,Eos Trans.AGU, 63,
Mt. St. Helens,In; P.W. Lipman,D.R. Mullineaux 601-602, 1982.
(eds.)The 1980 eruptionsof Mt. St. Helens,Washington, Sigurdsson,H. J. D. Devine, andA. N. Davis, The
U.S. Geol. Surv.Prof. Pap., 1250, 3-15, 1981. petrologicestimationof volcanicdegassing,Jokull,35,
Nielsen,C. H., andH. Sigurdsson,Quantitativemethods 1-8, 1985.
for electronmicroprobeanalysisof sodiumin naturaland Sirnkin, T., L. Seibert, L. McClelland, W.G. Melson, D.
syntheticglasses,Am. Mineral., 66, 547-552, 1981. Bridge,C.G. Newhall, andJ. Latter,Volcanoesof the
Oliver, R.C., On the responseof hemisphericmean World, Smithsonian Institution, Wash. D.C., 1981.
temperature to stratospheric
dust:An empiricalapproach, Solomon, S. and R.R. Garcia, On the distribution of
Jour.Appl. Meteor., 15, 933-950, 1976. long-livedtracersof chlorinespeciesin themiddle
Oskarsson,N., The interactionbetweenvolcanicgasesand atmosphere. J. Geophys.Res. 89, 11633-11644,1984.
tephra:Fluorineadheringto tephrafrom the 1979Hekla Sparks,R.S.J.,H. Sigurdsson, andL. Wilson,Magma
eruption,J. Volcanol. Geotherm.Res.,8, 251-266, 1980. mixing:a mechanismfor triggeringacidexplosive
Palais,J.M. andM. Legrand,Solubleimpurities
in theByrd eruptions,Nature,267, 315-318, 1977.
Stationice core,Antarctica:
Theiroriginandsources,
J. Stolarski, R.S., and D.M. Butler, Possibleeffects of
Geophys.Res., 90, 1143-1154, 1985. volcaniceruptionson stratospheric minorconstituent
Pollack,J.B.,O.B. Toon,C. Sagan,A. Summers,B. chemistry.Pageoph.117,486-497, 1978.
Baldwin,andW. Van Camp,Volcanicexplosions and Stolarski,R.S., andR.J. Cicerone,Stratospheric
chlorine:a
climatechange:A theoreticalassessment,
J. Geophys. possiblesinkfor ozone.Can. J. Chem. 52, 1610-1615,
Res,, 81, 1071-1083, 1976. 1974.
Porter,S.C., Recentglaciervariationsandvolcanic Stothers,R.B., Mysterycloudof AD 536, Nature,307,
eruptions,Nature,291,139-142, 1981. 344-345, 1984a.
Rampino,M.R. andS. Self,Historiceruptionsof Tambora Stothers,R.B., The greatTamboraeruptionin 1815 andits
(1815), Krakatau(1883), andAgung(1963), their aftermath, Science, 224, 1191-1198, 1984b.
stratosphericaerosols,andclimaticimpact,Quat.Res,, Stothers,R.B., J.A. Wolff, S. Self andM.R. Rampino,
18, 127-143, 1982. 1986:Basalticfissureeruptions,plumeheightsand
Rampino,M.R. andS. Self, Sulphur-richvolcanic atmospericaerosols.Geophy•. Res. Lett. 13, 725-728.
eruptionsand stratospheric
aerosols,Nature,310, Thompson,L.G., E. Mosley-Thompson, W. Dansgaard,
677-679, 1984a. andP.M. Grootes,The Little Ice Age asrecordedin the
Rampino,M.R., andSelf, S., The atmospheric
effectsof E1 stratigraphy of the tropicalQuelccayaice cap,Science,
Chichon, $•i, Am,, 48, 1984b. 234, 361-364, 1986.
Rampino,M.R., R.B. Stothers,and S. Self, Climatic effects Thorarinsson,
S., The Oraefajokulleruptionof 1362, Acta
of volcaniceruptions,Natur?,313, 272, 1985. Naturalia lslandica, 2, 1-101, 1958.
Reclus,E., NouvelleG6ographieuniv•rs½11•, 17, 488-489, Walker, G.P.L., The Taupopumice:Productof the most
Paris, 1891. powerfulknown (ultraplinian)eruption?,J. Volcanol.
Robock,A., The Mount St. Helensvolcaniceruptionof 18 Geotherm. Res., 8, 69-94, 1980.
May 1980: Minimal climaticeffect,Science,212, Walker, G.P.L., R.F. Herrting,T.J. Sprod,andH.R.
1383-1384, 1981. Walker, Latestmajoreruptionsof Rabaulvolcano,In'
Geophysical Monograph Series Understanding Climate Change Vol. 52
BIOGEOCHEMICAL CYCLES,
LAND HYDROLOGY,
LAND SURFACE PROCESSES
AND
Geophysical Monograph Series Understanding Climate Change Vol. 52
Department
of Meteorology,
University
of Stockholm,
Sweden
Berrien Moore
Abstract. Inverse methods have been used to centuries. We see this most clearly in the
deduce water circulation, spatial patterns of steadily increasing 032 concentrationin the
turbulent exchange and biological activity in atmosphere, from a most likely value of about
the Atlantic Ocean, by using a set of stationary 280 ppm in the middle of the 18th century to 344
tracers and a condition of quasi-geostrophic ppm in 1983 (cf. Siegenthaler and Oeschger 1987;
flow. The solution yields a direct meridional Figure 1). This means an increase of the amount
circulation cell with descending motion in the of carbon in the form of 03• in the atmosphere
northern Atlantic with an intensity of 20-25 by 135Pg (Pg=1015g). z. .
The em•ssxons, due to
Sverdrup, a reasonable distribution of vertical fossil fuel combustion, during this period of
turbulent transfer in the uppermost ocean layers time have been about 180 Pg C (cf. Rotty, 1987),
and cont•ratively large rates of detritus while estimates of the emissions due to man's
formation,about4.5 PgC yr-1. - Thesolution interference with the terrestrial biota
is used to compute the invasion of tritium (vegetation and soils) are much more uncertain
1955-1983, and the uptake of excess radiocarbon and range between 100-200 Pg C (cf. Bolin,
and carbon dioxide during the period 1760-1983. 1986). The airborne fraction during this period
A fair agreementbetween•x•u. ted and observed thus has been between 36-48% (cf. Table 1). If
changes of tritium and -=C is obtained, but we consider the two periods 1760-1957 and
the period of observations is too short to serve 1958-1985 separately we find that we are not
as a conclusive test of the model. - The uptake able to tell if there has been a significant
of carbon dioxide during the 220 years period c2k3ngeof the airborne fraction during this
into the Atlantic Ocean is 33 .+. 5 Pg and it is period of time (cf. Table 1).
further found that significant variations of the It has generally been accepted that a major
uptake fraction of the CO• emissions may have part of the emissionsof 032 into the atmosphere
occurred due to varying r•tes of emissions in has been absorbed by the world oceans.
course of time. The conclusion is tentatively Insufficient measuremen• are available that can
drawn that the ocean and its carbonate system quantitatively verify this view. It is rather
may not have been the only sink for primarily being based on simple models of the
anthropogenic emissions of carbon dioxide into circulation of carbon in the sea (cf. Bolin,
the atmosphere. - Means for how to further 1986). The most commonly used model is the one
improve the model and its capability to developed by Siegenthaler (1983), the so called
reproduce the ocean behaviour are discussed. box-diffusion model with polar outcrops. A
simulation of the fossil fuel uptake since 1860
Introduction yields an airborne fraction of 60%. Siegenthaler
(1983) argues, however, that the exchange
Burning of fossil fuels, deforestation and between the ocean surface layers and the deep
changing land use have c2k3ngedthe global carbon sea should be enhanced as c• with what is
cycle very significantly during the last two being obtained by determining the model
t• using steady state distribution of
modified model yields a value for the
Copyright 1989 by airborne fraction slightly above 50%. It is,
International Union of Geodesy and Geophysics however, questionable if such a modification of
and American Geophysical Union. the box-diffusion model is justified. We need
57
Geophysical Monograph Series Understanding Climate Change Vol. 52
350 - I I I I I I I I i I i I' I
330-
z
r¾" 310 -
I--
z
ILl --
Z
o
290 -
o
o o
270 , , , , , , , , , , , , ,
1720 1760 1800 1840 1880 1920 1960 2000
YEAR
Fig. 1. AtmosphericpCO
2 1760-1980as determinedby analysis of air bubbles in glacier ice
and since 1957 by direc• measuremen• (Cc•piled by Siegenthaler ar• Oeschger, 1987).
rather more detailed a•d realistic models to be physical oceanographers and about chemical and
able to determine more precisely the uptake biological processes that are of importance for
capability of the oceans. For this purpose we understanding the role of the oceans in the
should combine our knc•ledge about the global carbon cycle as is being studied by
circulation of the • as developed by the chemical c•eanc•3ra••. If using General Ocean
(1) Measurementsas reported by Keeling and colleagues (cf. Bolin, 1986) and
Siegenthaler and Oeschger (1987).
(2) Rotty (1981); Rotty and Masters (1985).
(3) According to Bolin (1986). The lower estimate (100 PgC) implies very small
emissionsbefore1860, as direct assessmentsof releases since 1860 hardly are less
than 100 PgC.Thepartitioning of the emissions to the two periods is approximate.
Geophysical Monograph Series Understanding Climate Change Vol. 52
BOLIN ET AL. 59
Circulation Models it is most important that the distributions of temperaS, salinity, total
rates of ocean overturning be carefully ssolved inorganic carbon (DIC), alkalinity,
validated by using tracer distributions, since "C, oxygen and phosphorus. Note that we do not
direct current • are hardly adequate consider the possible role of the formation of
for this purpose. It is obvious that dissolved organic carbon in the decomposition
determination of gaseous uptake in which we are process of organic matter. The following
interested is crucially dependent on careful analysis is still of interest as a study of what
validation of model characteristics in this the classical set of processes imply with regard
regard. Maier-Reimer and Hasselmann (1987) have to transfer patterns of carbon in the Atlantic
recently presented studies of the circulation of Ocean.
carbonin the sea (both12Cand14C)by using We divide our domain into eight
general ocean circulation models. It is, quasi-isopycnic layers (Figure 2) and twelve
however, obvious that a more detailed regions (Figure 3), based on our qualitative
discussion of their results is •nee•ed in order knowledge about the key features of water
to ascertain that the response characteristics circulation. Altogether 84 water reservoirs,
of the modelwith re• to •02 uptakeagree boxes, are defined in this way. An unknown
with those of the real ocean. vector x is specified having the 536 components:
In the present paper we shall attempt an
- advective fluxes of water between all
analysis of oceanuptake of (I) 2 using an inverse
methodology (cf. Bolin et al. I987) in which adjacent boxes (184 components)
case we shall employ the continuity equations
for a set of key tr•• (salinity, dissolved - rates of turbulent exchange of matter
inorganic carbon, radiocarbon, alkalinity, between all adjacent boxes (184 ccmpo•ts)
phosphorus, and oxygen). It is, however, also
important to make use of the extensive production or decomposition of organic
hydrographic data that have been collected for matter, one ccmpo•t for each box (84
many years. We accordingly also impose a c•mponents)
constraint of q•asigeostrophic flow. Our aim is
to develop a multiplebox model of the Atlantic - carbonate production or dissolution, one
Ocean that is carefully calibrated with the data component for each box (84 ccmpo•ts).
referred to and that has sufficient spatial
resolution so that the •02 uptakeduring the Assuming that the steady state concentration of
last few hundred years can be computed with the seven tracers can be determined for the 84
reasonable accuracy. boxes we are able to formulate 8 x 84 = 672
The present analysis will be restricted to equations, one for each of the seven tracers and
the Atlantic ocean because of data limitations each box and one additional set of 84 equations
and generalizations of our results to the world expressing water continuity for each box. We
oceans are not yet possible. A few general have adopted the finite difference approximation
observations will however be made questioning that advection between two boxes carries the
the view that the world oceans represent the mean concentration of these in the direction of
only significant sink for excess atmospheric flow and that turbulent transfers matter in the
carbon. We shall also learn that the emplo•t direction of the gradient between them. It is
of inverse methods to develop boxmodels requires further assumed that the change of the
a careful and detailed analysis in order to show horizontal flow as a function of depth
that the results are reliable. Our conclusions approximately agrees with the thermal wind
will thus not be final and firm, but will computed from hydrographic data, which yields
contribute to the fundamental problems we another set of 80 equations. This total set of
encounter when trying to assess the role of the 672 + 80 = 752 equations can be written in the
oceans in the global carbon cycle. form of a matrix equation (for details, see
Bolin et al., 1987)
Model Development
Ax = b (1)
Derivation of a Basic Steady State Pattern of
Circulation and Biochemical Processes where A is a 752 x 536 matrix, determined by the
concentration distributions in the sea, b
Bolin et al. (1987) have presented a general quantifies the water or tracer flux across
method for derivation of the steady state external bo•ies and radioactive decay (in
circulation pattern and biochemical processes by the case of ' •=-),
C and the prescribed vertical
using inverse methods. We ask the question: What change of horizontal advective flow from one box
steady state patterns of water circulation, to another in the equations for
turbulent transfer, new primary production, q•asi-geostrophic flow. We determine the air sea
decomposition of organic detrital matter and exchange by assuming that there is no net flow
dissolution of biogenic carbonates are required of •02 to or from the A•antic Oceanas a whole
to explain the observed q•si-steady and t/%at the inflow of •=C is balanced by
Geophysical Monograph Series Understanding Climate Change Vol. 52
O O O O O O O O
O O O O O O O
•D O O O O O O
I I I / I
o o o o o o o
•D o o o o o o
Geophysical Monograph Series Understanding Climate Change Vol. 52
BOLIN ET AL 61
lO 11
12
radioactive decay within the basin. In reality weighting of the equations in (1). Bolin et al.
there maybe a net through flow which implies an (1987) put equal weight on all tracer equations
error in the way we determine rates of air-sea by normalizing the equations, but up-weighted
exchange.Such an error will affect the computed the geostrophic equations in order to fulfill
uptake
of excess
carbon
and14Cin the the geostrophic condition to about 15%. It is
atmosphere.In addition the following inequality clear that the gross features of the circulation
and biochemical processes were reasonably well
reproducedas we knowthem from other studies,
- turbulent flux of matter is always in the but resolution as well as specification of the
direction of the concentration gradient key processes is not adequate to yield features
such as the boundary currents or the coastal
primary production only occurs in sunlit maximaof primary production. This is obviously
surface boxes and deccaposition and a short-com/ng of the model, that has been
dissolution of detrital matter take place at discussed by Bolin et al. (1987).
Schlitzer (1988) in a similar attempt to
inverse methods to deduce the role of ocean
Since the model is approximate, equation system circulation and biological processes for the
(1) is incompatible and a solution for the qp,mmi-steady
distribution of tracers applied a
vector x is obtained by minimizir• the norm in more detailed and probably also more accurate
the least square sense. The result of such a methodof analysis. It is, however, not possible
procedureis of course dependenton the relative in that way to arrive at a closed set of
Geophysical Monograph Series Understanding Climate Change Vol. 52
equations (1), which can be extended to a set of in the atmosph• and the chang• amount of DIC
time dependent equations (1), that in turn can in the surface boxes caused by influx fr•m the
be used for study of transient changes, which is atmosph• •nd will accord•ly vary with time
the prime objective of the present analysis. Similarlyb" in theradiocarbon
equation
will •
Derivation of the Time Dependent Model
••Cf• due
the to thetransfer
atmosphere, whileof
b•b•_produced
• will be given
as a function of time since return flow from the
We shall use the model and the solution x sea to the atmosphere can be neglected.
In order not to introduce artificial sources
derived in the previoussection•o determinehow and sinks in the course of deriving a tz-mmsient
excess concentrations of •0•, •=C and tritium solution we must d• that the solution x is
in the atmo•ere penetrate-into the ocean
assuming that the water motions and biological derived with the condition that the equations
for the particular tracer (n) to be considered
processes,i.e. the vector x, •main unchanged. be well satisfied in steady state, i.e.
•ne injections of tritium and •=C since 1954 are
tknrti•arly interesting, because there are sc•ne
data that shc• the invasion of these tracers Bn •= An x = bn (5)
into and within the Atlantic Ocean. If we are
able to ascertain that the model is able to This condition was not forcefully imposed for
any particular tracer in deriving the solutions
describe these processes adequately, we may also
given by Bolin et al. (1987), but the errors in
deduce with plausible reliability the likely
the incompatible set of equations (1) were
uptake of carbon dioxide, although direct
observations for verification are not available. rather distributed equally between all tracer
equations. We now rather derive a solution x,
It should be kept in mind, however, that the
transient data for the last 20-30 years do not d•ing that the •ts of equations
validate features of the solution that are referring to DIC and •=C be well satisfied i.e.
we u•wei•t them markedly. To the extent small
importantfor studyingthe •02 uptakeduring the errors remain, these will be added to the right
last centuries. We therefore implicitly still
will be relying on the adequacy of the hand side of equation (4) when integrating with
quasi-steady tracer distributions to derive the time making the implicit assumption that they
solution x, when deducing the response of the are due to processes that have not been
accounted for by the model and that these remain
systemfor the time scales r•evant for unchanged, i.e.
redistribution of •02 and •C in the Atlantic
Ocean during the last 100-200 years.
The time dependent form of equation (1) can be dqn/dt
= _ • qn+ bn +•1 (6)
written (Bolin et al. 1983)
whereBn•is the matrixfor the solutionx1 and
• is theXerrorfield in the set of equations
(5)
dq/dt = - Ax + b (2) for the tracer n. This procedure is of course
acceptable only if the solution x1 is reasonably
similar to the one previously derived with equal
weighting for all tracers.
It is useful to rather transform this equation The transient solution for tritium can be
into the equivalent form
obtained simply, s• the water continuity
equation is exactly satisfied and accordingly no
(3)
dq/dt =-Bq + b artificial sources and sinks are introduced.
It should be remarked that the finite
difference formulation of the advective terms as
where B is the matrix of coefficients that
used may imply a potential numerical instability
describes the way q will change and is defined when employing the equation (4) for tz-mmsient
by x. Equation (3) is a block diagonal matrix of cc•putations. The presence of turbulence, on the
the sets of tracer equations for the n tracers other hand, implies a stabilizing effect.
Whether such an instm_bility exists or not can
dqn/dt = -Bnq
n + bn (4) easily be determined from the signs of the
eigenvalues to the equation (4) and has only
appeared in a few experiments and not in any of
whereqn is the 84-component
vector of those described below.
concentrations of tracer n (tritium, Tr, carbon, The fact that the distributions of DIC and
C, or radiocarbon, *); Bnis the (84 x 84) matrix 14Chavebeenchang• duringthe last few
which represents the subset of equations for hundred years implies that we do not have
tracer n andbn correspon•ngly
the flux fromor accurate observations of steady state
to the exterior and, for •=C, radioactive decay. distributions of these tracers as required for
Note that Bn will not vary with time, while the the derivation of the vector x as outlined
vector bc for the DIC equationsis dependenton above. We assess, howler, that the changeshave
the changesof the partial pressureof •02, p, been rather small ( •=C data are available from
Geophysical Monograph Series Understanding Climate Change Vol. 52
BOLIN ET ALw 63
the time before major injections occttrred i.e. we use the observed concentration at
because of bomb-testing, see below) and we the time of G•0S•2S minus a correction that
therefore adopt the following iterative is •m] to the estimated • between
procedure to derive a proper steady state
soluti•9, x, andthe transient responseof DIC
and •=C due to the perturbations •ced by Thevalue Pc (to)' correspor•ting
to a
man.
state of no bet flux of 00•. between the
We introduce the notations atmosphere and the sea, is-deduced for the
t = tn = 1760, preindustrial time, before field q% (to) thusderived.
whichVanthropogenic influences are assumed to
have been negliable b) Similarlythe change
of A 14Cin the
sea during the period t o to t 1 is
t = t. = 1957, bomb testing had. Qot yet obtained and an estimated pre-industrial
significantly influ• the itc value
distribution
q*e
foimula
(to) is derived
usingthe
t = t•. = 1973, time for the Gq•OSE
obserPations
oftritium
DIC
and •4Cin
the Atlantic Ocean
qe(to) = qo(tl) - (qc(tl) - qo(tl) ) (9)
t = t• = 1983, the time to which transient
computations
wereextended
Wefurtherusethe notations
qno,qnc andqne 4. Thetracerdistributions
qCe (to) andq*e
(t_) together with the other sets of ti•cer
for observed, computed and estimated distributions can be used to derive a second
concentrations. The following iterative
steady state solution x''
procedure is then applied: The procedure described above can of course
1. A preliminary steady state solution x' is be repeated but the two solutions x' and x'' as
d .cedbyusingthe datasets qC
q o(t])
o (t2) and
as wasdoneby Bolin et aI. (1987)
well
that as Pe(to)
further and Po(•o
iterations o)'
notarethat
seem similar
necessary.
although they do not represent a true steady The solutions x' or x'' are obtained by
state. It should be noted that the atmospheric assuming that a steady state prevails. Bolin et
al. (1987) ascertained that no net flux of any
00•.associatedconcentration
Pc (t2) is computed tracer occurs to or from the domain by assigning
in-such a manner that no net aLr-sea exchar•e of
proper boundary conditions for exchange with the
002 betweenthe atmosphere and sea takes place. Pacific and Indian Oceans. In course of a
Pc (t2) differs little from Po (t2) ' transient computation this balance will be
2. Use the solution x' to derive transient disturbed because of chang• tracer
concentrations in those reservoirs from which
solutions during the time period t_ to t• there is outflow of water. In order to assess in
assuming
in this first iteration t•at th• the best possible way the role of the Atlantic
prescribed atmospheric carbon dioxide is given
Ocean for uptake from the atmosphere it seems
reasonable to change the tracer concentrations
Pc(t) = Pc (t2)+ (Po (t) -Do (to)) (7) of the inflowing water in such a manner that the
increased tracer outflow is balanced. It should
using data for Po (t) as given by be recognized, however, that the Atlantic Ocean
Siegenthaler and Oeschger (1987). may well be a pathway for uptake of tracers with
excess atmospheric concentrations and that in
b) •114C
for atmospheric
002,is reality a net flow from the Atlantic Ocean to
adjacent oceans may take place. It is of course
prescribed in accordance with •ata on the
Suess effect during the time period not possible to address such a question with a
t o < t < t 1 and as due to release of treatment of only the Atlantic Ocean as in the
bomb-
produced
14Cfor t 1 < t < t 2 present case.
(see further below).
Data on Tracer Distributions and Their Change
3.a) Thecomputed
valuesof qC
c (t)
during the period t o < t < t 2 are used We shall use the data given by Bolin et al.
to derive a most plausible distribution of (1987) for salinity, alkalinity, phosphorus and
DISat the timet = to, qC
e (to) , oxygen and assume that no changes have occurred
using the formula during the time period considered. We describe
C C C C belowb¾•efly the data for atmosphericpCO2,
qe(to)= qo(t2)- (qe(t2) - qo(t2)) (8) DIC, •=C and tritium that are used.
Geophysical Monograph Series Understanding Climate Change Vol. 52
Atmospheric
002 andDissolvedInorganicCarbon atmospheresince the mid• o•.the 18th century
in the Sea - led to a decreaseof the •=•/•ZC isotoperatio
in our environment, i.e. /% •C decreased.
The cbar•es of atmosphericCO
2 havebeen Stuiver and Quay.(1981) concludethat
summarized by Siegenthaler and •er (1987) atmospheric • •C decreasedby about 22%•from
ar• their data are shown in Figure 1. We notice an approximately constant value before the
that an incr•__se by about 10 ppm occurred before middle of last century to 1955, when bomb
significant emissionsof .(392due to fossil fuel testing began, cf Figure 4. Druffel and Linick
ccaubustion began in the midc]le of last century. (1978) and Druffel and Suess (1983) 'have shown
The distribution of DIC in the Atlantic Ocean by the analysis of growthr'•g dated corals from
as determined during the G•0SECS program in 1973 the Florida Straits that Zl •=C for ocean surface
suppleted by TIO data frc• 1981 in the water in this region decreased by about 12%.
northat parts of the sea have been used in during this •me ñod.
the •tations to be described (cf Bolin et Bomb-produced
•C injections
intothe
at. 1987). atm•ere began in 1952 but were hardly
significant before 1955. Due to t• major
testing activities 1958-1962 zl '•C for
Radiocarbon,14C atmospheric CO (lower troposphere) almost
doubled,
but• sincethendecreased
to a value
Broecker et at. (1960) have reported about of about 300 ø/oo abovethe pre-bombvalue in
200 -- ts in the Atlantic Ocean during 1980 (cf Figure 5; Bolin, 1986). A transfer into
1957. These data ar• G•DSECS data frcau the the oceanhas also occurred.as is clear from
deeper parts of the ooean have been used to direct observations.A A •4C concentrationof
db•k•ca most plausible pre-bomb about 100 ø/oo wasreachedaround1970 and a
distribution (cf Broecker and Peng, slight decrease seems to have ocoarred since
1982; Bol• et at., 1987). then. The increase of • 14C until 1973 within
The • •C distribution had, however, already the Atlantic Ocean as observed during the
been disturbe•.by humanaction at that time. The G•0SECS expeditions has been evaluated for the
emissions of •C-free carbon dioxide into the 84 boxes and is given in Table 2.
10 r I I I I I i i i i I
l]l] ,, 1[]
-10
-- _
-20
-3 0 I , I I i ! i I I i I
Fig. 4. C•anges
of atmo•ic /%14C1820-1955
as deduced
fromanalysisof woodsamples.
(stuiver ar• Quay, 1981).
Geophysical Monograph Series Understanding Climate Change Vol. 52
BOLIN ET AL 65
120
100
80
ß ,
ß
; ß . ß
. ß.
ß%.•' o o ' , ß z,c::.
oN....,....,......
ß
20 • ; o o o %o _
øøø.'.,, ,, o ,,o
= *
.., c
•'
½' 8 a
i o oo
o c ooo
o
o o
o oc• ½
ao o
,•o
ß'" '" _.
1960 1965 1970 1975 1980
Fig. 5. Changes
of Z•14Cin atmospheric
002(filled dots),andin dissolved
inorganic
carbon in ocean surface water (circles), 1954-1980 (Nydal and L6vseth, 1983).
TABLE
2. Changes
of 14C(in o/oo) in Ocean
Boxes1957-1973
Basedon DataFrom
Broecker et al. (1960) and GqHDSECS Data According to '0stlund et at. (1976) and
'0stlund et al. (1987).
Region Western Basin
Layer 1 2 4 6 8 10 12
Eastern Basin
1 3 5 7 9 11 12
state is not a correct one. There is in reality fossil fuel_•cm•_•stion.An air-sea exchangerate
a net inflc• of CO
2 into the ocean,while an of 14 motm "yr-• therefore see•_ moye
outflowof 14C•_ to compensate
for the appropriate than about 19 mot m-"yr -• as is the
decrease of Zl • •c in the atmosphere because of outcome of the first inversion. There is of
Region 1 2 4 6 8 10 12
Layer
1 6.8 2.6 2.2
2 6.5 1.8 1.3 1.5
3 17.5 3.8 0.2 0.2 0.6
4 11.0 11.3 1.0 0.1 0.1 0.2 0.7
5 3.7 7.3 0.8 0.0 0.0 0.3
6 3.9 0.6 0.0 0.1 0.1
7 1.5 0.1 0.1 0.1 0.0
8 2.9 0.2 0.1 0.0 0.0
Eastern basin
Region 1 3 5 7 9 11 12
Layer
1
2 7.2 1.6
3 7.5 5.0 0.6
4 11.0 6.9 0.7 0.0 0.7
5 3.7 4.1 0.3 0.0 0.3
6 2.1 0.2 0.0 0.1
7 0.1 0.0 0.0
8 0.0 0.0 0.0
Geophysical Monograph Series Understanding Climate Change Vol. 52
BOLIN ET AL. 67
! ! !
DEPTH
m
1000
2000
3000
/,000
5000
ar• the associat_•4_ transient solution as our 4. The northward flow in the surface and
reference case.
intermediatelayers of the westernbasin outside
The follc•ir• features of the reference the no_r• Americancontinent is about 0.7 10•
solution should be noticed (Figure 6): tons yr • ( 23 Sverdrup) and is obviously much
1. The mean meridional circulation with sinking weaker than the Gulf Stream. In reality some
motSon
in regions
1-3, is about0.701015ton recirculation of Gulf Stream water takes place
yr-• ( 23 Sverdrup) in generalagreem• with to the west of the mid-Atlantic ridge ar• can
the contain view. A weak cell ( 0.10 10'• ton partly explain the discrepancy. Sc•e of the
yr-1) in the oppositedirection is foundbetween tracer transfer by this major coastal current
the equator and 30øNbelow about 2.500 m. (as well as others elsewhere) is therefore
2. TWo meridional cells in the thermocline implicitly aco3unted for by the horizontal
region with u•ling in equatorial regio• and
poleward flow in surface layers are superim• 5. The turbulent components are generally
smaller than the advective ones, but do
described in the previous paragraph. The contribute to the solution. Vertical turbulent
northward flow of •ate water in the transfer plays a significant role only between
Geophysical Monograph Series Understanding Climate Change Vol. 52
BOLIN ET ALA 69
the uppermost boxes and those immediately below as revealed by acomulation of tracer material
thus accounting for exchange between the surface in course of transient invasions from e.g. the
mixed layer ar• the uppermost part of the atmosphere are remarkably similar for different
x solutions. The solution II is, however,
6. Large horizontal turbulent components (along plausibly one that overstimates the rate of
isopycnic surfaces) appear spotwise in our tracer transfer within our ocean domain.
solutions. These are most likely due to the fact
that tracer gradients often are comparatively Invasion of Tritium into the Atlantic Ocean
small along isopycnic surfaces leaving the rate 1955-1973
of turbulent exchange rather indeterminate.
Accordingly they only contribute marginally to Transient computations to determi/• the
the errors. It should be remarked, however, that uptake of bc•b produced tritium during the
this would not neoessarily be so, when computing period 1955-1973 w•re done using the transfer
tracer invasion into the sea that perferentially vectors x'' from exper•ts I and II. The total
takes place along isopycnic surfaces. For this computed uptake tnlrned out to be 25-30% larger
reason the rate of uptake that we will derive
later may be an overestimate. GqDOS•CS cruises in 1973, the difference between
7. The detritus flux s•tling c•_• of the the two experiments being due to somewhat
surfa_c•b•_•esis 4.7 10• g C yr • ( 50 different flc•s to adjacent oceans. Sarmiento
g C m-'yr •) which seems large as comparedwith (1983) found a corresponding discrepancy between
the basic case presented by Bolin et al., (1987) the given inflow acco• to weiss and Roether
and direct obeervations (cf Bx•echer and Peng, (1980) and observed storage of about 20%. The
1982). This may well be due to the comparatively correlations between c•served and compu•
large turbulent components derived (see above) acomulation into the model boxes where uptake
which transfer DIC upwards from the layers of has occurred are 0.91 and 0.88 for the two
high concentrations into the mixed layer, for transient runs I and II. Table 5 compares the
which a oompensat• detritus flux is derived. uptake by region. we note the unrealistically
8. The solution has been obtained without large values in region 1, and also seemingly too
taking use of the ten•pera• data (cf Bolin et large values in regions 6 and 8 (see further
al, 1987). It is, however, possible to compute a comments in section 4.5).
posteriori the heat transfer that the
steady-state solution implies. We find that the The Suess Effect
northwardtransfer (experimentII) from reg$•ns
4 and 5 into regions 2 and 3 is about • 10-• W The emissionsof 14C free carbondioxide due
compared to a range between 0 and 8 10 '= W as to fossil fuel bur•g have changedthe
assessed by Wunsch (1984). Our larger-•alue is distribution of A •C both in the atmosphere and
primarily caused by turbulent transfer due to the sea (the •es•. effect), already before
rather large horizontal turbulent components (cf injection of •C into the atmosphere due to bc•b
6 above). This implies that the •0 uptake that testing. Wehave usedthe data given • Stuiver
wecompute
below
(seesection
4.6)2may
alsobe and Quay (1981) for the changes of zl •=C in the
somewhat too large as the penetration in the atmo•ere since 1800 and assumed a constant
North Atlantic and horizontal transfer southward
is a majorpathwayfor (/)2 invasioninto the
TABLE
(in 5.
TU xU•10
• ton),
ofTritium 1955-1973 byRegion
Atlantic Ocean.
9. There are other details in the solution that Observed According to
deserve further study. It would obviously be of GHDSECSand Computed in Transient Exper•ts I
interest to relax the constraint due to the and II.
Region 1 2 4 6 8 10 12
Layer
1 -8 -5 -7
2 -5 -4 -5 -5
3 -5 -4 -2 -2 -4
4 -9 -4 -1 0 -1 -2 -3
5 -8 -4 -1 0 -1 0 0
6 -6 -2 -1 0 0 0 0
7 0 0 0 0 0 0 0
8 0 0 0 0 0 0 0
1 3 5 7 9 11 12
-5 -5 -7
-4 -4 -5 -7
-5 -3 -1 -1 -3
-9 -3 -2 0 0 -2 -3
-8 -2 0 0 0 0 0
-6 -1 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
value before that time to determine the • 14C reference solution II are given in Table 7. We
changes in the Atlantic Ocean during the period
1760-1955 with the aid of the solution II. A The area average increase of fa '•C for all
slow drift of less than 1 ø/oo occurredbefore surface boxes is computed
to have been 132 ø/oo
fossil fuel emissions began (about 1850). The if using the reference solution II as compared
comput• Suesseffect was acco•ly derived as with an observedchangeof 138 ø/oo. Wenote
the Zl •=C change during the period 1850-1955.
The result is shc•n in Table 6. The changes in thattheair
assl/med
to beseaex
14.6_ch_a•ge
-••ate
Mm-Zyr ofCO
on2ah•Cbeen
based
the surface waters vary between-4 and -9 ø/oo, balance between preindustrial inflow from the
excluding the southernmost region for which the atmosphere and radioactive decay. As has already
results may be questioned because of influence been pointed out this is less than the most
from the approximate treatment of exchange with commonly
acceptedvalue of about18 Mm-2yr
-1.
adjacent seas. The average Suess effect in (Broecker and Peng, 1982). The solution with
surface water deducedfor 1955 is -5.5 ø/oo, enhanced
air sea exchangein pola¾.regions(III)
co•paredwith -22 ø/oo for atmosphericcarbon yields almost the samevalue for •qC uptake. In
dioxide. In region 4 a value of -8 ø/oo is the case of a comparatively slow ocean
deduced which is significantly less than the circulation (I) the surface /%14Cvalues
observedvalue of -12 ø/oo as reported by increasemorerapidly 163 ø/oo, presumably
Druffel and Suess (1983). The differences can because of less effective transfer away from the
hardly be explained by the deficiancies of the ocean surface. The increase of surface
solution II as discussed in the previous concentrations depend more on the rate of
section. The question is raised of how transfer into deeper layers of the sea than on
representative the measur•m•n• may be. the precise value for the rate of air-sea
exchange.
Uptakeof Bcmbp•ced 14C The A 14Cchanges
in individualsurface
boxes show a somewhat more irregular pattern
Since1955large amounts
of 14Chavebeen (Figure 7). The computed changes with time in
injected into the atmosphere and a transfer into surface box 4, layer 1, are compared with
the oceans has occurred. G•OS• observations observed changes as deduced by analysis of
reveal the increase and Table 2 show the corals in Florida Straits and Bermuda, Figure 8
estimated •hanges during the period 1955-1973. (Druffel and Suess, 1983). The agreement between
The computed cha• as determined with the values computed with the reference solution (II)
Geophysical Monograph Series Understanding Climate Change Vol. 52
BOLIN ET AL. 71
Region
Layer 1 2 4 6 8 10 12
1 3 5 7 9 11 12
i
iiI •-200 somewhat smaller concentrations in exper•t
than in I which implies a larger pOD
II
2 difference
between the atmospheric concentration and the
\ iI mean pOOpof the ocean surface boxes.
B•cozdir•ly the cumulative002 flux fr•m the
atmosphere into the Atlantic Ocean was larger in
exper•t II (35.9 P•3Cin 1983) than in I (30.2
PgC in 1983). This is due to the more intense
ocean circulation in II and accordingly more
rapid transfer of 002 into the deeperstrata of
the ocean. We notice systematically higher
concentrations in II in deeper layers than in
exper•t I. As a matter of fact the rate of
air sea exchange is of less importance for the
2 4 6 8 10 12
uptake than is the rate of ocezunoverturning as
Equ. long as the air sea exchange is rapid enough not
Western basin
to let oceansurface pCO
2 lag too muchbehind
A14C the increasir• atmosphericpO02concentrations.
Figure 10 shows the verticaI distribution of
the uptake in a series of profiles for selected
\ -200
re•ions. The computed changes until 1860, 1955
and 1983 are reproduced. A few particular
features of the penetration into the ocean
(Figure 10) are worth noticing, even though
careful improv•t
of the solution by using independent information
may modify some of these (see further next
section).
- Rapid penetration down to the bottc• of the
ocean occurs in region 2, while it does not
reach layer 8 in re•ion 3 because of upwellir•
from the deepest layer of the eastern basin (cf.
also a similar feature of the water flow
obta• by Schlitzer, 1987).
- The vertical distributions in re•ions 4 and 6
Equ. shows clearly the grad•m] invasion of carbon
Eastern basin from the north into intermediate and deep layers
due to the advective flow, while turu•ent
penetration from above is considerably less
(dashed
•ine) changes
of zI•=Cin surface significant.
boxes as computed with the aid of transient
- The penetration of 002 into the intermediate
solution II during the period 1957-1973, for the and deep waters is remarkably similar whenusing
western (a) and eastern (b) basins. the different vectors x that we have derived,
in spite of significant differences between
these, particularly with regard to the turbulent
sufficient similarities to justify an analysis components. Clearly the small scale features of
of the modeluptake of CO).during an extended the solutions are not of prime importance for
period of time using the •olution II. the in--ted_ uptake.
Figure 11 shows some chara•istic features
Uptakeof 002During_
the •eriod 1760-1983 of the transient solution. It should be
emphasized that the integrations start from a
The annual rate of 002 uptak• has been state of no net transfer of O0 between the
c•,•fced using the solutions I and II. In the atmosphere
andtheseai.e. th2e
arealmean
of
light of the analysis in the previous sections 002 partial pressurein the sea is assumed
to be
we shall adopt the latter as one reference case. eqSal to the atmo•eric pO0•. in 1760. Figure 11
Before discussin• the details of these results a shows how a difference be• these develop in
few other features of the transient solutions course of time and accorditaly the uptake
areof •o /ncreases with time.
Figure 9 is a scatter diagram showing the The transient computations imply that we
diff• of the uptake by the different assume
a net emissionof 002 into the atmo•ere
r•servoirs in the experiments I and II. They are due to man, causing an increase of the
Geophysical Monograph Series Understanding Climate Change Vol. 52
BOLIN ET AL• 73
A•C
ø/oo
2OO
100
GEOSEC5
/
/
Fig. 8. Computed
changes
of A 14Cin the surfaoelayer of region4 as deduced
with the
transient solutions I, II and III (solid lines) cc• with measured changes as
determined from analysis of corals in Florida Straits (dashed line; Druffel and Suess,
1983). The average value in 1973 as det•_•l•iz•ed during the GI•SE• project is also
•• (*).
atmosphericpCO• and a responseof the Atlantic of the world oceans would behave in an analogous
Ocean in terms Bf a net uptake. We can perhaps way as described by our model of the Atlantic
better visualize the characteristics of what Ocean. Not knowir• the total emissions we
happens by assuming hypothetically that the rest explore how cases in which we assume that the
TABLE
8. Uptakeof Excess
Atmospheric
14Cby RegionDuringthe Period1955-1973.
Observations are from the Gt•SECS Expeditions. •tations are According to the
Transient Experiments II and III.
40
o
•'• o
30-
• 20-
• x
• x
• x x
• 10-
: /x • •x
•XXX
O]
0x
X•x
• I
10 20 3•
DICchange
Experiment
•g. •. ••ec •• oE •• •••L•on •• oE
average air-borne fraction during the period time in cases A and B. During the first 120
1760-1983 were 0.50 (case A) and 0.45 (case B). years ,• decreases from an initial value of 100%
This implies for this period to 35-40% in 1870 at which time the atmospheric
co•trations began to increase more rapidly,
Total emission (Pg) 278 310 but slowed dc•n after the turn of the century.
Fossil fuel emissions (Pg) 179 179 Because of the accelerating increase s•
Biosphere emissions (Pg) 99 131 1950, ,• increased significantly during the
Ratio of the total uptake to last 35 years. Although we cannot trust the
the Atlantic Ocean 3.9 4.8 precise values as shc•n in figure 11, it seems
likely that this general feature of our model is
We note that the ratio of the total ocean one which might well also characterize the
surface to that of the Atlantic Ocean is 4.0 or world oceans (cf also Oeschger and Siegenthaler,
4.5 dependent on whether region 12 is included 1985). Because of the fact that irregular
or not. Acco•ly the uptake by the Atlantic variations in reality are superi•posed on such
Ocean per unit area is larger than that of the slow trends, we usually cannot discover these
rest of the oceans in case A and vice versa in until a decade or two have gone by.
case B. Although the calculations described above are
The curves a in figure 11 shows how the based on an analysis of the Atlantic Ocean it
airborne fraction ( • ) would have varied with seems plausible that the airborne fraction of
Geophysical Monograph Series Understanding Climate Change Vol. 52
BOLIN ET AL. 75
DIC pMol/kg
10 20 30
DEPTH
km
10 20 30 10 20 30
1
2
10 20 30 10 20 30
N& ION 5
10 20 30 10 20 30
REGION 6 REGION 7
Fig. 10. A•ation of carbon (in the form of dissolved inorganic carbon in the Atlantic
Ocean) as a result of increasing atmospheric 002 concentrations as shownin Figure 1.
Vertical profiles of concentrations are shown f•r selected regions for 1880, 1955 and 1983
as deduced with the aid of the transient solution II.
man'semissionsof 002 into the atm•ere has 032 emissions to the atmospherebefore 1850
increased from 1950 t5 1973 and then decreased, due t5 fossi% fuel burningwere small probably
when the annual increase of 032 emissions by 0.05 PgC yr • or less and insignificant before
4.5% per year due to fossil fu•l combustion in 1830. The total emissions until 1860 can hardly
1973 decreased to only about 1% per year, which have been more than 4 PgC (cf Rotty, 1981). The
situation has prevailed until recently. increaseof the atmospheric
032 content
Geophysical Monograph Series Understanding Climate Change Vol. 52
50
340
330
320
310
300
0.5
0.4
0.3
6
, 0
1800 1850 1900 1950 2000
Fig. 11. Transient exper•t II illustrating CO
2 uptake by the Atlantic Ocean.
a) Airbornefraction of 002 emissions(in %) de•]h•edby consideringthe response
characteristics of the Atlahtic Ocean being rep•tative of the world oceans as a whole
and assuming that the mean airborne fraction 1760-1983 was 50% (case A; dashed line), 45%
(case B, solid-line (cf text).
b) Atmospheric
002•partial pressureas determined
by Oes•er andSiegenthaler(1987), cf
Figure 1.
c) Atlantic • uptake(PgCyr-1).
d) 002 emissionsandtheir partitioning betw•_nemissionsdue_•ofossil fuel combustion
and deforestation and char•ing land use (bioemissions), PgC yr • as computed for the two
cases A (dashed line), B (solid line).
1760-1880 was about 20 PgC, and the uptake by and expandingagriculture mayaccoxd•! • on the
the Atlantic Ocean about 6 PgC according to our average have been about 0.3-0.4 PgCyr -x during
computations. We assess that the total oceans this period.
will have taken up 20-25 I•. Thus, if the Figure 11d finally shows the emissions that
equilibrium between the atmosphere and the sea we deduce corresponding to a total airborne
was not markedly disturbed already before 1760, fraction of 45% and 50% (solid and dashed lines
the total emission to the atmosphere during the respectively). The values until 1950 seem
period 1760-1880 were above 35 PgC, of which plausible yielding a total emission due to
accordingly only about 10% was due to fossil deforestation and changing land use
fuel c•bustion. Emissions due to deforestation (bioemissions) of 115 and 97 PgC respectively,
Geophysical Monograph Series Understanding Climate Change Vol. 52
BOLIN ET AL• 77
deduced as the difference between the computed coefficients in the transfer terms could be
total emissions and those due to fossil fuel given values within ranges as determined by
combustion (Rotty, 1981). During the last 25 the errors in the data field. Some of the
years, however, the bioemissions become quite tracer continuity equations in A x = b could
small. This result leads to a suspicion that rather be given as inequality constraints in
other sinks than the oceans may have been the set G x > h. The Simplex method of
present to absorb the larger bioemissions that solution offers another suc•,possibility.
seems plausible. 5. The steady state solution x has • used
Let us finally return to the discussion in to compute the invasion of tritium, •C and
the introduction as summarized in Table 1. We carbon from the atmosphere.
have found that about 35 PgC may have been 6. In spite of the •inty of the basic
emitted into the atmosphere before 1880 due to oft•fer solution x the integra• uptake
deforestation and changning land use. In the C during the years 1955-1973 is quite
light of the assessments by Bolin (1986) 100 PgC stable and agrees reasonably well with
seems minimum of these emissions for the period observations during GEOSECS.
1860-1980. Thus the total emissions from 7. Themoredetailedpatternof 14C-penetration
deforestation and charging land use for the as computed show some differences co•pared
total period is likely to have been well above with observations. A further analysis of
100 PgC, most likely 135 PgC. Total emissions, these descrepancies might yield a more
i.e. including those due to fossil fuel reliable solution. Generally great care must
emissions therefore probably have been larger be exercised in the use of available data
than 300 PgC, i.e. Case B is applicable rather and their uncertainties in order to improve
than A. On the other hand it does not seem sucessively the solutions of the basic
likely that the uptake per unit area by the matrix equation.
Atlantic Ocean is less than that of the rest of
8. The Suess effect in the Atlantic•
1955 being due to the emission of C-free
in
the ocean. Other sinks for ODp.maybe of
importance than those considered here. carbon dioxide by fossil fuel combustion in
the past is obtained with the aid of a
Su•nary of Results and Conclusions transient computation. The computed values,
the averageof which is 5.5 o/oo, are
1. Inverse methods (Bolin et al, 1987) have considerably less than the observed being of
been used to deduce ocean circulations, a the order of 10 o/oo. The discrepancycan
spatial pattern of turbulent exchange and hardly be due to inadequacies of the
biological activity that best satisfy the transient solution and the result warrants
steady state destribution of a set of ocean further analysis.
9. The total uptake of carbon dioxide by the
2. An analysis of a series of such solutions Atlantic Ocean during 1760-1983 is computed
and comparison with some independant data to have been 33+ 5 Pg C, of which about 6
show fair agreement for the large-scale I•g occurred before 1860.
pattern (cf Bolin et al, 1987) but also 10. Considerable temporal clmarges of the uptake
significant uncertainties on the smallest in terms of a percentage of the atmos[i%eric
scale of resolution particularly with regard increase (equivalent to a varying air-borne
to the pattern of turbulent flux. fraction) are found. It seems plausible that
3. A prominent feature of the most plausible the air-borne fraction of the emissions was
solution is clumracterized by a direct mean considerably grea• in the 1950' and 1960 ' s
meridional circulation cell w_•than than during the earlier part of this century
intensity of 0.7 10 • ton yr • and it may have decreased since 1973.
( 23 Sverdrup) with sinking motion in the
north and two coupled circulation cells in References
the termocline region with 0.7
1015ton yr-1 ( 23 Sver•), upwelling Bolin, B., 1986. Howmuch032 will remain in the
in the equatorial regions. Turbulent and atmosphere? In Bolin B., D66s, B. Warrick, R.
detrital transfer seem too high and implies and JRger, J. (Eds) The greenhouse effect,
probably somewhat too rapid tracer transfer climatic chanqe and ecosystems. S03PE Report
in transient computations. 29, pp 93-155. J. Wiley, C•ichester, England.
4. The errors in satisfying the set of tracer Bolin, B., Bjdrkstrdm, A., Ho•, K. and Moore,
equations are due to the incompatibility B., 1987. On inverse methods for combinirg
because of crude spatial resolution, chemi• and physical ooeanogra•ic data: A
approximate description of ch•mti• and steady-state analysis of the Atlantic Ocean.
biological processes and above all Report C•-71, Dept. of Meteorology, Univ. of
inaccurate data fields. It would be Stockholm, 220 p.
desirable to obtain solutions in which both Broecker, W.S., Gerard, R., Ewing, M. and
the b-vector (primarily determined by the Heezen, B.C., 1960. Natural radiocartx• in the
prescribed boundary conditions) and the Atlantic Ocean. J. Ge•ys. Res., 6__5,2903-2931.
Geophysical Monograph Series Understanding Climate Change Vol. 52
Broecker, W.S. and Peng, T.H., 1982. Tracers in Rotty, R.M. and Masters, C.D. 1985. Carbon
the sea. Lamont-Doherty Geological dioxide from fossil fuel combustion: Trends,
Observatory, Columbia Univ. N.Y. resources, and technological implications. In:
Druffel, E. and Linick, T., 1978. Radian in Trabalka, J.R. (Ed.) Atmospheric carbon
annual coral rings of Florida. Geophys. Res. dioxide and the global carbon cycle. U.S.
Lett. 5, 913-916. Department of Energy, DOE/ER-0239, 63-80.
Druffel, E. and Suess, H., 1983. On the Sarmiento, J.L., 1983. A simulation of bomb
tritium entry into the Atlantic Ocean. J_•.
radiocarbon
record
inbanded
cry¬s:
Exchange Physical Oceanoqr. 13, 1924-1939.
atmosphere
andsurfaceocean.J. Ge•phys. Schlitzer, R. 1987. Renewal rates of east
•s. 88 C2, 1271-1280. lantic deepwater estimatedby inversion of
Maier-Reimer, E. and Hasselman, K., 1987. C data. J. Geophys. Res. 92 C3, 2953-2969.
Transport and storage of CO•.in the ocean - an Schlitzer, R., 1988. Modeling the Nutrient and
inorganic ocean-circulation-carbon cycle Carbon Cycles of the North Atlantic. Part I:
model. Climate Dynamics 2, 63-90. Circulation, Mixing Coefficients, and Heat
Fluxes. J. Geophys. Res. To be published.
in the atmosphere (1962-1980). J. Geophys. Siegenthaler, U. 1983. Uptakeof excessCO
2 by
Res., 8--8, 3621-3642. an outcrop-diffussion model of the ocean. J.
'0stlund, H.G., Dorsey, H.G. and Brescher, R. Geophys. Res. 8--8, C6, 3599-3608.
1976. G•OSECS Atlantic radiocarbon and tritium Siegenthaler, U. and Oesc•er, H. 1987.
results (Miami). Tritium Laboratory Data Biospheric CO•.emissions during the past 200
Report 5, Rosenstiel School of Marine and years reconstSructed by deconvolution of ice
Atmospheric Science, Univ. of M•ami. core data. Tellus 39B, 140-154.
'0stlund, H.G., Craig, H., Broecker, W.S. and Striver, M. and Quay, P.D., 1981. Atmospheric
Spencer, D. 1987. G•OSE• Atlantic, Pacific •=C changesresulting from fossil fuel CO2
and Indian Ocean expeditions. Volume 7: release and cosmic ray flux variability. Earth
Shorebased data and graphics. Superintendent and Planetary Science I•tters 5--3, 349-362.
of Documents, U.S. Government Printing Office, Weiss, W. and Roether, W. 1980. The rates of
Washington D.C. 20402. tritium input to the world oceans. Earth and
Rotty, R.M. 1981. Data for global CO
2 production Planetary Science Letters 4--9, 435-446.
from fossil fuels and cement. In: Bolin, B. Wunsch, C. 1984. An eclectic model of the
(Ed.) Carbon cycle modellinq. J. Wiley and Atlantic Circulation - Part I. The meridional
Sons, Chichester, 121-125. flux of heat. J. Phys. Oceanogr. 1--4,1712-1733.
Geophysical Monograph Series Understanding Climate Change Vol. 52
AFRICAN DROUGHT:
CHARACTERISTICS, CAUSAL THEORIES AND GLOBAL TELECONNECTIONS
Sharon E. Nicholson
Department
of Meteorology,
FloridaStateUniversity,
Tallahassee,
FL 32306
79
Geophysical Monograph Series Understanding Climate Change Vol. 52
,:.:,•?-.---:::":
ß .. :......,,
::...:•-..
,.:-
-'•.--' ---........:-:::.::.:;......:....
.... .
. ..
: .
....... -..--.*.;.
•.
-.** '-:.'.:
-.;,:
4.•
0 * '•'-•,;.,....;:!•
..-,,• ..'
...........
:'--",•.:
:*-,- .•\. - ':.-•..
.-
- 40 • ";'
.... '""--:"•
':"..
,"":
.....
:-'"'"'"'-%'•....
ß -20 ;:..-'-
• • . '""-?.
...:
'*•,%, ....
:->' -:.;
":::.;.......?:"
....<
..'.
,..
,,,
-.....',::.,::'•;!:'•i;•;:•,:*
.........
L..•;•-:•. '"*":>•:':":•"*
%%, -40 .-'. ,..
ß-:.:',-(-'::..' ':
.
•:.:. • • *'. , .: . .:
;...
"b / ""':;'*
'
., •.;
-- :!•.... ::..
?-:;..::!}? ,.... <"' :
, ,,. •7,:.•"
...
•½;::
*' %:':' - 20 - 40 , '½:*;•::'
'"<,-:.:.--::."
Fig. 1. Schematic of rainfall departures (% above or below normal) for 1983 superim-
posed on 14eteosat view.
apparently encompassed the equatorial regions and lihood that is no longer possible. Rift Valley
the subtropics of both hemispheres. h second humid lakes were several hundred feet deeper than at
period prevailed from about 6,000 to 4,000 B.P. present and expanded to form a huge hydrologic
The desert had all but vanished; lowland marshes system. Lake Chad, almost entirely desiccated in
forned in the northwestern Sahara while Neolithic recent years, was then over a hundred meters deep
nan herded cattle in the central Sahara, surrounded and expanded to ten times its normal twentieth
by fauna grazing on what was apparently a savanna century size.
landscape. Lakes dotted now arid regions of •auri- The onset of more arid conditions occurred about
tania; fish hooks uncovered in archaeological 3,000 years ago and since that time conditions
sites there attest to human occupation and a live- generally resembling those of the current century
Geophysical Monograph Series Understanding Climate Change Vol. 52
NICHOLSON 81
TABLE 1. Mean Annual Rainfall (mm) cators, including rainfall, river flow, lakes,
agriculture and harvests.
Fig. 2. African paleoclimates c. 18,000 B.P., 9,000 B.P. and 6,000 B.P. compared to
present conditions (rainfall' shading or plus signs, generally greater than at
present; dots or minus signs, less than at present; hatching, decreasing rainfall;
lakes: ß : high stands, O: intermediate stands, A : low stands; based on Street-
Pertoft et al., Nicholson and Flohn, Cockcroft et al., and others).
ß
the persistence can be used as a point of evi- frequencies. Quasi-periodic fluctuations on these
dence in support of the land-surface/atmosphere time scales are characteristic of rainfall through-
feedback hypothesis previously mentioned and de- out equatorial and southern Africa (see also Tyson
scribed in more detail in Section 6. et al., 1975; Rodhe and Virje, 1976; Dyer and
The temporal characteristics suggested by the Marker, 1978; Tyson, 1980), but are notably absent
time series in Fig. 4 are quantified using spectral elsewhere on the continent (Fig. 6). They also
analysis (Fig. 5). The spectra for East Africa and characterize numerous other atmospheric and oceanic
the Kalahari regions each show three significant phenomena in the tropics. All three are evident in
peaks in the range 2.2-2.4, 3.3-3.8 and 5-6 years the Southern Oscillation; the 5-6 year time scale
(Nicholson and Entekhabi, 1986). In contrast, that is the dominant one for SST variability in the
for the Sahel shows "red noise", which is indic- Atlantic and Indian Oceans (Fig. ?); and the Quasi-
ative of persistence, and little power at higher Biennial Oscillation is a well-known phenomenon.
Geophysical Monograph Series Understanding Climate Change Vol. 52
0 -+'t- i
+'1-
-I- .-I-
+o+O - _ .--
-I-
.... __
-I-
....
.._
-t-
o•
...
I
I0' O'
I0' 20' i•' 40'
i SO' I0' O' I0' 40' SO'
c. 1895-1920
I- -I- --
-I- --
Fig. 3. African rainfall anomalies for three historical periods (minus signs denote
evidence of drier conditions; plus signs denote evidence of above-average rainfall;
zeroes, near normal conditions; circled symbols denote regional integrators, such as
lakes or rivers).
Geophysical Monograph Series Understanding Climate Change Vol. 52
SAHEL
Lamb attributed the tower Saber rainfall to
southward displacement of the wind convergence zone
and equatorial trough, thermally forced by the
anomalous SST pattern, but other atmospheric anoma-
lies over the At]antic are also evident (Hasten-
rath, 1984a). Lough (1986), examining approxi-
-5O mately the same area and time period, found little
EAST AFRICA evidence of the consistent relationship between
SSTs and Sahel rainfall which Lamb suggested. She
11'Jl,
im,i also found that SSTs appear to lag Sahel rainfall,
a result which implies that SST patterns do not
- [lLr UJyL' ,-LJ uLr force the rainfall
Foiland et at. (1986a,b)
fluctuations.
found an Atlantic SST
-5O anomaly pattern for Sahel dry years which is mar-
50 KALAHARI NORTH kedly similar to that shown by Lamb; this is sig-
nificant, since only one of the five years in
I Rnr,.['•[]•Rn
I"!•
nnn•,•,..nn•l..•rl
I] Lamb's dry composite was also in their
posite. Foiland et al.
dry com-
also showed that the Atlan-
'il
global SSTs (explaining 4.7% of the variance) to
represent interhemispheric SST differences (Fig.
•0 8). Based on a correlation of .6 between EOF 3 and
I I I i i I I I
Sahe] rainfall (significant at the 99% confidence
1900 1910 t920 1930 19•0 t950 1960 1970 1980
level), an attempt was made to predict rainfall
from global SST anomalies. However, prediction may
Fig. 4. Rainfall fluctuations in four sectors of
be premature since SST changes often lag rainfall
Africa 1901-1984, expressed as a percent departure
(Fig. 8) and there is little correspondence between
from the long-term mean (ENSO years are shaded for
the two series during several periods of signifi-
Sahel and East Africa, ENSO +1 is shaded for the
cantly abnormal rainfall. As an example, the coef-
two Katahari series).
ficients of EOF 3 are similar for the 1910s drought
and the wet 1950s. This situation is well illus-
trated by a scatter diagram of the EOF coefficients
Tropical Teteconnections to versus rainfall (Fig. 9)' despite a general posi-
African Rainfall tive correlation, much scatter is evident except
for the most extreme years. Moreover, the reta-
The correspondence between the dominant time tionship appears to be break down prior to the mid-
scales of rainfall variability and those of other ]940s (Lough, 1986; Foiland et al., 1986a),
tropical parameters suggests possible forcing leading to speculation that it may be the steady
mechanisms of rainfall variability. Here, several downward trend in both Sahel rainfall and inter-
of these are examined, beginning with sea-surface hemispheric SST differences which is principally
temperatures. Most studies relating rainfall to responsible for the statistical correlations.
SSTs have been concerned only with the Sahet and a The mpst consistent result of all of these stu-
number of viewpoints have emerged. Only a handful dies ]s a pattern of anomalously high SSTs in the
of studies have dealt with other locations. In Atlantic south of c. 10'N and tower than normal
many cases, weak or strong statistical associations temperatures to the north during years of Sahel
have been documented but the dynamic mechanisms drought. The authors agree that such conditions
accounting for these associations have not been appear to favor decreased Saber rainfall, but are
established. unlikely to be uniquely the cause of drought.
One of the first studies to examine the rela- Recent papers such as Foiland et al. (1986a,b),
tionship between Saber rainfall and sea-surface Semazzi et at. (1988), Hsiung and Newell (1983) and
temperatures was that of Lamb (1978a,b). Using a Nicholson and Nyenzi (manuscript submitted to
Sahel rainfall index heavily weighted toward the Journal of Climate, 1989) show that the Atlantic
western Saber (west of 0øW), he concluded that anomalies are part of a global pattern. That pat-
Sahelian wet years are associated with abnormally tern in turn, or the atmospheric anomalies produc-
high temperatures throughout most of the tropical ing or responding to it, may be the critical factor
Atlantic, but below average SSTs to the north and in Saber rainfall variability. Numerical simula-
west of the convergence zone over the Atlantic. tions (Palmer, 1986' Semazzi, personal communica-
Geophysical Monograph Series Understanding Climate Change Vol. 52
NICHOLSON 85
SPECTRUM SPECTRUM
Sahel East Africa
.03
: .
..
. .
99% cg--
. .
. .
. .
. .
.02 ..•
-::::
95% CL.__
ii•i 2.6-3.8yrs ii
.01
.05 .10 .15 .20 .•5 ß05 .10 .15 .L•O .Z5
FREQUENCY FREQUEI•I•'
SPECTRUM
SPECTRUM
Kalahari South
Kalahari North
.04
:e•2.3 yrs
3.8 yrs
3.6 yrs
:
•: :
.O3
• ::
•! •i 99% CL
• 2.3yrs•i 99%
CL--
4 8 yrs • •
::
.
•-!!- - • - •i • c• _ _
• :• :. ::
.02 :
:
:::
:::
:::
:::
:
:.::
:: - yrs •:
i •i •:: ......
:::: ..... -- : .
:::
..
:. ::::::
......................
:: ::::::
::::::
:::::::::::::.:
:::::: : .....................
.........
........................ ...
.0!
...................................
•••z••z•z•z• .... •zz•z
...................................
..................................
:. ::::::::::::
........................................
........................................... ......................................
........................................... ........................................
...........................................
............................................
.............................................
.............................................
.............................................. :::::::::::::::::::::::::::
............................................... .............................................
tion) likewise suggest that global, Pacific Ocean, cially the relationship with the Indian Ocean,
and Indian OceanSSTs all have a greater influence since the systemswhich "fail" in the western Sahel
on the Sahel than Atlantic SSTs do. This finding in dry years develop from pressure perturbations in
is more compatible with Sahelian climatology, espe- the eastern Sahel and traverse most of the east-
Geophysical Monograph Series Understanding Climate Change Vol. 52
30
20
lO
2.2
-2.4
-2O
-3O
40
30.
20.
-lO.
5.0- 6.3yrs
-20.
-30.
west extent of the continent. These are more wind stress in the equatorial Atlantic. This
likely influenced by factors "upstream" in the implies remote forcing of BBT anomalies via the
Indian Ocean than several thousand kilometers down- Kelvin wave mechanism described by Adamec and
stream in the tropical Atlantic. O'Brien (1978). Surprisingly, they found that
A handful of studies have examined the relation- SST, pressure and wind anomalies accompanying wet
ship between sea-surface temperatures and rainfall conditions during the second rainy season of
variability elsewhere in Africa. Hirst and Basten- October and November were dissimilar to those of
rath (1983a, b) showed that increased rainfall wet Match/April seasons and in some ways (e.g.,
during the March/April rainy season along the high pressure and strong southeast trades) resem-
Angolan coast and in the western equatorial region bled the March/April dry composites.
(Zaire basin) is associated with a weakening of A strong relationship between SSTs and rainfall
the high pressure over the Atlantic and reduced along the Benguela coast, which includes the
strength of the southeasterly trades and westward Angolan sector studies by Hirst and Hastenrath, was
Geophysical Monograph Series Understanding Climate Change Vol. 52
N I CHOLSON
PERIOD (YEARS)
5010754 :•?..5 ?. 1.5 I 0.5
PERIOD (YEARS)
r I I I I i I I I I
I0 5 3, 2.5 1.5 .
.17. .15
ATLANTIC ......
: . INDIAN
SOI
.12
- 4 ;; -
,I t f I
- •0 , -
-•1•.• -
.O6
.:1 • •. ,,• _
.O3
'
;•1 •" •, , ! • -
25 IO 5 3 2 I
I I I '1 I I
.OO
.OO .05 . I0 .15 .20 .25
FREQUENCY (CYCLES/SEASON)
10%- SSTz
Fig. 7b. Spectra of the first principal component
of sea-surface temperatures in the tropical Atlan-
tic and tropical Indian Oceans (from Nicholson and
Nyenzi, 1988, manuscript submitted to the Journal
of Climate).
•.n•u.•
oo•st
o•
t•.
sout•..t.•
•t•.t•
(•0's
20'S,1948-72)
- ø i,,:,IV
andtheTahiti-Darwin
of the Southern Oscillation
pressure
index
(1935-73) ....
/ ,
demonstrated
by
Nicholson
and
Entekhabi
(1987).
and rainfa]]
-,oo
i i
Theyfurther showed
that the varianceo[ both SSTs
years (Pigs. 6 and 7).
is
As described in section
concentrated
3,
on time sca]es of 4-6
' -,
the variance of rainfal] throughout the equatoria]
zone and in much o[ southern Africa is similarly
dominated by 4-6 year quasi-periodicities and is Pig. 8. Time series of the third principal coN, on-
strongly coherent with SST fluctuations on this ent of g]obal sea-surface temperatures and Sahel
time scale (Pig. 10). Cadet (1987, personal com- rain[a]], 1901-1984 (from Polland et al., 1986a).
Geophysical Monograph Series Understanding Climate Change Vol. 52
I00
I i I ! I I
(Nicholson, 1980b). Likewise, SSTs may be more
important as high frequency forcing of the inter-
annual variability than as a mechanism producing
the ionget-term rainfall trends in the Sahe].
Evidence for a global teleconnection derives
5O directly from studies relating rainfall to tropical
ß
phenomena such as the Southern Oscillation/E1 Nino
ß ß (ENSO) and the 30-60 day oscillation. Only a few
ß
ß ß studies, however, have evaluated African rainfall
ß ß
ß
in a global text. The two most comprehensive are
ß
ß ,
ßß ß
ß
those of Nicholson and Entekhabi (1986) and Rope-
EOF3 ß ß ß
lewsk] and Halpert (1987). These studies used
basically the same data set but different analysis
ß ß ß
ß ß ßß ß
-5O .
ß
ß
ß
ß ß ß
.
techniques, the former dealing primarily in the
frequency domain and the latter solely in the time
ß ß
ß ß
ß
ß ß
.76
N I CHOLSON 89
40.
30.
20.
10.
O.
-10.
-20.
-30.
-40.
40.
30.
20.
lO.
O.
-10.
-20.
-30.
-40.
'11950
1933 1931 1971
1953 1955 1947 1972
1968
15ø 1956 1957 1973
,oø
5ø ,
AM J J AS 0 A MJ J A S 0 AMJ J AS 0 AMJ JA SO A MJ J ASO'
Fig. 11. Location of the ITCZ over West Africa during wet years and dry years, as
approximated by the "center of gravity" of the rain belt for the months of April to
October (three-year averages plus long-term mean; circled areas are rainfall exceeding
200 mm/month; shading marks latitude of maximumrainfall in August; from Nicholson,
•9S•b).
at 5-6 years in equatorial regions, at ~3.5 years droughts. Modern satellite-derived data sets,
in southern Africa, and at ~2.3 years throughout which are hard to compare with conventionally
the eastern half of the subcontinent. In no case observed data, are available only for recent years;
is the relationship with ENSOas strong as with for the Sahel, this includes only drought years and
Atlantic SSTs along the Benguela coast. This sug- no true "normal" or wet year for comparison. The
gests that the direct link with rainfall in equa- result is that moststudies have either considered
torial and southern Africa is probably sea-surface a very limited numberof years or were severely
temperaturesor atmosphericparametersin the restricted in the upperair parameterswhich could
marine sector, which in turn appearto be modulated be assessed. The results of these studies are
by the ENSO events. briefly summarized.Threeareas, the Sahel, East
A numberof other global studies illustrate the Africa and southernAfrica, are considered separ-
influence of majortropical phenomenon on the Afri- ately becausethe analyseshavetendedto do so and
can sector. The ENSO wind compositesof Arkin because different factors appear to operate in
(1982) and Selkirk (1984) illustrate consistent these locations.
sequencesof anomaliesover certain parts of Anearly hypothesisinterpretedchanges
of Sahel
Africa. Arkin, for example,shows a progression rainfall mainly as a function of the position of
from easterly to westerly 200-mbwind anomalies the Intertropical Convergence Zone(ITCZ) and sub-
over southern Africa during the course of ENSO tropical high (Kraus, 1977a, b; Greenhut, 1977;
years, and a progression from westerly to southerly Beer et al., 1977). The only actual evidence to
anomalies over northern Africa. Selkirk's results support the hypothesis came from Laml) (1978a,b),
likewise emphasize the change of the upper-level whose analysis was limited to conditions over the
zonal flow over Africa in relationship to high and Atlantic Ocean. Later, Newell and Kidson (1984)
low index periods. The 30-60 day oscillation stu- and Nicholson (1981b) showedthat over the African
dies of Madden (1987), Weickmannet al. (1985), continent itself there was no systematic southward
Knutson et al. 1986, Knutson and Weickmann (1987) displacement of the ITCZ. In many dry years, the
show variations in outgoing longwave radiation convergence zone is at least as far south as during
(OLR) and wind over the African sector that are droughts (Fig. 11) and it remains near its mean
coherent components of the global 30-60 day signal. position over the continent during most dry years.
Others (e.g., Namias, 1974; Miles and Foiland,
Atmosphere Circulation Changes 1974; Tanaka et al., 1975; Schupelius, 1976; and
Associated with Rainfall Fluctuations Nicholson, 1979b, 1980b) produced additional evi-
dence refuting the hypothesis.
Although numerousstatistical analyses and model A number of more plausible explanations for
simulations have linked African rainfall to sea- drought in the Sahel have since been proposed.
surface temperature fluctuations and tropical Various investigations have shown that, compared
atmospheric phenomena, few studies have dealt dir- with wet years, drought years in the sub-Saharan
ectly with drought mechanismsin terms of local region are characterized by a weaker 200 mb trop-
atmospheric factors (e.g., circulation changes, ical easterly jet (TEJ), a stronger mid-tropo-
atmospheric moisture, stability). This lack may be spheric (700 mb) African easterly jet (AEJ), weaker
due to the dearth of upper air data over Africa. shear in the 700 mb easterly jet south of the
Sufficient radiosonde and pibal data exist for only equator, enhanced Endicy-type overturnings (and
a few years, most prior to the recent severe weaker Walker-type overturnings), increased geopo-
Geophysical Monograph Series Understanding Climate Change Vol. 52
NICHOLSON
tential of the 700 mb surface, increased vertical 1984), but a dynamical explanation for this associ-
shear over West Africa (thus, enhanced horizontal etlon is not readily apparent. The interaction
temperature gradients), and the virtual disappear- between tropical and extra-tropical systems (Kumar,
ance of the 850 mb trough over West Africa (Kidson, 1978; Riehl, 1979) is likewise a significant
1977; Tanaka et el., 1975; Kanamitsu and Krishna- determinant of the character of the rainy season.
mufti, 1978; Newell and Kidson, 1979, 1984; and As an example, the coincidence of an eastward-
Dennett et al., 1985). Surprisingly, there appears moving westerly (extratropical) trough with a
to be no systematic difference in the depth of the westward-propagating trough in the tropical easter-
monsoon layer, its humidity or advective moisture lies can rapidly transform a stable tropical air
flux over West Africa between wet and dry years in mass into an intense disturbance. Knowledge of the
the Sahel (Lamb, 1983). In many cases, linkages relationship between such systems and interannual
with other tropical weather anomalies are evident variability of rainfall is woefully inadequate. A
(Kanamitsu and Krishnamufti, 1978; Krueger and better picture, however, is starting to emerge.
Winston, 1975; Kraus, 1977a; Fleer, 1981; and Several recent papers have shown the importance of
Hastenrath and Kaczm•rczyk, 1981). diagonal cloud bands (also a system manifested
Even fewer systematic studies of wet and dry by tropical/extra-tropical interactions) in deter-
years have been carried out for other parts of mining the seasonal character of rainfall (Her-
Africa. In East Africa, local forecasters gener- rison, 1983, 1984; Harangozo and Harrison, 1983).
ally associate westerly wind anomalies in the mid- These systems influence most of the southern sub-
or lower troposphere with abnormally high rainfall continent and are clearly linked to large-scale
but that is not always the case (Kiangi and Temu, patterns of tropical flow, as evidenced by their
1984). The two monsoon flows prevailing during relationship to the ENSOphenomenon.
most of the year are generally divergent and a
large-scale diffluence of the NE monsoonover East Land Surface Processes
Africa is readily observable (Flohn, 1964; Anyamba,
1984). Thesuggested causesof rainfall variabil- In the early 1970sit becameapparentthat human
ity are quite varied, as the factors controlling activities weredramaticallyaltering the natural
rainfall are complex anddiversein various parts global landscape.Semi-arid]ands,the predominant
of the region. Moreover, most investigations of African environment, wereshown in manycases to
rainfall variability havebeenconfined to case be undergoing a processof "desertification", the
studies of one or twoyears(e.g., Minja, 1985; result of overcultivation,deforestation,overgra-
Anya.•a, 1983, 1984; Agu•a, 1984; Anya.•a and zing, improper irrigation, slash/burn agriculture
Ogallo, 1986; Ogallo and Anyamba,1986; and Flohn, and other forms of land mismanagement(United
1988). Therefore, few generalizations can be made. Nations, 1977). The desertified landscape, rela-
Factors which have been associated with abnormally tively barren of vegetation with bare soil exposed
wet conditions, such as those of 1961/62 or and subsequently eroded, was illustrated by several
1977/78, include strong low-level westerlies from satellite photos showinga contrast between high
the Atlantic and simultaneous intensification of albedo in altered sectors and low albedo in adja-
the Mascareneand Arabian Highs. This is consis- cent protected areas (e.g., Glantz, 1977). Other
tent with Cadet's finding of a strengthening of physical changes expected to accompanythe removal
ITCZ over the western Indian Oceanduring wet years of vegetation include reduced soil moisture, evapo--
in East Africa. During droughts, such as 1972 or transpiration and surface roughness, higher surface
1983/84, low or mid-level easterly anomalies often temperatures, and increased soil erosion and dust
appear in association with a MascareneHigh which generation (Anthes, 1984). A drought would likely
is unusually intense or more zonally oriented. produce similar land surface changes.
There is also evidence that East Africa is influ- The potential feedbacks on the atmosphere in-
enced by fluctuations in the tropical divergent clude changes of latent and sensible heat transfer,
circulations (Kanamitsu and Krishnamufti, 1978; thermal stability and convergence. The hypothe-
Chen and van Loon, 1987). Anomalous
zonal diver- sized feedback was first modelled by Charney
gent circulations (Walker-type overturnings) may (1975). His initial study involved an albedo
account for the apparent inverse relationship change over the Sahara from 35% to 14%; the result
between the intensity of East African rainfall and was a significant reduction in rainfall over the
the Indian monsoon,as suggested by studies of Sahel and a southwarddisplacement of the rain zone
Stoeckenius (1981), Rasmusson and Carpenter (1982), in the high albedo case. In subsequentGCMsimula-
and Ropelewskiand Halpert (1987). tions (Charneyet al., 1975, 1977) the concept of
Although the synoptic systems of southern Africa reduced evapotranspiration was added and a similar
are fairly well described (Bhalotra, 1973; Acharya reduction in rainfall resulted (Fig. 12). A large
and Bhaskara Rao, 1981), in most cases our under- number of studies have similarly examined land-
standing of mechanismsof interannual rainfall surface feedbackmechanisms, simulating not only
variability in the region can be reduced to a few albedo and evapotranspiration or soil moisture but
generalizations. The importanceof easterly flow also surface roughness. These are beyondthe scope
in the upper-troposphere in producing high rainfall of this paper but summarizedin review articles
has been established (Taljaard, 1981a,b and Tyson, such as Mintz (1984) or Nicholson (1988). Despite
Geophysical Monograph Series Understanding Climate Change Vol. 52
8 .
E
z 5
.
% ALB •
z
f-I4%ALBEDO
_(2
k- 4•
a.:3-\,•/35Olo
ALBEDO
I
i i
I 2 3 4 I 2 3 4
NEGLIGIBLE EVAPORATION NEGLIGIBLE EVAPCRATICN NEGLIGIBLE EVAPORATION
14%ALBEDO
' ' '
• 4.l- /14'/o
ALBEDO
-I r- 14%ALBEDO
35%ALBEDO
% ALBEDO
I
2 3 4 1 2 3 4 2 3 4
TIME (weeks)- JULY TIME (weeks) TIME (weeks)
Fig. 12. Weekly average rainfall rates (mm/day) for July in three semi-arid regions
from model simulations using surface albedos of 14% and 35% in the test area (from
Charhey et al., 1977).
NICHOLSON 93
16'
•14 o
12 ø
,oo-
:
111]1
fIlljIll
IIII]]
I ',",'11''''111111
',_'
Iili
11II1
II!IIII
]lilliJllllll ]IIIIIii1111I iO o
Sø
:o,
%
_ ,.o Ili1 !111111
III1["
II1111IIIIII!1IIIII!
IIIIIII!
I Ill......
IIll ........
- , ,,,,,,, "III
4"- IT[, 4 •
ITn •.•-2o
i
2•_ i i i i i i i i i i i i i i i i i i i i i i i
16øW 14ø 12ø I0 ø 8" 6ø 4ø 2ø 0 2ø 4ø 6" 8" I0 ø 12" 14' 16øE
Fig. 13. Map of albedo changes over West Africa, pre-agriculture to present (from
Gornitz, 1985).
from a few local studies (e.g., Ibrahim, 1978; Otterman and Tucker (1985) provides some confirma-
Wendler and Eaton, 1983). Recent papers which have tion of the ear]]er result.
attempted to verify the claims of anthropogenic One of the shortcomings of the feedback hypothe-
desertification (Hellden, 1984; Olsson, 1983, 1985) sis is that the proposed feedback mechanisms do not
could not do so and have also demonstrat,ed the adequately provide a "memory"from one year to the
importance of rainfall fluctuations in altering next. If the ]and surface acts to promote multi-
surface characteristics. Similarly, Gornitz (1985), year drought, the surface characteristics respon-
in a monumental study of anthropogenic vegetation sible for the feedback must be related to rainfall
change over West Africa during the last century, in the previous year. This does not seem to be
has shown that only small changes of surface albedo true for soil moisture, since the soil is dry
have accompanied it (Fig. 13). throughout the the dry season. Nor does albedo
Questions concerning the physical impact of sur- show the anticipated relationship with rainfall of
face change also remain. For example, the assumed the previous year.
increaseof Sahelianalbedoaccompanying droughtor The surface-relatedparameterwhichappears to
a denuded land surface (Charney,1975) cannot be changemost consistently in responseto Sahel
firmly established. The albedo of "protected" rainfall is dust production. Prosperoand Nees
patchesin Niger and two areas of southern Tunisia (1977, 1986) haveshownthat African dust, whichis
(.34, .36 and .26 respectively) is considerably transported across the Atlantic and Caribbean,
higher than in surrounding overgrazedareas (.42, appears at Barbados (WestIndies) in dramatically
1983). Courelet al. (1984), combiningtheir own increased concentration during years of Sahel
albedo measurements with those of Norton et al. drought (Fig. 15). The correlation with Sahel
(1979), haveshown that albedoincreased signill- rainfall in the previousyear is exceedinglyhigh.
cantly as the drought p¾ogressed
from 1967 to 1973, The dust influences atmospheric stability, its
going from -.31 to -.38 in the more northern Sahel presence contributing to the formation of inversion
(Fig. 14). However, as the drought continued in layers above the Sahel/Sahara (Prospero and Carl-
the late 1970s, albedo decreased to values well son, 1972; Carlson and Prospero, 1972). Unfortu-
below those prior to the drought. Surface tempera- nately, the impact of dust on atmospheric dynamics
ture changes associated with decreased vegetation over West Africa has not been extensively modelled
are likewise complex and controversial. Otterman but is potentially quite significant (see Coakley
(1974) reported lower radiative temperatures over and Cess, 1985). This phenomenonhas received more
the overgrazed Sinai than in the Negev, where attention in recent years, in part because of the
grazing is more controlled. This result was chal- French ECLATSexperiment, and a number of studies
]enged by Jackson and Idso (1975), amongothers, as of dust mobilization and transport and its radia-
being a manifestation of the lower emissivity of five effects have appeared (Carlson and Benjamin,
the more denuded surface. A more recent study of 1980; ben Mohamedand Frangi, 1983, 1986; Druil-
Geophysical Monograph Series Understanding Climate Change Vol. 52
• 25
Reprinted by permission from Nature, Vol.307, p.530.
Copyright 1984, MacMillan Magazine Limited.
•z 20
1987). 0
Currently, few scientists accept the original 165188187188180170171172173174175178177178171
le01811e2183184
hypothesis that human-induced land surface changes
caused the recent drought in the Sahel. Several
points of evidence contradict such a suggestion. Fig. 15. Monthly meantrade wind mineEal .aerosol
For one, strongatmospheric
teleconnections
between concentrationat Barbados(Units 10-Vg/m•; from
the Sahel and other African locations are evident Prospero and Nees, 1986, compare with Figure 4).
and droughts often occur more or less synchronously
throughout the continent. Secondly, the Sahel Reprintedby permissionfrom Nature, Vol.320, p.736.
droughts are simply too large in scale to result Copyright 1986, MacMillan MagazineLimited.
Geophysical Monograph Series Understanding Climate Change Vol. 52
N I CHOLSON 95
McGinnis and Tarpley, 1985), and hydrology (Schnei- size or intensity of rain-bearing disturbances and
der et al., 1985; McCauley et al., 1982). how these are influenced by aerosols. Thus, fore-
casting drought in the Sahel is considerably more
complex than elsewhere in Africa since two very
Implications for Forecasting and different types of forcing must be evaluated. It
Future Research will be quite some time before these issues are
resolved.
NICHOLSON 97
d'Almeida, G., 1987' On the variability of desert sols and vertical radiative flux divergence.
aerosol radiative characteristics. J•lurnal of •9•• 9• Cl•m•9 and •pp.• Mp•9_9•g!sgy, •,
Geophysical Research, 92, 3017 3026. 38-52.
Deacon, J. and N. Lancaster. ]988' Late Quaternar3 Glantz, M.H., ed., 1977' •.9•eJ_t•%q•ODl
Palaeoenvironments of Southern Africa. Clarendon •9•%•! •9g•a•a•q• k• and •_o_u_D• Arid Lands.
P•ess, Oxiurd, 225 pp. Westview Press, Boulder, 346 pp.
Dennett, M.D., J• Elston and J.A. Rodgers, 1985' A Gornitz, V., 1985' A survey of anthropogenic vege-
reappraisal of rainfall trends in the Sahel. tation changes in West Africa during the last
Jou[nal q• •l•matology, •, 353-362. century--climatic implications. Climatic Change,
Druilhe•, A. and P. Durand, 1984' Etude de la 7, 285--326.
couche limite convective Sah•lienne en presence Greenhut, G.K., 1977' A new criterion for locating
de brumes seches/Experience (ECLATS). L_• M6t•_9•- the subtropical high in West Africa. Journal
9•9•9, 51-77. •P•!•9• •e•.9•9.lpg3, •, 727,734.
Druilhet, A., J.P. Frangi and P. Durand, 1982' Gregory, S., 1982' Spatial patterns of Sahelian
Donne•s statistiques sur le bilan d'energie de la annual rainfall. •b•¾es fuer
couche de surface Sah•lienne. La •t•oro19_g•e, •99pbysik und Bioklimatoloq•e, Ser. B., •, 273-
29--30, 227 237. 286.
Duvel, J.P. and R.S. Randel, ]985' Regional-scale Guedalia, D., C. Estournel, and R. Vehil, 1984'
diurnal variations of outgoing infrared radiation Effects of Sahel dust layers upon nocturnal
observed by METEOSAT. Jou•D• 9f •imate and cooling of the atmosphere (ECLATS) experiment.
•pP•e• M9•q•olog3_, •_•, 335-349. •9•r_n_•_[9• Climate and Appli__e_•
Meteorology, •,
Dyer, T.G.J., ]975' The assignment of rainfall 644-650.
stations into homogeneousgroups' An application Harangozo, S. and Harrison, M.S.J., 1983' On the
of principal componentanalysis. Qua•e•y J9u•i use of synoptic data indicating the presence of
na• of tb9 •9yal Meteor9•_qgical Society, 1_0•., cloud bands over southern Africa. South African
]005-1013. Journal of Science, 7•, 413-414.
Dyer, T.G.J., 1979' Rainfall along the east coast Harrison, M.S.J., 1983' The Southern Oscillation,
of southernAfrica, the SouthernOscillation and zonal equatorial circulation cells and South
the latitude of the subtropical high pressure African rainfall. Preprints of the 1st Inter-
belt. Qug[•9[_lyQQ.urng.)
of th9 Royal Meteoroloqœ. national Conference on SouthernHemisphereMet-
•q•l $Qci•3, !QS, 445--452. eorology, Americn Meteorological Society, Boston,
Dyer, T.G.J. and M.E. Marker, 1978' On the varia-
302-305.
tion of rainfall over southwest Africa. Th• •opt• Harrison, M.S.J., 1984' The annual rainfall cycle
•_[•q• G9og•phi•_ •o•a_l, •Q, 143-149. over the central interior of Africa. South
Fleer, H., ]98]' Large-scale tropical rainfall •r•½•n •9grn• of •c%•nqg, •õ, 47-64.
anomalies. Bonner Meteorolog. ische Abhandlun99_gn, Hastenrath, S.L. and E.B. Kaczmarczyk, 1981' On the
•, 114 pp. spectra and coherence of tropical climate anoma-
Flohn, H. 1964' Investigations on the Tropical lies. Tg.•l•, •, 453-462.
Easterly Jet. BonnerMeteorolog_•b••handlun- Hastenrath, S.L., 1984a' Interannualvariability
•en, 2•, 83pp. andannualcycle' mechanisms of circulation and
Flohn, H., 1988' East African rains of 1961/2 and climate in the tropical Atlantic sector. Monthly
the abrupt change of the WhiteNile discharge. W9•9• R9v%DW, •, 1097-1107.
Palaeoecoloq3
of Africa, •, in press. Hastenrath,S., 1984b'Predictability of north-east
Foiland,C.K., D.E.Parker,M.N.WardandA.W.Col- Brazil droughts.N•ure, •Q!, 531-533.
man '
1986a' Sahelrainfall' Northern
Hemisphere
Hastenrath, S., anomalies'
soon rainfall
1987' Predictability of Java
A case study.
mon-
Journal
circulationanomalies
andworldwide
seatempera- 9• •l•m•t••n• •PP•9••9t99_rp1993,
•6, 133--141.
turechanges.•ong-Range
Forecastinq
andClimateHelgren,D.M.andJ.M.Prospero,-1987'
Windveloc-
Research,!, Meteorological Office, Braeknell. fries associated with dust deflation events in
Folland, C.K., T.N. Palmer, andD.E. Parker, 1986b' the westernSahara. Journal of Climate and App-
Sahel rainfall and world-widesea temperatures lied Meteorology, .•, 1147-1151.
1901-85; Observational,modelling,and simula- He•i•en, U., 1984' Droughtimpactmonitoring' a
tion studies. Nature, 32•, 602-607. remotesensingstudyof desertification in Kordo-
Fouquart,Y., B. Bonnel,G. Brogniez, J.C. Burfez, fan, Sudan. LundsUniversitets Naturgeografiska
L. Smith, J.J. Morcrette and A. Cerf, 1987a' Institution, •, 61 pp.
Observationsof Saharan aerosols' Results of Hirst, A.C. and S. Hastenrath, 1983a' Atmos-
ECLATSfield experiment. Part I' Optical thick- phere-oceanmechanisms of climate anomalies in
nessesandaerosolsize distributions. Journalof the Angola-tropicalAtlantic sector. Journal
ClimateandAppliedMeteorology,26, 28-37. •hysical Oce•n_9o•faphl,
•, 1146-1157.
Fouquart, Y., B. Bonnel, M. Chaoui Roquai and R. Hirst, A.C. and S. Hastenrath, 1983b' Diagnostics
Santer, 1987b' Observations of Saharan aerosols' of hydrometeorological anomalies in the Zaire
Results of ECLATS field experiment Part II' (Congo) basin. Quarterly Journal of the Royal
Broadband radiative characteristics of the aero- Meteorological Socie•3, •9•, 881-892.
Geophysical Monograph Series Understanding Climate Change Vol. 52
]{slung, J. and R. Newell, 1983' The principal non- 1986' The Southern Oscillation and South African
seasonal modes of variation of global sea-surface rainfal 1. _so_ _u_t__h
African Jou__r__n_a_l .of Science
temperature. Journal of ._P_h_ysical
Ocea_n_o_g_ra•_ht, 196-198.
1__3,1952-1967. Lough, J.M., 1986' Tropical Atlantic sea-surface
Ibrahim, F., 1978' Anthropogenic causes of deserti- temperatures and rainfall variations in Sub-
fication Jou_r_n_a_l_, saharan
in Western Sudan. Ge_oir_a_ph•i_½a_l_ Africa. Mo__nth!_yWe_ a__th_e_
r_ Re_v._iew,_1_14,
23, 243-254. 561-570.
Jackson, R.D. and S.B. Idson, 1975' Surface albedo Madden, R.A., 1987' Seasonal variations of the
and desertification. Science, 189, 1012-1013. 40-50 day oscillation in the tropics. Journal
Kanamitsu, M. and T.N. Krishnamufti, 1978' Northern the _Atmo_sp•_er.i_c.
Sc_iences, 43, 3138-3158.
summer tropical circulations during drought and McCaul ey, J.F., G.G. Schaber, C.S. Breed, M.J.
normal ra]nfal 1 months. ea_t_her R_e_v_iew, Grolier,
Mpn_.t._h_ly
_W. C.V. Haynes, B. Issawi, C. Elachi, and
_1_06, 331-347. R. Blom, 1982' Subsurface valleys and geoarchae-
Kiangi, P.M.R. and J.J. Temu, 1984' Equatorial ology of the eastern Sahara revealed in shuttle
westerlies in Kenya. Are they all rain laden? radar. Science, _218_,1004-1017.
Ex_t.e_nded a_b__st_ra_c_t_sof kape_r_sp_r_e_s__e__nt_e_d at_ th_e_WM_O_ McGinnis, D.F. and J.D. Tarp]ey, 1985' Vegetation
Reg_ion_alS½ientif_ic c_o.n__fer_egc_e on GATE, •AME_X_a_n_d cover mapping from NOAA/AVHRR. A_dv_a_n__c_e_s i_n_S_pa_ce
T_r_.opicalm_e_t_e__o__ro..lo_G3 in_ A_f_r
i__c
a_, World Meteorolog- Research, 5, 359-369.
ical Organization Tropical Meteorology Programme Miles, M.K. and C.K. Folland, 1974' Changes in the
Report Series No. 16, Geneva, 144-146. latitude of the climatic zones of the Northern
Kidson, J.W., ]977' African rainfall and its rela- Hemisphere. N_a_•_u_r_e_,_2_5_2,616.
tion to the upper air circulation. Minja,
Qu_a_rt_e__r__1I W.E.S., ]985' A comparative investigation
_JQu_rn__a_! of the _Roya_! Met_e_or_p.l_9g_i__c_a_l .S.
oci etj, of weather anomalies over West Africa. M.S.
103, 441-456. Thesis, Department of Meteorology, University of
Klaus, D., 1978' Spatial distribution and periodi- Nairobi.
city of mean annual precipitation south of the Mintz, Y., 1984' The sensitivity of numerically
Sahara. Archives _fu__er M_eteorol_pG•_e_•g_eo•p_blsi_k un_d simulated climates to land-surface boundary con--
Sioklimatolo_g_i_e_ S_e__r_. B_., 2_6, 17-27. ditions. The G]9b_al Cl.ima_te, J.T. Houghton, Ed.,
Knuison, T.R., and K.M. Weickmann, 1987' 30-60 day Cambridge University Press, 70-105.
atmospheric oscillations' composite life cycles Motha, R.P., eta]., 1980' Precipitation patterns
of convection and circulation anomalies. Monthly [n West Africa. Month_ly Weathe_r Rev.ie_w, 1QS_,
1567- 1578.
Weat_h_e_r_R e_v i _e_w_
, _1.1_5_,1407- 1436.
Knutson, T.R., K.M. Weickmann and J.E. Kutzbach, Namias, J., 1974' Suggestions for research leading
1986' Global-scale intraseasonal oscillations of to long-range precipitation forecasting for the
outgoing longwave radiation and 250 n•b zonal wind tropics. Preprints of the Internat_i_ona] Tr_op_ical
during northern hemisphere summer. Month_l_yW_e_az Meteorol_ogy Meeting, Nairobi, American Meteoro-
th_e_• Rev.i_ew, 114. logical Society, 141-144.
Kraus, E.B., 1977a' Subtropical droughts and cross- Newell, R.E. and J.W. Kidson, 1979' The tropo-
equatorial energy transports. MQ_n•hJy W_e__a_ther spheric circulation over Africa. Saharan Dust
Review, 10_5, 1009-1018. (Scope Report 14), C. Moral es, Ed., Wiley and
Kraus, E.B., 1977b' The seasonal excursion of the Sons, ]33-370.
intertropical convergence zone. Monthly th_e_r. Newell,
W_e_a. R.E. and J.W. Kidson, 1984' African mean
Rev.iew, _105, 1009-1018. wind changes in Sahelian wet and dry periods.
Krueger, A. and J.S. Winston, 1975' Large-scale Journal of Climatology, 4, 1-7.
circulation anomalies over the tropics during Nicholls• N., 1983' Predicting Indian monsoon
1971-72. Monthly Wea_ther Re_v_i_gw, _10_3_, 465-473. rainfall from sea surface temperature in the
Kumar, S., 1978' Interaction of upper _westerly Indonesia North Australia area. Natu_•re, 306,
waves with intertropical convergen_ce MoDe and 576-577.
the_if e_f_fect on the weather ov_e_rZanib_ia _during Nicholson, S.E., ]978' Climatic variations in the
the rainy season. Government Printers, Lusaka, Sahel and other African regions during the past
36 pp. five centuries. Journal of Arid Environments, 1,
Lamb, P.J., 1978a' Case studies of tropical Atlan-- 3-24.
tic surface circulation patterns during recent Nicholson, S.E., ]979a' The methodology oi histor
sub-Saharan weather anomalies' 1967 and 1968. ical climate reconstruction and its application
Mon_th_lyWeather Review, _106, 482-491. •o Africa. Journal of African History, 20, No. 1,
I. amb, P.J., 1978b' Large-scale tropical Atlantic 31 49.
surface circulation patterns associated with Sub- Nicholson, S.E., 1979b' Revised rainfall series for
saharan weather anomalies. Tellus, 30, 240-251. the West African subtropics. Monthly weat_her
Lamb, P.J., 1982' Persistence of Sub - saharan R__evi_ew_, _107, 620 623.
drought. Nature, 299, 46-47. Nicholson, S.E., 1980a' Saharan climates in his-
Lamb, P.J., 1983' West African water vapor varia- toric times. Tb_e_Sahara an.d_ the N_i•.9, M.A.J.
tions between recent contrasting Subsaharan rainy Williams and H. Faure, Eds., A.A. Ba]kema, Rot-
seasons. Te_l._lus., 3__5_A_, 198-212. terdam, 173-200.
Lindesay, J.A., M.S.J. Harrison and M.P. Haffner, Nicholson, S.E., ]980b' The nature of rainia]]
Geophysical Monograph Series Understanding Climate Change Vol. 52
N I CHOL$ON •
lation. MO8•y •9•h• Review, •, 1606-1626. Tanaka, M., B.C. Weare, A.R. Navato and R.E. New-
Schneider, S.R., D.F. McGinnis, Jr. and G. Step- ell, 1975' Recent African rainfall patterns.
hens, 1985' Monitoring Africa's Lake Chad basin •ture, 255, 20]-203.
with LANDSAT and NOAAsatellite daf.a. Inter- Tucker, C.J., C. Vanpraet, E. Boerwinkel and A.
•9•.• J9•g•.• q• Rem9•e Sensing, •, 59-73. Gaston, 1983' Satellite remote sensing of total
Schupelius, G.D., 1976' Monsoon rains over West dry matter production in the Senegalese Sahel.
Africa. T..•J•u•, •8, 533-537. R9mote Sensing of Environment, 13, 461-474.
Schultze, G.C., ]983' 'N Moontlike verband tussen Tucker, C.J., C. Vanpraet, M.J. Sharman and G. Van
die Suidelike Ossilasie/E1Nino verskynsels en Ittersum, 1985a' Satellite remote sensing of
droogtes oor die somerreenva]streke van Suid- total herbaceous biomass production in the Sene--
Afrika. 'N Voorlopige studie. Sgut• •_f•fg8 galese Sahe]. Remote Sensing of Envirqnment, ]7,
,
Duzheng Ye
Instituteof AtmosphericPhysics,AcademiaSinica,Beijing,China
lol
Geophysical Monograph Series Understanding Climate Change Vol. 52
DRIP
Jl
cast coast of Asia there is an increaseof precipitation. It is also inter-
csting to note that decreasein precipitation is not limited to the con-
EVAPORATION
tinents; over the oceans,especiallyover the western tropical Pacific,
{IR
DOWN the decrease of precipitation is also very pronounced; up to - $
mm/day in large areas. Kurbatkin ctal. [1979] attributed this to
suppressof vertical motion over the oceans. The vertical motion over
the continentsin the initially dry caseis increasedalthoughthe precip-
FLOOD
FLOW itation rate is decreaseddue to lack of humidity. The increaseof the
continental vertical motion is probably associatedwith the decreasein
vertical motion over the ocean.
To further relax the constraints of the simulation we discard the
global condition and go t6 regional experiment with time-dependent
PERCOLATION soil moisture. Many authors have done it, e.g., Rowntrcc and Bolton
[1983], Sud and Fenncssy[1982, 1984], Wardlow [1985], Yeh, ct al.,
• TO
GROUNDWATER [1984], and many others. In the global experimentsabove we have al-
ready sccn that the climatic responseto uniform global condition is
Fig.l Schematicdiagramof the processes
that needto be considered
in a canopymodel of surfaceevapotranspirationand energybalance
[Dickinson, 1984]. 8CN
60N
must be replenishedby winter precipitation. Thus we can sccthat the
annual cycle of soil moisture content must be connected with the 40N
YE 103
80N
60N
40N
20N
20S
zo
IS
40S o IO
60S
80N
60N
40N
20N
20S
40S
60S
180 120W 60W 0 60E 120E 180
Fig.3. Ground-surface temperature(E) in wet- soil case(top) and dry-soil case(bottom), in experimentofShukla Mintz[1981].
not everywere the same. This means that the influence of hydrological intensity of the increase is much smaller than in the other two cases.
processesdependson the climate regimes.We shall further discussthis Further, there is no well-defined maximum of precipitation increase
point by usingYeh, et al./s [1984] results.They irrigated the continents within the irrgatcd region; instead, a relatively samall increase takes
of 30-60N zone to its saturation (15cm) on July 1 and two other place throughout the entire zone. But the intensity of the evaporation
zones, 0-30N and 15S-15N, to saturation on Jan. 1, and then inte- increase for 0-30N is of comparable value with that for 30-60N or
grated each case for a period of five months. The influence of the 15S-15N cases. This indicates that the major part of water vapor
large-scale irrigation on climate was obtained by analyzing the differ- gained from evaporation in the zone 0-30N is transported out and on-
ence between the irrigated and the standard run. ly a small part is used to increase the precipitation. Ych, ct al. [1984]
Fig.5a-c show the latitude-time distribution of the mean zonal dif- invoked an increasein the descendingbranch of the Hadlcy circula-
ference in precipitation (cm/day) between the irrigated and control tion to explain this phenomenon.
experiments. One common feature in all the three figures is the in- Figs.5a and 5b share another characteristic. Both show two belts of
creaseof precipitation in the regions where the soil is instantaneously maximum increase, one to the north and the other to the south of the
saturated with water. However, it is clear that the period of enhanced irrigated region. In the 30-60N case the northern belt of maximum
precipitation last longest in the 30-60N case (about 5 months) and increasesis found at 60-65N, lasting about two months. The southern
shortest in the 15S-15N case (about 3 months). These difference in bclt is in the tropics and persistsfor a much longer period. It is intcr-
precipitation among the three casesare qualitatively similar to the dif- csting to note that the northern belts of maximum increase in precipi-
ference in the persistenceof soil moisture anomaly or enhanced evap- tation in Figs.5a (summer) and 5b (winter) arc near the midlatitude
oration (not shown here). It can also be seen that in the 30-60N case rainbelts in summer and winter, rcspcctivcly, and that the southern
(Fig.5a) the increase is maximum near the center of the irrigated re- bclts of maximum increase in precipitation in Figs. 5a (summer) and
gion and that it decreasesboth northward and southward. The corre- 5b (winter) arc situated over the ITCZ rainbelt which is north of the
sponding distribution for 15S-15N case (Fig.5c) is less symmetric. Its equator and south of the equator, rcspcctivcly. This indicates the
maximum is displaced into the Southern Hemisphere in January and largc-'scalc increase of soil moisture can induce certain physical pro-
shifts equatorward during February and March. For the 0- 30N case cesscswhich tend to enhance the activity of the ITCZ and midlatitude
(Fig.5b) there is a notable difference from the other two cases,i.e., the rainbelt cvcn outside the region of soil moisture increase. Ych, ct al.
Geophysical Monograph Series Understanding Climate Change Vol. 52
90N
90N
60
30
ß
i.... '::!::.:.:.:.:..
•.:..-•:.:.:.:.:.:.:.: .... ........':::::'::
...::i:
...:.:.:.:.:.:. !..:.::i]'.-•.•i::'•/"
.'.'.¾'
'.:.'.'
....
.•....:.:.:.:.:.:.:.:.:.:.:.:
:.:...: 3O
-- EQ
30 6O
..::::iiiiiiiiiiiii!:::"
"'"'i:i:i:i:i:!:i:i:i:i:i:iiii::iiii:".•.•i!iiiii!!iiiiiiii!...
3O
90S 90S
0 90E 180 90W 0
90N 90N
6O 6O
... . ...._:.__
! ß....:.>:.:.:.:.>.'.. . ß======================
/•....:'......
ß :.
....
.• .......:.:.:.:.:.:.:.:
........•.._ ._,.•....•..;.
ß
•i.
:..'..
.'•::iii::::::iii!•i::•i"
ß ....... :"
3O 3O
EQ EQ
3O 6O
6O 3O
,....::::
•"' '""'•i•:.::•:.?•: ••?:::::::::'"'
90S ..:.:.:.:.;.:.1.:.:.;.:.:,:.:1:.:.>;.,.,.,.;i;,;( I i .....
:'.••
::•..
:'":
:'-
.•.
' :'.'
".
[-•----.•.•.-..,,..,..._.. ß.J:.:,:.:.:.:.:,I.;.;.;.;.:.:.:.1,:.:.:.;._,.:,_..1.:.:.:.:.:._..:.
90S
o 9o[ ] 8o 9ow 0
Fig.4. Rate of evaporation difference(a) and rate of precipitationdif- experimentsand are horizontally smoothedto emphasizethe large-
ference (b). Difference (Arid Exp.-Control Exp.) distributions are scalefeatures. Isolines are drawn for -5, -2, -1, 0, 1, 2, 5, mm/day
constructedfrom the July-August seasonalmeans of the two model in both distributions. [after Kurbatkin, ct al., 1979].
[1984] postulated thc following physical reasonsfor thc enhancements: Ych, ct al. [1984] also pointed out that for all thc threccasesthcre is
The strengthening of evaporation due to irrigation will produce a a deercaseof precipitation just south of thc area of maximum increase
sheetof moist and cool air near the groud. This sheetwill spread from in the tropical rainbits. They speculatedthat this reductionmight be
the region and supply moisture to the neighboring rainbelts and en- due to the changeof upward motion in theseregions.Yeh, et al. [1984]
hancesthe already existing convectiveactivity and precipitation there. also calculated the total increase of precipation (AP) and the corre-
a b
90øN ; ', , . , 60ON 45øN
45ø 15ø' oo
30ø 0o 15ø
oß 30ø5 45øS
J A S O N J F M A M [ M A M
Fig.5. The latitude-time distributin of thc zonal mcan differcnce of precipitation rate (cm/ day) between thc
perturbedand the normal experiment.(a) the 30 ø N-60 ø N case;(b) the 0-30 ø N case;(c) the 15 ø S-15 ø N
½•e. [after Yeh, Wetheraid and Manabe, 1985].
Geophysical Monograph Series Understanding Climate Change Vol. 52
YE 105
TABLE 1. The integrated increase in the water balance components AE, AP, and AR due to an increase of soil
moisture AWs for three latitude zones averaged over the period where the values of AP were Positive (unit: cm)
and the ratios of water balance components to AWs. (After Yeh, Wetheraid and Manabe, 1984).
sponding changes of evaporation (AE) and runoff (R) for the whole of Yeh, et al.'s [1984] experiment. The picture shows negative differ-
irrigated zone during the period when the average regional change of ence in the lower layers of the atmosphere above the irrigted zone and
precipitation is positive. The results are shown in Table 1. The total positive difference higher up. This indicates that in the lower atmos-
water (AWs) used to irrigate each zone to saturation is also given in phere the increaseof latent heat flux by irrigation from the earth de-
the table. The values of AWs of the three zonesare comparable as seen creasesthe temperatures while at higher levels the increase of latent
in the table. The ratios AE / AWs, AP / AWs and AR / AWs in Table heat release in precipitation increase the temperature. Even two
1 can be used as a measure of efficiency of irrigation in inducing months after the large- scale irrigation, the maximum decrease of
changesin the hydrological processes.The larger the ratio AP/AWs, temperatureis still nearly 4U and the maximum increasein tempera-
the more benefit is gained from irrigation. When AE/AWs• ture at high levels 1.8U The correspondingchang in zonal-mean
AP/AWs the benefit attains its optimum. In the region where westly wind is given in Fig.7.
AP / AWs<<AE / AWs the efficiency of irrigation is least. From Table
I it is seen that the highest efficiency of irrigation is in the zone Sensitivity of Climate Models
30-60N and the least is in zone 0-30N. The physical reason for the ef- to Surface Albedo
ficiency being least in the 0-30N is that this zone is located under the
descendingbranch of the Hadley cell. Another interestingpoint seen The presistentdrought in Sahel in recent years aroused great con-
in the table is that in the 0-30N zone both AP / AWs and AR / AWs cern among the people of the world including scientists and
are the lowest among the three zones.This indicatesthat the transport politicians. Meteorologists studied the causeof the drought from va-
of moisture obtained from the irrigation out off the irrigation area rious points of view and made simulations by GCM under different
should be highest in 0-30N zone. Table 1 also tells that land surfaceforcings.In discussing
the simulationof the Sahelian
(AP / AWs)(30-60N) >(AP / AWs)(15S-15N). The reason for this is drought we should keep in mind that this region is situated roughly in
that (AR/AWs)(30-60N)<(AR/AWs)(15S-15N). Thus the mois- the descendingbranch of Hadley cell and that its spatial scaleis on the
ture obtained from irrigation going to runoff is greater in the order of a few thousand kilometers.
15S-15N zone than in the 30-60N zone, making lessmoisture availa- Charney [1975] first proposed a model for maintaining the drought
ble for precipitation in the 15S-15N zone. The above discussionclear- in the Sahel. Charney•sdynamicsdepend upon a feedbackmechanism
ly showsthe high dependenceof the influenceof the hydrological pro- involving radiation, subsidenceand albedo. A lack of vegetation gives
cesseson the climate regimes. Many other authors have obtained simi- a higher surface albedo which leads to a net radiative heat loss at the
lar resultsbut they cannot be disussedhere due to limited space. top of the atmosphere. This will induce a subsidenceof dry air aloft
Although the physics of the hydrological processesmainly lies at with a subsequentreduction of precipitation, thus maintaining the de-
the ground surface, the upper air will also be influenced through the sert. This hypothesis was subsequently tested by a numerical
atmosphericboundary layer. To illustrate this in an example, we give simulation with a general circulation model [Charney, et al. 1977].
Fig.6 which shows the latitude-height distribution of zonal mean Following Charhey et al. Chervin [1979], Sud and Fennessy [1982],
temperature difference between the irrigated and the standard experi- Planton [1986], Cunnington and Rowntree [1986], Laval [1986], and
ment for the zone 30-60N in the secondmonth of irrigation (August) Carson and Sangster[1981] using different GCM's obtained similar
95
350
680
F,30
•0
90o• ?5 ø 60 ø 45 ø 30 ø 15 ø 0ø 15ø 30 ø 45 oS
! I
Fig.6. The latitude-height distribution of the zonal mean differenceof air temperature(U) over land between
the perturedand normal experiment,August of the 30 ø N-60 ø N case[after Yeh, Wetheraid and Manabe,
19851.
Geophysical Monograph Series Understanding Climate Change Vol. 52
95
20.•
350
515 ,
680
830
940
9'}0
90•N 75 ø 60 ø •5" 30 ø 15" 0ø
I
Fig.7.Thelatitude-height
distribution
of thedifference
of theAugust z•nal mcanwind(m/s) betweenthe perturbed
30 o N-60 ø N and the standardexperiment.[after Yeh, Wetheraid and Manabe, 1985].
resultsof decreasingprecipitation when increasingthe surfacealbedo. well with the observed seasonalcycle of rainfall. Changes in rainfall
The resultsof theseexperimentssupportedCharney'shypothesis. with deforestation are not obvious because of high natural variability
All above experiments undoubtdly indicate that a high albedo is of model convective rainfall. The deforested case has more runoff, es-
important in maintaining deserts.However to prove this hypothesisas pecially during the dry seasonand lessinterception of rainfall. The pe-
the main cause of the Sahelian drought, we need observational evi- riod during the season of driest soil is widened from one month to
dence of the actual albedo change. Charhey, et al. [1977] also found several months in southern Amazonia. The soil and air become warm-
that changesin the surfacehydrology can have comparable effectsas er by several degrees, especiallyduring the dry season;sensibleheat
a change in surfacealbedo from about 0.1 to 0.3, which is the value fluxes are increased and evapotranspiration is reduced by
usedby Charney for explainingthe Saheliandrought. Rowntree and deforestation.'
Sangster[1985] recentlyalso simulatedthe decreaseof sahelianprecip- With emphasis on tropical deforestation, Henderson-Sellers et al.
itation by only decreasingthe soil mosturethere. [1985] studied the possible climate impacts of land cover transforma-
In 1986 Cunnington and Rowntree [1986] went further to test the tions. Hansen et al. [1983] have also, but in a much simpler way, in-
influenceof the initial atmosphericmoisture state on the precipitation corporated the parameters associatedwith different vegetation types
and hydrologicalfactors of the Sahara. into their climate model, model II. The vegetation distribution over
To close this section we should say that although the soil moisture the earth were compiled by Mathews [1983]. Using this model Rind
content and the albedo are indeed important factors determining the [1982] investigatedthe influenceof vegetation on the hydrological cy-
Saharan climate we sould be very cautions to use the above experi- cle. In this model the function of vegetation is much simplified and the
mental resultsto state that we actually explained the recent drought in soil is consideredas composedby two layers, an upper layer of 10 cm
the Sahel. There have been many long-term natural variations of the depth and a lower layer down to 4 m depth.
averagedprecipitationoverAfrica [Nicholson,1985].We mustbe able Using this model Rind [1984] studied further the albedo, water
to detect the albedo- or soil moisture content- induced effects against holding capacity and diffusion for the vegetation types in the climate.
this large background noise. The albedo effect is most effective in areas with plentiful moisture as-
sociated with evaporation from the ocean surface and during the sea-
son when large-scale dynamics or convection control the precipita-
Sensitivity of Climate M odeIsto Vegetation tions. The water holding capacity influence dominates where local
evaporation is essential for precipitation. The diffusion experiment al-
As pointed out in the Introduction, vegetation is another important tered the rate at which water is made available from deeper levels. Its
factor determining the surface moisture availability which controls the main effect was a stretching out of time of the soil moisture loss, with
partition between the surface fluxes of latent heat and sensiblehent. In affected the seasonalphase of the precipitation.
all the foregoing described experiments evaporation and transpiration
were combined as cvapotranspiration. Dickinson [1985] strongly em-
phasized the need for separation of the roles of vegetation and soils in The Dependence on The Spatial
the GCM's. A comparatively detailed description of the land cover Scale of The Anomaly
and soil has been introduced into NCAR Community Climate Model,
and several sensitivity studies have been completed with major em- So far we have been concerned with very large- scale anomalies, ei-
phasis given to the effect of the Amazon deforestation. The prelimina- ther on the global scale or on the continental scale. Certainly the im-
ry conclu- sions drawn from comparisons of the control and the portance of the surface hydrology on climate will depend on the spa-
deforested runs for the Amazons region were given by Dickinson tial scale of the surface anomaly. It may be estimated relative to the
[1985]. They arc: •the control and deforestedruns both agree fairly role of advection. Rowntree [1984] estimated the importance of
Geophysical Monograph Series Understanding Climate Change Vol. 52
YE 107
advected moisture on continental rainfall under the assumptions of three-dimensional global climate models. 58-72, In Climate Pro-
nondivergent flow of air acrossa land mass with a humidity acquired cessesand Climate Sensitivity, Geophys. Mon., 29,/I.E. Hansen, and
while crossingthe land mass, an evaporation rate of 3 mm/day (ap- T. Takanhashi editors.
propriam for moist tropics), and a mean flow of 5 m / s. Under these Dickinson, R.E. (1985),GCM sensitivity studies- implications for
assumptions he showed that for an island of 100km horizontal scale parameterizations of land surface processes, Proc. ISLSCP. Con-
CLo), the prescribedabvectedmoisture flux convergenceplays a dom- .•rence, Rome, Italy, 2-6, Dec. 1985, EAS. SP-248, 127- 129.
inant role, and local evaporation a very minor role. However for Lo• Eagleson, P.S. (1982), Dynamical hydro-thermal balance at
1000 km (e.g., Penisular India), local evaporation is a little more im- macroscale, In land Sur•ce Processesin Atmospheric General Circu-
portant while for Lo > 3000 km (e.g., Africa), the contribution from lation, Ed. by P.S. Eagleason, Cambridge University Press.
the horizontal moisture flux convergenceis relatively small. Although Gadd, A./I and/I.F. Keers (1970), Surface exchangesof sensitiveand
Rowntree'scalculationwas basedon very idealizedassumptions,they lantent heat in a 10-level model atmosphere, Quart. J.R. Met. Soc.,
give a qualitative earlmate of the scaleeffect in the influenceof evapo- 96, 297-308.
ration anomalies. Hansen,/I.E., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R.
Ruedy and L. Travis: (1983), Efficient three dimensional global
models for climate studies: Models I and II. Mon. Wea. Rev., 111,
Concluding Remarks
609-662.
Henderson-Sellers, A. and H. Gornitz (1985), Possible climatic im-
The above discussionsshow that the hydrological porcessesare very
pacts of land-cover transformations with particular emphasis on
important in climate. The inclusion of hydrology in GCM's is not on-
tropical deforestation, Climate Change, 6, 231- 257.
ly necessaryfor climate studiesor predictions,it can also improve me-
Kurbatkin, G.P., S. Manabe End D.G. Hahn (1979), The moisture
dium-range weather forecast [Blondin, 1985], a few days forecast
content of the continents and the intensity of summer monsoon cir-
[Tada, 1985] and even very short-range weather forecast [Gadd and
culation, Meteorlogiya i Gidrologya, 11, 5-11.
Keers, 1970].
Laval, K. (1984), Modeling the impact of soil properties on European
Another potentially possible practical application of the surface
climate. In Current Issues in Climate Research, A. Ghazi and R.
hydrological processeswould be the modification of local weather or
Fantechi, editors. D. Reidel Publishing Company.
climate. Anthes [1984] made a study showing the possibility of
Laval. K. (1986), General circulation model experiment with suface
enhancement of convective precipitation in semi-arid regions by
albedo changes, Climate Change, 9, Special issue: Climate and
planting alternating bands of dense vegetation with width of
Desertification, 91-102.
25-50km. Segal et al., [1983] suggestedto modify the local climate by
Laval, K., A. Perfie and Y. Serratini (1984), Effect of
filling in the QattaraDeppression
(about 18,000km2 in size)with
parameterization of evapotranspira- tion on climate simulated by a
Mediterranean water which is about 90km from this deppression.
GCM, In New Perspectivesin Climate Modeling, Ed by A.C. Berger
There are, of course, many other possibleapplications.
and C. Nicolis, 223-247.
Since the two crucial factors in hydrology are the soil moisture con-
tent and the vegetation, we strongly recommend that soil moisture ob- Mathews, E. (1983), Global vegetation and land use: New high resolu-
tion data basis for climate studies, J. Clim. and A ppl Meteor., 22,
servations be made by satellite and that the astellite data be validated
474- 487.
by ground-truth observationsand global mapping of vegetation.
Mintz, Y. (1984), The sensitivity of numerically simulated climates to
land-surface boundary conditions, 79-105, In The Global Climate,
References Ed. by/i. T. Houghton.
Namias, /i. (1962), Influences of abnormal surface heat sourcesand
Anthes, R.A. (1984), Enhancement of convective precipitation by sinks on atmospheric behavior, Proc. Internat. Symp. Num. Wea
mesoscale variations in vegetative covering in semiarid regions, J. Predict. Tokyo, Nov. 7-13. 1960, Met. Soc./iapan, 615-627.
Climate & Appl. Met., 23, 541-554. Namias, /i. (1963), Surface-atmosphere interactions as fundamental
Blondin, C.A. (1985),Treatement of land-surface properties in causes of drought and other climatic fluctuations, Arid Zone Re-
ECMWE model, Proc. ISLSCP Con.•grence,Rome, Italy, 2-6 Dec. search Changes of Climate Proc, Rome Symp. UNESCO and
1985. WMO, 345-359.
Carson, D./I. and A.B. Sangster (1981), The influence of land-surface Nicholson, S.E. (1985), African drought: an example of the influence
albedo and soil moisture on general circulation model simulations, of land-surface processeson climate. Proc. ISLSCP Con•rence,
GAPP / WCRP: Research Activities in Atmospherical and Oceanic Rome, Italy, Dec. 1985, ESA SP-248,405-410.
Modelling, (Ed. by J.D. Rutherford) Numerical Experomentation Planton, S. (1986): Sensitivity of the annual cycle simulated by a
Programme. Rport No.2, PP5.14-5.21. GCM to a change in land surface albedo, Proc. ISLSCP
Charney,J.G., (1975),Dynamics of deserts drought in the Sahel, Con•rence, Rome, December, 1985, ESA SP-248, 135-142.
Quart. J.R. Met. $oc., 101,193- 202. Rind, D. (1984), The influenceof vegetationon the hydrologicalcycle
Charney, J.G.,W.J. Quirk, S.H. Chow and J. Kornfield (1977), A in a global climate model, 73-91, In Climate Processesand Climate
comparative study of the effects of albedo change on drought in Sensitivity, Geophys. Mon., 29, Ed. by /I.E. Hansen and T.
semi- arid region, J. Atemos. Sci., 34, 1366-85. Takahashi.
Chervin, R.M. (1979),Respones of the NCAR general circulation Rowntree, P.R. and/I.A. Bolton (1983), Simulation of the atmospher-
model to changed land surface albedo, Report o.fthe JSC study Con- ic response to soil moisture anomalies over Europe, Quart. J. R.
.•grenceon Climate Models: Per.•rmance, Intercomparison and Sen- Met. $oc., 109, 501-526.
sitivity Studies. Washington, DC, 3-7, April, 1978, GAPP Publ. Se- Rowntree, P.R. (1984), Review of general circulation models as a ba-
ries, No. 22, Vol. 1 563-81. sisfor predicting the effect of vegegationchange on climate, Met. O,
Cunnington, W.M. and P.R. Rowntree (1986), Simulations of 20, Technical Note. II ? 225.
Shaharan atmosphere-dependence on moisture and albedo, Quart. Rowntree, P.R., W.F. Wilson and A.B. Sangster (1985), Impact of
J.R. Met. $oc., 112, 971-999. land surface variation on African rainfall in general circulation,
Dickinson, R.E. (1984),Modelling evapotranspiration for DCTN 30 Met. O.
Geophysical Monograph Series Understanding Climate Change Vol. 52
Segal, M., R.A. Pielke, and Y. Mahrer (1983), On climatic change due spectral model, Proc. ISLSCP Con.•rence, Rome, Italy, 2-6, Dec.
to deliberate flooding Qatara Deppression (Egypt), Climatic 1985, ESA. SP-248.
Change, 5, 73-83. Warrilow, D.A (1985), The sensitivityof the UK Meteorological Of-
Shukla, J. and Y. Mintz (1982),Influence of land- surface fice atmospheric general circulation model to recent changes in the
evapotranspiration on the earth's climate, Sciece,21:$, 1498-1501. parameterization of hydrology, Proc. ISLSCP Con.Igrence,Rome,
Sud, Y.C. and M. Fennessy (1982), A study of the influence of surface Italy, Dec. 1985, ESA. SP- 248, 143-149.
albedo on July circulation in semi-arid regions using GLAS GCM, Yamazaki, K. (1986), The sensitivity experiment to land-suface
J. Climatology, 2, 105-125. boundary conditions with the M.R.I.G.C.M. Proc. ISLSCP Con-
Sud, Y.C. and M. Fennessy (1984), Influence of evaporation in .Igrence,Rome, Italy December,1985, ESA. SP-248, 151-157.
semi-arid regions on the July circulation - a numerical study, J. Yeh,T.-C., R.T. WetheraldandS. Manabe(1984),The effectof soil
Climate, 4, 383-398. moisture on the short-term climate and hydrology change-a nu-
Tada, K. (1985), Land-suface parameterization in JMA operational merical experiment, Mon. Wea. Rev., 112, 474-490.
Geophysical Monograph Series Understanding Climate Change Vol. 52
Peter S. Eagleson
Abstract. The average annual water balance of landsurface area in Figure 1. The area of bare
tree/grass vegetation systems is modeled as an soil is indicated by Ms . The area shaded by
interactive competition for water and energy. grass (with the sun directly overhead) is Mg
Ecological optimality hypotheses are introduced and that by the trees is Mw. This is often
which allow specification of the fractional area called the "projectlye foliage cover" or the
covered by woodland canopy and by grass canopy "canopy density". Following the observations of
under conditions of natural equilibrium. Three Sarmiento [1984] we assume no runoff.
equilibrium states are found. Two are monocul- We further assume that the taller canopy
tures, i.e., grassland and closed forest which suppresses the atmospheric water vapor transport
are shown to be unstable with respect to per- capacity ep that is effective for the shorter
turbations of canopy cover. The third, a canopy. Accordingly and using a simple linear
tree/grass mixture, is shown to be stable with relation, we have for the grass
respect to perturbations of canopy cover but
metastable with respect to climate change.
epg= (1 - Mw) epw (1)
Introduction
and for the bare soil
109
Geophysical Monograph Series Understanding Climate Change Vol. 52
PA
Mg
•M w • ß Ms ,
epw'kvw
TREES
t
epg-
kvg
j . eps
Rs=O
NO
RUNOFF
_-
Rg=O
NO DEEPPERCOLATION
I
in which
G: Mw
+ Mg(1
K
-Mw)+ (1 - Mg)R
So5/2(9)
m6 = fraction of year during which soil
moisture in lower layer can support
where
transpiration by trees-
Mw : (KS/s) so
5 (6) R = Jo/(epwkvg
) (11)
Geophysical Monograph Series Understanding Climate Change Vol. 52
EAGLESON
x =CONSTANT
G--K=1
/ \
/ \
/ \
/ \
/ \
/
/
dMw /
dt EQUILIBRIA / EQUILIBRIA
o I Mw
/
/
/
/
/
/
/
/
/
/
in which PA(Cm) is the average annual precip- a closed forest (Mw = 1, Mg = 0) will use the
itation rate. the same amount of water annually as will grass-
land (Mw= 0, Mr = 1), and G = 1 specifies that PA
Considering
kvgandtheproduct
m6kvw
to be is exactly sufficient for both of these extreme
known parameters of the system there are three
states. An intermediate tree/grass equilibrium
unknowns,So, Mw and Mg but only two equations, is given by (14).
(6) and (9). Invoking the hypothesis [Eagleson,
1978b, 1982] that natural vegetation systems will
attempt to minimize water demand stress, we add Importance of Seasonality
the third equation
Note that as seasonality vanishes (i.e., ml
bSo/•Mg= 0. ( 12) and hence m• approach unity) equations (7), (10)
and (13) give
Eqs.
feasible
(6),
solution
(9) and
only
(12)
for
have a physically
kvg/kvw
+1
and
G=K= 1 (13)
PA/(epwkvg)
+1
for which
which can be satisfied either for Mw = 0 or
for
kvg= kvw
. Since under water-limited
Mw = [-X + (X2 + 1)1/2] (14)
conditions
kvg>kvw[Eagleson
andSegarra,1985],
where the latter condition will occur only with
unlimited moisture when = 1. Under
kvg= kvw
X = R/[2(S/•)]1/2 (15) such conditions the vegetation will probably seek
to optimize biomass rather than soil moisture
[Eagleson, 1982] and trees will replace the
with Mg being arbitrary.
There are thus three equilibrium states of the grass. This situation is outside the range of
savanna system. The condition K = 1 implies that the current analysis.
Geophysical Monograph Series Understanding Climate Change Vol. 52
1 12 SAV•NA S•A•œ•ILI f¾
0.5
K=I
0.4
Mw
0.3
so
G=I (
0.2 /
/ 0.8,.
/
/
/
0.1
I I I I I I I
00 0.2 0.4 0.6 1.0
MgandMw
Fig. 3. Sub-optimal equilibria; R -- 24, S/• = 250, K = 1.
We thus see that seasonality of the moisture order-of-magnitude analysis we assume the
supply is a necessary condition to have an absolute value of the first term on the right
equilibrium tree/grass system. hand side of (16) exceeds that of the second.
The sign of dMw/dt will then be that of dSo/dt.
Stability to Vegetation Perturbations
Shouldwe be at the Mw = 0, Mg= I equilibrium
and somehow
decreaseMgwe replace grass,
We begin by considering perturbations to the transpiring at the potential rate, by bare soil
equilibrium vegetation under the assumption that of lower evaporation efficiency. The soil
these cause no change in the parameters of the moisture so will thus rise making dSo/dt , and
climate and/or soil.
hence dMw/dt positive. Should we be at the
In the phase plane diagram of Figure 2 the Mw = 1, Mg = 0 equilibrium and somehowdecrease
three equilibria in M are located along the Mw we will have moisture loss by bare soil evap-
dMw/dt - 0 axis. The solid and dashed lines oration and from growth of grass thus decreasing
are qualitative representations of the only two t o and making dMw/dt negative. This reasoning
possible forms of dMw/dt - f(Mw). The slope establishes f(Mw) as the solid curve (a) in
of f(M w) at dMw/dt - 0 determines the stability Fig. 2 and the intermediate tree/grass equilib-
of the associated equilibria. Writing the rium is seen to be unconditionally stable to per-
conservation equation for water in the lower soil
turbations in Mw. If fire should decrease Mw
layer where Zw is the water table elevation and from this equilibrium value, Mw would be in a
differentiating with respect to time gives region of positive dMw/dt and the system would
return to equililbrlum. Such restoration would
not follow perturbations of the Mw = I and Mw = 0
=m6epwkvw
dMw
dt
= 5m•
K(1)So4
dsø
dt
- n d2Zw
dt 2
(16) equilibria
equilibria
and they
under this analysis.
are therefore
Slash-and-burn
unstable
EAGLESO• 1 1•
Stability to Climate Change solar energy there are three equilibrium states:
closed forest, grassland, and a tree-grass
Consider Fig. 3 in which for K = 1 and for mixture. Only the last of these appears stable
representative constant values of R and S/• with respect to perturbation in the vegetation
various solutions of the equilibrium equations components but it is roetastable with respect to
(6) and (9) are presented. The dashed curve climate change.
represents Mw(so) as given by (6) and is
independent of the climate parameter G. The Acknowledgments. This work was supported by
solid curves represent Mg(So;G)as given by (6) the National Science Foundation under Grant No.
and (9). The horizontal line G = 1 is the only ATM-8114723. Prior publication of these results
solid line that also satisfies (12) and thus its has been made first, in a more complete form by
intersection with the dashed line represents the the American Geophysical Union [Eagleson and
intermediate tree/grass equilibrium discussed Segarra, 1985], and second, essentially as
above. presented here by NASA [Eagleson, 1986] .
For G > 1 we see that Mg increases with so
until s o reaches the value (s o = 0.33 for
G = 1.2) at which Mw becomes unity. At this References
Kevin Trenberth
Abstract. A brief outlineis givenof the TOGA (TropicalOceans 3} Modeling:Atmospheric, with specifiedseasurfacetemperatures
andGlobalAtmosphere} Programalongwith a moredetaileddiscus- (SSTs);oceanographic, with a specifiedsurfaceatmosphere(or
sion of the relationship of the atmosphericcirculation to sea surface fluxes);andcoupledocean-atmosphere models.
temperatures {SSTs}in the tropics. The best knownphenomenon 4) Prediction: Statistical-dynamical,
andusingmodels.
that is part of TOGA is E1 Nifio-Southern Oscillation(ENSO} and 5) Data Management
this paper focuseson the atmosphericcomponentof ENSO. The sim- Theseaspects
are dealtwith in-depthby WCRP (1985a,b) and
ilaritiesand differencesamongdifferentENSO eventsare reexamined NationalAcademyof Sciences
(1986).
as seen through a Southern Oscillation index and indices of SST in The best known phenomenonrelevant to TOGA is El Niio-South-
the tropical Pacific. Variations from event to event are marked and ernOscillation(ENSO). In this paperthe focuswill be on the atmo-
phaselocking of ENSO eventsto the annual cycle is only weak. The sphericcomponentof ENSO and its link to the tropical SSTs and,
1986-87 ENSO has been especiallyanomalouswith regard to timing. in particular, why convectionoccurswhere it does. Many recent
However, it provides an excellent illustration of the link between at- studieshave usedthe tool of compositingto bring out the common
mosphericconvectionand SSTs. The reasonswhy convectionoccurs features
of pastENSOevents[e.g.Rasmusson
andCarpenter,1982];
where it does and the importance of warm water greater than 28øC [vanLoon,1984];[vanLoonand Shes,1985,1987]. Herewe will
are discussed. It is shown that there is a need to better understand
emphasizethe differencesamongthe eventsand focusespeciallyon
the atmosphere-oceanlinks in the tropics and a need to measuremore recent developments,the 1986-87 ENSO event, which turns out to
accurately and understand changesin SSTs. havesomeuniquefeatures,but which nicely illustratesthe relation-
shipsbetweenSSTs and atmosphericconvection.
Introduction
El Nifio-Southern Oscillation Events
117
Geophysical Monograph Series Understanding Climate Change Vol. 52
•o • C3
o • o o o
0 0 0 0 0 0 0 0
C) C) C) C) C) C) C) C)
Geophysical Monograph Series Understanding Climate Change Vol. 52
TRENBERTH 1 19
DARWIN SIP 2 UoN'rH nUNN,NGUSxN tent to which the phenomenonis phaselockedto the annual cycle, as
impliedin manystudiesfollowingthe landmarkpaperby [Rasmusson
,Me .,-'% andCarpenter,1982];and 2} the extentto whichthe eventsaresim-
2.5 toe0 , ilar or differ from one another through their life cycle. Subsequently,
------ Ig'2•
....... 198• s thesepoints are reexamined using SST data.
..,_'
1.5 Figure 2 showsthe three year sequencesof anomaliesof Darwin sea
level pressure,as 3-month running means, during all ENSO events
\ ,o since1939. Each eventis labelledby the centralyear [year 0) and
0.$ ',
extendsfromthe yearbefore(year-1)to the yearfollowing(year+1),
usingthe terminolog7 in Rasmussonand Carpenter. The large events
are shown in the top panel, the smaller events in the central panel,
and the compositeplus and minusone standarddeviation(•r} (ex-
-I.5
ß
cluding1986}is givenbelow. Of note is the lack of significance
of
the compositewhich is nowhere more than 2a from the mean. This
-2.5 I ........... i ........... is due mostly to the inclusion of the smaller events in the central
YEAR -1 YEAR 0 YEAR +1 panel, which are of the order of one •r events, both in the SO and
SST fields. However, another factor is that the phase locking with
3.5
;...........
;...........
the annual cycle is not very strong. Phase locking seemsreasonable
1863
in Fig. 2a, but the weaker events do not follow that pattern and the
2.5 ....... 1063
1986-87 sequenceis the most unusual on record. It did not really
---,0so ,/,,,
•,••/....
.......
begin until late in 1986, six months later than usual, and is the most
• 1.5 extreme case on record in May-July of year +1. It has been associ-
ated with a very poor summer monsoonin India, one of the worst in
several decades.
• 0.5 • /
[-o.,
'_
&.
-1.5
-2.5 • ...........
YEAR -]
• ...........
YEAR 0 YEAR +
.%_/
Although many studieshave used atmosphericand oceanicparam-
eter anomaliesin their statistical analyses,the link between the two
is complex. Instead, the simplest link appears to be between the
atmosphericanomaliesand the total SST field, especiallyin models
[Shuklaand Wallace,1983]. This is illustratedin a generalway in
•.5 ; ........... ; .........
Fig. 3 which showsthe annual mean SST and outgoing longwavera-
diation{OLR) fieldsfrom [Shea,1086]and [Janowiak
et al., 1085].
2.5 Low OLR in the tropicsis associatedwith high top clouds,and OLR
is therefore an index of convectionin the tropics. The stippled re-
gionscorrespond
to the highSSTs(waterwarmerthan 28ø0) and
lowOLR (the convergencesones).Thereis a closerelationship
over
the oceanswhich is alsopresent,althoughwith migration north and
,,<%,.,•c,.:.. ..;.;.>:.....>;.....;........:.......
south, in the individual seasons.
• .:•:•:•:•:•:•:•:•.•.•:•:•:.:•:•:•:•:•:•:•'
..:•.• '-,,•?•:•:•:•:•:•:•:•::;•.••: Low OLR valuesalso occur over the warm tropical continents. In
• •.s fact, the absenceof a tropical continent in the Indonesianregion is
.,<:•..{•:•:•:•:•:•:•:•:•:•:•:•:.
one thing that makesthe Pacific unique. The warmestwater on the
globe,averagingover 29øC, occursin the tropicalwesternPacificand
is actuallylocatedslightlysouthof the equator,corresponding to the
northernmost
part of the SouthPacificConvergence
Zone(SPCZ}.
YEAR -1 YEAR 0 YEAR + The 1986-87ENSO has beenrelatively modestby somestandards.
Fi•. 2. Sequencesof anomalies in sea level pressure aZ D•in The SST anomalieshave not been huge, or even closeto the magni-
over aH El N•o events s•ce 1939. Each tudesexperiencedduring 1982-83. However,they havebeen sufficient
central%e• (%e• 0). The l•er events•e to significantlychangethe location of the warmest water.
5, •d •he compositemean (excludh• 198•) plusand m•us one This is illustratedin Fig. 4 whichshowsthe total SST field and the
s•nd•d deviation •e • c. anomaliesfor April 1987. The pattern is typicalof that for January-
July 1987. Positiveanomaliesalongthe equatorialsoneof the Pa-
ThusENSO eventsare thosein whichboth an SO extreme(Dar- cificare only ~IøC and are largestin the east. However,in the Pa-
win pressure
high, Tahiti low} and an EN occurtogether. A point cific the warmest water of ~30øC is shifted from west of the dateline
madeby TrenberthandShea[1987]is that the SO andEN are not to ~170øW. The correspondingOLR fields in Fig. 5 show that the
necessarilylinked on a one-to-onebasis. Abovenormal SSTs can oc- SPCZ is shifted north and east, as is characteristicof ENSO events
cur without an SO swing, suchas during 197g. In particular, South [Trenberth,
1976],andmerges
with the ITCZ (InterTropical
Conver-
American coastalevents are not very closelylinked to the SO except genceZone}near 170øW.The maximumtropicalconvection in the
during the major El Nifios where SSTs in the central Pacific are af- westernPacifichas shifted from 150øE to 170øW, as is shownalso by
fected. This is an aspectillustrated later here. ENSO eventsare also the OLR anomalyfield. The OLR anomalies
exceed45 W/m2 and
referredto as 'Warm Events"by vanLoonandShea[1985,1987]. are highly significant. Note the closelink with the total SST field
WeuseDarwinas an indexof the SO to showtwopoints1} the ex- but the relationshipwith the SST anomaly field is lessclear.
Geophysical Monograph Series Understanding Climate Change Vol. 52
o 3OE 60E 90E ! I$OE I•0 15011 I•01,1 go• 60• •0•
go
6O
•0N
$05
605
•OS ß
0 30E 60E 90E
50N 5O
40N 4O
•:)N
EQ EO
20S
40S 4O
50S
TRENBERTH 121
o.e,
,• . ,•,.>• •& •, ,., . - o
•
0 LU
I I I •
n
•:::::::I '
•
•::• ,
::::::::. i,
• •.'.
• z z z o • • • z z z o u3 (/3 •3
o o o • o o o o o o b.i o o o
Geophysical Monograph Series Understanding Climate Change Vol. 52
Z Z Z C• 0'•
0 0 0 La 0 0 0
uJ U'• UJ
z z z o (•1 •1 (•1
o o • I•J o o o
Geophysical Monograph Series Understanding Climate Change Vol. 52
TRENBERTH •2•
The evolution of the SST field in several critical regions is moni- beencomputedrelativeto the overallmeans1970-1986(whichdiffer
tored by the U.S. National MeteorologicalCenter'sClimate Analysis somewhat fromthoseusedin the CAC ClimateDiagnostics Bulletin).
Center(CAC)throughaverages overregions
labeledNino1+2 (0. As in Fig. 2, there is someevidencefor similar evolutionof SST
10øS,90-80øW),Nino3 (5øN-5øS,150.90øW)andNino4 (5øN-5øS, anomaliesin each event but the differencesamong the eventsare also
160øS-150øW),
seeFig. 6. marked. Once again, phaselockingwith the annualcycleis not very
strong. The 1986-87event is quite modestalongthe coastof South
America(Fig. 7). Until the middleof 1987,it alsofeaturedmediocre
anomaliesin the Nifo 3 region,but by July anomalieswereexceeding
those of the 1982-83 event. Although the anomaliesin Nifo 4 are
small(Fig. 8), for 1987they are the largestby far of all the ENSO
11to' 133'w eventsshownand, as seenfrom Fig. 4, they are sufficientto move
160'E I 150'w I
the location of the warmest water into this region. Consequently,it
o'q NINO-4I NINO-3 is the modest anomalies in Nifio 4 which turn out to be critical to
the current event.
This makesa very strong casefor knowingthe details of the SST
patternswell. It is especiallyimportant to know SSTs accurately,
to better than 0.5øC, in regionsof water warmer than 28øC since
Fig. 6. Areasaveragedto produceSST indicesfor the Nifio 1+2 subtle changescan alter the region where the warmest water occurs
region and for the Nifo 3 and Nifio 4 regions. by thousands
of km. Gadgilet al. [1984]notedthat deepconvection
mainly occurswhen SSTs exceed 28øC.
"
øJ ßß vection tends to break out in that region and the convergencezone
moves back over the warm water. Then the wind is reduced over
the warm water and the surfacefluxes actually becomeminiraised,
JAN JUL • JUL JAN
which explainswhy it is not possibleto understand the answerto
YEAR -1 YEAR 0 YEAR this questionfrom a budget study standpoint. In the real atmo-
sphere,there are many transients, ranging over time scalesfrom
Fig. 7. Sequences
of SST anomalies
for the Nifo 1+2 region,for all individualconvection,to easterlywaves,to 40-50 day oscillations.
E1 Nifio events since 1970 plus 1939-41. We suspect,however, that the nature of the transientsin atmo-
Geophysical Monograph Series Understanding Climate Change Vol. 52
;,i-- /. ',•', - -. •
•' b ."/" ß
[- ,-'½
x.,J•.\
['l. ,(...'"•//'."•::]:
:1
• "
•' \.,; •,, _.k-.:"-' i:,-•..
V :/.. ; '.'•/ I
':A••...,
..../..,.. eeeeeß
ßße •
"-:
• '; /?....;
L; '."• \..'/
; : '..
ß':k.. ' /' ½•
Ii lillllllltlllllllllllll
..j
Illllllll :..:.
YEAR -1 YEAR 0 YEAR YEAR -1 YEAR 0 YEAR + 1
Fig. 8. Sequences
of SST anomalies
for Nifio 3 and Nifio 4 regions(seeFig. 6) for all E1 Nifio eventssince1970.
Deeper
•,ction
More More
Moisture Latent Heat
genc Release
Fig. 10. Cartoon illustrating how the atmospherefeels the high
SSTs when the region of convergence,given by the cumuluscloud,
is shifted away from the warm water by transients. The vertical
arrows indicate enhanced sensible and latent heat fluxes into the
Fig. 9. Feedbackcycle showingwhy warm water is important. atmosphereand the thin arrow showsthe atmosphericflow.
Geophysical Monograph Series Understanding Climate Change Vol. 52
TRENBERTH •25
sphericgeneralcirculationmodets(GCMs) is rather differentand National Academy of Sciences,U.S. participation in the TOGA Pro-
GCM transients tend to be dominated by the 'convective adjust- gram: A researchstrategy. National Academy Press, 24 pp, 1986.
ment" process. The impressionis that in GCMs, the convection Neelin, J. D. and I. M. Held, Modeling tropical convergencebased on
is tied to the warmest water even more strongly than in the real the moist static energy budget, Mort. Wea. Rev., 115, 3-12, 1987.
atmosphere. Rasmusson,E. M. and T. H. Carpenter, Variations in tropical sea
The m•in point to be made here then, is that the responseof the surface temperature and surface wind fields associatedwith the
atmosphereto SSTs is a highly nonlinearprocess.Sincethe area of SouthernOscillation/ElNifio, Mort. Wea. Rev., 110, 354-384,
1982.
fairly warm (>28øC) water is extensivea smaJlchangein SST can
alter the configurationand is capableof producinga major shift in Rasmusson,E. M. and J. M. Wallace, Meteorological aspectsof the
the locationof the convergence soneswith consequencesfor wherethe El Nifio/SouthernOscillation.Science,œœœ,
1195-1202,1983.
releaseof latent heat occurs. In turn, this changesthe atmospheric Shea, D. J., Climatological Atlas 1950-79: Surface air temperature
forcingof Rossbywavesand the teleconnections into mid-latitudes. precipitation,
sea-level
pressure,
andsea-surface
temperature{45øS
-
The needfor accurateSST fields,and the needto understandchanges 90øN),NCAR Tech.NoteNCAR/TN-I•69-I-STR,1986.
in SSTs can be seento be paramount, and thus these are central to Shukla, J. and J. M. Wallace, Numerical simulation of the atmo-
the TOGA program. However, it should be noted that other, in sphericresponseto equatorial sea surface temperature anomalies.
particular radiative, effectsare atso important. J. Atmos. Sci., 40, 1(•13.1(•30, 1983.
Finally,sinceatmospheric windsdrivethe oceanand, as discussed Trenberth, K. E., Spatial and temporal variations of the Southern
here,SSTsare critical in determiningthe atmospheric response,it is Oscillation. Quart. J. Ro•. Meteor. Soc., 102, 639-653, 1976.
essentialto considerthe coupledsystemin order to fully understand Trenberth, K. E. and D. A. P•olino, Jr., Characteristic patterns of
atmosphere-oceaninteractions. variability of sea level pressurein the Northern Hemisphere. Mort.
Wea. Rev., 109, 1169-1189, 1981.
Trenberth, K. E. and D. J. Shea, On the evolution of the Southern
Acknowledgments. I am grateful to Vern Kousky and CAC for Oscillation, Mort., Wea. Rev., 115,3078-3096, 1987.
supplyingme with the SST indexvalues. The NationalCenterfor van Loon, H., 1984: The Southern Oscillation. Part III. Associations
Atmospheric Research is sponsored by the NationalScience
Founda• with the trades and with the trough in the westerliesof the South
tion. Pacific Ocean. Mort. Wea. Rev., 11œ,947-954, 1984.
References van Loon, H. and D. J. Shea, The Southern Oscillation. Part IV. The
precursorssouth of 15øS to the extremes of the oscillation. Mort.
Branstator,G., Analysisof generalcirculationmodelseasurfacetem- Wea. Rev., 115, 2063-2074, 1985.
peratureanomalysimulationsusinga linear model. Pt. 1. Forced van Loon, H. and D. J. Shea, The Southern Oscillation. Part VI.
solutions. J. Atrnos. Sci., 4œ,2225-2241, 1985. Anomalies of sea level pressureon the Southern Hemisphere and of
ClimateDiagnosticsBulletin, Climate AnalysisCenter, NOAA, Wash- Pacificsea surfacetemperature during the developmentof a Warm
ington, D.C. Event. Mort. Wea. Rev., 115, 370-379, 1987.
Gadgil, S., P. V. Josephand N. V. Joshi,Ocean-atmosphere coupling WCRP, Scientific plan for the Tropical Ocean and Global Atmo-
over monsoonregions. Nature, $1œ,141-143, 1984. sphereProgramme. WCRP No. $, WMO/TD No. 64, 147 pp,
Janowiak,J. G., A. F. Kruger, P. A. Arkin and A. Gruber, Atla.s 1985a.
of outgoinglong-waveradiationderivedfrom NOAA satellitedata. WCRP, International conferenceon the TOGA Scientific Program.
NOAA Atlas No. 6, U.S. Dept. of Commerce, 44 pp, 1985. WCRPNo. 4, WMO/TD No. 65, 1985b.
Geophysical Monograph Series Understanding Climate Change Vol. 52
David Halpern
Abstract. Reliable estimates of the evolution Episodes of high SST in the equatorial Pacific
of large scale sea surface temperature (SST) vat- are strongly correlated with regional and, often-
iations up to several months in advance is a prim- times, global atmospheric circulation changes
ary goal of the 1985-94 Tropical Oceans & Global (Rasmusson and Wallace, 1983). The decade-long
Atmosphere (TOGA) Program. Since the beginning of (1985-94) Tropical Ocean & Global Atmosphere
the TOGA, significant innovative accomplishments (TOGA) program, which is one of six subprograms of
include (1) an increase in the quantity of in situ the World Climate Research Program, was formed
data and efficiencies of data management, (2) focus upon the coupled ocean-atmosphere system.
rapid distribution of real time ocean products, The primary aim of the TOGAprogram is the devel-
(3) effective utilization of global observations opment of an operational capability for dynamical
from satellites, and (4) assimilation of data into prediction up to several months in advance of the
an ocean general circulation model to simulate time averaged (month-to-season) anomalies of the
monthly mean features of upper ocean thermal and coupled tropical ocean-atmosphere system.
flow fields. TOGA accomplishments are demonstrat- Variations of SST in low latitudes are related
ed with a discussion of oceanographic conditions to the responses of upper ocean thermal and flow
during June 1987. fields to large scale changes in surface winds, in
addition to local influences due to air-sea mo-
Introduction mentum and surface heat fluxes. Prediction of the
onset of large scale, long period ocean-atmosphere
Episodes of anomalous warm surface water, which interactions up to several months in advance re-
typically exist for about l-year, occur in the quires knowledge of the evolution of equatorial
equatorial Pacific Ocean at irregular intervals SST variations on time scales of days to a
every 4-7 years. During the extremely warm epi- month. This time scale is short compared to the
sode of 1982-83, the monthly mean equatorial sea one applicable for middle latitudes. Philander
surface temperature (SST) rose above its climato- (1979) reported that basin wide density gradients
logical-mean monthly value by as much as 4øC. in the upper portion of a resting ocean would be
This high SST covered a wide area (perhaps 40 ø established in about a decade in middle latitudes
longitude by 10ø latitude) centered approximately and only weeks near the equator. Remote wind
along the equator, intensified the Hadley circula- effects in the tropical ocean are important on
tion, and displaced the Walker circulation (Ras- time scales of weeks because of the rapid propaga-
musson and Wallace, 1983). The upward, eastward tion of dynamic signals, such as Kelvin wave mo-
slope of the thermocline along the equator de- tion along a narrow equatorial zone (Knox and
creased (Halpern, 1987). Sea level decreased in Halpern, 1982).
the western equatorial Pacific, and increased by At the time of onset of E1 Nino of 1982-83,
more than 20 cm in the eastern region (Wyrtki, surface wind measurements from the World Meteoro-
1984). The strengths of the eastward flowing logical Organization volunteer observing ship
Equatorial Undercurrent and westward flowing South (v.o.s.) network and SST observations from v.o.s.
Equatorial Current were substantially reduced and satellite radiance retrievals were the only
(Halpern et al., 1983; Halpern, 1987). The zone oceanographic parameters recorded throughout the
of intense atmospheric convection, which normally equatorial Pacific and distributed as data or data
resided west of about 160øE, moved eastward and products in (oceanographic) real time. Within the
traversed the entire width of the Pacific (Liu, ocean sciences, real time means approximately 30-
1988). days (sometimes up to 60-days) of measurement.
Expendable bathythermograph (XBT) measurements of
Copyright 1989 by the vertical profile of temperature between the
International Union of Geodesy and Geophysics surface and about 450 m were made from a small
and American Geophysical Union. subset of about 25 ships within the vast v.o.s.
127
Geophysical Monograph Series Understanding Climate Change Vol. 52
network. but five •ears aao these data were •ot several publications (Table 1). This innovative
distributed in real time. To achieve the TOGA development of real time viewing of oceanic condi-
objective of dynamical prediction up to several tions will be discussed in this paper. The Paci-
months in advance, two activities required devel- f[c Ocean is emphasized because distribution of
opment: (1) an operational capability of record- oceanographic data products in real time has de-
ing, transmitting, and distributing data on sea veloped more rapidly there than in the Atlantic or
level, subsurface thermal and flow fields, and net Indian Oceans. This paper is not an exhaustive
air-sea heat and momentum fluxes; and (2) assimi- treatment of all oceanographic research withi. n the
lation of oceanographic data into an operational TOGA program.
ocean general circulation model (until a reliable
interactive ocean-atmosphere general circulation TOGA Real Time Oceanography
model becomes available) for simulation and pre-
diction of oceanographic conditions. The fullness of TOGA real time oceanography was
It was recognized immediately by TOGA scien- demonstrated at the XIX General Assembly of the
tists that the oceanographic data base required International Union of Geodesy and Geophysics
improvements. One of the perpetual contributions (IUGG) in August 1987 when a preliminary version
of TOGA will be the substantial increases in quan- of this paper was presented. Attention was focus-
tity and quality of the subsurface oceanographic ed upon June 1987 oceanographic conditions because
fields. TOGA fostered an awareness of real time real time data products become available with a
oceanography and of prediction of large scale, delay of about 1-month, making June the period
long period SST variations and related features closest in time to the IUGG Assembly when data
such as thermocline depth, zonal slope of thermo- would be available. Also, the warm episode in the
cline along equator, and anomalous behavior of Pacific, which began in 1986 (Bergman, 1987) was
surface current. This situation was virtually continuing, which provided an opportunity for
nonexistent in oceanography before TOGA. This too discussion of ocean dynamics associated with this
is an everlasting contribution of TOGA. As a moderate E1 Nino.
result of the emphasis upon rapid transmission of The 16 June - 1 July 1987 averaged SST anomaly
TOGA data, a number of monthly publications or pattern (Figure 1) resembled the August - October
bulletins regularly display oceanographic data. anomaly distribution of the composite E1 Nino
Examples are listed in Table 1. described by Rasmusson and Carpenter (1982), but
As an indication of the progress due to TOGA, the 16 June - 1 July 1987 maximum amplitude of 2øC
many oceanographic aspects of the evolution of the was more than 0.5øC larger. Levitus (1987) showed
warm episode of 1986-87 were described every month that the magnit,ide of the SST anomaly was depend-
from its beginning via data products displayed in ent upon the climatology and data analysis tech-
Bulletin Address
1. Climate Diagnostic Bulletin Climate Analysis Center/NMC, National Weather
Service, National Oceanic and Atmospheric
Administration, Washington, DC 20233, UNITED
STATES OF AMERICA
2. Climate Monitoring Bulletin - Bureau of Meteorology, National Climate Centre,
Southern Hemisphere P.O. Box 1289K, Melbourne 3001, AUSTRALIA
3. Climate System Monitoring Monthly World Climate Program, WMO Secretariat, Case
Bulletin Postale No. 5, CH-1211 Geneva 20, SWITZERLAND
4. Darwin Tropical Diagnostic Statement Bureau of Meteorology, P.O. Box 735, Darwin,
Northern Territory 5794, AUSTRALIA
5. ERFEN Boletin de Analisis Climatico Comision Permanente del Pacifico Sur, Calle 76
No. 9-88, Apartado 92292, Bogota, COLUMBIA
6. MEDS Realtime Data Monthly Marine Environmental Data Service, 200 Kent
Monitor Report/Drifting Buoys Street, Ottawa, Ontario K1A 0E6, CANADA
7. Monthly Report on Climate System Long-Range Forecast Division, Japan
Meteorological Agency, 1-3-40te-machi,
Chiyoda-ku, Tokyo, JAPAN
8. Oceanographic Monthly Summary National Ocean Service, National Oceanic and
Atmospheric Administration, 5200 Auth Road,
Camp Srings, bid 20746, UNITED STATES OF AMERICA
9. Veille Climatique Satellitaire Centre de Meteorologie Spatiale, B.P. 147,
Lannion 22303, FR•CE
Geophysical Monograph Series Understanding Climate Change Vol. 52
HALPERN 129
20ON
| I I J I /I I I I / I I ..l I I 120ON
•0øS • - • • • • • • • • • • • • • • • •0øS
140øE 160 ø 180 ø 160 ø 140 ø 120 ø 100 ø 80øW
Fig. 1. Sea surface temperature (SST) anomaly during 16 June - 1 July 1987. Climatologi-
cal-mean monthly SST was defined by Reynolds (1982). Contour interval is IøC and a dashed
line represents a negative value, which means that SST was lower than normal. The 15-day
SST product represents a blend of in situ, but not drifting buoy SST observations, and
satellite data recorded during the previous 15 days. Redrafted from a chart distributed
by the NOAA Ocean Product Center.
nique used to compute the climatology. Comparing ed with XBTs launched from the v.o.s. network. As
the SST anomaly distributions for the 15-day in- a result of the TOGA program, XBT lines are now
tervals before and after the 16 June - 1 July 1987 more uniformly distributed throughout the tropical
period indicated that the location of the SST Pacific. Nearly 10,000 XBTs are launched annually
anomaly was not constant. in the tropical Pacific. Although there seems to
Near surface currents are recorded by the move- be enough XBT observations to produce an adequate
ments of freely drifting buoys drogued at about 15 realization of the monthly mean thermal field,
m depth and tracked several times each day by the bimonthly data products are generated by Pazan et
Argos data collection and platform location system al. (1987). The depth of the 14øC isotherm, which
on-board NOAA polar orbiting satellites. This usually occurs near the bottom of the thermocline,
technique was used successfully during E1 Nino of is an indicator of thermocline depth. During May
1982-83 (Halpern et al., 1983). During June 1987 - June 1987 the thermocline along the equator was
there were 29 dr•fters afloat •n the 20øN - 20øS shallower (deeper) than a 4-year mean depth, which
Pacific (Figure 2). The comparison between the was computed from data recorded during 1979, 1980,
May 1987 surface current field and the June 1987 1981, and 1984, •n the western (eastern) region
distribution indicated that the near equatorial (Figure 3). Throughout the region 10øN - 10øS
eastward current observed in May between 140øW and westward (eastward) of ]30øW, the thermociine
110øW was now confined to a smaller region near depth was less (greater)than the mean depth.
130øW. The eastward flow in May 1987 may have In the tropical oceans a close relationship
contributed to the development of the SST anomaly exists between sea level and the depth of the
in June by advection of warm water from the west, thermocline. Along the equator, changes in zonal
a situation similar to that found during the 1982- sea level slope mirror, to a large degree, varia-
83 E1 Nino (Halpern, 1987). tions in zonal slope of the thermocline (Wyrtki,
Since about 1978, numerous temperature profiles 1984). Two sources of sea level data existed: in
within the upper 500 m have been routinely measur- situ and satellite data. Tide gauge measurements
20øS
I '•I I ! I I I I
140øE 160
ø 180
ø 160
ø 1410
ø
I I I I I I\• 20øS
120
ø 100
ø 80øW
Fig. 2. Trajectories of satellite-tracked drifter buoys drogued at about 15 m depth
during June 1987. Redrafted from a chart which appeared in the Climate Diagnostic Bulle-
tin.
Geophysical Monograph Series Understanding Climate Change Vol. 52
140øE 160• 180 ø 160 ø 140 ø 120 ø 100 ø 80øW in situ data was caused by the absence of station
data. The prominent -20 cm sea level departure at
about 5øN, 135øW was typical of the annual cycle
associated with the North Equatorial Countercur-
rent.
H•LPERN 131
(A)
140" E 160•' 180 •' 160" 140 120 100' 80 'W
• •. -10 -5
'• . -20 -15- _/ / _ /
I I I I I I I I I I I• I 20•S
(B)
20ON [ I I I I I I I I I I I I I 20ON
0 0
20•S I 20øS
(c)
20øN I i I ! I 20ON
20øS 20øS
140øE 160ø 180ø 160ø 140ø 120ø 100ø 80øW
Fig. 4. (A) Sea level during June 1987 estimated from tide guage measurementsrecorded at
island and coastal stations. This map represents departure of sea level from 7-year (1975
- 81) mean. Solid dots represent tn situ sea level stations. Dashed lines represent
negative values, which mean that sea level was lower than the mean. Redrafted from a
chart appearing in the Climate Diagnostic Bulletin. (B) Sea level anomaly during 1 June
1987 estimated from Geosat sea surface topography data using daily data, a 15-day decorre-
lation time, and a grid of 8ø longitude by 2ø latitude. Anomaly is defined as difference
from 1-year (April 1985 - March 1986) mean. Dashed contours represent negative values.
Redrafted from a chart which appeared in the Climate Diagnostic Bulletin. (C) Surface
dynamic height anomaly during June 1987 estimated from an ocean general circulation mod-
el. Anomaly defined as difference between monthly mean distribution and climatological-
mean distribution determined from Hellerman and Rosenstein's (1983) wind field. Dashed
contours represent negative values. Redrafted from a diagram kindly provided by Dr. Ants
Leetmaa (Climate Analysis Center).
are resolvable by primitive equation models run on cline depth and of the zonal pressure gradient
currently available supercomputers. Equatorial force in the upper ocean. Simulations of the
wave-like structures were simulated by Philander monthly difference of the depth of the 20øC iso-
et al. (1986) with a 33-km latitudinal by 100-km therm relative to the climatological-mean monthly
longitudinal model geometry near the equator. distribution computed with the Hellerman and Ros-
The depth of the 20øC isotherm along the equat- enstein (1983) wind field were computed by Leetmaa
or, which usually occurs in the middle of the (personal communication). Figure 5 indicated that
thermocline, is a suitable indicator of thermo- maximum departures from the mean may have been
Geophysical Monograph Series Understanding Climate Change Vol. 52
140OE 160ø 180 ø 160 ø 140ø 120 •' 100ø 80•W April 1987 the westerly wind anomaly was confined
o I i i i ! I I I / I I I i I -o to the 170øE to 170øW region, but in May the west-
1986[x •0_• • • •- %• • L I•10 -N
1986
erly wind anomaly increased in strength and longi-
tudinal coverage from 160% to 140øW (Figure
•,r"•J•/'•
D , '-'--:'C--'
30' ,- -D
6B). In June the westerly wind anomaly was limit-
-3 ed to the region from the date line to 100øWand
19871./ •0 -•10••.•,,•• '• 1987
- F the easterlies in the far western Pacific were
-M stronger than normal. If this situation had con-
" 30 20 ttnued and expanded towards the east, it would
-A signal the demise of E1 Nino conditions: this did
-M not occur and SST continued to be warm.
It is tempting to associate April and May 1987
surface wind patterns with the June 1987 oceano-
J-•{)
N,
-20
•k._ .,... •,5\ •('/..---/._..30
-20 •"•
•'•• - AJ graphic conditions.
that reduced
Conventional wisdom dictated
easterlies in May would produce in
S June a decreasedeast-west tilt of the thermocline
along the equator (Figures 3 and 5), elevated and
'-'l I I I I I I I I I I I I I I I o depressed sea level in the eastern and western
140øE 160ø 180ø 160ø 140ø 120ø 100ø 80øW
equatorial regions (Figure 4), respectively, and a
reduction in the strength of the South Equatorial
Fig. 5. Time - longitude section of the model Current along the equator (Figure 2). The east-
simulation of monthly mean anomaly of the 20øC ward direction of the buoy drifting near 0 ø, 130øW
isotherm depth. Anomaly defined as difference was anomalous because the South Equatorial Current
between monthly mean depth of 20øC isotherm and is usually well developed in June (Halpern,
climatological-mean monthly depth determined from 1987). This eastward motion was indicative of
a wind-driven ocean model using Hellerman and advection of warmer water from the west, which
Rosenstein's (1983) wind field. Dashed contours contributes significantly to the generation and
represent negative values, which mean that the maintenance of the anomalousSST pattern (Figure
20øC isotherm was shallower than normal. Re- 1). The reduced equatorial upwelling caused by
drafted from a chart which appeared in the Climate the westerly wind anomaly also contributed toward
SystemMonitoring Monthly Bulletin. the anomaly of high SST, though no measurements
exist to quantify this assertion.
HALPERN 133
(A)
t I I I I !
MHR HPR MHY JUN JUL
1987
(B)
140oE 160 180 160• 140• 120 100• 80•W
I I •1 7;rl / I I ! / • 7 I I I I I I I 20ON
/ , • , . • • • • • • • • - • t • • • •• ..... t_ L
20øS F:< i
140øE 160 180 160 TM,'- : ' ' 120 o "' ': -
•8. 6. (A)Time
varla[ions
offsurface
zonat
windcomponent
(• s-t) measured
at Christmas
Island (20N• I57"W)• which 1s one of several opera[ional• rear time ?OOAequa[oriat •acif-
tc 1sland and mooredbuoy stations. Dashed t•nes represent 25-year (t947-72)
cal-mean monthly zonat componentwind determinedby Wyr[k[ and Meyers (t976). D[asram
kindly producedby •aut Frei•a8 (Pacific Marine Envlronmen[atLaboratory). (B) Surface
wlndstressanomaly
durin8May1987. A windstressvectoris de•t½ed_½s
the w•ndcompo-
dynecm assumln8
a dra• coefficiento• t.4xtO• andair densityo• t.2 k8 m- ß Monthly
meananomal• deffinedas departure from 23-year (t96t -83) mon[htymeandls[rlbu[ton.
wes[erl• wlnd anomalymeansa reduction &n the speedo• the normally occurtin8 easterly
westward wind. Adapted •rom a chart which appeared in the Cttma[e Olasnos[[c Butte[tn.
monthlyaveragedsurface wind was demonstratedby than the total numberof v.o.s. surface wind ob-
Harrison et al. (1988). A 20% error in wind servations measuredthroughouthistory. Unfortu-
stress, which is about the minimum
error expected nately, these satellite scatterometers do not
with presenttechnology,
wasequivalent
to a 2øC yield accura_•e
estimates
of the windspeed
below
uncertainty in $ST. Large wind speed variations about 3 m s ß A new, innovative satellite scat-
occur on short time scales of 1- to 5-days (Figure terometer designed to measurethe tropical surface
6A). Th•
0.5 ms- , accuracy
isof
which monthly
mean
necessary
for v.o.s.studies,
climatedatato wind
•20%field
for with
wind a root-mean-square
lessthan3 ms-1accuracy
speeds of
is current-
is doubtful, except along a few well-traveled ly being considered for developmentby NASA.
shipping routes becauseof inadequatesampling A major obstacle remainingin the development
(Halpern, 1988). Rosati and Miyakoda(1988) ira- of ocean modelsfor simulation of SST is the par-
provedthe SSTsimulationby usingwindswith a 1- ameterizationof horizontal and vertical turbulent
day Nyquist period, which substantially reduced mixingsin the upper ocean. Philanderand Seigel
the underestimation of the net surface heat flux. (1985) used a vertical mixing parameterization
In the early 1990s, every day there will be based upon the Richardsonnumber. Rosati and
morethan 50,000 surfacewindvectorswith a reso- Miyakoda(1988) demonstratedthat a turbulence
lution of nearly 50-kmavailable from the National closure schemewas more appropriate than constant
Aeronautic and SpaceAdministration's (NASA)scat- eddy viscosity. Evaluating different mixing par-
terometer (called NSCAT),whichmaybe launchedin ameterizations is exceedinglydifficult becauseof
1993 on Japan's ADEOSsatellite, and from the the scarcity of suitable in situ measurements.
EuropeanSpaceAgency's(ESA)scatterometersched- Peters et al. (1988) reported substantial dtffer-
uled to be launchedin 1991 on the ERS-1 satel- ences betweenPhilander and Seigel's (1985) Rich-
lite. The number of satellite surface wind vec- ardson number formulation and Ln situ turbulent
tots to be recorded in one monthwill be greater mixing data recorded at 0ø, 140øW. However, the
Geophysical Monograph Series Understanding Climate Change Vol. 52
in situ mixing observations should not be applied tory, California Institute of Technology, under
beyond the time and place of their measurement contract with the National Aeronautics and Space
until more is learned about the causes of their Administration.
variability because dissipation rates were 100
times larger at night than during daytime. References
HALPERN 135
Hellerman, S. and M. Rosenstein (1983) Normal Philander, S. G. H., W. J. Hurlin and R. C. Pacan-
monthly wind stress over the world ocean with owski (1986) Properties of long equatorial
error estimates. Journal of Physical Oceano- waves in models of the seasonal cycle in the
•raphy, 13, 1093-1104. tropical Atlantic and Pacific Oceans. Journal
Knox, R. A. and D. Halpern (1982) Long range Kel- of Geophysical Research, 91, 14207-14211.
vin wave propagation of transport variations in Rasmusson, E. M. and T. H. Carpenter (1982) Varia-
Pacific Ocean equatorial currents. Journal of tions in tropical sea surface temperature and
Marine Research, 40 Supplement, 329-339. surface wind fields associated with the South-
Legeckis, R. (197Y) Long waves in the eastern ern Oscillation/E1 Nino. Monthly Weather
ß
MODELLING CLIMATE,
PAST, PRESENT AND
Geophysical Monograph Series Understanding Climate Change Vol. 52
J.-C. G•rard
Institutd'Astrophysique,Universit•de LiC•ge
B-4200Ougr•e-Li&ge,Belgium
139
Geophysical Monograph Series Understanding Climate Change Vol. 52
sphericrisein temperatureis associated with the absorp- tally frozen solutiondependson severalfactorssuchas the
tion of solarradiation shortwardof 300 nm. In the prim- global albedo-temperaturefunction and the tropospheric
itive atmosphere,the only sourceof oxygen•vasthe pho- lapserate (figure1). Saganand mullen(1972)initially
todissociation
of watervapor(Kastinget al., 1980). This suggestedthat large amountsof NH3, an active green-
mechanism wasshownto yield only very small(10-8) of housegas, would be able to elevate the surfacetempera-
the presentatmosphericlevel or PAL) amountsof free ture and compensatefor the past lower solar luminosity.
oxygenand negligiblequantitiesof ozone. Later, as pho- However,due to the short lifetime of ammonium(Kast-
tosynthetic activity began to develop at the ocean sur- ing, 1982) in the primitive atmosphere,this moleculewas
face about 3.5 Ma ago. (Schopf,1983), the oxidationof abandonedand CO2 appearedas the most likely infrared
the reducing seawaterprobably acted as an efficient sink absorber. Hart (1978) calculateda scenariofor the CO2
for atmosphericoxygen and preventedthe accumulation atmosphericlevel which is compatiblewith the absenceof
of 02 in the atmosphere. The simultaneouspresenceof global glaciation in the Precambrian. His work was based
ferrous and ferric iron observed in the banded iron forma- on a simplifiedglobal climatic model coupledto a geo-
tions(BIFs) is usuallyinterpretedasan indicationof the chemicalapproximation.This CO2 historywasfrequently
existence of smallamounts(< 10-6 PAL) of atmospheric used in conjonction with radiative-convectivemodels to
02 duringthe precipitationof the oxidizediron (Holland, test the validity of the thermostatic effect of this con-
1984, Fran(•oisand G•rard, 1986). Photooxidationof Fe stituent (Owen et al., 1979; Kasting et al., 1984; Kuhn
2+ionsmayalsohavebeena significant source of a abiotic and Kasting,1983;Kiehl and Dickinson,1987). The re-
formationof the BIFs (Bratermanet al., 1983;Fran(•ois, sults of Kasting et al. (1984) and Kiehl and Dickinson
1987 ). More recently,oxidationof continentalcrust and (1987)(figure2) showthat the combination of increasing
solarluminosityand decreasingCO2 mixing ratio is able
formation of red rocks contributed to limit the level of
oxygenin the atmosphere.Roughly400 Ma ago,the oxy- to maintain the global temperature of the planet within 4-
gen reachedapproximatelyits presentatmosphericmix- 15K. However,the treatmentby Owenet al. (1979)yields
ingratio(Cloud,1983).Therefore, mostof its an averagesurfacetemperature 3.15Ga ago approximately
throughout
hasvariedcontinuously 7 K warmer than Kiehl and Dickinson who found that the
history,the Earth's atmosphere
in composition and thermal structure. The abundance pressurescaling of the mean band halfwidth of the CO2
of otherminorconstituents
(N20, NO2,CH4,NH3) has bands at 961 and 1064 cm -1 accounts for most of the
probablyalso changedsubstantiallyduring the Earth's temperature difference with Owen et al.'s calculations.
historybut they appearto haveplayeda lesscrucialrole Adopting the samescenario,we usethe one-dimensio-
in the climatic evolutionof the planet. nal radiative-convectiveglobalmodeldescribedby Gdrard
Finally,if evidenceof a controlof solarcycleactivity and Francois(1988) and Francois(1988) to calculatethe
overthe climatehas remainedelusivein the corntempo- evolutionof the mean surfacetemperature. Briefly, this
rary atmosphere,precambriansedimentaryrockshavere- model solvesthe thermodynamicequationusinga for-
vealeda clear signatureof the ancientsolar variations. It ward time-marchingmethod. The absorptionof solar ra-
is therefore important to examine whether the different diation by CO2,H20, O3 and cloudsis calculatedusing
atmospheric
composition whichprevailedin the pastmay the Lacisand Hansen(1974) formulation,includingthe
havefavoreda strongerresponse
of the globalclimateto effects of Rayleigh scattering and surfacealbedo. The
solar forcing. cloud coveris characterizedby a singlelayer with fixed
Carbon Dioxide and Precambrian Climate altitudeand opticaldepth. Rossowet al. (1982)demon-
strated the important potential role of cloud feedbackon
The responseof the surfacetemperatureto variations the stabilizationof the Earth's climate duringits evolu-
in solar luminosity was examined in details with one- tion. In particular, they demonstratedthat the strong
dimensionalenergybalancemodels(North et al., 1981; negative cloud feedback present in their model is able
Endal and Schatten,1982). More recently,Gdrardand to partly compensatethe lower ancient sun luminosity.
Francois(1988) useda radiative-convective modelto test The greenhouseeffect due to infrared absorptionby O3,
the sensitivity of the climate responseto decreasesof N20 and CH4 is treated following the method by
the solarluminosity.In particular,they shoxved that, in manathan(1976) and Donner and Ramanathan(1980).
agreementwith energy-balance models,irreversibleglobal The contribution of water vapor is calculated from the
glaciations•vouldbe predictedif the solarconstantdrop- parameterization
givenby Sasamori(1968). The varia-
ped by a fexvpercent belo•vits presentvalue. The exact tion of the CO2 absorptancewith the CO2 columngiven
position of the discontinuityfrom the unfrozen to the to- by Kasting et al. 1984) is adopted. This expressionwas
Geophysical Monograph Series Understanding Climate Change Vol. 52
G•3RARD 141
6O
20
ß
ß ß
ß ß
ß ß
-2O
-6O
-100 I I I I I I I I I
0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6
RELATIVESOLARLUMINOSITY(L/L
O)
Fig. 1. Globalmeansurface temperature asa function
of theluminosity
relativeto thepresent one
for differentcases.In thefirstcase,the tropospheric
temperaturegradientis fixedto 6.5 K/km (full
line),whilein thesecond
curve
it isfixedtothemoistadiabatic
value(dashed
line).Forbothcurves,
the temperature dependence
of the surface
albedois adopted
fromGdrardandFrancois (1988)and
correspondsto an icealbedoof 0.5. In the third case(dash-dotted
line),a moistadiabaticgradient
is usedandthe albedo-temperature relationfromWangand Stone(1980)is adopted.The locations
of the jumpsto and from the frozensolutionsare alsoindicated.
obtainedby parameterizingresultsof laboratory absorp-the water vapor feedback, account for the differencesin
tion measurements. Whenever the temperature gradient ZXT8 . Overall, all models lead to the conclusionthat it
is possibleto find a plausible CO2 scenarioable to coun-
tends to exceed the moist adiabatic lapse rate, a con-
vectiveadjustementis appliedto restorethe convective teract the effect of the reduced past solar luminosity and
profile. This convective
adjustement is madefollowing thus to avoid global glaciation. Yet, it is important to
the methodof Manabe and Wetheraid(1967). Our cal- stressthat, at this point, the only detailed geochemical
culatedsurfacetemperatureT• alsoremainwithin5 K of model extendsback in time only 100 Ma ago (Lasaga
the value calculatedfor presentconditions,but the vari- et al., 1985). Theseglobalcarbonate-silicate geochemical
ationsare slightlylargerthan thoseobtainedin previous cyclecalculationspredict a significantincreasein the past
studiesIt is likely that differences
in the radiativecode, CO2 atmosphericcontent and a parallel rise in the global
and in the convectiveadjustmentscheme,amplifiedby surfacetemperature reaching 8 K, 100 Ma ago. Devel-
Geophysical Monograph Series Understanding Climate Change Vol. 52
GI•RARD 143
lO20 I I I i I ! I
288
• 286
1018
• 28•,
1016 282
__//•////, This study• 280
• -'-'--KQsting
.... Levine
et
al. •278-
(1981)
lO1•. I I I 1
10-•' 10-2 1 10-•. 10-2
0 2 LEVEL ( PAL) 02 LEVEL (PAL)
Fig. 3. Left ß dependenceof the ozone vertical column upon the oxygen level at the surface. Com-
parison is made with previousstudies. Right ß calculated variation of the global surfacetemperature
with the 02 atmosphericlevel.
drop by another 0.9 K is predicted when the 02 level in Responseof the Precambrian Atmosphere
the atmosphereis set to a vanishinglysmall value. Con- to Solar Cycle Activity
sequently,a global temperature decreaseof 8.(3 K is pre-
dicted if the oxygen was removed from the atmosphere, Analysisof periodicitiesobservedby Williams (1981)
all other conditionsbeing held constant. and Williams and Sonett (1985) in the annual deposits
The vertical distribution of the global mean temper- (varves)of a precambrianlakein SouthAustraliastrongly
ature calculation with this model levels of 02 from 10-4 suggesta solar control of the mean annual temperature
to 1 PAL are shown in figure 4. The progressivedevel- by solar activity. These (380 Ma old deposits were found
opment of the stratosphereis observedas the oxygenand on the Elatina formation in the Flinders Range, North
the ozone abundances are increased. At 0.1 PAL, the of Adelaide. They were formed during the Marinoan
stratospherictemperature peak is lower in magnitude and glaciation, one of the series of glacial. episodeswhich
altitude. This change is due to the decreaseof ozone in characterizethe end of the Precambrian era. The pale-
the upper stratosphere. By contrast, a small increase of ographic setting of the site indicates a marked seasonal,
the temperature is obtained in the region near 20 km as arid, periglacial climate. These circumstancesimply a
a result of the slightly enhancedozone concentrationsin strong seasonalcontrol of the glacier meltwater discharge
the lowerstratosphere.At 10-2 PAL of 02, no strat0- into the periglacial lake. Laminae characteristic of dis-
spheric maximum is predicted by the model due to the tal clasticvarveswere obtained, avering a total period of
low stratosphericozone density. As the 02 level is further 19, 000 va (varve years). Non-randomcyclicitiesin the
decreasedto 10-a PAL, an inflexionin the temperature thickness of the annual varves were found in the sam-
near 20 km is the only effect of vanishinglysmall 03 con- ple showing characteristic periods of 8 to 1(3va, with a
centrations.Finally, at 10-4 PAL, the role of ozonein the mean period of 12 years. Recent re-analysis by Sonett
thermal structure of the atmospherebecomesnegligible. and Williams (1987) indicatethat the actual averagepe-
It is also noted that, in spite of the important changes riod closeto 11.7 va. Longerperiodsof 22 and a 314 years
in the vertical temperature distribution, the top of the were also obtained in the Fourier analysis of the time se-
convectiveregion is located in each casebetween 10 and ries. Comparisonwith modern analoguesindicatesthat
14 km. the Elatina varve thickness may be interpreted as an in-
Geophysical Monograph Series Understanding Climate Change Vol. 52
•o
•..•
•'•. I
I !
/ x PAL PAL
\
\
\
l
/
.-.30 /
ß..10 PAl
..
'x
._-L' ' /
10 PAL' -.. \t
' ' 'x,\
o I I
120 lZ.O 160 180 200 220 2/,0 260
TEMPERATURE (K)
dicator of the mean annual temperature. Therefore, the Finally, 2 Ga-old varve recordsfrom the BIFs found
Elatina series can be consideredas a proxy of the time- in the Harmersleybasinin Australia, showclearevidence
variation of the mean annual temperature. of a 23.3 yr cyclewhich may be associatedwith the Hale
A weak solar influenceon glacial climate was also dis- (22-year)solarcycle.However,WalkerandZahnle(1986)
coveredin the analysis of a 236-year varve sequencefrom interpreted this periodicity as reflecting the climatic in-
Skilak Lake, Alaska (Sonett and Williams, 1985) . The fluence of the lunar nodal tide, which is weakly present
geographiclocation of the lake, the climatic environment in modern climate records with a period of 18.6 years.
and seasonalityof meltwaters make it a good analogy In this theory, it would arise from the precessionof the
with what appear to have been the conditions prevail- Moon'sorbital plane occuringwith a longerperiod at the
ing at the time of the Elatina deposition. However, the Precambrian as a result of the smaller Moon-Earth dis-
Elatina periglacial environment was dryer and probably tance in the past.
free of glacial ice whereas Skilak Lake is in a relatively The clarity of the solar-cyclesignaturein the Elatina
humid periglacial climate with permanent glaciers as a Formation and the remarkablelength of the recordmake
source of meltwater. A correlation study between the this set of data unique at the present time. Compared
varve sequencethicknessand the sunspotindicesbetween to the generalabsenceof solar signalin publishedvarve
years 1700 to 1930 reveals a good correlation between series,it is legitimateto wonderwhetherspecificenviron-
the two series with common periods of 11 and 22 years. mental conditionsprevailednear the end of the Precam-
By contrast, analysis of three time seriesof varves from brian era which couldexplain the specificityof the Elatina
the EoceneGreen River formation (Crowleyet al., 1986) responseto the solar cycle forcing.
showedonly a ;veak 11-yearsignalduring a restricted por- It is striking that the Elatina varveswereformeddur-
tion of the 7496-yearsamplecoredepositedapproximately ing a period characterizedby a seriesof glacialagesoccur-
50 million years ago. ring between0.9 and 0.6 Ma ago: a mostunusualclimatic
Geophysical Monograph Series Understanding Climate Change Vol. 52
GERARD
event. The distribution of continental masses is difficult models of UV solar cycle modulation were tested. The
to reconstruct for this remote period. However, paleo- first one (caseA), assumedno variability above200 nm
magnetic studies indicate that the three major phasesof and a max/min irradianceratio varying linearly from 2.6
extensive glaciations occured at low latitudes. Indeed, at 120 nm to 1 at 200 nm. The mean solar irradiance
most regionsof the Earth, with the possibleexception of distributionreportedby Brasseurand Simon(1981) was
Antarctica, containing Precambrian rocks show evidence adopted. In this case,the calculated surface temperature
for glaciation during this phase of the Earth's history variation remains negligiblysmall ( 0.03 K). In the sec-
(Crowley,1983,Christie-Blick,1982). More specifically,
a ond model, the solar cycle modulation extends to 300 nm.
recent study of paleomagneticmeasurementsof the rema- In this case(caseB) the surfacetemperatureresponseis
nent magnetization of the Elatina Formation was made muchlarger and exceeds0.5 K at 3 x 10 -3 PAL of 02.
by Embletonand Williams (1986). The resultsclearly A significant responseis only obtained in case A if the
demonstrate that the deposition was made at a paleo- variability ratio max/min is substantiallyincreased.
magnetic latitude of 5 degrees,in agreementwith previous Figure 5 illustrates the dependenceof Ts and the ozone
determinationsfrom other paleomagneticAustralian late vertical column on the 02 level for both cases. At low
Precambrianrocks(Huimin and Wenzhi, 1985). Thus, 02 levels, the peak of ozone is located near 10 km, inde-
the occurenceof ice sheetsand periglacial climate near pendently of the oxygen concentration. The magnitude
sealevel at low latitudes, consideredtogether with the ap- of the temperature increasesin parallel to the ozone col-
parent absenceof glaciation during the late Precambrian umn since the "greenhouse"fluctuation is progressively
is in itself, as a major enigma. The climatic seasonality enhanced.However,for 02 > 10-3 PAL, the ozonemax-
responsiblefor the deposition of varves is also difficult to imum and the altitude of the maximum ozone variation
explain if the region was located at low latitudes at this shift toward higher altitudes. This altitude changeof the
period. solar cycle responsetends to reduce the amplitude of the
The periodicities associated with the Elatina varves surfacetemperature responseuntil, at still higher levels,
have been interpreted as being of solar origin since the the sign of the Ts modulation itself is inverted. This de-
11-yr and longer periods known to occur in the mod- pendenceon 02 is complex and dependson the thermal
ern sun were found in the sediments. Could the differ- structure as well as the altitude of the ultraviolet energy
ent atmospheric conditions at the late Precambrian be deposition. Numerical tests also indicate that these re-
responsiblefor the larger climatic responseto the solar suits are nearly unaffectedby an increaseof the CO2 par-
forcing? A possiblemechanismlinking the global climate tial pressure. By contrast, it is found that during cold
and the solar activity cycle was analyzed by G•rard and climatic conditions,correspondingto a dryer troposphere
Francois(1987). In this scenario,the lowerozonelevelin and to a larger ozone troposphericcontent, the solar ul-
the atmosphere680 Ma ago and its different vertical dis- traviolet radiation penetrateslessdeepin the atmosphere.
tribution could possiblyhave made the tropospheremore Both effects contribute to increase the variation of the
responsiveto the 11-yr modulation of the UV solar irra- ozone column during a solar cycle. The model indicates
diance. Indeed, fluctuations of the temperatures associ- that the temperature variation responseincreasesby a
ated with the 27-day and 11-yr cycles are observedto- factor of 3.5 when the mean surfacetemperature drops
day in the stratosphere.(Keating et al. 1986; Chandra, from 18ø C to 0ø C. This factor may explain, in part, the
1984). It may be expectedthat in an atmospherepoorer specificity of the Elatina periglacial environment to the
in ozone and with an ozone peak closerto the ground, solar cycle activity.
the responseof the surfacetemperature would be ampli-
fied compared to the modern atmosphere. This idea was Summary
quantitalively examinedusing the coupledchemical-R-C
model described above. Since the Elatina varves were de- Energy-balanceand radiative-convectivemodels show
posited during a periglacialperiod, the model cloud layer that the past reducedsolarluminosity shouldhave gener-
is fixed at low altitude with an optical thicknessr = 10 ated global irreversible glaciationsas a result of the ice-
to produce surfacetemperaturesin the range 268-275 K, albedofeedback. Geologicand paleontologicevidencein-
about 15 to 20 K lower than for the present atmosphere. dicatesthe absenceof a totally frozen Earth at any stage
The responseof the global surfacetemperature to an of the planer'sevolution. Increasedlevelsof active green-
imposed variation of the solar UV irradiance associated house gases such as CO2 appear as the most plausible
ßrith the 11-yr solar cycle is strongly dependenton the compensatingfactor to maintain the global climate within
sp•tral distribution of the modulation. Therefore, two the limited range by palcoclimates. Numerical models
Geophysical Monograph Series Understanding Climate Change Vol. 52
I I
0.5 _.•.•. ß\
...... low attitude cloud layer.
cold climate and solar % ! \,,.....
ß
"'•
\\
•\
variabilityfor),< 200nm
( enhanced
20-fold
) •10 / //• -
LU
•8 i • :. •
',: • •
ß
o •,
o
LL,I
z
o 2 -
N
O
m -0.5 I ! o I I
11j
3 162 fi• 1 163 •2 16• 1
OXYGEN I_EYEL. (F'•L.) OXYGEN L.EYEL. (F'•L.)
Gi•RARD •47
Donner L. and V. Ramanathan, Methane and nitrous oxide: effects of enhanced atmospheric CO2 and CH4 on early
their effects on the terrestrial climate, J. Atmos. Sci., earth surface temperatures, J. Geophys. Res., 92, 2991-
37,119-124, 1980. 2998, 1987.
Embleton B.J.J. and Williams G.E., Low palaeolatitude of Kuhn W.R. and J.F. Kasting, Effectsof increasedCO2 con-
depositionfor late precambrianperiglacialvarvites in South centrations on surface temperature of the early earth.,
Australia: implications for palaeoclimatology,Earth Planet. Nature, $01, 53-55, 1983.
Sci. Letters, 79,419-430, 1986. Lacis A.A., J.E. Hansen, A parameterization for the ab-
Endal A.D. and K.H. Schatten, The faint young sun-climate sorptionof solarradiation in the Earth's atmosphere,J.
paradox: continental influences. J. Geophys. Res., 87, Atmos. Sci., $1, 118-133. 1974.
7295-7302, 1982. Lasaga A.C., R.A. Berner and R.M. Garrels R.M., An im-
Fran(;oisL.M., Reducingpowerof ferrousiron in the Archean proved geochemicalmodel of atmospheric CO2 fluctua-
Ocean, 1. Role of FeOH+ photooxidation,Paleoceanogra- tions over the past 100 million years, in The carbon cy-
phy, 2 , 395-408, 1987. cle and atmospheric C02' natural variations Arcbean to
Fran(;oisL.M., The effectof cloudparametrization and tem- present, E.T. Sundquist and ]V.S. Broeckereds., pp. 397-
perature profile on the climate sensitivity to ozonepertur- 411, GeophysicalMonograph 32, American Geophysical
bations, J. A tmos. Res., 21, 305-321, 1988. Union, •Vashington, 1985.
Fran(;oisL.M. and J.C. Gdrard, Reducingpower of ferrous Levine J.S. and R.E. Boughner, The effect of paleoatmo-
iron in the Archean Ocean, 1. Contribution of photosyn- spheric ozoneon surfacetemperature, Icarus, $9, 310-314,
thetic oxygen, Paleoceanography, 1, 355-368, 1986. 1979.
Fran(;ois L.M. and J.C. Gdrard, Ozone, climate and bio- Levine J.S., T.R. Augustsson,R.E. Boughner, hi. Natara-
spheric environment in the ancient oxygen-poor atmo- jan and L.J. Sacks, Comets and the photochemistry of
sphere, Planet. SpaceSci., in press,1988. the paleoatmosphere, in Comets and the origin of life.,
Gdrard J.C. and Fran(;oisL.M., A model of solar-cycleef- edited by C. Ponnamperuma, pp. 161-190, D. Reidel,
fects on palaeoclimate and its implictions for the Elatina Dordrecht, 1981.
Formation, Nature, $2t7, 577-580, 1987. Manabe S. and R.T. Wetheraid R.T., Thermal equilibrium
Gdrard J.C. and L.M. Francois, A sensitivity study of the of the atmosphere xvith a given distribution of relative
effects of solar luminosity changeson the Earth's global humidity, J. Atmos. Sci., 2•, 241-251 ,1967.
temperature, Ann. Geophys.,t7, 101-112, 1988. North G.R., R.F. Cahalan and J.A. Coakley Jr., Energy
Hart M.H., The evolution of the atmosphere of the Earth, balance climate models, Rev. Geophys. Space Phys., 19,
Icarus, 33, 23-39, 1978. 91-121, 1981.
Holland H.D., The chemicalevolutionof the atmosphereand Owen T., R.D. Cess and V. Ramanathan, Enhanced CO2
oceans Princeton University Press, Princeton, 1984. greenhouseto compensatefor reduced solar luminosity
Huimin Z. and Z. X'VenzhiZ., Paleomagnetic data, late pre- on early earth, Nature, 277, 640-642, 1979.
cambrian magnetostratigraphy and tectonic evolution of Ramanathan V., Radiative Transfert within the earth's tro-
eastern China, Precambrian Res., 29, 65-75, 1985. posphereand Stratosphere' a simplified radiative-convective
Kasting J.F., Stability of ammonia in the primitive terres- model, J. Atmos. Sci., $$, 1330-1346, 1976.
trial atmosphere, J. Geophys. Res., 87,3091-3098,1982. Rossow \V.B., Henderson-Sellers A. and Weinreich S.K.,
Kasting J.F., Theoretical constraintson oxygen and carbon Cloud feedbackß a stabilizing effect for the early Earth?,
dioxide concentrations in the Precambrian atmosphere, Science, 217, 1245-1247, 1982.
Precambrian Res., $•, 205-229, 1987. Sagan C. and G. lXlullen, Earth and •lars' Evolution of
Kasting J.F. and T.M. Donahue, The evolution of atmo- atmospheresand surface temperatures, Science. 177, 52-
spheric ozone, J. Geophys. Res., 85, 3255-3263, 1980. 56, 1972.
Kasting J.F., J.B. Pollack, T.P. Ackerman, Responseof the Sasamori T., The radiative cooling calculation for applica-
Earth's atmosphere to increasesin solar flux and impli- tion to general circulation experiments, J. Appl. Meteo-
cations for loss of water from Venus, Icarus, 57, 335-355, rol., 7, 721-729, 1968.
1984. Schatten K.H. and A.S. Endal, The faint youngsun - climate
Keating G.M., M.C. Pitts, G. Brasseurand A. De Rudder, paradox' volcanic influences, Geophys. Res. Left., 9,
Responseof middle atmosphereto short-term solar ultra- 1309-1311, 1982.
violet variations, 1, observations,J. Geophys. Res., 92, Schopf J.W., Earth's earliest biosphere:its origin and evo-
889-902, 1986. lution, Princeton University Press, Princeton, N.J., 1983.
Kiehl J.T. and R.E. Dickinson, A study of the radiative Soneft C.P. and G.E. Williams, Solar aflqnityof sedimentary
Geophysical Monograph Series Understanding Climate Change Vol. 52
cyclesin the late precambrian elatina formation, Aust. J. climate model, J. Atmos. Sci., 37, 545-552, 1980.
Phys. 38, 1027-1043, 1985. Wang W.C., J.P. Pinto and Y. L. Yung, Climatic effectsdue
Soneft C.P. and G.E. Williams, Frequency modulation and to halogenatedcomponentsin the Earth's atmosphere,J.
stochasticvariability of the Elatina varve record: a proxy Atmos. Sci., 37, 333-338, 1980.
for solar cyclicity?, Solar Phys., 110,397-410, 1987. Williams G.E., Sunspot periods in the late precambrian
Walker J.C.G. and K.J. Zahnle, Lunar nodal tide and dis- glacial climate and solar-planetaryrelations, Nature, 291,
tance to the Moon during the precambrian,Nature, 320, 624-628, 1981.
600-602, 1986. Williams G.E. and C.P. Soneft, Solar signaturein sedimen-
'•VangW.C. and P.H. Stone,Effect of ice-albedofeedbackon tary cyclesfrom the late precambrian Elatina Formation,
globalsensitivityin a one-dimensionalradiative-convective Australia, Nature, 318, 523-527, 1985.
Geophysical Monograph Series Understanding Climate Change Vol. 52
Eric J. Barron
149
Geophysical Monograph Series Understanding Climate Change Vol. 52
60 ,
3,0
0
30
60
Cretaceous warmth (Barron et al., 1981a; Barron Climate Models and Cretaceous Temperatures:
and Washington, 1984) demonstrated that a The Major Discrepancies
hierarchy of models fail to achieve Cretaceous
warmth given paleogeography as the primary The initial efforts to determine the role of
forcing factor. Cretaceous geography (Figure 1) in explaining
However, these contributions did not refute Cretaceous temperatures (Figure 2) focussed on
hypotheses that geography explained Cretaceous experiments with an Energy Balance Model
warmth. Rather, they left as an open question (Barron et al., 1981a) and with a mean annual
whether the discrepancy between the model General Circulation Model (Barron and
experiments and climatic reconstructions could Washington, 1984).
be explained because of (1) model limitations, Barron et al. (1981a) utilized a model based
(2) the need for climatic forcing factors in on a zonally averaged energy balance of the
addition to geography and/or (3) the need to vertically integrated earth-atmosphere system
reinterpret Cretaceous paleoclimatic data. computed for the annual cycle. The energy
This contribution summarizes a series of transported poleward was approximated as a
subsequent efforts to demonstrate or eliminate diffusion process. The zonally averaged land
one or more of these points as an explanation distribution was derived from Figure 1. Two
for the discrepancy between model results and conclusions were apparent. First, without
climatic data. At the end of this effort the major prescribed changes in cloud cover or some
same three questions are pertinent, however the additional net global heating, the Cretaceous
evidence is clearer: (t) geography is a planetary temperature could not be achieved (a
substantial climatic forcing factor but (2) 6-12øC increase in comparison with the present
geography is unlikely to be the sole day). The EBM achieved a 1.6øK warming as a
explanation for Cretaceous climate. function of geography alone. Second, in each
Geophysical Monograph Series Understanding Climate Change Vol. 52
system. In fact, the role of ocean heat of a simple thermodynamic equation for heat
transport in climatic change is rather poorly storage in an ocean layer 50 meters thick (see
known (see review by Covey and Barron, 1988). Washington and Meehl, 1984 for additional
However, for the specific case of the Cretaceous discussion). The heat flux into the ocean is
a series of sensitivity experiments with the CCM given by the surface energy balance.
were completed by Schneider et al. (1985) to The results of this experiment are in contrast
evaluate whether a greater role by the ocean with the results for the mean annual simulation
could explain the model-observation discrepancies with Cretaceous geography described by Barron and
cited above. Washington (1984). First, the globally averaged
The Schneider et al. (1985) experiments did surface temperature warming due to the change in
not utilize a dynamic ocean circulation model geography and removal of permanent ice caps is
coupled to the CCM. Instead, extreme near 3.0øC in the annual cycle experiment
assumptions on the efficiency of ocean heat compared to 4.8øC in the mean annual experiment.
transport and how these assumptions could alter Interestingly, using the same two models, the
sea surface temperature were used to specify sensitivity to a doubling in atmospheric CO
sea surface temperatures in CCM experiments. (Washington
andMeehl,1984)wasgreaterin2the
The experiments were not designed to be annual cycle experiment (3.5øC) than in the mean
realistic, but rather to evaluate the impact of annual experiment (1.3øC). The greater
a greater role by the ocean and reduced sensitivity of the model to higher CO• levels
equator-to-pole temperature gradients on the evidently occurred because of the effect of
climate. In particular, a greater role by the incorporating the mixed layer on the latitude of
oceans and warmer polar oceans (even if set at the sea ice margin.
20øC) did not prove sufficient to maintain high The primary question to be addressed is why
latitude continental regions above freezing in the annual cycle experiment with Cretaceous
winter. This result is in apparent contrast to geography •llustrated less sensitivity than the
data of extreme warmth and lack of winter mean annual model. The seasonal cycle
freezing at high latitudes. In fact, experiment resulted in a greater discrepancy
experiments with an assumed or implied very between model results and observations. A
efficient ocean heat transport resulted in series of sensitivity experiments demonstrated
substantially colder continental interiors in which aspect of Cretaceous geography resulted
winter. This result occurred because of a in the warming in the mean annual experiments
decreased vigor of the atmospheric circulation, (Barron and Washington, 1984). Reduced
which in turn provides insufficient northern hemisphere land area at high latitudes
ocean-to-land atmospheric heat transport to in comparison with the present day explained
mitigate mid-winter continental radiative the majority of the Cretaceous northern
cooling. hemisphere warming illustrated in Figure 3.
If the above results are valid, a greater role Given mean annual insolation and an equilibrium
by the oceans is unlikely to solve the problems simulation for an ocean without heat capacity,
presented by comparing Cretaceous climate model the differences between land and ocean surface
experiments w•th Cretaceous temperature data. characteristics dominate the model-derived
northern hemisphere warming. Surface
The Role of the Seasonal Cycle in temperature isotherms are displaced poleward
Explaining Cretaceous Temperatures over land and ocean due to a warming related to
the decrease in total land area.
The role of the annual cycle of insolation was The seasonal simulation included ocean heat
not addressed in the first general circulation storage. In this case the difference in land and
model experiments •o investigate geography as an ocean surface characteristics is much less
explanation of Cretaceous wa•nth. Crowley et al. significant than is the thermal inertia of the
(1986) emphasize that, in particular, summer system. Summer heat storage and winter heat
temperatures may be a critical parameter in loss result in northern hemisphere ocean mixed
initiating glaciation. Depressed summer layer temperatures which are not substantially
temperatures on high latitude continents may different from the present day control
allow winter snow to remain throughout the year. experiment. The increased area of ocean
Of further importance, few geologic indicators resulted in a zonally averaged temperature at
respond to mean annual conditions. Seasonal mid to higher latitudes in the northern
i.nfo•at•on is essential to compare model results hemisphere which is cooler in summer and warmer
with observations and to ascertain the real in winter. This model response is directly
nature of any discrepancy. comparable to the change in land area (Figure
A full annual cycle simulation was performed 4) or the degree of continental flooding.
with mid-Cretaceous geography using the version In fact, the majority of the globally-averaged
of the CCM described by Barron and Washington surface temperature increase in the annual cycle
(1984) and discussed above, coupled with a experiment is in response to the removal of the
mixed layer ocean. The ocean model makes use Antarctic ice sheet. If this aspect is ignored,
Geophysical Monograph Series Understanding Climate Change Vol. 52
BARRON 1 53
% CHANGE IN LAND FRACTION warmth-are not restricted to the ocean and the
seasonal cycle. The discussions on the lack of
PRESENT DAY TO CRETACEOUS
agreement
betweenmodelsensitivities to a CO
2
I I I I ' 0 doubling described by Washington and Meehl
(1986) and Schlesinger and Mitchell (1987)
indicate a number of additional factors. The
- -10
two most important factors are (1) cloud
formulations and cloud-climate feedbacks and
(2) snow and sea ice parameterizations and
ice-albedo feedbacks of the control õimulation.
- -20
If the comparison of GCM studies for a CO_
doublingare indicative, the GCM
utilized2by
Barron and Washington (1984) can be
- -30 characterized by a comparatively low
sensitivity. In addition, this model has
little negative lapse rate feedback at high
- -40 latitudes in comparison with Manabe and
Wetheraid (1975). The model by Manabe and
Wetheraid (1975) had a stability-dependent
10-
- -50 vertical diffusion as do more recent versions of
• DJF the Community Climate Model. This high latitude
factor may be very significant for the degree of
polar amplification of warming. As pointed out
by Dickinson (1985) these factors, and the
ß ' JJA differences between climate models, suggest that
:• -10- potential model limitations cannot be eliminated
!
as a solution to the discrepanies between model
90 6'0 3'0 0 experiments and Cretaceous observations. Still
N
LATITUDE the results of the annual cycle simulation and
the lack of global temperature sensitivity in
ZONALLY AVERAGED comparison
with the samehierarchyof CO
2
experiments, indicates that other forcing
SURFACE TEMPERATURE DIFFERENCE factors, in addition to paleogeography, must also
CRETACEOUS - PRESENT DAY be considered as explanations of Cretaceous
warmth.
Fig. 4. Zonally-averaged surface temperature
(o Kelvin) differences (Cretaceous minus the
Are Other Climatic Forcing Factors Important
present day) for annual cycle simulations using
the NCAR CCM coupled with a mixed layer ocean.
With the advent of plate tectonic theory,
DJF - December, January, February average. JJA -
other climatic forcing factors which might
June, July, August average. The surface
explain the contrast between warm episodes and
temperature differences are compared with the
glacial time periods received little attention.
percentage change in land fraction for each 10 ø Atmospheric CO_ variations are the one
latitude belt from the Present day to the
significant exception. Variations in atmospheric
Cretaceous.
carbon dioxide levels were suggested by
Chamberlin (1899), Budyko and Ronov (1979) and
Fischer (1982), but received little credible
the change in geography resulted in a large quantitative or observational support. Then, the
change in the amplitude of the seasonal cycle, geochemical models of Berner et al. (1983) and
but little change in mean, global surface Lasaga et al. (1985), which are based on the
temperature. Evidently, the addition of the carbonate-silicate geochemical cycle, provided
annual cycle of insolation does not solve the much stronger support for the prospect of large
discrepancies between model simulations and variations in atmospheric CO_ levels.
Cretaceous observations. Further, this study Specifically thesemodelsli•ked plate tectonics
continues to develop the theme noted by Rind (rapid sea floor spreading), continental area
(1986), that model sensitivity to geography may (flooding of continents is equated with rapid sea
be quite different frommodelsensitivity to CO
2 floor spreading), higher rates of volcanic
variations. degassing (with rapid sea floor spreading),
geochemical weathering rates and fluxes, CO_
Other Model Factors levels andclimate. Thegeochemical
models
2
predicted a several-fold increase in atmospheric
Potential model limitations which have a
CO
2 during the Cretaceousdependingon a set of
bearinq on the problem of simulating Cretaceous assumptions.
Geophysical Monograph Series Understanding Climate Change Vol. 52
i I i I I / exception.
Alligator-rela%ed
reptiles,
lake
310
BARRON
Cretaceous isotopic record, the occurrence of The model studies summarized in this
substantial ice is unlikely or can be rejected. contribution suggest that Cretaceous geography is
If substantial ice enriched in 0-16 were present, an insufficient forcing factor because (a) the
the ice would be detectable in the oceanic oxygen magnitude of the global warming is insufficient
isotopic record. Warmth at the Antarctic margin and (b) the degree of polar amplification is
and the oxygen isotopic record support a case for insufficient. At least two of the major
the lack of permanent ice on Antarctica. limitations in the model studies (the role of the
A number of cases can be presented which favor oceans and of the annual cycle) were addressed
some permanent ice or at least seasonally without abrogating these deficiencies. However,
subfreezing continental temperatures, either of the strength of the negative response to the
which would substantially reduce the nature of first question is still constrained by model
the problems described here. limitations. Three limitations have been
First, waxing and waning of ice sheets identified: (1) surface-albedo feedbacks, (2)
produces fast and large variations in sea level. cloud-climate feedbacks, and (3) polar lapse rate
We might expect that the character of the record feedbacks. In addition, the nature of the
of sea level variations for glacial and non- sensitivity experiments which attempted to
glacial climates would be different. However, examine the role of the oceans is far too crude
differences are not noted in reconstructions of to eliminate completely a different role by the
sea level by Haq et al., 1987 or Vail et al., oceans as a major causitive factor in explaining
1977) throughout the last 100 million years. Cretaceous climate.
Second, the Antarctic warmth is restricted to Cretaceous geography might also be the major
data from the continental margin. Numerous areas explanation of Cretaceous climate if the nature
(e.g. New Zealand) can be cited as examples of of the Cretaceous warmth is mis-interpreted. Two
coastal, maritime warmth adjacent to interior key climatic characteristics have been identified
glaciers. Certainly, the condition of coastal in this regard: (1) the amplitude of the
warmth on a large polar continent is not seasonal cycle in continental interiors and (2)
necessarily indicative of interior conditions. the nature of polar climates especially in
Third, the oxygen isotopic effect could be winter. In both cases, the data are either
arguable if ice caps were small or the product of sparse or inferential. The Cretaceous was
evaporation from adjacent warm oceans, with a substantially warmer than the present climate,
short path length through the atmosphere (little but the condition of cold continental interiors
distillation and fractionation of isotopes) and during winter at higher latitudes and permanent
if Antarctica was characterized by little ice in the interior of Antarctica would
topographic expression (see Covey and Haagenson, substantially reduce the key differences between
1984, for a discussion of variables which model experiments and observations. The model
influence oxygen isotopic composition of snow and studies in this case have directed attention at
glacial ice). Certainly, the oxygen isotope the two weakest points in the Cretaceous climatic
record does not eliminate isolated glaciers, and reconstructions.
might not eliminate small polar ice caps. If the stated model limitations and the
Even the condition of sub-freezing winter weaknesses in the climatic record are not the
temperatures and summer warmth (a mean near zero) answer to the problems associated with simulating
would substantially reduce the need for polar sufficient global warmth and the degree of polar
amplification of Cretaceous warming in model amplification of temperature then the alternative
experiments. However, all the evidence described is additional or alternative climatic forcing
above is circumstantial and speaks only to factors. The series of sensitivity experiments
possibilities. A substantial increase in indicate that alternatives must be considered and
knowledge from continental interiors and may be likely. The most plausible additional
specifically from Antarctica is required to factor is the possibility of higher COplevels.
determine if the nature of Cretaceous warmth has The level of CO_ required to simulate •ufficient
been correctly interpreted. globalwarmth(•-10x presentday) is within the
range suggested by geochemical models as
reasonable or likely. Even in this case,
Geography as a Forcing Factor however, the polar amplification of warmth may be
for Climatic Change problematic and the degree of tropical warming
becomes a critical factor.
The role of geography in climatic change The weight of the arguments support a
should be divided into two questions: (1) Is conclusion, with several caveats, that Cretaceous
Cretaceous geography the explanation for geography is insufficient to explain fully the
Cretaceous warmth and (2) is geography an observations of Cretaceous warmth. However,
important factor in climatic change? The answer geography remains a substantial climatic forcing
to the first question is probably negative and factor. A hierarchy of models yield an increase
the answer to the second question is almost in globally averaged surface temperature of
certainly positive. several degrees. The amplitude of the annual
Geophysical Monograph Series Understanding Climate Change Vol. 52
BARRON 157
Chamberlin, T. C., An attempt to frame a working Geophysical Monoqraph Series' 32, edited by E.
hypothesis of the cause of glacial periods on T. Sundquist and W. S. Broecker, pp. 397-411,
an atmospheric basis, J. Geol., •, 545-585, AGU, Washington, DC, 1985.
1899. Luyendyk, B., D. Forsyth, and J. Phillips, An
Covey, C. and E. Barron, The role of ocean heat experimental approach to the paleocirculation
transport in climatic change, Earth Sci. Rev., of oceanic surface waters, Geol. Soc. Am.
24, 429-445, 1988. Bull., 83, 2649-2664, 1972.
Covey, C. and P. L. Haagenson, A model of oxygen Manabe, S. and R. T. Wetheraid, The effects of
isotope composition of precipitation' doubling the COA2
concentration on the climate
implications for paleoclimate, J. Geophys. of a general circulation model, J. Atmos. Sci.,
Res., 89, 4647-4656, 1984. 32, 3-15, 1975.
Crowell, J. and L. A. Frakes, Phanerozoic Rind, D., The dynamics of warm and cold climates,
glaciation and the causes of the Ice Ages, Am. J. Atmos. Sci., 43, 3--24, 1986.
J. Sci., 268, 193-224, 1970. Sandberg,P. A., Nonskeletalaragonite andpCO
2
Crowley, T. J., D. A. Short, J. G. Mengel, and G. in the Phanerozoic and Proterozoic, in The
R. North, Role of seasonality in the evolution CarbonCycle and AtmosphericCO?• Na•ura•
of climate during the last 100 million years, Variations Archean to Present• Geophysical
Science, 231, 579-584, 1986. Monograph Series' 32, edited by E. T. Sundquist
Dickinson, R. E., Climate sensitivity, Advances and W. S. Broecker, pp. 585-594, AGU,
Geoph¾s., 28A, 99-129, 1985. Washington, DC, 1985.
Donn, W. and D. Shaw, Model of climate evolution Schlesinger, M. E. and J. F. B. Mitchell, Climate
based on continental drift and polar wandering, model simulations of the equilibrium climate
Geol. Soc. Am. Bull., 88, 390-396, 1977. response to increased carbon dioxide, Rev.
Emiliani, C. and N.J. Shackleton, The Brunhes Geophys., 24, 760-798, 1987.
Epoch' Isotopic paleotemperatures and Schneider, S. H., S. L. Thompson, and E. J.
geochronology, Science, 183, 511-514, 1974. Barron, Mid-Cretaceous continental surface
Fischer, A. G., Long-term climatic oscillations temperatures: Are high COA concentrations 2
recorded in stratigraphy, in Climate in Earth needed to simulate above freezing winter
History, edited by W. H. Berger and J. C. conditions?, in The Carbon Cycle and
Crowell, pp. 97-104, National Academy Press, Atmospheric
CO2' Natural Variations Archeanto
Washington, DC, 1982. Present• Geophys. Mono•r. Ser. 32, edited by E.
Frakes, L. A. and E. Kemp, Paleogene continental T. Sundquist and W. S. Broecker, pp. 554-560,
positions and evolution of climate, in AGU, Washington, DC, 1985.
Implications of Continental Drift to the Earth Spicer, R. A. and J. T. Parrish, Pañeobotanical
Sciences, edited by D. Tarling and S. Runcorn, evidence for cool north polar climates in
pp. 535-550, Academic Press, New York, NY, middle Cretaceous (Albian-Cenomanian) time,
1973. Geoloqy, 14, 703-706, 1986.
Frakes, L. A. and E. Kemp, Influence of Tarling, D. H., The geological-geophysical
continental positions on Early Tertiary framework of ice ages, in Climatic Chanqe,
climates, Nature, 240, 97-100, 1972. edited by J. Gribben, pp. 3-24, Cambridge Univ.
Hag, B. U., J. Hardenbol, and P. R. Vail, Press, Cambridge, 1978.
Chronology of fluctuating sea Levels since the Vail, R. R., R. M. Mitchum, Jr., and S.
Triassic, Science, 235, 1156-1167, 1987. Thompson, III, Seismic stratigraphy and global
Hays, J. and W. Pitman, Lithospheric plate changes of sea level, in Seismic Stratigraphy
motion, sea level changes, and climatic and and Global Chanqes of Sea Level, AAPG Memoir
ecological consequences, Nature, 246, 18-22, 26, edited by C. E. Payton, pp. 49-212, AAPG,
1973. Tulsa, Oklahoma, 1977.
Kasting, J., Comments on the BLAG model- the Volk, T., Feedbacks between weathering and
carbonate-silicate geochemical cycle and its atmosphericCO
2 over the last 100 million
effect on atmospheric carbon dioxide over the years, Am. J. Sci., 287, 763-779, 1987.
past 100 million years, Am. J. Sci., 284, Washington, W. M. and G. A. Meehl, General
1175-1182, 1984. circulation model CO• sensitivity experiments'
Kennett, J.P., Cenozoic evolution of Antarctic snow-sea ice albedo •arameterizations and
glaciation, the circum-Antarctic Ocean, and globally averaged surface air temperature,
their impact on global paleoceanography, J. Climatic Change, •, 231-241, 1986.
Geoph¾s. Res., 82, 3843-3860, 1977. Washington, W. M. and G. A. Meehl, Seasonal cycle
Lasaga, A. C., R. A. Berner, and R. M. Garrels, experiment on the climatic sensitivity due to a
An improved geochemical model of atmospheric doubling of CO
2 with an atmosphericgeneral
CO2 fluctuations over the past 100 million circulation moael coupled to a simple
years, in TheCarbonCycleandAtmospheric CO2i mixed-layer ocean model, J. Geophys. Res., 89,
Natural Variations Archean to Present, 9475-9503, 1984.
Geophysical Monograph Series Understanding Climate Change Vol. 52
SylvieJoussaume
•, JeanJouzel2, and RobertSadourny•
Abstract. Simulations of the Last Glacial well understood by simple isotope models. Howe-
Maximum have been performed with an atmospheric ver, the isotope atmospheric cycle is quite
general circulation model. We focus on a new ap- complex and is sensitive to the entire atmosphe-
proach of the problem with the modeling of ric circulation. To investigate the sensitivity
important climatic tracers: water isotopes and to a glacial/interglacial oscillation, AGCMs are
desert dust particles. The mean dependency of the only appropriate tools.
the water isotope content of precipitation with As regards desert dust particles, a large in-
temperature is similar to the present-day one, crease in the amount of dust in ice cores has
but can depart locally. The global amount of been observed for the Last Glacial Maximum
dust is little changed for the ice age, but im- [Petit et al., 1984]. Thus AGCMscan help to
portant changes are simulated over some regions. investigate the link between climate (aridity,
atmospheric circulation) and dust deposits.
Introduction
159
Geophysical Monograph Series Understanding Climate Change Vol. 52
..
. >- 2
Do
.,..
... - 4
III
-1& D- ID. - 8
Fig. 1. Simulated oxygen 18 ratio of precipitations for the estimated annual mean
(February + August) present day climate. Isolines every 2% 0 down to -12°/ , every
4% 0 for lower values. Dotted areas above -2 % 0 , hatched areas between -48 / 00 and
-8% ,
0
.
10.
_ 6•8o(O/oo) a 6•8o(O/oo)
-10.
-20.
_
-30. _
-40.
-50.
-60. -40. -20. O. 20. •0. -qO. -20. O. 20.
lies some caution when interpreting regional associated with changes in the atmospheric
results. circulation. The simulated change in dust depo-
sits over East Antarctica is weak compared to
Desert dust observations [Petit et al., 1981]; this discre-
pancy is more likely to be due to an under-
The simulated extent of arid regions, con- estimation in the change of source regions for
sidered as source areas of dust, is reasonable, the LGM.
except for an underestimation of the Australian
desert. Simulated modern desert dust results show Acknowledgments. This work was supported by
a clear seasonal cycle, with a double amount of the Programme National d'Etude de la Dynamique du
the atmospheric content of dust in August compa- Climat. The computer time was contributed by the
red to February [Joussaume et Sadourny, in Centre de Calcul Vectoriel pour la Recherche.
press]. During the Last Glacial Maximum, a small
increase is simulated by the model both for the References
source regions (+18%) and the atmospheric dust
content (+8%), with a stronger increase in Feb- Joussaume S., Mod•lisation des cycles des esp•ces
ruary, leading to a weaker seasonal contrast. isotopiques de l'eau et des a•rosols d'origine
However, important changes are simulated in some d•sertique dans un module de circulation
regions, for example over the tropical Atlantic g•n•rale de l'atmosph•re, Th•se de 3•me cycle,
Ocean, and over Europe (Figure 3) and are mainly Universit• de Paris VI, 1983.
•iii'
ß'•- <•',•
' ..'
....:b'ii::
::
.....ii!• -.• lO
=============================================
.....:::......
:::::;:
....................
g:....................
;2;.......
Fig. 3. Simulated total removal of dust (dry and wet) change ß ice age/present day
ratio for the estimated annual means. Isolines 1/2, 1, 2, 5 (shading for values above
1.)
Geophysical Monograph Series Understanding Climate Change Vol. 52
Joussaume S., J. Jouzel and R. Sadourny, A past 18,000 years, J. Atmos. Sciences, 43,
general circulation model of water isotope 1726-1759, 1986.
cycles in the atmosphere, Nature, 311,24-29,1984. Petit J.R., M. Briat and A. Royer, Ice age aero-
Joussaume S., Simulation of airborne impurity sol content from East Antarctic ice core
cycles using atmospheric circulation models, samples and past wind strength, Nature, 293,
Annals of Glaciology, l, 131-137, 1985. 391-394, 1981.
Joussaume S. and R. Sadourny, Simulation of the Rind D., Components of the ice age circulation,
desert dust cycle using an atmospheric general J. Geophys. Res., 92, 4241-4281, 1987.
circulation model,IAMAP Conference on Sadourny R. and K. Laval, January and July
"Aerosols and Climate", Vancouver 1987, publi- performance of the LMD general circulation
shed by A. Deepak, in press. model. In A. Berger and C. Nicolis (eds) New
Kutzbach J.E. and P.J. Guetter, The influence of perspectives in Climate Modelling, Develop-
changing orbital parameters and surface bound- ments in Atmospheric Sciences, 16, Elsevier,
ary conditions on climate simulations for the 173-198 1984.
Geophysical Monograph Series Understanding Climate Change Vol. 52
P. R. Rowntree
DynamicalClimatologyBranch,MeteorologicalOffice
Bracknell,England, RG12 2SZ
163
Geophysical Monograph Series Understanding Climate Change Vol. 52
ROWNTREE 165
a
1024
1032•
1024•
b
•,992 984
103 16
1024
1008
C _
!ooo
1024
1024 1016
1Ol
oo8
Fig. 1. December-February sea level pressure: (a) observed, based on Meteorological
Office (MO) operational analyses for 1983-86; (b) modelled, average for 8 winters
without gravity wave drag; (c) modelled, average for 4 winters with gravity wave drag.
long run of the model. The pattern of zonal mean simulation with a much higher resolution (40
winds is quite close to observations, though the level) model. One might also suspect radiation as
jets are too strong. Differences greater than 5 a cause. The ICRCCM(Intercomparison of Radiation
m/s are shownin Figure 3 -mostly they are Codesfor Climate Models) organised a valuable
confined to the stratosphere. The other contours comparison of line-by-line models and
here are for temperature errors - mostly these are parametrizations for climate models; Figure 5
less than 4 K except in the polar stratospheres. A comparestwo Line-by-line comparisonsof long wave
similar result is obtained in June to August cooling - GFDLand LMD. Though this shows
(Figure 4). This stratospheric coolness is not a generally good agreement, there are differences.
problem peculiar to our model - all three of the Apart from the coarse resolution, the MO model's
models used for CO2 experiments r•viewed by differences from the line-by-line calculations are
Schlesinger and Mitchell •1985] suffered from such of comparable magnitude; however there is too much
errors, especially in the Southern Hemisphere. A cooling from 60 to 370 mb. Note that these
possible cause is the rather poor resolution of schemesomit the water vapour continuum. Though
the stratosphere - we have only three layers above this is a poorly understood area of radiation
200 mb - in view of the important role of vertical theory, a better representation would not affect
motions in the stratospheric heat balance. the stratospheric cooling.
However,Mahlmanand Umscheid[1984] have error• Further intercomparisonprogrammes are underway
approaching 10 K in the January of their - for boundaryand surface fluxes, as discussed by
Geophysical Monograph Series Understanding Climate Change Vol. 52
200
600
90N 60 30 0 30 60 90S
Latitude
200
600
90N 60 30 0 30 60 90S
Latitude
Fig. 2. December-February zonally averaged zonal winds: (a) modelled 4 year average;
(b) observed 3 year mean (1983/4-1985/6). Contour interval 5 m/s; easterlies shaded.
MeBean (this symposium). In this context, it is stress on the ocean surface, averaged over the
r•levant to •xamlne the surface winds which are of year, compared with Hellerman and Rosenstein's
importance for driving an ocean model. Figure estimate from observed data. There are
shows the zonal mean of the eastward surface differences, but mostly these probably owe as much
Geophysical Monograph Series Understanding Climate Change Vol. 52
ROWNTREE 167
200 k----1I---+---+-+:i=--'""""+-'-+--+
600 F4rl-----1r--+---l\--#-+---+---+--+--+-----lf----+--+---+---It--H-----1r--I-'~
90N 60 30 o 30 90S
Latitude
Fig. 3. December-February zonally averaged model errors for temperatures (solid) and
zonal winds (dashed). Contour interval 2K for temperature, 5 m/s for winds.
to uncertainties in the data and the differences through Arctic Canada. The result (Figure 8) was
in the stress formulation as to errors in the disappointing with too strong a summer snow melt
model winds. An exception is probably near 50-60° peak and too 11 ttle the rest of the year.
S, where the model's peak westerlies are a few However, the model is not necessarily so wrong in
degrees too far from the pole, and north of 60° N, high latitudes. Another comparison (Figure 9) we
where some easterly bias is present. made was for the region of Siberia from 60-105° E
Rainfall (Figure 7) is important for north of 50° N, which closely apprOKimates the
agriculture over land and over the ocean because basins of the Ob'Irtysh and Yenisei. Here, the
of salinity's role in driving ocean currents. model had nearly 50% too much run off, but the
This simulation, though not as good in several apportionment through the year is close to
respects as some of our earlier ones, nevertheless observed.
serves to demonstrate that broadly the patterns An important need in climate change studies,
are realistic. The 'desert' areas with less than especially those involving CO 2 , may be sea level
1 rom/day and the rainy areas with more than 5 information. To obtain this, it is essential to
rom/day are broadly similar in the simulated and have an ice 'sheet model and a prerequisite for
observed maps. Features which were worse here that is realistic snow accumulation. This is
than in most of our experiments included several obviously very dependent on getting precipitation
parts of the tropics, for example the excessive right but other factors are also important, as
rain over South America and northern Australia and shown by the ice budget analysis for the North
the deficient rain over the tropical Atlantic and American ice sheet at the last glacial maximum in
southern tropical Africa. For many purposes, it Manabe and Broccoli [1985]. Their results show
is also important to simulate the runoff well - we the sublimation to be nearly half as large as
have recently started to calculate runoff for snowfall. We have calculated snow budget data for
selected basins, simply summing the runoff of grid Antarctica and compared it to the estimates by
bOKes within the catchments. One basin we Schwerdtfeger [1970] (Figure 10). The broad
considered was the Mackenzie River which runs pattern of a minimum of less than 5 g cm- 2 yr- 1
Geophysical Monograph Series Understanding Climate Change Vol. 52
2OO
600
90N 60 30 0 30 60 90S
Latitude
over most of eastern Antarctica6 with peak values northwest Atlantic and cool northeast Atlantic.
near the coast of over 40 g cm
-= yr-1 in places, Most of these errors are, I think, not well
is simulated quite well. The South Pole understood.
apparently has too little - there is a maximum on
the south edge of the Ross Ice Shelf, but like Future
much of western Antarctica, magnitudes are too
small. Now I want to touch on a few of the many
The results discussed so far are from challenges and problems in modelling climate.
atmosphere models. It is obvious that the Table 4 indicates some likely future developments
simulation of the surface climate from coupled
models cannot, except through chance cancellation
of errors, be expected to be as good as that for
atmospheric GCMs, which use climatologically TABLE 4. Future Developments, Problems
correct sea surface temperatures (SSTs) and sea
ice. So the crucial question is how good are the
SST and ice simulations in coupled models - I will (a) Ocean: - eddies
discuss only SST, referring to Han et al. [1985]'s Tropical (El qino)
useful comparison of annually averaged SST errors. Extratropical (glacial cycles)
Their maps show rather large areas with errors (b) Cloud: - water content -> albedo
exceeding 4 K -i.e., greater than the largest (c) Mountains: - elevation
observed anomalies - and there is much in common (barrier or heating)
between the simulations which is only partly gravity waves
disguised by the range of up to 2 K in the mean (d) Land surface: - soil, etc datasets
errors. These common features include a warm (e) Other possible developments:
Antarctic Ocean and eastern southern hemisphere vegetation models
ice sheet models.
oceans, cool water in the northwestern tropical
and sub-tropical oceans, warm north Pacific and
Geophysical Monograph Series Understanding Climate Change Vol. 52
ROWNTREE 16 9
GFDL
200
variations of North Atlantic circulation, for
example around the Younger Dryas period, suggest
another delicate balance.
600
_-I
oo_- ,
I), - so omit a possibly important
as suggested by Somerville
albedo tends to increase
negative feedback
1.5
.. == o.s /
I
90 ø N
.........
........
.,.,./.` o
90 øS
-I.0
Fig. 6. Zonally averaged annual mean zonal wind stress (sea only), observed (Hellerman
and Rosenstein, 1983) (dotted), and modelled (• year mean for 80 mode•) (solid).
Geophysical Monograph Series Understanding Climate Change Vol. 52
b ::::::::::::::::::::::::::::::::::
::•::..-...••:;.:-•
'•--•'.-'-•'
' -.-•.•
'....''
:--.-.. :i•::i:::"
'• •::::::::::::::::....•:::::::::::::::::
:::::::::::::::::::::::::::::::
...........-...-.......-...-..•........••.•
ß'.'.'.'.'.'.'.'.'.'.'.'.'.'..• .'.'.' ß ß•.-.-.-...---•......................•-
ß ß '.'.'. ß ß ß -•.... .... • - j •
..2 ••...-.-.-...-.-.•.
ß .'.'.'.' ' ' ----•.....-.-..•...........-•
ß .'..'.'.:.:-:.:.......- ... -.-. '-', ..•=•,•.-.......-.-.-.-...-.
..... .'.'-'.'-:-:..'.:-:-:.
ß...=====================
:::::::::::::::::::::::::::::::
.........•
;'.... '- o -:-,.....
:::::::::::::::::::::::
======================
:.'.-
"'-'•1• '•) liiii!!ii!::.:!ii;,•.fc---'• '• '• $ • '__ '.-:-'•:::'•i:i•i:i:i:i::':'•f.:
:::::::::::::::::::::::::::::::::::
::.,•.
)_.."i-,
• • 0":'::
i::
i:::::
:'-"
?.0-.•.••
••t•"'"•i•;;•
...•...]
i '"::i:i:i:i:i:i:i:i:i:i:!
•-• '.i:i:i:i:i:i:i:i:!:i:::i:::'
:::::i:i:i:i
'.i''
''"'
:"•
'f': :"".'•
..............
, =========================
.• .:.:.:.:.:.:.:.:.:.:.:.i:i:i:i:i::
<.:" 2 •--_ - ? . . ;.
.::'.!:!:!:
::::::::::::::::::::::::::::::::::
.ii:::i:
-
:- o
.':•" _,,'--
•i!i!!.!.!.!..!.!.........
""::.. ":':':':'"
":::i:i:!:i:!
i:i:':
..i:i:i :,?'-',
•: _
-•
"'.'.'-:-:-:.:.:-:-:-:-:.:.'..i'
:•"
'"'::{!!iiiiii
•
iiii::i!•'
-,
============================i ========================:••
:::::::::::::::::::::::::::::::::::::::::
ß ' ' '--",,/ :::::::::::::::::::::
i:i:i: --" fi/'/' .....:.:.:.:::::::
!:!:!:i:i:i:i:i:i::'
ROWNTREE 171
,oF
40
Model
"20
•Observed
I0
o
Aug Oct Dec Feb Apr Jun Aug
Fig. 8. Modelled (solid) and observed (Mackay and Loken, 1974) (dashed) runoff
for the Mackenziebasin (103m3s
-1).
30
= 20
•' I0
0 I I
Aug Oct Dec Feb Apr Jun Aug
Fig. 9. As Figure 8 but as percentage of annual total and for Ob'-Irtysh and
Yenisei basins (observed) and 60-105 E north of 50 N (modelled).
Geophysical Monograph Series Understanding Climate Change Vol. 52
40
•0
20
40
30
2O 3O
Snowaccumulation
(gcm-2yr
-')
0ø
b
90øW• 90øE
60
70 ø S-
30
180 ø
ROWNTREE 173
240 -
220
2OO
•'• 180 F \
I
/
140
• 120
.--.
'-- I00
o
/
-• 80
' /
o 60 - /
/
40
7
20
I I i I I I I I I I I I I I IN• I I I
90 80 70 60 50 40 30 20 I0 0 -I0 -20 -30 --40 --50 -60 -70 --80 --90
Latitude (degreesnorth)
too fast compared with the satellite data. These results are consistent with paleoclimatic
However, the water path must still be interpreted data indicating high lake levels over the Saharan
in terms of cloud radiative properties. This is region at about 9000-6500 years BP. Mitchell et
not straightforward as Figure 12 (W. Ingram, al. E1988] used the UK Meteorological Office
personal communication), showing several proposed model with a 50 m slab mixed layer ocean to
relationships between reflectivity and water simulate the effects, but although rainfall
content, illustrates. Clearly, estimates of drop increased over southern Asia and eastern Africa,
size effective radii are needed if uncertainties they obtained little change, even some decreases
of up to 10 per cent in albedo are to be avoided. over the central and western Sahara. The
Land surface parametrizations are becoming much explanation of this difference between the models
more comprehensive. However, the land surface was traced to the formulations in evaporation: the
poses some other interesting problems. Let me NCAR model used one quarter of the potential
illustrate these with the simulation of the evaporation, however little soil moisture there
climate 9000 years ago when the solar radiation at was, so that evaporation was able to respond to
the top of the atmosphere differed from that of increases of radiation and allow the model to
the present day by about as much as it has in the develop a moist climate. In the Meteorological
last few tens of thousands of years. There were Office scheme, any increase in evaporation had to
increasesof 30-40Wm
-2 overparts of the Northern be metfromsoil moistureand, as there usually is
Hemisphere in summer. This was due mainly to the little over' the model Sahara, no evaporation
different time of perihelion - in July compared to increase occurred. However, if the albedo was
January at present. Kutzbach and Guetter [1986] decreased to values typical of grassland,
simulated the effects of this in the NCAR CCM and precipitation did increase fairly widely as a
obtained precipitation increases in summer across result of the Charney albedo feedback mechanism.
the Sahara and Arabia, as well as southern Asia. Now, such a vegetation change could occur
Geophysical Monograph Series Understanding Climate Change Vol. 52
1.0
As
Stephens .... Cu
.8 eee
eeeee ••x' St
7#m
ß_ .......... 30 #m
0.6
Ci
0.4
0.2
0.0
0 I O0 200 300 400 500 600 700 8O0
Watercontent
(g/m:)
Fig. 12. Estimates of the relation between reflectivity and cloud water content for
different clouds (stratus, cumulus, altostratus and cirrus from Lieu and Wittman
(1979), Charlock and Ramanathan (1985) 7 and 30um effective drop radius and Stephens'
(1984) spectral average.
ROWNTREE 175
Acknowledgments. The author acknowledges the T. Matsuno), (Terra Sci./Reidel) pp. 501-525,
considerable help from staff of the Dynamical 1984.
Climatology Branch who developed and ran the model Manabe, S., and Broccoli, A.J., The influence of
and diagnostic programs used for many of the continental ice sheets on the climate of an ice
results •n this paper, and also the constructive age, J. Geophys. Res., 90, 2167-2190, 1985.
suggestions made by Dr Howard Cattle. Mitchell, J.F.B., Grahame, N.S., and Needham,
K.J., Climate simulations for 9000 years before
present; seasonal variations and the effect of
Re fe re nce s
the Laurentide •ce sheet, J. Geophys. Res., in
press, 1988.
Charlock, T.P., and Ramanathan, V., The albedo Njoku, E.G., and Swanson, L., Global measurements
field and cloud radiative forcing produced by a of sea surface temperature, wind speed and
general circulation model with internally atmospheric water content from satellite
generated cloud optics, J. Atmos. Sci., 42, microwave radiometry, Mon. Weath. Rev., 111,
1408-1429, 1985. 1977-1987, 1983.
Han, Y.-J., Schlesinger, M.E., and Gates, W.L., An Palmer, T.N., Shutts, G.J., and Swinba•k, R.,
analysis of the air-sea-ice interaction Alleviation of a systematic westerly bias in
s•mulated by the OSU coupled atmosphere-ocean general circulation and numerical weather
general circulation model, Coupled prediction models through an orographic gravity
Ocean-Atmosphere Models (ed. J. C. J. Nihoul), wave drag parametrization, Quart. J.R. Met.
Elsevier, pp 167-182, 1985. Soc., 112, 1001-1039, 1986.
Hellerman, S., and Rosenstein, M., Normal monthly Schlesinger, M.E., and Mitchell, J.F.B., Model
wind stress over the world ocean with error projections of the equilibrium climatic response
estimates, J. Phys. Oc., 13, 1093-1104, 1983. to increased carbon dioxide, Projecting the
Hills, T.S., Sensitiv•ty of numerical models to climatic effects of increasing carbon dioxide,
mountain representation, ECMWF Workshop on US Dept. of Energy, pp 81-147, 1985.
mountains and numerical weather prediction, pp Schwerdtfeger, W., The Climate of the Antarctic,
139-161, 1979. Climate of the polar regions, World Survey of
Jaeger, L., Monthly maps of precipitation for the Climatology, Elsevier, Vol. 14, pp 253-355, 1970.
whole world, Berichte des Deutschen Slingo, A., and Pearson, D.W., A comparison of the
Wetterdienstes, 18, No. 139, 1976. impact of an envelope orography and of a
Kutzbach, J.E., and Guetter, P.J., The influence parametrization of orographic gravity-wave drag
of changing orbital parameters and surface on model simulations. Quart. J. R. Met. Soc.,
boundary conditions on climate simulations for 113, 847-870, 1987.
the past 18000 years, J. Atmos. Sci., 43, Slingo, A., and Wilderspin, R.C., Development of a
1726-1759, 1986. revised longwave radiation scheme for an
Liou, K.-N., and Wittmann, G.D., Parametrization atmospheric general circulation model, Quart. J.
of the radiative properties of clouds, J. Atmos. R. Met. Soc., 112, 371-386, 1986.
Sci., 36, 1261-1273, 1979. Somerville, R.C.J., and Remer, L.A., Cloud optical
Luther, F., Intercomparison of radiation codes in thickness feedbacks in the CO• climate problem,
climate models (ICRCCM): Longwave clear sky J. Geophys. Res., 89, 9668-9672, 1984.
calculations, WMO/ICSU WCP-93, 1984. Stephens, G.L., The parametrization of radiation
Mackay, D.K., and Loken, O.H., Arctic hydrology, for numerical weather prediction and climate
Arctic and alpine environments (ed. J.D. Ives models, Mon. Weath. Rev., 112, 826-867, 1984.
and R.G. Barry), Methuen, pp. 111-132, 1974 Wallace, J.M., Tibaldi, S., and Simmons, A.J.,
Mahlman, J.D., and Umscheid, L.J., Dynamics of the Reduction of systematic errors in the ECMWF
middle atmosphere; Successes and problems of the model through the introduction of an envelope
GFDL 'SKYHI' general circulation model, Dynamics orography, Quart. J. R. Met. Soc., 109, 683-718,
of the middle atmosphere (ed. J.R. Holton and 1983.
Geophysical Monograph Series Understanding Climate Change Vol. 52
Michael E. Schlesinger
177
Geophysical Monograph Series Understanding Climate Change Vol. 52
introduced with the aid of the block diagram for the AT.- I .of AQ, (3)
climate system shown in Fig. 1. In this figure AQ
represents the "input" to the climate system in the where f = GoF is the feedbackfactor [Bode, 1975, p.
form of a radiative forcing due to an external 32] or, more simply, the feedback. Equation (3) can
perturbation such as a doublingof CO2, and AT. is also be written as
Geophysical Monograph Series Understanding Climate Change Vol. 52
SCHLESINGER 179
6 I
I /
I ß
-0.6
" I /
I '
/
I
-I.0
-
I
Fig. 2. The feedback/no-feedback
responseratio Rf= AT./(AT.)o versusthe feedbackf,
where AT. is the surface-temperatureresponseof the system with feedback and (AT.) o is
surface-temperature response without feedback. There is a change in the scales of both
axes at the origin.
For f < 0 the feedback is negative and 0 _<Rf < 1. The Feedback f For The Climate System
Consequently, when there is negative feedback,
regardless of its magnitude, the sign of the The net radiation at the top of the Earth's
response is the same as the sign of the forcing. atmosphere N can be expressed from the viewpoint
This is in contrast to what has sometimes been of a planetary energy balance model as N = N(E, T.,
erroneously inferred for the outcome of negative I). Here E is a vector of quantities that can be
feedback. For 0 < f < 1, the feedback is positive and regarded as external to the climate system, that is,
Rf > 1. However, for positive feedbackwith f > 1, Rf quantities whose change can lead to a change in
< 0 and the sign of the response is opposite to the climate, but which are independent of climate. I is
sign of the forcing. This outcome, while a vector of quantities that are internal to the climate
mathematically possible (and actually obtained by system, that is, quantities that can change as the
one improperly formulated EBM; see Schlesinger climate changes and, in so doing, feed back to
[1985, 1988a]), is not physically consistent and must modify the climatic change. The external
therefore be rejected. quantities include, for example, the solar constant,
The classical expression for feedback given by the optically-active ejecta from volcanic eruptions
Eq. (4) can be contrasted with the definition of the and the CO2 concentration(although eventually it
"feedback parameter k" used by Dickinson [1981], may change as a result of climatic change). The
namely, internal quantities include all the variables of the
climate system other than the surface temperature
AT,= 1 AQ. (7) T. such as the atmospheric temperature, water
vapor and clouds. BecauseT. is the only dependent
From Eqs. (4), (5) and (7) it can be seen that variable in this model, the internal quantities must
be represented therein by I = I(T.).
I 1-f
• - - (8)
A small change in the energy flux, AN, can be
Gf G o '
expressed as
Consequently, as f increases and the climate system
becomes more sensitive, k decreases, while k AN= AQ- (G•- F)AT., (9)
increases as f decreases and the climate system
becomes less sensitive. This "upside-down"
where
behavior of k is unnecessarily confounding and can
be avoided by the use instead of the gain of the
AQ
=i• 0NAE. (10)
climate system with feedback, Gf, together with Eq.
(4). In fact, Cess and Potter [1984] employed the ß •E. '
1
SCHLESINGER
I - O•
where
N= 4
P S - eo"Id,
'
(16)
aN dI.
f.=G J (14) whereS is the solarconstant,
ap the planetary
j o 3I. dT.'
albedo, e the effective emissivity of the Earth-
atmosphere system, and • the Stefan-Boltzmann
Equation (13) with constant. From this equation and Eq. (11) we can
calculate the zero-feedback gain as
f = Z f. (15)
J
(17)
is identical to Eqs. (4) and (5) which were obtained Go:I•T,J=(1-•p)S
'
solely from our consideration of the feedback block
diagram of Fig. 1. Taking
S= 1370Wm-2,ap= 0.3andT. = Ts= 288K,
It should be noted that this classical feedback
where T s is the observed global-mean surface air
analysis, together with its representation by Fig. 1,
temperature, Eq. (17) gives GO = 0.3oC/(Wm-2).
is linear because it neglects the second- and higher-
order derivatives of N. Because this is equivalent to Thus, for AQ = ART = 4 Wm-2 for a doublingof the
CO2 concentration, where ART is the change in the
theassumption
that32N/aIjaIk
= 0,where
j andk
are any two physical processes, the effects on N of net longwave radiation at the tropopause, Eq. (1)
the individual physical processes are considered to gives(AT,)o = 1.2øC. (The reasonsfor evaluatingAQ
be independent. at the tropopause rather than at the top of the
As a result of the classical linear feedback atmosphere are described subsequently.) This
analysis, Eq. (14) shows that the feedback of a value of (AT,) o for a CO2 doubling is in agreement
physical process j depends on three quantities: 1)
with what has been obtained by RCMs without
the sensitivity of the net flux N to the process as
feedbacks (see below and Schlesinger [1985, 1988a],
measured by 3N/aIj,2) the sensitivity
of the process and Schlesinger and Mitchell [1985]).
to the surfacetemperature as measured by dIj/dT.,
and 3) the zero-feedback gain of the climate system Methods Of Feedback Analysis For
Go. Consequently, the feedback of any physical Climate Model Simulations
process is zero if either the net flux N is
independent of that physical process or the physical Three methods have been used to evaluate the
process is independent of the surface temperature. feedbacks in climate model simulations, one
As shown by Eq. (13) and Fig. 2, the influence of method for simulations by radiative-convective
any particularfeedback• on the response
of the models, and two other methods for simulations by
climate system depends nonlinearly on the sum of general circulation models. In the following we
the other feedbacks. For example, the addition of a describe each of these feedback analysis methods.
Geophysical Monograph Series Understanding Climate Change Vol. 52
(ATs)j-
1=1- j• fi ' i=1 al.
Use of a radiative-convective model. Hansen et
[1984] used a radiative-convective model to
analyze the feedbacks in the GISS (NASA Goddard
Institute for Space Studies, New York, New York)
where (ATs)j and (ATs)j_x are the surfaceair
temperature changes from two 1xCO2-2xCO 2 general circulation model simulation of the global-
simulation pairs, the first with physical process j mean surface air temperature change induced by a
and the second without. (Here and in the following
doubling of the C02 concentration. The basis for
this analysis is Eq. (6) which can be written as
we replace the surface temperature change AT.
with the surface air temperature change ATs. We J
AT -(ATs)
ø= J=E• •(ATs)
do this because the results for RCMs and GCMs are
SCHLESINGER
rerunning the RCM lxCO2 simulation with this globally. Then this radiative computation was
fixed modified value. (During this rerun of the repeated individually for each feedback process by
RCM lxCO 2 simulation, the temperaturelapse rate replacing the appropriate monthly-mean quantity
in the troposphere was adjusted to the global-mean from the GCM lxCO2 simulation with the
value of the GCM lxCO2 simulation if it exceeded corresponding monthly-mean quantity from the
that value.) The difference between this RCM GCM 2xCO2 simulation. The difference between
lxCO2 simulationwith the prescribedchangein the the latter and former calculations then gives
appropriate quantity and the initial RCM lxCO2 [•jN]ave.
simulation
withoutfeedback
gives•j(ATs). FeedbacksFor A C02 Doubling
Use of a radiative transfer model. The
principal shortcomingof analyzing the feedbacksof In this section we present results, obtained by
a GCM simulation with an RCM is the need to use the above analysis methods, of the feedbacks in
the global-mean values of the GCM as input to the radiative-convective and general circulation model
RCM. This would be of no importance if the climate simulations of the temperature changes induced by
system were linear because then [T (X)]ave = a doubling of the C02 concentration.
T(Xave),wherex and T representthe input to and
output from the RCM, respectively, and "ave" Radiative-Convective Models
Feedback
Mechanism RCM a GCM
GISS b GFDL c
effect of the 33% increase in total water vapor inserting a uniform cloud change in the RCM equal
simulated by the GCM was estimated by increasing to the total change in the GCM. Lastly, the effect of
the water vapor at each level of the RCM by 33%, the cloud altitude change,õ5(ATs),was obtained by
thereby giving the result for õi(ATs). To determine subtractingõ6(ATs)from the total cloudeffect.
the effect of the change in the vertical distribution of The results of the feedback analysis using the
water vapor simulated by the GCM, the latter was method given by Eq. (22) together with the Hansen et
inserted into the RCM and the resulting
al. [1984]valuesof•(ATs)arepresented
in Table1.
temperature change was decreasedby õ1(ATs) to The feedback due to the changes in water vapor
obtain•2(ATs). To determinethe effectof the change amount and vertical distribution is fw = 0.66. This
in lapse rate simulated by the GCM, the latter was is considerablylarger than the fw = 0.3 to 0.4 given
inserted into the RCM and gave õ3(ATs). Similarly, by the RCMs. The much larger fw estimated for
õ4(ATs) for the GCM-simulated change in surface the GISS GCM indicates that the relative humidity
albedo was obtained. The total cloud effect on increased with doubled C02 in that model, unlike
temperature was obtained by changing the cloud the constant relative humidity assumed by the
amounts at all levels in the RCM in proportion to RCMs. Indeed, Hansen et al. [1984] state that the
the changes obtained in the GCM. The effect of average relative humidity increased by 1.5% with a
changingonly cloud cover,õ6(ATs),was obtainedby maximum of 6% at the 200 mb level. The estimated
Geophysical Monograph Series Understanding Climate Change Vol. 52
SCHLESINGER
lapse rate feedback,fLR = -0.26, lies at the smaller 1, this is a very large value of the feedback, much
limit given by the RCMs for the moist adiabatic larger than the maximum value for the RCMs, f =
lapse rate case, perhaps because the change in 0.71, and the same value for the GISS GCM. The
lapse rate of-0.2øC/km in the GCM simulation is reason for this large value is the use above of
less than the change in the moist adiabatic value of [AQTOA]av
e, that is, the changein the net radiation
-0.5oC/km. The cloud altitude feedback,fCA = 0.12, at the top of the atmosphere. Consequently, for the
also lies at the lower limit given by the RCMs. The feedback analysis of the Wetheraid and Manabe
cloud cover feedback estimated for the GCM is [1988] results to be consistent with the two other
positive. This indicates that the global-mean low feedback analyses, it is necessary to convert the top-
and middle clouds, whose albedo effect dominates of-the-atmosphere results, [AQTOA] and
their longwaveeffectgiving•N/•Ij < 0 in Eq. (14),
decreased and/or the global-mean high clouds,
[•jNTOA]ave, to the correspondingtropopause
whose longwave effect dominates their albedo effect results,[AQTROP] and[•ijNTROP]av
e. (Wetheraid
giving•N/•Ij > 0, increased.Finally,the surface and Manabe [1988] presentedvalues of [AQToA] and
albedo feedback, due largely to the reduced sea ice, [•jNTOA]av
e insteadof [AQTRO
P] and[•jNTROP]ave
is estimated as fSA = 0.09 which is somewhat because the profiles of the changed quantities
smaller than the estimates given by the RCMs. The extended throughout both the troposphere and
total feedback estimated for the GCM is f = 0.71, of stratosphere, and because it is difficult to define the
which water vapor feedback,fw = 0.66, is the single tropopause in a GCM simulation. However,
most important positive contributor, followed by because: 1) the change in the net radiation flux is a
cloud feedback, fc = 0.22, and surface albedo maximum at the tropopause, 2) the sign of the C02-
feedback, fCA = 0.09, with the lapse rate feedback, induced change in the stratospheric temperature is
fLR = -0.26, making a negative contribution. opposite to that of the tropospheric and surface
Analysis by use of a radiative transfer model. temperature changes, and 3) the changes in the
Wetheraid and Manabe [1988] present results for: 1) stratospheric quantities have little effect on the
the change in the net radiation at the top of the troposphere and the surface [Schlesinger 1985,
atmosphere (TOA) due to the doubling of the C02 1988a], it is more correct to use [AQTRo P] and
concentration,[AQTOA]av
e = 2.28 Wm-2;2) the total [•jNTROP]av
e than [AQToA]and[•SjNTOA]ave.)
We
do this as described below.
C02-induced temperature change [ATs]ave = 4.0øC,
and 3)the [•ijNTOA]av
e values obtainedfor the For the forcing we take [AQwRoP]av e =
2xCO2-1xC02 change in surface albedo and the [AQTOA]ave + C, where C is a conversionfactor.
changes in the vertical profiles of temperature, The value of C was obtained from the
water vapor and fractional cloud cover. From Eqs. Intercomparison of Radiation Codes used in
(3) and (23) the total feedback can be expressedby Climate Models (ICRCCM) study in which the
radiation model of the GCM was used to compute
G
the longwave fluxes for five atmospheric profiles
1- o[AQTOA]ave
(24a) [Ellis, 1987, personal communication]. From the
[ATs]av
e average over the five profiles of these ICRCCM
results for the difference between the changesin the
fTOA-- longwave fluxes at the tropopause and at the top of
the atmosphere, C = 2.03 Wm -2. (The contribution
o
[ATs]av
e j=l
[•jNTOA]av
e (24b)to [AQTROP]av
e and [AQTOA]av
e by the changein the
absorbedsolar radiation due to the C02 doubling is
UsingEq. (24a) togetherwith Go= 0.3oC/(Wm
-2) and negligible.) Thus for [AQTOA]av
e = 2.28 Wm-2, the
the above values of [AQTOA]av e and [ATs]ave gives above conversiongives [AQTROP]av e = 4.31 Wm-2.
fTOA = 0.829. As can be seenfrom Fig. 2 and Table Using this value in
Geophysical Monograph Series Understanding Climate Change Vol. 52
G [AQTROP]av
o e Conclusion
fTROP
=I - (25)
[ATs]av
e In the Introduction we recommended that the
quantitative analysis of the feedbacks in climatic-
gives fTROP = 0.677 which is comparable to the change simulations should be the first step of a
maximum value for the RCMs and the same value systematic approach to answering the question:
for the GISS GCM. Why do models, even within the same category of
Unfortunately, we do not have the analog of the the climate model hierarchy, simulate different
ICRCCM results with which to convert the climatic changes for the same forcing? To support
[•jNTOA]ave values to [•jNTROP]ave values. this recommendation we have presented in this
Consequently,we computethe components
of fTOA paper: 1) the concept and terminology of classical
from Eq. (24b) and convert them into the feedback analysis; 2) three different feedback
componentsof fTROPby analysis methods, one for radiative-convective
models and two others for general circulation
models; and 3) the application of these feedback
fTROP,
j TMfTOA
fTOA,
j =0.817 j (26)analysis
fTOA, methodsto the simulationsof CO2-induced
climatic change.
The intercomparison of the results of these
Although this conversion is by no means unique, at feedback analysis shows that the contribution of the
least the resultantvaluesof fTROP,
j andfTROP individual feedback processes to the simulated
satisfy Eq. (15). climatic change is not the same for different
The results of this feedback analysis of the models, even when these models are within the
Wetheraid and Manabe [1988] study are presented same category of the climate model hierarchy. This
in Table 1. It can be seen that the water vapor is a clear demonstration of the utility of the analysis
feedback for the GFDL GCM is comparable to that of of the feedbacks in climatic-change simulations.
the RCMs, and is about two-thirds that of the GISS To improve our understanding of the behavior of
GCM. Interestingly, the total lapse rate feedback is feedbacks,the feedbacksin the simulations of CO2-
positive in the GFDL GCM in contrast to the
induced climatic change obtained by the GISS,
negative lapse rate feedback of the GISS GCM. The NCAR, OSU and UKMO (NCAR is the National
combined cloud altitude and cloud cover feedback is
Center for Atmospheric Research, Boulder, CO;
positive for the GFDL GCM but is only about 40% of
OSU is Oregon State University, Corvallis, OR; and
the corresponding positive feedback of the GISS UKMO is the United Kingdom Meteorological
GCM. Finally, the surface albedo feedback of the Office, Bracknell, Berkshire) GCMs should be
GFDL GCM is comparable to that of the GISS GCM
determined using the radiative-transfer-model
and is on the low side of the RCM results.
method and compared with the feedbacks of the
A rigorous intercomparison of the quantitative GFDL model which have been determined by this
values of the feedbacks for the GFDL and GISS method. However, the correct application of this
GCMs is made difficult by the fact that the feedback-analysis method requires use of the
feedbacksof these models have been analyzed by two changes in the net radiative fluxes at the
different methods, namely, the radiative-convective- tropopause, not at the top of the atmosphere.
model method for the GISS model and the radiative-
After completion of these additional feedback
transfer-model method for the GFDL model. This
analyses, the feedback processes can be ranked in
notwithstanding, the intercomparison indicates terms of their contributions to the simulated
that, although these models simulate similar climatic changes. Then it will be time to take the
valuesof [ATs]av
e, they do so with •valuesof their secondstep in the systematic approach, namely, the
feedbacks
whichdifferin bothmagnitude
andsign, intercomparison of the parameterizations of the
this despite the approximate agreement of their highest-ranked processes among themselves, with
simulatedvalues oft s for the lxCO 2 climate. more-detailed models, and with observations.
Geophysical Monograph Series Understanding Climate Change Vol. 52
SCHLESINGER • s7
Acknowledgments. I thank John Mitchell for models, in The Potential Climatic Effects of
his constructive review of this paper. This study Increasing Carbon Dioxide, M.C. MacCracken
was supported by the U.S. National Science and F.M. Luther (Eds.), U.S. Department of
Foundation and the Carbon Dioxide Research Energy, DOE/ER-0237, pp. 280-319, 1985.
Division, Office of Basic Energy Sciences of the U.S. (Available from NTIS, Springfield, Virginia.)
Department of Energy under grant ATM 87-12033. Schlesinger, M. E., Equilibrium and transient
warminginducedby increasedatmo.spheric
CO2,
References Climate Dynamics, 1, 35-51, 1986.
Schlesinger, M. E., Quantitative analysis of
Bode, H. W., Network Analysis and Feedback feedbacks in climate model simulations of CO2-
Amplifier Design, Krieger, New York, 577 pp., induced warming, in Physically-Based
1975. Modelling and Simulation of Climate and
Cess, R. D., and G. L. Potter, A commentary on the Climatic Change, M.E. Schlesinger (Ed.), NATO
recent CO2-climate controversy, Climatic Advanced Study Institute Series, Reidel,
Change, 6, 365-376, 1984. Dordrecht, pp. 653-736, 1988a.
Dickinson, R. E., Convergence rate and stability of Schlesinger, M. E., Model Projections of the
ocean-atmosphere coupling schemes with a zero- Climatic Changes Induced by Increased
dimensional climate model, J. Atmos. Sci., 38, Atmospheric CO2, in NATO Climate/Geo-
2112-2120,1981. Sciences Symposium, A. Berger and S.H.
Hansen, J., A. Lacis, D. Rind, G. Russell, P. Stone, Schneider (Eds.), Kluwer Academic (in press),
I. Fung, R. Ruedy and J. Lerner, Climate 1988b.
sensitivity: Analysis of feedback mechanisms, in Schlesinger, M. E., and J.F.B. Mitchell, Model
Climate Processes and Climate Sensitivity, Projectionsof the EquilibriumClimatic Response
Maurice Ewing Series, 5, J. E. Hansen and T. to Increased Carbon Dioxide, in The Potential
Takahashi (Eds.), American Geophysical Union, Climatic Effects of Increasing Carbon Dioxide,
Washington, D.C., pp. 130-163, 1984. M.C. MacCracken and F.M. Luther (Eds.), U.S.
Manabe, S., and R. T. Wetheraid, On the
Departmentof Energy,DOE/ER-0237,pp. 81-147,
distribution of climate change resulting from an 1985. (Available from NTIS, Springfield,
increase in CO2 content of the atmosphere, J. Virginia.)
Atmos. Sci. , 37, 99-118, 1980. Wetheraid, R. T., and S. Manabe, Cloud feedback
Schlesinger, M. E., Feedback analysis of results processes in a general circulation model. J.
from energy balance and radiative-convective Atmos. Sci., 45, 1397-1415, 1988.