FLO-2D Model Validation

for Version 2009 and Up

Prepared for:
Federal Emergency Management Agency
June 2011

Prepared by:
FLO-2D Software, Inc.
Nutrioso, Arizona

Endorsements
The following agencies have reviewed this document, endorse the findings presented herein and
support the recommendation that the FLO-2D flood routing model (Versions 2009 and up) to be
included on FEMA's list of approved hydraulic and hydrologic models for Flood Insurance Studies.
Flood Control District of Maricopa County, Phoenix, Arizona (FCDMC)
U.S. Army Corps of Engineers, Sacramento District (Corps)
California Department of Water Resources, Sacramento, California (DWR)
Truckee River Flood Project, Reno, Nevada (TRFP)
U.S. International Boundary and Water Commission, El Paso, Texas (IBWC)
City of Camarillo, California
Mohave County Flood Control District, Kingman, Arizona (MCFCD)
Cochise County Flood Control District, Bisbee, Arizona (CCFCD)
Some of these agencies have a very long history and current projects using the FLO-2D model (Corps,
DWR, FCDMC, IBWC, TRFP). The other three agencies have FIS current projects using Version 2009.
Some of the agencies provided formal letters of endorsement sent to FEMA (FCDMC, MCFCD, CCFCD),
other agencies sent a formal endorsement by email to FLO-2D Software, Inc. (Corps, DWR, FRFP). These
letters and email endorsement are presented in the Appendix. The City of Camarillo and IBWC indicated
their endorsement with a simple email message that was not included in the Appendix.

Executive Summary
Since 1999 and through Version 2007 FLO-2D has been listed as a FEMA approved hydraulics model. In
June 2010 FLO-2D Software, Inc. petitioned the Federal Emergency Management Agency (FEMA) to
include Version 2009 of the FLO-2D model on FEMA's list of approved hydraulics models. This petition
procedure to list a new model version was accomplished with previous versions according to a
Memorandum of Understanding agreement between FEMA and FLO-2D Software, Inc. In February
2011, FEMA informed FLO-2D Software, Inc. that supporting validation documentation was required for
listing both new models and to update model versions. This document presents information that the
FLO-2D model for Versions 2009 and subsequent releases accurately predicts flood hydrology and
hydraulics and distributes flood volume. Specifically, this document was prepared to support a FEMA
listing of the FLO-2D Model (Versions 2009 and up) as an approved model for both hydraulic and
hydrologic Flood Insurance Studies (FIS).
Following International Association for Hydro-Environment Engineering and Research (IAHR) guidelines
for documenting model validation, this document discusses the FLO-2D model representation of the
flood physical system, the validation approach, the computational core validation, range of model
applicability, and the accuracy of the model results. The model background information is presented in
Chapter I along with validation goals and objectives. An overview of the model is presented in Chapter II
including model formulation, algorithm solution, numerical stability criteria, model organization and
assumption. Model validation for hydrology and hydraulics is discussed in Chapter III focusing on the
validity of the computation core for overland flow, river channel flow and channel-floodplain exchange.
Chapter III is organized according to the IAHR format for documenting the validity of computational
modeling software which presents a validation claim followed by a section on the substantiation of the
claim. The substantiations for the claims are based on a series of case study projects that validate
specific model components. Results from each case study project are detailed to verify model accuracy
and consistency. Further details of hydraulic validation case studies are presented in Chapter IV. Model
certification letters are presented in the Appendix.
This document demonstrates that the FLO-2D model provides accurate hydrology and hydraulic results
for a broad range complex flood studies from river flooding to urbanized alluvial fans. This document
also asserts that the computation core has provided consistent hydraulic results throughout the last
eleven years of new releases. Benchmark testing ensures that the model engine will provide future
accurate and reliable results.

ii

Table of Contents
Endorsements .............................................................................................................................................................i
Executive Summary ................................................................................................................................................. ii
FLO-2D Model Validation .......................................................................................................................................1
I.

Introduction .......................................................................................................................................................1

1.1
1.2
1.3
1.4
II.

Background .................................................................................................................................... 1
Some Comments on Modeling Free Surface Flows ....................................................................... 2
Validation Goal and Objectives ...................................................................................................... 2
Model Speed, Accuracy, Stability and Sensitivity .......................................................................... 3

Overview of the FLO-2D Model.......................................................................................................................5

2.1 Model formulation ......................................................................................................................... 5

2.2 Solution Algorithm - How the Model Works.................................................................................. 6
2.3 Numerical Stability Criteria Controlling the Computational Timestep ..................................... 10
2.4 Courant Number Guidelines ........................................................................................................ 11
2.5 The Importance of Volume Conservation .................................................................................... 12
2.6 Organization of the FLO-2D Model .............................................................................................. 12
2.6.1 Grid Developer System (GDS) ............................................................................................... 13
2.6.2 Input Data ............................................................................................................................. 13
2.6.3 Output Options ..................................................................................................................... 13
2.7 FLO-2D Model Assumptions and Approximations ....................................................................... 14
III. FLO-2D Model Validation .............................................................................................................................. 15

3.1 Overview ...................................................................................................................................... 15

3.2 Representing Physical System...................................................................................................... 15
3.3 Validity of Computational Core.................................................................................................... 17
3.3.1 Validity of the Computation Core - Overland Flow ............................................................... 19
3.3.2 Validity of the Computational Core - Channel Flow.............................................................. 35
3.3.2.1 Operational Flows in the California Aqueduct ............................................................... 35
3.3.2.2 Design flow in Camarillo Hills Drain ............................................................................... 38
3.3.2.3 Tracking Green River Flaming Gorge Dam Power Release ............................................ 41
3.3.3 Channel and Floodplain Flow Exchange ................................................................................ 46
3.4 Hydrology ..................................................................................................................................... 48
3.4.1 Overview ............................................................................................................................... 48
3.4.2 Hydrology Validation............................................................................................................. 48
3.4.3 Summary of the Hydrology Component ............................................................................... 57
3.5 Component Applicability.............................................................................................................. 58
3.6 Accuracy of Computational Results ............................................................................................. 61

iii

3.7 Benchmark Testing....................................................................................................................... 64

IV. Summary of Validation Studies ...................................................................................................................... 66

4.1
4.2
4.3
4.4
4.5
4.6

Case Study 1. Overland Flow-Flume with Analytical Solution ..................................................... 66

Case Study 2. Overland Flow - Truckee River............................................................................... 67
Case Study 3. City of Camarillo, California FIS Study ................................................................... 68
Case Study 4. Middle Rio Grande 2005 Spring Runoff Monitoring Project ................................ 71
Case Study 5. California Aqueduct Operational Design Flows .................................................... 73
Case Study 6. Green River FLO-2D Model - Flaming Gorge Dam to Colorado Confluence ........ 75

V. References ....................................................................................................................................................... 78

Appendix
U.S. Army Corps of Engineers Original Two Letters of Certification
Letters of Endorsement
Email Endorsements
List of FLO-2D Model Revisions

List of Figures
Figure 1. Unconfined Discharge Computations .......................................................................................................6
Figure 2. FLO-2D Numerical Stability Flow Chart .................................................................................................9
Figure 3a. SUMMARY.OUT File Report. ............................................................................................................ 18
Figure 3b. SUMMARY.OUT File for the Camarillo Project ................................................................................ 18
Figure 4. FLO-2D Version 2009 Results for Maximum Velocity and Depth ........................................................ 21
Figure 5. Version 2007 Flume Case Results Showing Identical Results as Version 2009 in Figure 4. ................. 21
Figure 6. Truckee River Observed (Black Outline) and FLO-2D Predicted Area of Flood Inundation ............... 22
Figure 7. Predicted Truckee River Water Surface Profiles for HEC-2 and FLO-2D............................................. 23
Figure 8. FLO-2D Predicted Stage for the 1997 Flood Event at the Reno Gage .................................................. 24
Figure 9. FLO-2D Predicted Stage for the 1997 Flood at the New Vista Gage .................................................... 25
Figure 10. FLO-2D Version 2009 Predicted Stage for the 1997 Flood at the New Vista Gage ............................ 26
Figure 11. FEMA 2005 FIRM MAP for Ventura County ..................................................................................... 27
Figure 12. FLO-2D Predicted Maximum Flow Depths for the City of Camarillo ................................................. 28
Figure 13. FLO-2D 2011 Model FIRM for the City of Camarillo ......................................................................... 28
Figure 14. USGS San Felipe Gage 1998 (FLO-2D Predicted vs. Measured Discharge) ....................................... 30
Figure 15. FLO-2D Predicted vs. USGS Gage Measured Discharge at Chama River Confluence ....................... 31
Figure 16. FLO-2D Predicted Area of Inundation Compared with the Corps' Digitized Shape File for the 2005
Area of Inundation Monitoring Project ................................................................................................................... 32
Figure 17. FLO-2D Predicted vs. USGS Gage Discharge at San Felipe May 24-June 1, 2005............................. 33
Figure 18. FLO-2D Predicted vs. USGS Gage Discharge at Albuquerque May 24-June 1, 2005 ......................... 33
Figure 19. FLO-2D Predicted vs. USGS Gage Discharge at Isleta Lakes May 24-June 1, 2005 .......................... 34

iv

Figure 20. FLO-2D Predicted vs. USGS Gage Discharge at San Acacia May 24-June 1, 2005............................ 34
Figure 21. 1997 High Flow Season Hydrograph at Jensen, Utah (FLO-2D vs. Measured Discharge) .................. 42
Figure 22. November 1998 Power Plant Discharge Release from Flaming Gorge Dam ....................................... 43
Figure 23. FLO-2D Predicted vs. Measured Discharge, Jensen USGS Gage for Power Plant Releases at Flaming
Gorge Dam ............................................................................................................................................................. 44
Figure 24. Version 2009 FLO-2D Predicted vs. Measured Discharge, Jensen USGS Gage for Power Plant
Releases at Flaming Gorge Dam............................................................................................................................. 45
Figure 25. Channel-Floodplain Flow Exchange ................................................................................................... 46
Figure 26. NEXRAD Cell Total Precipitation July 31, 2006 on the Catalina Mountain Watersheds .................... 49
Figure 27. Lower Soldier Canyon Alluvial Fan Aerial Photo. Agua Caliente Wash crosses ............................... 50
Soldier Trail (center photograph). Photograph by Terry Hendricks (PCRFCD; 7-31-06). .................................... 50
Figure 28. Soldier Canyon Alluvial Fan FLO-2D Predicted Area of Inundation for the July 31, 2006 Flood ...... 51
Figure 29. FLO-2D Predicted Maximum Depth for Rainbow Wash (FCDMC, 2010).......................................... 52
Figure 30. FLO-2D Predicted Hydrograph vs. Gage Data and HEC-2D Results (FCDMC, 2010) ....................... 52
Figure 31. FLO-2D Predicted Inundation Area vs. the Observed Flooding from 2005 Storm (FCDMC, 2010) ... 53
Figure 32. FLO-2D Model Predicted Depth and Velocities for White Tanks Fan 36 (JE Fuller, 2010) ............... 56
Figure 33. FLO-2D Mode Predicted Depth and Velocities for White Tanks Fan 36 ............................................ 57
Figure 34. Rio Grande Overbank Flooding Confined by Levees (red outline in Maxplot) ................................... 59
Figure 35. Simple Levee Example Project with 1,000 cfs Steady Discharge ........................................................ 60
Figure 36. Corps' Monroe Alluvial Fan Project with a Defined Channel Using Version 2007 .............................. 62
Figure 37. Corps' Monroe Alluvial Fan Project Difference in Maximum Depth .................................................... 62
(all the differences in maximum depth are 0.2 ft or less) ........................................................................................ 62
Figure 38. Sparks Truckee Meadows Area General Peak Flow Path (Google Earth, 2011) .................................. 68
Figure 39. City of Camarillo 10 ft FLO-2D Grid System with the Building Shape File ....................................... 69
Figure 40. Detail of the City of Camarillo 10 ft FLO-2D Grid System with LiDAR Data .................................... 69
Figure 41. Spatially Variable n-value Shapefile .................................................................................................... 69
Figure 42. Flooding Differences Between the Models (100 ft model left and 10 ft model right) .......................... 70
Figure 43. Flooding along Ponderosa Drive (100 ft model left and 10 ft model right) ......................................... 70
Figure 44. 100 ft Model - Flooding across Highway 101. ...................................................................................... 71
Figure 45. 10 ft Model - Flooding is confined by Highway 101. ........................................................................... 71
Figure 46. FLO-2D Predicted vs. USGS Gage Discharge at San Marcial April 3-11, 2004 ................................. 73
Figure 47. FLO-2D Predicted vs. USGS Gage Discharge at San Marcial May 5-22, 2004 .................................. 73
Figure 48. Aerial Image of the FLO-2D Reach of the California Aqueduct Near Huron ...................................... 74
Figure 49. California Aqueduct Near Huron (from Kewaneh, Google Earth, 2011) ............................................. 74
Figure 50. California Aqueduct Bedslope Near Huron .......................................................................................... 75
Figure 51. Green River Bed Slope from Flaming Gorge Dam to the Colorado River Confluence ........................ 76
Figure 24. FLO-2D Predicted vs. Measured Discharge, Jensen USGS Gage for the Flaming Gorge Releases ..... 76
Figure 52. Flaming Gorge Power Plant Spike Release Hydrograph at the Ouray Gage ........................................ 77
Figure 53. Flaming Gorge Power Plant Spike Release Hydrograph at the end of Uinta Valley ............................ 77
Figure 54. Flaming Gorge Power Plant Spike Release Hydrograph at the Colorado River Confluence ............... 77

List of Tables
Table 1. FLO-2D Models for the Middle Rio Grande ........................................................................................... 29
Table 2. Floodplain Area of Inundation Calibration .............................................................................................. 32

Table 3. FLO-2D Results for the California .......................................................................................................... 37

Table 4. WSPG Hydraulic Results for the Camarillo Hills Drain Los Posas Road to Highway 101..................... 39
Table 5. FLO-2D Results for Camarillo Hills Drain 100-year Design Discharge ................................................. 40
Table 6. SUMMARY.OUT file for the Diamond Project to Verify Rainfall Volume Conservation ..................... 48
Table 7. SUMMARY.OUT File for the Green-Ampt Verification........................................................................ 54
Table 8. Validation of the SCS Curve Number Computations .............................................................................. 55
Table 9. Reported Levee Overtopping Discharge at Hour 4.0 ............................................................................... 60
Table 10. Difference in Max Velocity Between v2007 and v2009 for the Corps' Monroe Project ....................... 63
Table 11. FLO-2D Model Revisions Reporting Procedure................................................................................... 64

vi

FLO-2D Model Validation

for Version 2009 and Up
I. Introduction
1.1 Background
This document presents information that the FLO-2D model for Version 2009 and subsequent releases
accurately predicts flood hydrology and hydraulics. Specifically, this document was prepared to support
a Federal Emergency Management Agency (FEMA) listing of FLO-2D Version 2009 and up as an approved
model for both hydraulic and hydrologic Flood Insurance Studies (FIS). Since 1999 and through Version
2007 FLO-2D has been listed as a FEMA approved hydraulics model.
FLO-2D model validation tests were initiated prior the first release of the model to the Sacramento
District of the Corps of Engineers in September, 1995. As the FLO-2D project list grew to thousands of
projects worldwide, the model has been subjected to rigorous independent testing. This has included
calibration, verification and validation by consultants, agencies, and universities. The FLO-2D model has
been the basis for several Masters thesis and Ph.D. dissertations (e.g. Miller, 1996 and Nalesso, 2009).
The model has been verified with flume studies in Europe, used for FEMA Flood Insurance Studies
throughout the United States, and is the support model for the Albuquerque District of the Corps of
Engineers Upper Rio River Operations Model (URGWOM) covering over 400 miles of Rio Grande Valley
where it was applied to replicate 30 years of gage record. There was a large sample of projects to
choose from in preparing this document. The selected FLO-2D projects herein represent those which
were readily available, well documented and perhaps validated more than one component of the
model.
The FLO-2D model has a relatively long history with FEMA. FEMA funded the development of the very
first version of the model in 1989 with a FIS study of Telluride, Colorado. FEMA contracted with Simons,
Li and Associates (SLA) of Fort Collins, Colorado to develop a model that would simulate both water and
mudflow on the Cornet Creek alluvial fan in Telluride. SLA contracted with Jim OBrien to create the
code to simulate hyperconcentrated sediment flows. The original model was called MUDFLOW and its
development proceeded over the next 10 years by FLO Engineering (FLO is an acronym for Fullerton,
Lenzotti and OBrien). During this period the model name was changed to FLO-2D and it was applied on
FIS CLOMR and LOMR alluvial fan studies in the western United States.
In 1999, FEMA suggested that the FLO-2D model be added to FEMAs list of approved hydraulic models
because an increasing number of complex flood studies could not be accurately analyzed with the single
discharge, 1-D HEC-RAS model. Negotiations were concluded with a Memorandum of Understanding
Agreement between FEMA and FLO-2D Software, Inc. that enabled the FLO-2D model to be listed. Two
letters of certification (presented in Appendix A) were written by the Corps of Engineers to support the
use of FLO-2D on FIS studies.

1.2 Some Comments on Modeling Free Surface Flows

With faster computers and high resolution digital terrain models, flood routing models are becoming
very detailed. As flood models have become more versatile with expanding computer resources, the
accuracy of the predicted flooding is still limited by the available hydrographic data bases. When adding
detail to a two-dimensional flood routing model, a number of factors have to be considered including
accuracy of the flood hydrology, topographic data resolution, cross section spacing, and calibration data.
Digital terrain models are becoming the foundation of high resolution mapping, but post-flood event
surveys of high water marks and aerial photography of the area of flood inundation are typically
unavailable or were collected long after the flood waters have receded. Often when correlating the area
of inundation with flood peak discharge it is discovered that aerial imagery is not timed with the peak
flow. Uncertainty in flood hazard delineation not only involves model applicability and assumptions, but
also a failure to calibrate the model because of limited or missing data bases for infrequent flood events.
As predictive flood inundation mapping advances with hydrograph routing, more extensive topographic
data bases, higher resolution graphics, and unconfined 2-D flood simulations, it may appear that flood
modeling complexity is becoming overwhelming. The comments of Cunge et al. (1980) prior to high
speed desktop computing are enlightening:
The modeler must resist the temptation to go back to one -dimensional
schematization because of lack of data otherwise necessary for an
accurate two-dimensional model calibration. If the flow pattern is truly
two-dimensional, a one -dimensional schematization will be useless as a
predictive tool... . It is better to have a two -dimensional model
partially calibrated in such situations than a one -dimensional one which
is unable to predict unobserved events. Indeed, the latter is of ver y little
use while the former is an appr oximation wh ich may always be improved
by complimentary survey.
1.3 Validation Goal and Objectives
This document describes the validation of the FLO-2D model according to the 1994 IAHR guidelines for
computational software. Model validation refers to the theoretical foundation and computational
techniques to produce accurate numerical and graphical results. FLO-2D model verification refers to the
capability of the model to replicate known field or laboratory measurements. The document was
prepared for FLO-2D Version 2009 which provides the computational foundation for all subsequent
model releases and enhancements.
The validation of the FLO-2D model has evolved over the years. The general testing of FLO-2D began in
1989 when the model was called MUDFLOW and a mudflow verification test using the Rudd Creek, Utah
hyperconcentrated sediment flow was implemented. Several tests with HEC-2 results were developed
and formal water flood propagation testing of the FLO-2D began with the implementation of the full
dynamic wave momentum equation. The Sacramento District of the Corps in 1995 became interested in
the model and applied it to a reach of the California Aqueduct. Simulating flow cases with analog

solutions began in 1999 in support of litigation in Federal Court with the California Department of Water
Resources and federal agencies over the release of flood waters from the Aqueduct onto farmland.
Benchmark testing started with Version 2002 and became more extensive with Version 2004. This
testing program expedited the development and modification of model components.
The goal of model validation is to have consistent results with the development of new features that
may also improve model simulation accuracy and stability. The validation objectives are:

Identify problems with code development

Ensure consistency of applications;
Verify the accuracy of model improvements;
Minimize error propagation between components;
Reduce or eliminate the need for quality control of model applications.

FLO-2D validation testing now includes:

Improved record keeping;

Benchmark testing;
Identified error procedure.

1.4 Model Speed, Accuracy, Stability and Sensitivity

Many of the FLO-2D model enhancements are focused on increasing computational speed while
preserving accuracy and numerical stability. There is a fine balance between improving model speed
and suppressing numerical surging. An unstable model is one in which certain types of numerical errors
grow until the solution begins to oscillate. In the FLO-2D model, it is possible to have a minor amount of
numerical surging and still have the model provide a complete solution with reasonable results;
however, the user focus should be to eliminate all the numerical instability.
Model accuracy is defined as the level of precision with which the model approaches the true solution.
A true or analytical solution can be obtained in the case of simulating flume conditions, but replicating
field conditions or historical flood events can only be approximated. Convergence is defined as the
models capability to reach a known solution while numerical instability represents a departure from the
true results. Numerical accuracy and model stability are dominant themes in this document and it will
be shown that the FLO-2D model is accurate for analog flow cases and that numerical surging can be
eliminated for all simulations.
Sensitivity is defined as the scale to which model results vary when input parameters are adjusted.
Sensitivity analyses can help to both speed up the model and maintain model stability. Sensitivity
analyses can also play an important role in the convergence to a known condition as well as to reduce
uncertainty in the model. Of particular interest is the sensitivity of the limiting Froude number and the
Courant number to maintain numerical stability. The automated adjustment of the roughness
Manning's n-value to the limiting Froude number reduces the user's reliance on n-value sensitivity
analyses to eliminate model instability. This stability tool is unique to the FLO-2D model. The following
factors affect the model speed, accuracy, and stability:

Grid element size;

Inflow discharge flux;
User selection of the roughness in relationship to flow area and slope;
Water surface controls such as hydraulic structures.

There are several ways to detect and control numerical instability that will be discussed in Chapter II.

II. Overview of the FLO-2D Model

2.1 Model formulation
FLO-2D is a two-dimensional flood routing model. It is a valuable tool for delineating flood hazards,
regulating floodplain zoning or designing flood mitigation. The model will simulate river overbank flows,
but it can also be used on unconventional flooding problems such as unconfined flows over complex
alluvial fan topography and roughness, split channel flows, mud/debris flows and urban flooding.
Technically it is a finite volume model that moves the flood volume around on a set of square tiles for
unconfined overland flow or through stream segments for channel routing. Floodwave progression in
two-dimensions is accomplished through a numerical integration of the equations of motion that
maintains volume conservation.
The constitutive fluid equations include the continuity equation and the equation of motion (dynamic
wave momentum equation):

h h V
+
=i
t
x
S f = So -

h V V V V 1 V
x g x g x g t

where h is the flow depth and V is the depth-averaged velocity in one of the eight flow directions x. The
excess rainfall intensity (i) may be nonzero on the flow surface. The friction slope term Sf is based on
Mannings equation. The other terms include the bed slope So, pressure gradient and convective and
local acceleration terms. If the pressure term and acceleration terms are negligible (Sf = So), the flow is
considered steady and uniform (referred to as the kinematic wave equation). The addition of the
pressure gradient term (h/ x) to the kinematic wave equation results in the diffusive wave equation,
which is required for floodwave attenuation and change in storage on the floodplain. The remaining
local and convective acceleration terms are important to the flood routing for flat or adverse slopes,
very steep slopes, or unsteady flow conditions. Including these additional terms results in the full
dynamic wave momentum equation. Only the full dynamic wave representation of the momentum
equation is used in the FLO-2D model.
For unconfined overland flow, the equations of motion in FLO-2D are applied by computing the average
flow velocity across a grid element boundary one direction at time. There are eight potential flow
directions, the four compass directions (north, east, south and west) and the four diagonal directions
(northeast, southeast, southwest and northwest). Each velocity computation is essentially onedimensional in nature and is solved independently of the other seven directions. The stability of this
explicit scheme is based on strict numerical criteria to control the magnitude of the variable
computational timestep. For a discussion of the supplemental equation necessary for
hyperconcentrated sediment flow refer to the FLO-2D Reference Manual.

2.2 Solution Algorithm - How the Model Works

The differential form of the continuity and momentum equations in the FLO-2D model are solved with a
central, finite difference numerical scheme. This explicit algorithm solves the momentum equation for
the flow velocity across the grid element boundary one element at a time. The solution of the
differential form of the momentum equation results from a discrete representation of the equation
when applied at a single point. Explicit schemes are simple to formulate but are limited to small
timesteps by strict numerical stability criteria and this is the case for FLO-2D.
The solution domain in the FLO-2D model is discretized into uniform, square grid elements. The
computational procedure for overland flow involves calculating the discharge across each of the
boundaries in the eight potential flow directions (Figure 1) and begins with a linear estimate of the flow
depth at the grid element boundary. The estimated boundary flow depth is an average of the flow
depths in the two grid elements that will be sharing discharge in one of the eight directions. Other
hydraulic parameters are also averaged between the two grid elements to compute the flow velocity
including flow resistance (Mannings n-value), flow area, slope, and wetted perimeter. Using the
average flow area between two elements, the discharge for each timestep is determined by multiplying
the velocity times flow area.

Figure 1. Unconfined Discharge Computations

The full dynamic wave equation is expressed as a second order, non-linear, partial differential form in
the FLO-2D model. To solve the equation, initially the flow velocity is calculated at the grid element
boundary with the diffusive wave equation using the energy slope (bed slope plus pressure head
gradient). This velocity is then used as a first estimate (or a seed) in the second order Newton-Raphson
tangent method to determine the roots of the full dynamic wave equation (James, et. al., 1985).
Mannings equation is used to compute the friction slope. If the Newton-Raphson solution fails to
converge after 3 iterations, the algorithm defaults to the diffusive wave solution.
In the full dynamic wave momentum equation, the local acceleration term is the difference in the
velocity over the previous timestep for a given flow direction. The convective acceleration term is
evaluated as the difference in the flow velocity across the grid element from the previous timestep. For
example, in the notation below, the local acceleration term (1/g*V/t) for grid element 251 in the east
(2) direction converts to:
(Vt Vt-1)251 (g * t)
where Vt is the velocity in the east direction for grid element 251 at time t, Vt-1 is the velocity at the
previous timestep (t-1) in the east direction, t is the timestep in seconds, and g is the acceleration due
to gravity. A similar construct for the convective acceleration term (Vx/g*V/x) can be made where V2
is the velocity in the east direction and V4 is the velocity in the west direction for grid element 251:
V2 * (V2 V4)251 (g * x)
Subsequent to Version 2009, the FLO-2D model also includes diffusion terms in the full dynamic wave
solution which helps to dampen the velocity oscillations for certain flow conditions:
Di (V2 - 2. * Vnew + V4)251 Length2
where Di = hardwired coefficient (~0.25), Vnew = seed velocity, and Length is either the channel length or
flow path within the grid element.
The discharge across the grid element boundary is computed by multiplying the velocity times the cross
sectional flow area. After the discharge is computed for all eight directions, the net change in discharge
(sum of the discharge in the eight flow directions) in or out of the grid element is multiplied by the
timestep to determine the net change in the grid element water volume (see Figure 1). This net change
in volume is then divided by the available surface area (Asurf = storage area) on the grid element to
obtain the increase or decrease in flow depth h for the timestep. The channel routing integration is
performed essentially the same way except that the flow depth for next timestep (i+1) is a function of
the channel cross section geometry and there is usually only one upstream and one downstream
channel grid element for sharing discharge.

i+1
Qx = Qn + Qe + Qs + Qw + Qne + Qse + Qsw + Qnw Asurf h/t

where: Qx = discharge across one boundary

Asurf = surface area of one grid element
h/t = change in flow depth in a grid element during one timestep

To summarize, the solution algorithm incorporates the following steps:

1. The average flow geometry, roughness and slope between two grid elements are computed.
2. For computing the velocity across a grid boundary for the next timestep (i+1), the flow depth dx is a
linear estimate (average depth between two elements) from the previous timestep i.
i+1
i
i
d x = d x + d x1

3. The first estimate of the velocity is computed using the diffusive wave equation. The only unknown
variable in the diffusive wave equation is the velocity for overland, channel or street flow.
4. The predicted diffusive wave velocity for the current timestep is used as a seed in the NewtonRaphson solution to solve the full dynamic wave equation for the solution velocity. The NewtonRaphson convergence criteria for the root solution is 0.01.
5. The discharge Q across the boundary is computed by multiplying the velocity by the cross sectional
flow area (flow width times depth for overland flow). The flow width is adjusted by the width
reduction factors (WRFs).
6. The incremental discharge for the timestep across the eight boundaries (or upstream and
downstream channel elements) are summed:

Qix+1 = Qn + Qe + Qs + Qw + Qne + Qse + Qsw + Qnw

and the change in volume (net discharge x timestep) is distributed over the available storage area
within the grid or channel element to determine an incremental increase in the flow depth.

d ix+1 = Qix+1 t /Asurf

where Qx is the net change in discharge in the eight floodplain directions for the grid element for
the timestep t between time i and i + 1.
7. The numerical stability criteria is then checked for the new grid element flow depth. If any of the
stability criteria are exceeded, the simulation time is reset to the last successful simulation time, the
timestep increment is reduced, all the previous timestep computations are discarded and the
velocity computations begin again.
8. The simulation progresses with increasing timesteps until the numerical stability criteria are
exceeded.
In the overland flow computational sequence, first the grid system inflow discharge and rainfall is
computed, then infiltration, evaporation and rainfall runoff in 8-directions is calculated. The channel and
street flow are computed by separate components. After the flood routing has been completed, the
numerical stability criteria are tested for every floodplain, channel or street element. If stability criteria
of any element is exceeded, the timestep is reduced by a functional relationship that depends on the
previous history of stability success and the computation sequence is restarted. If all the numerical
stability criteria are successfully met, the timestep is increased for the next grid system computational
sweep. A flow chart of the solution sequence is shown in Figure 2. At the end of each timestep, the
volume conservation is computed.

FLO-2D Flow Chart

Start

Read Input Data

Update Hydraulics,
Volumes, Output Files,
Increase Timestep

Decrease Timesteps,
Reset Hydraulics, Restart
Flood Routing

Initialize Variables

Start Flood
Routing Loop
Channel Subroutine,
Hydraulic Structures,
Mudflow, Sediment,
Transport, Infiltration

Yes

Channel/Street and
Floodplain Interface

Channel
Flow
Channel Stability
Criteria Satisfied

No

No

Sediment Distribution on
Channel/Floodplain Bed

Yes
Rainfall Runoff
and Evaporation

Yes

Rainfall and Evaporation

Subroutines

No

Output Interval
Complete

Overland Flow
Sediment Transport
Infiltration, Gully Flow,
Hydraulic Structures,
Mudflow

Yes

Simulation
Time Complete
Numerical Stability
Criteria Satisfied

No
Yes

Yes
Street Flow

No

End

Yes

Numerical Stability
Criteria

No

No

Yes

Figure 2. FLO-2D Numerical Stability Flow Chart

No

2.3 Numerical Stability Criteria Controlling the Computational Timestep

The FLO-2D timestep varies with the numerical stability criteria. The FLO-2D flood routing scheme
proceeds on the basis that the timestep is small enough to insure numerical stability and yet is
sufficiently large to complete the model simulation in a reasonable time. The numerical stability criteria
are checked for the every grid element on every timestep to ensure that the solution is stable. If the
numerical stability criteria are exceeded, the timestep is decreased and all the previous hydraulic
computations for that timestep are discarded. Most explicit schemes are subject to the CourantFriedrich-Lewy (CFL) condition for numerical stability (Jin and Fread, 1997). The CFL condition relates the
floodwave celerity to the model time and spatial increments. The physical interpretation of the CFL
condition is that a particle of fluid should not travel more than one spatial increment x in one timestep
t (Fletcher, 1990). FLO-2D uses the CFL condition for the floodplain, channel and street routing. The
timestep t is limited by:
t = C x / (V + c)
where:
C is the Courant number (C 1.0)
x is the square grid element width
V is the computed average cross section velocity
c is the computed wave celerity
The coefficient C will typically vary from 0.3 to 1.0 depending on the type of explicit routing algorithm.
When C is set to 1.0, artificial or numerical diffusivity is theoretically zero for a linear convective
equation (Fletcher, 1990). The Courant criteria is a sufficient condition to control numerical surging for
almost all flood simulations. The default C value of 0.6 is recommended for the FLO-2D model, but the
user can reduce the Courant number in 0.1 increments for subsequent simulations where numerical
instability is noted. A value of C less than 0.3 is unnecessary for virtually all simulations.
Previous to Version 2009, the Courant value C = 1.0 was hardwired in the FLO-2D model. This was an
attempt to allow the model to run as fast as possible. For nonlinear equations, it is not possible to
completely avoid artificial diffusivity or numerical dispersion with C = 1.0 (Fletcher, 1990). Other
numerical stability criteria was necessary. One stability criteria that was used in the FLO-2D model was
a simple grid element percent change in depth (DEPTOL) from the previous timestep. A percent change
in depth greater than 20% typically resulted in numerical instability. A second stability parameter that
was applied in Versions 2009 and earlier was developed by Ponce and Theurer (1982). This criteria is a
function of bed slope, specific discharge and grid element size and is referred to as the dynamic wave
stability criteria. It is expressed as:
t < So x2 / qo
where qo is the unit discharge, So is the bed slope and is an empirical coefficient (WAVEMAX). The
coefficient was created as a variable unique to the grid element and is adjusted by the model during
runtime. Similar to the CFL criteria, when this numerical stability is exceeded, the hydraulic

10

computations for that timestep are discarded and the timestep is decreased. The dynamic wave
stability criteria was determined to be useful for maintaining numerical stability in channel confluences
and other unique channel flow conditions.
It has been determined that the Courant Number is more effective in controlling numerical surging
parameter for most FLO-2D flood simulations than the combined application of the percent change in
depth DEPTOL and WAVEMAX stability. While the Courant condition is a necessary condition for
solution convergence, it is not always sufficient to guarantee numerical stability. The WAVEMAX
stability parameter may still have some value in controlling numerical instability for some unique flow
conditions such as channel confluence flow or split flows. With Version 2009 for most flood simulations,
however, the DEPTOL and WAVEMAX stability criteria are not necessary if the CFL criteria is properly
applied.
The key to efficient computational flood routing is to have the largest timesteps possible and still avoid
numerical instability. Timesteps for most FLO-2D simulations typically range from 0.1 to 30.0 seconds.
Each flood simulation starts with a timestep set to 1.0 second and increases it until the numerical
stability criteria is exceeded, then the timestep is decreased until the minimum timestep is reached. If
the minimum timestep is not small enough to maintain numerical stability, then the minimum timestep
can be reduced, the numerical stability criteria can be adjusted or the input data can be modified. The
timesteps are a function of the discharge flux for a given grid element and its size. Small grid elements
with a steep rising hydrograph and large peak discharge require small timesteps. Accuracy is not
compromised for small timesteps, but the computational time can be long for a large grid system.
2.4 Courant Number Guidelines
The Courant Number is assigned by the user in the TOLER.DAT file line 2 as follows:
Line 1.

0.1

0.2

Line 2.

0.6

1.0

(TOL, DEPTOL, and WAVEMAX values)

(Line 2 is optional; C is a line character and 0.6 is the Courant
Number)

Line 2 is optional and if left unassigned, the default Courant Number is 0.6. A typical range of the
Courant Number is 0.3 (for a slower more stable model) to 1.0 (for a faster less stable model). The
default value of 0.6 is recommended as a starting value. For models that appear to be unstable,
reducing the Courant Number will help to control or eliminate the numerical surging.
Using the VELTIMEFP.OUT and VELTIMEC.OUT files that list the grid element or channel element
maximum velocity in descending order, unreasonable velocities can be identified. Based on the model
results, the following guidelines are recommended:
1. For the initial simulation:

Courant Number
DEPTOL
WAVEMAX

= 0.6
= 0.0
= 0.0

2. If the model has no numerical surging or unreasonable maximum velocities and it is desired to
have the model run faster increase the Courant Number to 0.7 or 0.8.

11

3. If the model has some numerical instability, decrease the Courant Number to 0.3 - 0.5.
4. If the model has numerical surging in the channel, set the WAVEMAX to 0.25 (this will reduce
the timestep and slow the model down).
5. After a flood simulation is complete, review the TIME.OUT file to determine which of the
stability criterion is slowing down the model.
6. It may be necessary to experiment with short duration simulations to determine which
combination of the stability criteria results in the fastest stable model.
2.5 The Importance of Volume Conservation
A review of a flood model results begins with volume conservation. Volume conservation is an
indication of model accuracy and numerical stability. Many numerical models either do not conserve
volume or do not report on volume conservation. In the FLO-2D model, the inflow volume, outflow
volume, change in storage and infiltration and evaporation losses from the grid system are summed at
the end of each time step. The difference between the total inflow volume and the outflow volume plus
the storage and losses is a measure of the volume conservation. FLO-2D volume conservation results
are written to the output files and to the screen at run time for the user specified output time intervals.
Any FLO-2D flood simulation not conserving volume must be revised. It should be noted that volume
conservation in any model flood simulation is not exact. Data errors, numerical instability, or poorly
integrated components may contribute to a loss of volume conservation. While some numerical error is
introduced by rounding numbers, approximations or interpolations (such as with rating tables), volume
should be conserved within a fraction of a percent of the inflow volume. The user must decide on an
acceptable level of error in the volume conservation. Most FLO-2D simulations are accurate for volume
conservation within a few millionths of a percent. Generally, volume conservation within 0.0001
percent or less can be considered as a successful flood simulation.
2.6 Organization of the FLO-2D Model
The FLO-2D software package includes a grid developer system (GDS) for generating data input files, a
Mapper program that creates plots of output data and flood maps, and the FREQPLOT program to
analyze flood frequency. The GDS will filter DTM points, interpolate the DTM data and assign elevations
to grid elements. It also graphically creates and edits component data such as channels, streets and
buildings. The Mapper program automates flood hazard delineation. Mapper will plot very detailed
flood inundation color contour maps and shape files. It will also replay flood animations and generate
flood damage and risk maps.
The FLO-2D manual is divided into a series of four documents. The Reference Manual is devoted to
model theory and component description. The Data Input Manual is subdivided into a series of data file
sections devoted to variable definitions, ranges, descriptions and data input instructional comments.
Separate manuals are devoted to the application of the GDS and Mapper.

12

2.6.1 Grid Developer System (GDS)

The Grid Developer System (GDS) creates and edits the FLO-2D data files and provides a platform for
running the model and other pre- and post-processor programs. The GDS is a pre-processor program
that will overlay the grid system on the DTM points, and interpolate and assign elevations to the grid
elements. Geo-referenced aerial photos, shape file images or maps can be imported as background
images to support the graphical editing. The GDS also provides important editorial features including
the assignment of spatially variable grid element attributes such channels, levees, streets, infiltration,
area and width reduction factors, floodplain elevation and roughness, inflow and outflow nodes and rill
and gully geometry. It allows selection of individual elements or large groups of nodes for editing.
Rainfall can also be spatially varied. Detailed application instructions are presented in the GDS Manual.
2.6.2 Input Data
The FLO-2D model requires data in a series of ASCII text files representing model control, topography or
component attributes. The ASCII text files are generated by the GDS or can be viewed and edited from
any ASCII editor such as MS WordPad or TextPad. The model and the processor programs can be run
from the GDS, but they do not need the GDS to run a flood simulation. Specific data input instructions
are presented in the Data Input Manual.
2.6.3 Output Options
The FLO-2D model organizes the results into a series of output files. Floodplain, channel and street
hydraulics are written to file. Hydraulic data include water surface elevation, flow depth, velocities, and
discharge in the eight flow directions. Channel output data is extensive and is reported by channel
element and timestep. Discharge for specified output intervals (hydrographs) are written to various
files. A mass conservation table comparing the inflow, outflow and storage in the system is presented in
the SUMMARY.OUT file. Complete descriptions of the output files are presented in the Data Input
Manual.
Most of the FLO-2D output data can viewed graphically with data organized by spatially variable
maximum values or temporal output for animations. Graphical display of the flow depths can be viewed
on the computer screen during a simulation to visualize the progression of the floodwave over the grid
system. In addition to the predicted flow depths, an inflow hydrograph or rainfall cumulative
precipitation will be plotted.
Mapper is the primary program for graphically displaying the FLO-2D results. It can create high
resolution color contour plots. Several types of maps can be created: grid element or DTM point plots,
line contour maps and shaded contour maps. Maps can be created for ground contours, maximum
water surface elevations, maximum floodplain flow depths, maximum velocities, maximum static and
dynamic pressure, specific energy, and floodway delineation. One of the most important features of
Mapper is its capability to create flood depth plots using the DTM topographic points. When the user
activates this feature, Mapper will subtract each DTM ground point elevation from the grid element
floodplain water surface elevation. The resultant DTM point flow depths can then be interpolated and

13

plotted as shaded color contours. GIS shape files (*.shp) are automatically created with any plotted
results. The GIS shape files can be then be imported into ArcView or other GIS programs. The Mapper
features and functions are described in its own manual. Some of the Mapper features include:

Importation of multiple geo-referenced aerial photos in various graphic formats such as TIFF,
BMP, JPG, etc.;

Multiple layer capability including control of layer properties;

Cut and view flow depth and topography profiles;

Flood damage assessment component to compute the flood damage as a function of the FLO-2D
predicted maximum depths, building shape files and building value tables (dbf file);

Flood animation or the floodwave progression over the grid system;

Flood maximum deposition and scour;

Maximum flow velocity vectors;

Hazard maps based on flood intensity and frequency.

For creating FEMA Flood Insurance Rate Maps, the Mapper processor program has a Digital Flood
Insurance Rate Map (DFIRM) optional tool that was created by Anderson Consulting Engineers of Fort
Collins, Colorado. This tool will convert the FLO-2D water surface line contours to those used on a
DFIRM base map. All the required DFIRM symbols and map annotation tools are available. This is an
excellent and precise tool to create the FEMA DFIRMs.
Graphical displays of the output hydrographs and water surface profiles are provided in the HYDROG
and PROFILES post-processor programs. HYDROG will plot the hydrograph for every channel element or
selected floodplain cross section. HYDROG can also be used to evaluate the average channel hydraulics
in a given reach. The user can select the reach upstream and downstream channel elements and the
program will compute the average of all the channel hydraulics in the reach including: velocity, depth,
discharge, flow area, hydraulic radius, wetted perimeter, top width, width to depth ratio, energy slope,
and bed shear stress. The PROFILES program plots channel water surface and bed slope profiles.
2.7 FLO-2D Model Assumptions and Approximations
The basic assumptions and approximations associated the FLO-2D model are:

Steady flow for the duration of the timestep;

Grid element uniform elevation and roughness;

One-dimensional, depth averaged channel flow;

A channel element is represented by uniform channel geometry and roughness;

Hydraulic roughness that is based on steady, uniform turbulent flow resistance;

Hydrostatic pressure distribution;

Hydraulic structure (bridge, culvert, weir) discharge that is approximated by a rating curve or
table;

Pressure pipe flow is not simulated except by a hydraulic structure rating curve or table.
These assumptions and approximations are self-explanatory and do not limit the model from accurately
predicting flood inundation and water surface elevations.

14

III. FLO-2D Model Validation

3.1 Overview
Following IAHR (1994) guidelines for documenting model validation, this chapter presents the FLO-2D
model representation of the flood physical system, the validation approach, the computational core
validation, range of model applicability, and the accuracy of the model results. This section is organized
in the IAHR format of first presenting a validation claim which is then followed by a section on the
substantiation of the claim. There will be twenty-one validation claims followed by a proof or
substantiation of the claim. For further discussion, additional details of the validation case studies are
presented in Chapter IV.
FLO-2D is a comprehensive modeling tool with numerous interactive components. It is the most
comprehensive flood routing model available. The purpose of this document is to discuss the reliability
and accuracy of model results. It is not meant to provide the mathematical proof of the results for a
specific model application although several analytical solution results are presented.
The FLO-2D model has been used on thousands of projects worldwide. The model has been the subject
of masters theses and doctoral dissertations and university and institute laboratory tests. It has been
used by federal agencies and consultants for flood mitigation design, FEMA flood insurance studies,
replication of historical events, dam safety emergency action plans, and for local flood district landuse
regulation and zoning.
3.2 Representing Physical System
The FLO-2D model simulates all of the important physical processes of the hydrologic system from
rainfall-runoff to the prediction of river hydraulics and sediment transport. This includes:

Overland, unconfined flow in two dimensions

Channel flow as a one-dimensional depth averaged flow
Street flow as a shallow, rectangular channel
Channel floodplain exchange
Levee flow confinement
Levee and dam breach
Hydraulic structures
Floodplain surface storage loss and flow path obstruction
Rainfall and runoff hydrologic cycle
Rainfall Infiltration using Green-Ampt and SCS
Transmission loss using Green-Ampt
Evaporation
Alluvial fan distributary channels
Sediment transport
Mud and debris flows
Groundwater surface water exchange (using MODFLOW)

Each FLO-2D application involves a unique combination of physical components involving hydrology
(rainfall and inflow hydrographs), flow hydraulics (roughness or water surface control) and topographic
variability (elevations, slope and cross sections). The validation of the primary physical process
15

components will be discussed. The main issues in validating and verifying the FLO-2D model for the
various combinations of the physical components relate to:

Appropriate spatial and temporal discretization;

Complex interaction of component details (e.g. urban flooding);
Lack of measured data to support model verification (especially for 2-dimensional flow).

The crucial questions regarding model validation and verification using project data bases are:
1.
2.
3.
4.
5.

Does the data base support model validation/verification?

How accurate was the hydrology (both rainfall and inflow hydrographs)?
Does the model conserve volume?
Is the model numerically stable?
Was the model adequately calibrated for both low and high flows?

Accurate model replication of a historical flood event depends on the rainfall or flood inflow
hydrograph(s) data base. The area of inundation is more dependent on the volume of the inflow
hydrograph (overall shape) and rainfall than the peak discharge that may define only a small portion of
the inflow hydrograph. For example, in some alluvial fan applications, the fan area may be comparable
to the watershed area and the fan rainfall may be critical to estimating the total area of inundation.
Infiltration and transmission losses can have a similar impact on the area of inundation and could affect
both model calibration and verification.
Model numerical instability surging occurs when the timestep is too large for the relationship between
the discharge flux and the potential surface area of a grid element. A high discharge flux on a small
surface area can lead to growth of numerical oscillation. The velocity and therefore the discharge may
become unreasonable for one or more timesteps. This instantaneous surge may not affect the overall
distribution of the flood, but it does require model adjustments to avoid reporting invalid maximum
flow depths or velocities. The model automatically reduces the computational timestep based on the
selected stability criteria that can be adjusted to preserve numerical stability. Very small timesteps,
however, can lead to long flood simulation computer runtimes. Long simulations can be avoided by
selecting an appropriate size grid system. There is a tradeoff between model speed (large grid
elements) and mapping resolution (small grid elements). When selecting grid element size, the FLO-2D
manual suggests that:
QpeakAsurf < 1.0 cfs/ft2
where Qpeak = peak discharge and Asurf = surface area of one grid element.
To verify the model for a given set of known or measured flow conditions, it may be necessary to adjust
some of the components that represent water surface control, available floodplain storage, flow path
obstruction, flow distribution, volume gains and losses, and flow confinement. Data bases with
highwater marks, areas of inundation and gage recording may be influenced by:

Volume gains and losses rainfall, tributary inflow, infiltration, evaporation

Flow confinement levees and berms
Loss of storage buildings or other manmade features
16

Water surface control - hydraulic structures

Flow distribution streets, canals, channels

Complex urban or 2-D unconfined flood simulations typically cannot be calibrated and it is necessary to
perform either validation or verification with simpler models that isolate a specific component.
Furthermore, it is rare that the combination of discharge measurements and corresponding high water
marks are available for an infrequent flood event. A data base where there is corresponding flood cross
section surveys with measured discharge data is exceedingly rare. Validation is often limited to known
analog solutions or flume studies.
3.3 Validity of Computational Core
The flood routing portion of the model essentially includes three core subroutines: channel routing,
floodplain routing and the channel floodplain exchange. The street routing component uses the channel
routing component to simulate a shallow rectangular channel and does not require separate validation.
An assessment of the validity of the computation core routines starts with a volume conservation
report. Every FLO-2D flood simulation must conserve volume or the simulation is inaccurate due to user
input data errors. FLO-2D reports on volume conservation to the computer screen during the flood
simulation and then writes the volume conservation results in a comprehensive output file
(SUMMARY.OUT) that can be reviewed after each computer run. The first FLO-2D validation claim and
substantiation proof is follows:

Claim 1: FLO-2D conserves volume and reports on

volume conservation.
Substantiation of Claim 1: In the SUMMARY.OUT file (Figure 3a) that is generated for every FLO-2D
simulation, the volume conservation is listed by output interval and the ultimate disposition of all of the
volume at the end of the simulation is reported. This particular project has a number of components
including channel routing, floodplain inundation, infiltration and evaporation. The complexity of
interactive FLO-2D components requires that volume conservation be reviewed at the end of each
simulation. It should be noted that any flood routing model that does not generate a volume
conservation report could be suspect of generating or losing volume. FLO-2D tracks the volume for the
complete model. Isolated project control surfaces or prescribed areas for volume conservation are not
required.
For the case of very large urban project (discussed later in the report), the SUMMARY.OUT file is
presented for the City of Camarillo FEMA flood mapping simulation in Figure 3b. This table indicates a
volume conservation error of only 61 millionths of 1 percent. With over 1 million grid elements, this
project represents one of the largest flood simulations ever performed and the volume conservation
error in the model is negligible.

17

Figure 3a. SUMMARY.OUT File Report.

Figure 3b. SUMMARY.OUT File for the Camarillo Project

18

3.3.1 Validity of the Computation Core - Overland Flow

Overland flow is computed in 8-directions (4 compass and 4 diagonal directions) to predict unconfined
floodplain or alluvial fan flooding as shown in Figure 1. The full dynamic wave momentum equation is
applied to one flow direction at a time. Overland flow velocities and depths vary with topography and
the grid element roughness. The assignment of overland flow roughness must account for vegetation,
surface irregularity, non-uniform and unsteady flow. Roughness is also a function of flow depth in the
FLO-2D model.
FLO-2D can simulate an unconfined floodwave progression over a dry flow domain without specifying
any boundary conditions. No hot starts or prescribed water surface elevations are required. There is no
limit to the number of inflow hydrograph elements. Outflow nodes can be designated but they require
no specific boundary conditions or water surface control. Outflow from the grid system is approximated
as normal depth flow using a weighted average flow conditions of the contiguous upstream nodes.
If no inflow flood hydrograph data is available, FLO-2D can perform as a watershed model. Rainfall can
occur on the floodplain surface resulting in sheet runoff after infiltration losses have been computed. It
is possible to simulate rainfall while routing a flood event and have the rain fall on wetted surfaces. To
improve concentration time, rill and gullies can be modeled to exchange flow between grid elements.
This will reduce the travel time associated with sheet flow exchange between grid elements. Uniform or
spatially variable rainfall such as NEXRAD storm data can be simulated.
The objective of an unconfined flood simulation is typically to assess the area of inundation which is
primarily controlled by the flood volume. Important flood routing details include topography, spatial
variation in infiltration and roughness, flow obstructions, levees, hydraulic structures and streets. The
speed of the floodwave progression over the floodplain can be controlled by floodplain roughness. In
urban areas, street flow may affect the shallow flood distribution and buildings can reduce floodplain
storage. Floodwalls or privacy can obstruct flow paths. The levee routine can be used to simulate
berms, elevated road fill, railroad embankments or other topographic features to confine or redirect the
flow on the floodplain. Hydraulic conveyance facilities such as culverts, bridges and weirs may control
the local water surface elevations and cause backwater effects. Alluvial fan flooding is a specific case of
unconfined flooding that may be also be urbanized.
Sediment transport and mobile bed scour and deposition can be modeled for overland flow. For
infrequent flood events such as the 100-year event, sediment transport is generally inconsequential to
the flood distribution and water surface elevations. Since alluvial channels reformulate with the
channel forming discharge on the order of the 2-yr to 5-yr flood, sediment transport is more important
to the more frequent floods. For design floods such as the 100-yr flood, sediment transport modeling is
unnecessary. Flood events with high sediment concentration can evolve into mud and debris flows,
usually on alluvial fan surfaces.
For mud flow simulations, there are two methods for loading the hydrograph with sediment:. 1) A
sediment concentration by volume can be assigned to a discretized time interval of the inflow
hydrograph; 2) Load the inflow hydrograph increments with a volume of sediment. Once the

19

hydrograph is bulked with sediment, the mudflow is routed as a water and sediment continuum over
the hydrograph. The same water routing algorithm is used for mudflows but the momentum equation is
solved with the additional viscous and yield stress terms. The bulked sediment hydrograph is tracked
through system conserving volume for both water and sediment. Flow cessation and flow dilution are
possible outcomes of the mudflow routing.
The only distinction between unconfined floodplain and alluvial fan flooding is slope and as such it is not
necessary to have any special considerations for alluvial fan flooding. For design flood events, alluvial
fan flood inundation is dictated by topography and roughness and where appropriate by urban features
or hydraulic structures. Thus FLO-2D is an appropriate model for alluvial fan flooding as well as riverine
applications. The first FLO-2D 1999 certification for FEMA Flood Insurance Studies (FIS) flood studies by
the Corps of Engineers was for unconfined flooding on alluvial fans (see the Appendix for the Corps'
Certification Letter).

Claim 2: The FLO-2D model accurately simulates

unconfined overland flow and will yield an exact
solution for steady, uniform overland flow.
Substantiation of Claim 2: A simple flume model validates that the FLO-2D model will generate a known
analog solution for overland flow. An independent consulting firm prepared this Case Study model. The
flume is 4 elements wide by 60 elements long with a constant inflow discharge and uniform roughness.
Any grid element size, slope, n-value and steady inflow discharge can be used for this case study. The
steady, uniform flow solution can be determined from the full dynamic wave equation by setting the
diffusive pressure term and the acceleration terms to zero and evaluating the flow depth and velocity
from Manning's equation. These additional momentum acceleration terms are actually computed in the
model but are negligibly small. The predicted final steady, uniform flow depth and velocity are 6.82 ft
and 5.86 fps respectively (see Figure 4). Case Study 1 is further discussed in Chapter IV where the FLO2D results are compared with the analog solution.

Claim 3: The FLO-2D model Version 2009 can

replicate the exact solution from Version 2007
and previous versions for the flume case.
Substantiation of Claim 3: The flume Case Study 1 was run with both FLO-2D Versions 2007 and 2009
and the results are identical matching the analytical solution. This flume example was prepared by an
independent consulting firm in 1999 and the subsequent model versions have been in training classes
since then to demonstrate that FLO-2D can replicate a known analog solution for steady uniform flow.
The comparative results shown in Figure 5 prove that the computation core for overland flow has been
correct and consistent throughout all versions since 1999.

20

DEPTH = 6.82 ft

VELOCITY = 5.86 fps

Figure 4. FLO-2D Version 2009 Results for Maximum Velocity and Depth

Figure 5. Version 2007 Flume Case Results Showing

Identical Results as Version 2009 in Figure 4.

21

Claim 4: FLO-2D can replicate unconfined overland flow

in urban areas if the flood volume is accurately known.
Substantiation of Claim 4: In early January 1997, rainfall on snowpack and unusually warm weather
conditions resulted in flooding of the Truckee River through downtown Reno, Nevada and the nearby
City of Sparks. Fifteen miles of river were simulated with FLO-2D to predict the area of inundation
(Figure 6). An extensive hydrographic data base was compiled including over 100 channel post-flood
cross section surveys. Data for model calibration included two USGS river gage flood stages and
discharges, a relatively accurate Truckee River inflow hydrograph from an upstream age and numerous
surveyed highwater marks through the inundated area. The inflow from three tributaries was estimated
by the Corps of Engineers. In the Sparks Truckee Meadows area, significant overbank flooding occurred
in a sparsely populated area before the river entered a canyon to the east. The overbank flows from the
river and two tributaries commingled on the Truckee Meadows floodplain. Rainfall, infiltration, 18
bridges and a number of streets and buildings added detail to the simulation. Channel roughness values
were estimated from available HEC-2 models. Spatial variation in floodplain n-values and storage loss
ARF values were assigned for buildings. The filling of a large gravel pit was also simulated.

Figure 6. Truckee River Observed (Black Outline) and FLO-2D Predicted Area of Flood Inundation

22

The observed area of inundation in Figure 6 is displayed as a black outline. A comparison of fifty
surveyed high water marks and the FLO-2D predicted maximum water surface elevations resulted in
an average difference of only 0.2 ft (0.06 m). There are three important verification anchors to this
project:

The n-values were calibrated to match the original HEC-2 water surface elevation results (Figure
7) and the recorded stage at the USGS gages (Figure 8 and 9);

The predicted area of inundation matches the flooded developed from aerial photographs and
observations very well as shown in Figure 6;

The timing of the floodwave movement through the system matches the downstream Vista gage
recorded stage almost exactly (Figure 9).

Figure 7. illustrates that the FLO-2D predicted water surface elevation matches the HEC-2 water surface
profiles almost exactly through the urban areas. The City of Sparks, the City of Reno and the Truckee
Meadows are located at the downstream end of the profile in Figure 7 (the last 20,000 ft of the model
indicated by the flat slope). Figures 8 and 9 display the predicted versus measured stage reading for the
two USGS gages in operation during the 1997 flood. These stage recordings represent an accurate
assessment of the rise and fall of the hydrograph through the urban areas.

Bed and Water Surface Profiles:

HEC-2
FLO-2D

Figure 7. Predicted Truckee River Water Surface Profiles for HEC-2 and FLO-2D

23

Claim 5: FLO-2D accurately computes the overland

flood routing and channel-floodplain exchange.
Substantiation of Claim 5. Continuing with the Truckee River case study as an indication of the accuracy
of the channel-floodplain interface, Figures 8 and 9 display the predicted versus the measured stage
reading for the two USGS gages in operation during the 1997 flood. The Reno gage is located about the
beginning of the flat profile reach just downstream of the City of Reno and upstream of the Truckee
Meadows area. The Vista gage is located at the downstream end of the simulated reach just before the
Truckee River enters the canyon. Upstream of the Reno gage, there is limited overbank flooding and
some return flow from downtown Reno. The new Vista gage includes almost all of the return flow from
the overbank discharge in the Truckee Meadows area. There is some ungaged inflow that reaches the
new Vista gage that is not simulated in the model, nevertheless, the predicted stages correlate well with
the measured stage at both gages in both magnitude and timing. This is an key model verification
because the progression of the floodwave over Truckee Meadows from both the Truckee River and
Steamboat and Boynton Slough has a significant flood crest lag time of up to 6 hours when compared to
the floodwave movement in the river. The channel conveyance capacity is less than half of the
estimated peak discharge at the new Vista gage. In other words, most of the flood hydrograph is
moving over the floodplain between the gages and yet the predicted FLO-2D stage at the lower gage
matches the measured stage almost exactly in magnitude and timing. The FLO-2D model must
accurately predict the volume of overbank flow, the channel flood routing, the overland flood routing
and channel-floodplain exchange in order to match the stage at both USGS gages. This project verifies
the model capability to compute channel-floodplain discharge exchange both as overbank flooding and
return flow to the channel. For more discussion of this project, see the Case 2 description in Chapter IV.

Figure 8. FLO-2D Predicted Stage for the 1997 Flood Event at the Reno Gage

24

Figure 9. FLO-2D Predicted Stage for the 1997 Flood at the New Vista Gage

Claim 6: FLO-2D Version 2009 replicates previous

version results.
Substantiation of Claim 6. FLO-2D Version 2009 was used to simulate the 1997 flood and the results are
compared with Version 2000 results as shown in Figure 10. The computed hydrograph results are
shown for the New Vista gage near the end of the model. There is virtually an exact replication of the
measured New Vista gage stage and the previous version FLO-2D simulated stage. The more stable
Version 2009 also eliminates some of the minor surging near the peak discharge and on the recessional
limb of the hydrograph. In addition, the use of depth integrated n-values slightly delays the flood at the
New Vista gage on the rising and recessional parts of the hydrograph as shown. Nevertheless, the
Version 2009 model accurately replicates both the earlier FLO-2D version results and the measured
data.

25

25

Stage Depth (ft)

20

15

10

New Vista Gage

FLO-2D Data
v2009

0
0

10

Time (days)
Figure 10. FLO-2D Version 2009 Predicted Stage for the 1997 Flood at the New Vista Gage

Claim 7: Small grid elements models used for

urban flood simulations are numerically stable.
Substantiation of Claim 7: The City of Camarillo funded an FEMA FIS Study with a 10 ft grid system
model consisting of 1,093,307 grid elements (Case Study 3 Chapter IV). The original 2005 FEMA FIRM
mapping for Ventura County (Figure 11) depicted an area of inundation that encompassed most of the
project area.

26

Figure 11. FEMA 2005 FIRM MAP for Ventura County

In February 2006 the City appealed and in April 2006 FEMA agreed with the appeal and requested more
information and data. The new FLO-2D model FIS study with a 10 ft grid element resulted in the
predicted maximum flow depths in Figure 12 and the FIRM mapping in Figure 13. This highly detailed
model included loss of storage for 5,000 buildings, spatially variable n-values (including the streets),
channel modeling of numerous drains (2,395 channel elements) and numerous hydraulic structures to
simulate the culverts and bridges. The model conserved volume and a review of predicted maximum
velocities and depths indicated that there was no surging in the channel or on the floodplain. Output
hydrographs didn't display any numerical instabilities.
The level of detail in the Camarillo project is unmatched by any comparable two-dimensional flood
project. The City closely tracked and scrutinized this project with their consultant (Kasraie Consulting)
and made recommendations for adjustments based on their flood experience. They reviewed and
concurred with the FLO-2D predicted area of flood inundation. The City of Camarillo approved this FIS
FIRM map submittal to FEMA. For a more detailed discussion of this project see Study Case 3 in Chapter
IV.

27

Figure 12. FLO-2D Predicted Maximum Flow Depths for the City of Camarillo

Figure 13. FLO-2D 2011 Model FIRM for the City of Camarillo

28

Claim 8: FLO-2D accurately predicts the area of inundation for

unconfined overland flooding for a large river system.
Substantiation of Claim 8. The FLO-2D model was first applied to the Middle Rio Grande (MRG) in 1996.
Over the years, the model developed through the cooperation and funding from the U.S. Fish and
Wildlife Service (USFWS), the Albuquerque District of the Corps of Engineers (Corps), the Bureau of
Reclamation (BOR) and the New Mexico Interstate Stream Commission (ISC). Later the U.S.
International Boundary and Water Commission (IBWC) collaborated with the Corps to apply the model
to the Rio Grande south of Caballo Reservoir. Over 470 miles of the Rio Grande have been modeled
with FLO-2D from the confluence with the Rio Chama in New Mexico to Fort Quitman, Texas. Some of
the FLO-2D models developed since 1996 to investigate flooding of the Rio Grande are listed in Table 1
in chronological order.
Table 1. FLO-2D Models for the Middle Rio Grande
Project Name and Sponsor

Year

Location

Purpose

Isleta Mean Annual Flood Inundation FWS

1996

Isleta Dam to Belen Bridge

Floodplain Inundation

Refuge Flooding BOR & FWS

1998

BDANWR

Floodplain Inundation

MRG Model Corps, FWS, ISC

2002

Cochiti to Elephant Butte Res.

Flood Routing

Low Flow Model and High Flow Models - Corps

2002

Cochiti to Elephant Butte Res.

Model Calibration

AMAFCA Overbank Flooding - AMAFCA

2003

North Div. Channel To I-40

Flooding near Montano Br.

Levee Failure Project BOR

2003

Cochiti to Highway 44

Levee Failure Inundation

URGWOM Corps

2003

Cochiti to Elephant Butte Res.

URGWOM Support

RG Floodway Damage Reduction - Corps

2004

San Acacia to San Marcial

Levee Flood Hydrology

Albuquerque Levee Assessment - Corps

2004

Cochiti to Albuquerque

Levee Failure

Conceptual Restoration Plan SOBTF

2005

San Acacia to San Marcial

Restoration Plan

Levee Analysis and Redesign - IBWC & Corps

2006

Caballo Dam to American Dam

Flood Routing and Hazard

River Restoration Plan - Collaborative Program

2007

SA to SM, Isleta to Belen

Restoration Plans

Levee Analysis/ Channel Capacity - IBWC & Corps

2008

American Dam to Fort Quitman

Flood Routing

2009-10

Caballo Dam to American Dam

Levee Design and Failure

Levee Improvements - IBWC

Notes: SA = San Acacia, SM = San Marcial, BDANWR = Bosque del Apache National Wildlife Refuge, AMAFCA = Albuquerque
Metropolitan Arroyo Flood Control Authority, SOBTF = Save Our Bosque Task Force, Corps = Albuquerque District Corps of
Engineers, URGWOM = Corps Upper Rio Grande Water Operations Model, FWS = U.S. Fish and Wildlife Service, BOR = U.S.
Bureau of Reclamation, IBWC = U.S. International Boundary and Water Commission.

Since the completion of the Middle Rio Grande FLO-2D model from Cochiti Dam to the headwaters of
Elephant Butte Reservoir in 2003, the model has been used to support various Corps and IBWC flood
hazard and river operation studies. The FLO-2D model from Cochiti to Elephant Butte was used by the
Corps of Engineers to support their Upper Rio Grande Water Operations Model (URGWOM). Thirty
years of record were replicated with the calibrated FLO-2D model to predict water delivery to Elephant
Butte Reservoir. The Middle Rio Grande is about 173 miles in length from Cochiti Dam to Elephant Butte
Reservoir. Discharge hydrographs for seven gages throughout the reach for the 30 years of record were
developed. Figure 14 shows an example of the gage data replication. The model was also extended
upstream of Cochiti Dam to Abiquiu Reservoir on the Chama River. The release from the reservoir was
predicted at the confluence of the Chama River and the Rio Grande. The results are shown in Figure 15

29

which depicts the diurnal variations of the power plant release at Abiquiu Dam. FLO-2D predicted this
variation exactly in both magnitude and timing.
In 2005, the Corps implemented an intensive flow monitoring and data collection program for the spring
runoff Cochiti Dam releases that included aerial photography, video, cross section surveys and water
surface elevation measurements. This data base was used to develop a more detailed FLO-2D model.
The model was calibrated for future prediction of overbank flooding and floodwave attenuation. The
250 ft grid element model has 167,308 elements with 3,242 channel elements. This extensive data
collection effort during the controlled dam release enabled the FLO-2D model to be applied to the area
of floodplain inundation. The application used previous calibration infiltration and evaporation
parameters. For more discussion of this project, see Case Study 4 in Chapter IV.

4500

4000

3500

Discharge (cfs)

3000

2500

2000

1500

1000

500

USGS Discharge at San Felipe

FLO-2D Predicted Discharge at San Felipe

0
04/05/98

04/15/98

04/25/98

05/05/98

05/15/98

05/25/98

06/04/98

06/14/98

06/24/98

07/04/98

07/14/98

07/24/98

Date, MM/DD/YY

Figure 14. USGS San Felipe Gage 1998 (FLO-2D Predicted vs. Measured Discharge)

30

1600
1400
1200

Discharge (cfs)

1000
800
600
400
Gage Discharge
FLO-2D Predicted
Discharge

200

5/9/2003

5/8/2003

5/7/2003

5/6/2003

5/5/2003

5/4/2003

5/3/2003

5/2/2003

5/1/2003

4/30/2003

4/29/2003

4/28/2003

4/27/2003

4/26/2003

4/25/2003

4/24/2003

4/23/2003

4/22/2003

Date

Figure 15. FLO-2D Predicted vs. USGS Gage Measured Discharge at Chama River Confluence
Rio Grande aerial photographs were flown between roughly 10 am and 12 noon on June 1, 2005 when
the Cochiti Dam release was 5,300 cfs with the mean daily flows ranging from 5,040 cfs to 5,650 cfs. A
typical area of inundation comparison with the shape file and aerial photograph in the background is
shown in Figure 16. Flooded overbank areas shown in magenta were digitized by the Corps. The Corps
shape file area estimates involved interpretation of wet floodplain area, estimates of flooded areas that
were obscured by dense vegetation, and the timing of the photography with relationship to the
floodwave movement. The FLO-2D area of inundation is depicted by colored grid elements (mostly blue
and green). The accuracy of the of the FLO-2D predicted area of inundation is limited by the 250 ft grid
element size. Table2 compares the FLO-2D predicted area of inundation against the Corps shape file
estimates. The difference in the two estimates of the floodplain area of inundation is only 6 percent.
The first reach (Cochiti Dam to I-25) had the largest discrepancy because the Corps had difficulty
discerning the flooded area from the anatomizing channel in the aerial photos. In addition, the FLO-2D
model channel data did not accurately represent the complex channel braids in all areas because of lack
of cross section data.

31

Figure 16. FLO-2D Predicted Area of Inundation Compared with the Corps' Digitized Shape File for the
2005 Area of Inundation Monitoring Project

Table 2. Floodplain Area of Inundation Calibration

Reach

Corps Estimated Area of

Inundation (acres)

FLO-2D Estimated Area of

Inundation (acres)

% Difference

Cochiti Dam to I-25

155

207

33.8

I-25 to Belen

1746

1931

10.6

Belen to San Acacia

1013

999

1.4

San Acacia to San Marcial

4746

5026

5.9

7660

8163

6.2

Total

The FLO-2D prediction of the USGS gage hydrographs for the spring release flow monitoring period from
May 24 through June 1 are shown in Figures 17-20. The model had an initial ramp-up to the release
discharge as shown in each hydrograph plot. The FLO-2D model accurately replicated the discharge
hydrographs recorded at USGS gages for the June 2005 Cochiti Dam release, in part, because the
overbank storage volume has been accurately assessed. It should be noted that the overbank flows are
highly variable along the 128 mile reach of river and yet the model accurately predicts the discharge
within about 100 cfs out 6,000 cfs at all four gages located throughout the reach.

32

Discharge (cfs)
Discharge (cfs)

Figure 17. FLO-2D Predicted vs. USGS Gage Discharge at San Felipe May 24-June 1, 2005

Figure 18. FLO-2D Predicted vs. USGS Gage Discharge at Albuquerque May 24-June 1, 2005

33

Discharge (cfs)
Discharge (cfs)

Figure 19. FLO-2D Predicted vs. USGS Gage Discharge at Isleta Lakes May 24-June 1, 2005

Figure 20. FLO-2D Predicted vs. USGS Gage Discharge at San Acacia May 24-June 1, 2005
The model replication of the gage discharge deviates slightly from the measured data because the gages
are located in a sand bed channel and the gage rating curves shift over time. Scour and deposition in
the sand bed channel limits the accuracy of the reported gage data. Other causes of variability are
attributed to:

Too much or too little predicted floodplain storage;

Unmeasured or poorly measured irrigation diversion and return flow;

Unmeasured inflows from arroyo tributaries;

Variable infiltration storage;

Variation in bed form roughness;

Predicted evaporation losses based on mean monthly condition.

34

Evaporation during very hot or very cloudy days may vary significantly from the mean monthly
conditions, and this could result in several percent error in the discharge predictions at downstream
gages. It was concluded from these results that the model accurately replicated the 2005 overbank
monitoring data base. This project exemplifies that if the channel flood routing is accurate in time and
space, then the relationship between the volume of water in the channel and the predicted area of
floodplain inundation will be accurate.
3.3.2 Validity of the Computational Core - Channel Flow
The FLO-2D 1-D channel component has been validated in numerous studies using HEC-RAS and HEC-2
comparisons, flume studies and actual river data (as previously discussed for the Rio Grande). For FEMA
riverine Flood Insurance Studies, the Corps of Engineers prepared a certification letter in 2001
(presented in the Appendix) and the model has been used on numerous river FIS mapping projects. The
most difficult verification test is replicating gage data in both discharge magnitude (or stage) and timing.
Matching gage data is especially difficult for many rivers with dams because depth variable roughness
can play a role in timing of diurnal power releases. Three case studies are presented to validate the
channel routing hydraulics:

Operation Discharge of the California Aqueduct

Design flow in Camarillo Hills Drain

Tracking Green River Flaming Gorge Dam Power Release

It should be noted that since the street routing algorithm (as a rectangular channel) in the FLO-2D model
uses the channel routing subroutine, it is not necessary to present separate validation tests for the FLO2D street component.

Claim 9: FLO-2D accurately computes known

channel hydraulics.
3.3.2.1 Operational Flows in the California Aqueduct
Substantiation of Claim 9. Discharge in a 10 mile reach of the California Aqueduct near Huron, California
was simulated using 96 channel elements in a 500 ft square grid system. The aqueduct is essentially a
large outdoor flume and is one of the most difficult hydraulic simulations for a flood routing model
because of the high discharge flux, high flow depths, very low Froude number flow and very mild slope
(5.7 x 10-5). Numerical surging is often difficult to control in such cases. The design discharge for this
concrete lined trapezoidal channel was approximately 8,150 cfs. The roughness (Mannings n-values)
for most of the 10 mile reach ranged from 0.020 to 0.024. The roughness variation is attributed to
irregularities in the canal bed slope due to ground subsidence resulting from long term groundwater
extraction. A simulation time of 25 hours was assigned to ensure uniform flow throughout the reach.
At the end of the simulation, both the inflow and outflow discharge were approximately 8,150 cfs.
Operation conditions for the canal reach were an average flow depth of approximately 26 ft and a
depth-averaged velocity of about 3 fps. The design operating water surface elevations were 317.26 at

35

the start of the reach and 314.26 at the end of the reach. The FLO-2D predicted water surface
elevations were 317.28 and 314.31 for the inflow and outflow nodes. FLO-2D Version 2009 results in
Table 3 show consistent discharge throughout the reach. The predicted velocities and flow depths are
uniform and replicate the hydraulic operating conditions. The actual flow hydraulics deviate slightly
from the design conditions because of the bed slope variability that has evolved with the channel
subsidence (see Case Study 5). FLO-2D Version 2007 results are shown for comparison. The FLO-2D
Version 2009 predicted more consistent discharges and more accurate starting and ending water
surface elevations than Version 2007 because the Version 2009 stability criteria is more refined. This
application validated the FLO-2D dynamic wave routing algorithm for a large channel with a very mild
slope.

36

Version 2007

MODEL TIME = 25.00 HOURS TOTAL TIMESTEP NUMBER =

INFLOW RATE AT NODE 3865 IS EQUAL TO

Version 2009

41902.

MODEL TIME = 25.00 HOURS TOTAL TIMESTEP NUMBER =

8150.00

INFLOW RATE AT NODE 3865 IS EQUAL TO

AVERAGE HYDRAULIC CONDITIONS FOR THE OUTPUT INTERVAL

AVERAGE HYDRAULIC CONDITIONS FOR THE OUTPUT INTERVAL

NODE ELEVATION DEPTH VELOCITY OUT DISCHARGE OUT

NODE ELEVATION DEPTH VELOCITY OUT DISCHARGE OUT

3865
3960
4055
4150
4245
4340
4436
4532
4531
4530
4626
4721
4720
4719
4815
4911
4910
4909
4998
5086
5085
5084
5171
5258
5257
5256
5342
5341
5340
5425
5424
5423
5507
5506
5505
5587
5586
5585
5584
5664
5663
5662
5661
5738
5737
5736
5810
5809
5808
5807
5806
5805
5804
5803
5802
5801
5800
5799
5798
5797
5796
5795
5794
5793
5792
5791
5790
5789
5788
5787
5786
5785
5784
5783
5782
5706
5627
5626
5544
5543
5459
5458
5372
5285
5284
5283
5282
5281
5280
5279
5191
5102
5101
5012
5011

316.78
316.53
316.50
316.47
316.45
316.42
316.39
316.37
316.34
316.32
316.30
316.27
316.25
316.22
316.20
316.17
316.15
316.13
316.10
316.06
316.04
316.01
315.98
315.94
315.89
315.84
315.81
315.77
315.74
315.71
315.68
315.66
315.63
315.60
315.57
315.54
315.51
315.49
315.46
315.44
315.41
315.39
315.37
315.34
315.32
315.29
315.27
315.24
315.21
315.18
315.15
315.12
315.09
315.06
315.03
315.01
314.98
314.95
314.93
314.91
314.89
314.87
314.85
314.83
314.81
314.79
314.77
314.75
314.73
314.71
314.69
314.67
314.65
314.63
314.61
314.59
314.57
314.54
314.52
314.50
314.48
314.46
314.43
314.41
314.39
314.37
314.35
314.33
314.31
314.28
314.26
314.24
314.21
314.19
314.17

26.52
26.31
26.32
26.30
26.29
26.27
26.25
26.24
26.21
26.20
26.16
26.11
26.07
26.02
25.98
25.93
25.89
25.86
25.80
25.74
25.70
25.63
25.63
25.66
25.67
25.70
25.73
25.81
25.91
26.04
26.13
26.16
26.16
26.15
26.13
26.13
26.12
26.12
26.12
26.11
26.10
26.10
26.11
26.13
26.15
26.15
26.18
26.19
26.19
26.20
26.20
26.21
26.19
26.16
26.16
26.18
26.20
26.21
26.23
26.25
26.27
26.29
26.31
26.34
26.36
26.38
26.40
26.42
26.44
26.47
26.49
26.51
26.53
26.55
26.54
26.55
26.56
26.55
26.56
26.56
26.57
26.57
26.57
26.59
26.59
26.60
26.60
26.60
26.60
26.59
26.60
26.61
26.60
26.61
26.61

2.73
2.75
2.75
2.75
2.76
2.76
2.76
2.76
2.77
2.77
2.78
2.78
2.79
2.80
2.80
2.81
2.82
2.82
2.83
2.84
2.85
2.86
2.86
2.92
3.06
3.13
3.12
3.11
3.09
3.07
3.06
3.05
3.05
3.05
3.05
3.05
3.06
3.06
3.06
3.06
3.06
3.06
3.06
3.06
3.05
3.05
3.05
3.05
3.04
3.04
3.04
3.04
3.04
3.05
3.05
3.05
3.04
3.04
3.04
3.03
3.03
3.03
3.02
3.02
3.01
3.01
3.01
3.00
3.00
2.99
2.99
2.99
2.98
2.98
2.98
2.98
2.98
2.98
2.98
2.98
2.98
2.98
2.98
2.98
2.97
2.97
2.97
2.97
2.97
2.97
2.97
2.97
2.97
2.97
2.97

25153.

8150.00

8150.54
8158.49
8159.69
8156.24
8160.54
8158.80
8156.42
8153.06
8154.96
8154.77
8154.92
8153.67
8151.70
8154.46
8154.53
8153.75
8154.63
8155.47
8157.89
8161.16
8160.14
8161.48
8160.33
8159.09
8159.27
8161.08
8160.43
8161.72
8162.23
8161.16
8159.82
8160.12
8159.51
8160.55
8159.09
8159.73
8160.62
8160.48
8164.11
8165.50
8164.84
8165.92
8167.59
8165.12
8165.93
8166.05
8164.24
8163.69
8165.67
8166.55
8165.55
8165.37
8163.04
8161.77
8159.96
8159.38
8158.63
8159.64
8161.95
8160.76
8159.55
8159.54
8163.16
8163.08
8159.30
8158.60
8158.71
8159.30
8160.53
8159.85
8156.63
8156.08
8156.58
8154.97
8158.40
8158.06
8158.95
8160.02
8161.15
8161.10
8162.95
8162.79
8162.18
8161.18
8160.86
8161.31
8162.84
8163.19
8161.24
8162.85
8166.84
8168.79
8168.72
8169.45
8171.22

3865
3960
4055
4150
4245
4340
4436
4532
4531
4530
4626
4721
4720
4719
4815
4911
4910
4909
4998
5086
5085
5084
5171
5258
5257
5256
5342
5341
5340
5425
5424
5423
5507
5506
5505
5587
5586
5585
5584
5664
5663
5662
5661
5738
5737
5736
5810
5809
5808
5807
5806
5805
5804
5803
5802
5801
5800
5799
5798
5797
5796
5795
5794
5793
5792
5791
5790
5789
5788
5787
5786
5785
5784
5783
5782
5706
5627
5626
5544
5543
5459
5458
5372
5285
5284
5283
5282
5281
5280
5279
5191
5102
5101
5012
5011

Table 3. FLO-2D Results for the California

Aqueduct

37

317.28
317.04
317.01
316.98
316.95
316.93
316.90
316.88
316.85
316.83
316.80
316.78
316.75
316.73
316.70
316.67
316.65
316.62
316.59
316.56
316.54
316.51
316.48
316.46
316.42
316.37
316.33
316.30
316.27
316.23
316.20
316.17
316.14
316.11
316.07
316.04
316.01
315.98
315.95
315.92
315.89
315.86
315.84
315.81
315.78
315.75
315.72
315.69
315.66
315.63
315.59
315.56
315.52
315.49
315.46
315.42
315.39
315.36
315.33
315.30
315.27
315.24
315.21
315.18
315.15
315.12
315.09
315.07
315.04
315.01
314.98
314.95
314.92
314.89
314.86
314.83
314.80
314.76
314.73
314.70
314.67
314.64
314.61
314.57
314.54
314.51
314.48
314.45
314.42
314.39
314.35
314.32
314.28
314.25
314.21

27.03
26.82
26.83
26.81
26.79
26.78
26.76
26.75
26.72
26.71
26.66
26.62
26.57
26.53
26.48
26.43
26.39
26.35
26.29
26.24
26.20
26.13
26.13
26.18
26.20
26.23
26.25
26.34
26.44
26.56
26.65
26.67
26.67
26.66
26.63
26.63
26.62
26.61
26.61
26.59
26.58
26.57
26.58
26.60
26.61
26.61
26.63
26.64
26.64
26.65
26.64
26.65
26.62
26.59
26.59
26.59
26.61
26.62
26.63
26.64
26.65
26.66
26.67
26.69
26.70
26.71
26.72
26.74
26.75
26.77
26.78
26.79
26.80
26.81
26.79
26.79
26.79
26.77
26.77
26.76
26.76
26.75
26.75
26.75
26.74
26.74
26.73
26.72
26.71
26.70
26.69
26.69
26.67
26.67
26.66

2.66
2.67
2.67
2.68
2.68
2.68
2.68
2.69
2.69
2.69
2.70
2.70
2.71
2.72
2.72
2.73
2.74
2.74
2.75
2.76
2.77
2.77
2.78
2.84
2.97
3.03
3.03
3.01
3.00
2.98
2.96
2.96
2.96
2.96
2.96
2.96
2.97
2.97
2.97
2.97
2.97
2.97
2.97
2.97
2.97
2.97
2.96
2.96
2.96
2.96
2.96
2.96
2.97
2.97
2.97
2.97
2.97
2.97
2.96
2.96
2.96
2.96
2.96
2.95
2.95
2.95
2.95
2.95
2.94
2.94
2.94
2.94
2.94
2.93
2.94
2.94
2.94
2.94
2.94
2.94
2.94
2.94
2.94
2.94
2.94
2.95
2.95
2.95
2.95
2.95
2.95
2.95
2.96
2.96
2.96

8150.00
8150.05
8150.01
8150.00
8150.00
8150.00
8149.99
8149.99
8149.98
8149.98
8149.97
8149.97
8149.96
8149.96
8149.94
8149.95
8149.94
8149.94
8149.92
8149.93
8149.92
8149.91
8149.89
8149.89
8149.89
8149.88
8149.87
8149.86
8149.86
8149.85
8149.84
8149.84
8149.83
8149.82
8149.82
8149.81
8149.81
8149.80
8149.79
8149.78
8149.78
8149.77
8149.76
8149.75
8149.74
8149.74
8149.73
8149.73
8149.73
8149.72
8149.72
8149.71
8149.71
8149.70
8149.70
8149.69
8149.69
8149.69
8149.69
8149.68
8149.67
8149.66
8149.66
8149.66
8149.65
8149.64
8149.64
8149.63
8149.62
8149.62
8149.62
8149.61
8149.61
8149.60
8149.61
8149.59
8149.58
8149.57
8149.57
8149.56
8149.55
8149.55
8149.53
8149.54
8149.55
8149.53
8149.53
8149.53
8149.53
8149.54
8149.51
8149.52
8149.51
8149.52
8149.51

Claim 10: FLO-2D accurately computes channel

hydraulics in comparison with other 1-D channel models.
3.3.2.2 Design flow in Camarillo Hills Drain
Substantiation of Claim 10. The FLO-2D FIS Study for the City of Camarillo required the simulation of
flow in a drainage canals for the 100-year flood. See Case Study 3 in Chapter IV for more details on this
project. Most of the drainage canals had concrete lined rectangular and trapezoidal cross sections. A
number of the drains had bridges and culverts which were simulated with rating curves. The City of
Camarillo FIS project had a number of drains that had been analyzed with the Water Surface Pressure
Gradient (WSPG) program required by the Los Angeles Department of Public Works for hydraulic design.
The program computes uniform and non-uniform steady flow water surface profiles in open channels.
WSPG was applied to the Camarillo Hills Drain by CH2M Hill for a 30% Design Completion Study. The
results from this study for the drain reach from Los Posas Road to Highway 101 (about 1,100 ft) are
presented in Table 4. The rectangular drain (37 ft wide by 8 ft deep) was designed for critical flow at
bankfull discharge (4,786 cfs) as shown with a velocity of 16.0 fps and flow depth of 8.0 ft for the
downstream half of the channel. The design Manning's n-value was 0.015. The results are listed from
downstream to upstream. For the downstream portion of this reach (550 ft long) was designed for a
normal depth of about 8.0 ft and a depth averaged velocity of 16.2 fps.
Using this isolated reach of drain with uniform channel geometry and roughness to test the channel
hydraulics, the FLO-2D model predicted results are shown in Table 5. Since the flow is entered directly
into the initial channel element in the reach, it takes approximately 10 channel elements for the flow to
accelerate to normal depth conditions (red box in Table 5). The results are listed from upstream to
downstream. The lower 600 ft indicates a depth averaged velocity of 16.3 fps and a normal depth of
about 7.95 ft almost exactly what the WSPG program predicted. This substantiates the claim that FLO2D predicts accurate channel hydraulics.

38

Table 4. WSPG Hydraulic Results for the Camarillo Hills Drain Los Posas Road to Highway 101
39

INFLOW RATE AT NODE 1015765 IS EQUAL TO

NODE

1015765
1017063
1018358
1019653
1020946
1022236
1023525
1024814
1026100
1027385
1028667
1029949
1031229
1032506
1033783
1035059
1036332
1037604
1038874
1040143
1041388
1042614
1043820
1045005
1046168
1047312
1048434
1049536
1050618
1051679
1052721
1053741
1054740
1055720
1056678
1057617
1058535
1059432
1060309
1061164
1061999
1062813
1063609
1064382
1065134
1065868
1066580
1067271
1067942
1068593
1069223
1069832
1070421
1070989
1071536
1072063
1072062
1072569
1073055
1073054
1073520
1073965
1073964
1074388
1074791
1074790
1075173
1075537
1075536
1075879
1076201
1076200
1076501
1076785
1076784
1077066
1077065
1077344
1077620
1077619
1077894
1078166
1078165
1078434
1078700
1078699
1078964

ELEVATION

DEPTH

102.93
99.35
98.33
97.82
97.53
97.32
97.17
97.04
96.94
96.85
96.76
96.69
96.62
96.55
96.50
96.45
96.39
96.34
96.29
96.24
96.19
96.14
96.10
96.06
96.01
95.97
95.92
95.89
95.85
95.81
95.76
95.73
95.69
95.65
95.62
95.59
95.56
95.53
95.50
95.47
95.45
95.43
95.41
95.39
95.38
95.37
95.35
95.33
95.30
95.28
95.25
95.22
95.19
95.16
95.13
95.10
95.07
95.04
95.01
94.98
94.94
94.90
94.86
94.82
94.78
94.75
94.71
94.67
94.63
94.59
94.55
94.52
94.48
94.45
94.41
94.38
94.34
94.31
94.27
94.24
94.20
94.16
94.13
94.09
94.05
94.01
93.98

13.83
10.30
9.33
8.85
8.60
8.42
8.31
8.21
8.15
8.09
8.04
8.00
7.97
7.93
7.92
7.90
7.88
7.86
7.85
7.83
7.82
7.80
7.80
7.79
7.78
7.77
7.76
7.76
7.76
7.75
7.74
7.74
7.73
7.73
7.73
7.74
7.74
7.75
7.75
7.76
7.77
7.79
7.80
7.82
7.84
7.87
7.88
7.90
7.90
7.92
7.92
7.93
7.93
7.94
7.94
7.95
7.95
7.96
7.96
7.96
7.96
7.95
7.95
7.94
7.94
7.94
7.94
7.93
7.93
7.92
7.92
7.92
7.92
7.92
7.92
7.92
7.92
7.92
7.92
7.92
7.92
7.91
7.92
7.91
7.91
7.90
7.91

VELOCITY OUT

10.72
11.54
13.72
14.45
15.00
15.34
15.57
15.74
15.88
15.99
16.08
16.16
16.23
16.29
16.34
16.38
16.41
16.44
16.48
16.51
16.54
16.57
16.59
16.61
16.63
16.65
16.66
16.67
16.68
16.69
16.71
16.72
16.74
16.74
16.74
16.73
16.72
16.71
16.69
16.68
16.66
16.63
16.59
16.56
16.51
16.47
16.43
16.40
16.38
16.36
16.34
16.32
16.31
16.30
16.29
16.29
16.27
16.27
16.25
16.25
16.26
16.27
16.27
16.28
16.29
16.29
16.30
16.31
16.31
16.32
16.33
16.34
16.34
16.34
16.34
16.34
16.34
16.34
16.34
16.34
16.34
16.34
16.35
16.35
16.36
16.36
16.36

4786.00

DISCHARGE OUT

4786.00
4786.04
4786.18
4786.21
4786.13
4786.10
4786.08
4786.07
4786.07
4786.06
4786.06
4786.06
4786.05
4786.05
4786.05
4786.05
4786.05
4786.05
4786.05
4786.05
4786.04
4786.04
4786.05
4786.05
4786.05
4786.04
4786.04
4786.04
4786.05
4786.05
4786.04
4786.04
4786.03
4786.04
4786.03
4786.03
4786.03
4786.03
4786.03
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.02
4786.01
4786.01
4786.02
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.02
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01
4786.01

Table 5. FLO-2D Results for Camarillo Hills Drain 100-year Design Discharge

40

Claim 11: FLO-2D accurately computes floodwave

movement in a river system.

3.3.2.3 Tracking Green River Flaming Gorge Dam Power Release

Substantiation of Claim 11. A FLO-2D model was developed for the Green River system from Flaming
Gorge to the Colorado River confluence in Utah, a river distance of 412 miles. This project was
supported by the National Park Service, Fish and Wildlife Service and the Bureau of Reclamation and it
provided the opportunity to test a calibrated model with an actual reservoir power release. For further
discussion, review Case Study 6 in Chapter IV. The discharge data included all the available USGS
tributary inflow data (six major tributary gages), Flaming Gorge Dam releases and level logger discharge
monitoring data at several locations throughout the river system. The FLO-2D model consisted of 3,671
2,000 ft grid elements with 1,388 channel elements. The initial calibration run encompassed 100 days of
the 1996 high flow season. Following calibration of the roughness values and infiltration parameters
and the adjustment of floodplain elevations and channel geometry, the model was applied to simulate
100 days of the 1997 Green River hydrograph (Figure 21).
Based on the excellent correlation between the 1997 recorded discharge data and the FLO-2D predicted
discharge hydrograph at the USGS Jensen gage (95 miles downstream of Flaming Gorge), the model was
applied to various flow scenarios involving the regulated releases of Flaming Gorge Dam. In November
1998, two days of power generating spike flows were planned to be released from Flaming Gorge Dam
(Figure 22). The two spike releases of 4,400 cfs would be twice the normal daily release. As a test of
model accuracy, the powerplant spikes were routed downstream with the calibrated FLO-2D model
prior to the scheduled release. At the USGS Jensen gage, the shape and timing of the predicted
floodwaves match the measured hydrograph almost exactly (Figure 23). The predicted flow is about 200
cfs less than the measured discharge because of ungaged tributary inflow that is not accounted for in
the model. Jones Creek in Dinosaur National Monument is responsible for most of the unmeasured
flow. The predicted early arrival of the two peaks at the Jensen gage can be attributed to the increased
hydraulic roughness at lower flows through the steep Lodore Canyon. At low flows, numerous boulders
are exposed in the canyon channel.
By accurately predicting an apriori dam release event, the FLO-2D model channel routing algorithm was
validated. Baker (2001) recommended that the FLO-2D model be added to FEMA's list of approved
models for riverine studies and commented that "...because this application includes calibration,
verification and reasonable production runs, the FLO-2D model for this application accurately
demonstrates the capability to replicate known hydraulic conditions for channel flow." This application
was rerun with Version 2009 using the depth integrated n-value adjustment for the channel. The
roughness n-value is assigned for bankfull discharge (~ 20,000 cfs), and the n-value adjustment increases
the bankfull n-value by 1.5 - 1.8 times when the flow depth is 0.5 ft. This adjustment slows down the
low flow discharge movement through the canyons and improves the timing of the peaks and troughs
arriving at the Jensen gage as shown in Figure 24.

41

Figure 21. 1997 High Flow Season Hydrograph at Jensen, Utah (FLO-2D vs. Measured Discharge)

42

Figure 22. November 1998 Power Plant Discharge Release from Flaming Gorge Dam

43

Figure 23. FLO-2D Predicted vs. Measured Discharge, Jensen USGS Gage for Power Plant Releases at Flaming Gorge Dam

44

Figure 24. Version 2009 FLO-2D Predicted vs. Measured Discharge, Jensen USGS Gage for Power Plant Releases at Flaming Gorge Dam

45

3.3.3 Channel and Floodplain Flow Exchange

FLO-2D exchanges discharge between the channel and floodplain on a grid element basis representing
the left and right bank elements. The discharge exchange occurs in a separate routine after the channel,
and floodplain routing subroutines have been completed. When the channel conveyance capacity is
exceeded, an overbank discharge is computed. Conversely, return flow to the channel is also predicted
when the floodplain water surface exceeds the channel water surface. The overland flow can enter a
previously dry channel. The channel-floodplain flow exchange is limited by the available exchange
volume in the channel or by the available storage volume on the floodplain bank element. This
subroutine also computes the flow exchange between the street and the floodplain. The channelfloodplain exchange is based on the potential water surface elevation difference between the channel
and the floodplain bank element (Figure 25). The exchange velocity is computed using a diffusive wave
representation of the momentum equation and the floodplain roughness. It is assumed that the
overbank flow velocity is relatively small and thus the acceleration terms in the momentum equation are
negligible.

Figure 25. Channel-Floodplain Flow Exchange

46

Claim 12: Channel floodplain exchange conserves volume and

accurately represents the flood distribution between them.
Substantiation of Claim 12. This substantiation essentially revisits the Claim 5 substantiation regarding
with the accuracy of overland flow for the Truckee River in Reno. This project is highlighted again to
demonstrate the accuracy of the channel-floodplain exchange. A total of 5,089 500 ft grid elements
with 221 channel elements constituted the FLO-2D grid system. Rainfall, infiltration, 18 bridges and a
number of streets and buildings were simulated. Overbank flow ensues at approximately 11,000 cfs in
Truckee Meadows downstream of the Reno Gage. In the 1997 event (over 20,000 cfs), a large portion of
the flood flows overland through the Meadows area with a lag time of up to 6 hours while the river flow
takes less than an hour to reach the New Vista Gage (Figure 9). Water exchange between the channel
and floodplain occurs along most of the 5.8 mile reach between the two gages. The entire Meadows
area is inundated along the 2.7 mile segment of the Truckee River. By matching the area of inundation
(Figure 6) as well as the Reno and New Vista gage stage and timing, the channel floodplain exchange
must be accurately computed. The accuracy of the volume in the channel is depicted by matching the
channel water surface profile with the HEC-2 model.
The Middle Rio Grande spring runoff release monitoring project previously discussed presented a similar
opportunity to test the accuracy of the channel-floodplain exchange. The compilation of the area of
inundation from digital aerial photography was correlated with the data from the seven USGS gaging
stations. Again these results support the assessment of the volume in the channel versus the volume on
the floodplain. By accurately predicting the area of inundation in Table 2 and matching the timing of the
floodwave movement at the downstream gaging stations, by continuity the channel-floodplain exchange
are proven to be accurate.

47

3.4 Hydrology
3.4.1 Overview
Water inflow to the FLO-2D model can include an unlimited number of inflow hydrographs, rainfall,
stage-discharge or time-stage data. Rainfall can be uniform or spatially variable using either depth area
reduction values or NEXRAD radar data. For uniform rainfall, the storm is discretized as a cumulative
percent of the total rainfall. There are a number of options to simulate variable rainfall including a
moving storm, spatially variable depth area reduction, or grid based rainfall data from an actual storm
event. A rainfall distribution can be selected from a number of predefined distributions. The rainfall
runoff can be routed off the watershed and into to the channel system in the same flood simulation.
3.4.2 Hydrology Validation

Claim 13: Volume is conserved for rainfall on a FLO-2D

grid system.
Substantiation of Claim 13. A simple rainfall model in a small Las Vegas, Nevada watershed (Diamond
Project) was used to prove that the rainfall volume is conserved. The project has 10,374 300 ft grid
elements. The total rainfall is 3.1 inches. The total volume of water applied on the grid system is:
10,374 x 300 ft x 300 ft x 3.1 inches/12 in/ft /43,560 sq ft/acre = 5537 acre-ft
The reported volume of rainfall in the SUMMARY.OUT file for this simulations is shown in Table 6.

Table 6. SUMMARY.OUT file for the Diamond Project to Verify Rainfall Volume Conservation
Once the rainfall on the grid system has ponded to a depth greater than the tolerance value (TOL in
TOLER.DAT file), the water can be exchanged between grid elements as overland flow.

48

Claim 14: Storms can be accurately replicated using

NEXRAD rainfall data.
Substantiation of Claim 14. Historical storms can be accurately replicated with calibrated NextGeneration Radar (NEXRAD) data obtained from the National Weather Service. The NEXRAD rainfall
pixels for a given time interval (typically 5 minutes or 15 minutes) are automatically interpolated to the
FLO-2D grid system using the GDS. Each grid element is assigned a rainfall total for the NEXRAD time
interval and the rainfall is then interpolated by the model for each computational timestep. The result is
spatially and temporally variable rainfall-runoff from the grid system.
A FLO-2D model was developed for the Pima County Regional Flood Control District (PCRFCD) to
delineate the flood hazard on the Soldier Canyon alluvial fan. The model of the July 31, 2006 Soldier
Canyon flood event was calibrated using post-flood event aerial photos, NEXRAD rainfall data,
observations of the flood stage at the Catalina Highway culvert, and the PCRFCD estimates of sediment
yield. The USGS provided the NEXRAD precipitation data (in millimeters) for the July 31, 2006 storm in
15 minute intervals. The NEXRAD data covered much of the Catalina Mountain front including the
Soldier Canyon basin (Figure 26 shown in red). The NEXRAD cells are discretized radial radar bands on
the order of 1 km square. The precipitation values were derived from second level NEXRAD reflectivity
data using a Z-R transformation from a best model of 19 rain Catalina Mountain gages. Applying the
NEXRAD 15 minute data to the 100 ft grid system, the July 31, 2006 storm was simulated. This spatial
discretization of the storm enables the FLO-2D to replicate the storm in significant detail (Figure 27).

Figure 26. NEXRAD Cell Total Precipitation July 31, 2006 on the Catalina Mountain Watersheds
(Provided by the USGS email correspondence August 7, 2007)
49

Since the infiltration losses were minor because of the high antecedent moisture conditions, it was not
necessary to calibrate the FLO-2D predicted rainfall runoff. As a result, the FLO-2D generated flood
hazard maps closely replicated the aerial photo taken on the morning of July 31, 2006 (Figure 27). The
predicted flood paths in blue in Figure 28 can be precisely correlated with the green flow paths in Figure
27. It was concluded that the FLO-2D model hydrology component accurately predicted the flood
inundation as a result of the detailed temporal and spatial discretization of the storm event.

Figure 27. Lower Soldier Canyon Alluvial Fan Aerial Photo. Agua Caliente Wash crosses
Soldier Trail (center photograph). Photograph by Terry Hendricks (PCRFCD; 7-31-06).

50

Figure 28. Soldier Canyon Alluvial Fan FLO-2D Predicted Area of Inundation for the July 31, 2006 Flood
In a second NEXRAD rainfall project, the Flood Control District of Maricopa County (FCDMC) applied the
FLO-2D model on the Rainbow Wash watershed (Figure 29) to replicate a 2005 storm. The watershed
was gaged and had measured discharge data for the storm. FCDMC funded the development of the GDS
tool that interpolates the NEXRAD data to FLO-2D grid elements. The FLO-2D model results prepared by
the FCDMC were compared with the measured data and with comparable HEC-1 model results (Figure
30). The HEC-1 model was the previously required project hydrologic model for the FCDMC. The FLO2D model is also now an accepted hydrologic model for FCDMC projects. The discrepancy between the
gage data and the FLO-2D predicted recessional limb of the basin outflow hydrograph were attributed to
gage cross section scour. Modeling the concentrated flow as channel discharge later also improved the
timing on the rising limb. The FCDMC (2010) stated, "(a)s part of this study, temporal-based NEXRAD
radar data was used with FLO-2D and HEC-1 to model a significant storm that occurred over the 18
51

square mile Rainbow Wash watershed in 2005. Using the FCDMC standard modeling parameters
without calibration, FLO-2D very closely replicated the measured runoff hydrograph storm event, while
HEC-1 significantly over predicted the runoff volume and peak discharge." Figure 31 indicates that the
FLO-2D accurately replicated the area of inundation for this 2005 flood event.

Figure 29. FLO-2D Predicted Maximum Depth for Rainbow Wash (FCDMC, 2010)
7,000

Observed
FLO-2D Saxton 2005
HEC-1 2005

6,000

Discharge, cfs

5,000

4,000

RAINBOW WASH at SR-85

MEASURED DATA COMPARED WITH FLO-2D RESULTS
Storm: August 9, 2005; 5:00pm - 6:45pm
NEXRAD Locally-Adjusted Precipitation: 4.15" max, 2.43" Avg
Rainbow Wash Gage Measurement: 1.90" total
NOAA Atlas 14:
Frequency for Watershed Avg: 250-yr, 2-hr
Frequency for Gage: 50-yr, 2-hr
Observed: V = 169 ac-ft, Q = 1,919 cfs
FLO-2D 2005: V = 226 ac-ft, Q = 1,788 cfs
HEC-1 2005: V = 774 ac-ft, Q = 4,201 cfs

3,000

2,000

Volume Comparisons:
a. HEC-1 1993 Runoff Volume: 458 % of measured
b. Saxton 2005 Runoff Volume: 138 % of measured

1,000

Time

Figure 30. FLO-2D Predicted Hydrograph vs. Gage Data and HEC-2D Results (FCDMC, 2010)
52

Figure 31. FLO-2D Predicted Inundation Area vs. the Observed Flooding from 2005 Storm (FCDMC, 2010)

Claim 15: Infiltration Using the Green-Ampt method is

accurately computed and volume is conserved
Substantiation of Claim 15. Precipitation losses, abstraction (interception) and infiltration are simulated
in the FLO-2D model. Similar to HEC-1, the initial abstraction is filled prior to simulating infiltration.
Infiltration is simulated using either the Green-Ampt infiltration model or the SCS curve number
method. The infiltration parameters can be assigned uniformly or as spatially variable. No infiltration is
calculated for assigned streets, buildings or impervious surfaces in the grid elements. Channel
infiltration (seepage) can be also be simulated.
The Green-Ampt (1911) equation was selected to compute infiltration losses in the FLO-2D model
because it is sensitive to rainfall intensity. When the rainfall exceeds the potential infiltration, then
runoff is generated. The infiltration continues after the rainfall has ceased until all the available water
has run off or has been infiltrated. To utilize the FLO-2D Green-Ampt infiltration component, hydraulic
conductivity, soil suction, volumetric moisture deficit, the percent impervious area and the infiltration
soil depth must be specified. The volumetric moisture deficiency is evaluated as the difference between
53

the initial and final soil saturation conditions. The FLO-2D Green-Ampt code was provided to the
FCDMC and was manually verified by FCDMC engineering staff using HEC-1 and an Excel spreadsheet
with interactive code.
The FCDMC's Rainbow Wash simulation demonstrates the accuracy of the Green-Ampt infiltration
method. Nevertheless, a simple infiltration model was created to validate the Green-Ampt component.
Using the Diamond project that demonstrated rainfall volume accuracy, a uniform rainfall of 0.5 inches
in one hour was applied along with a Green-Ampt infiltration hydraulic conductivity of 0.5 inches/hr.
There was zero abstraction, no soil suction and no impervious area. The correct result of this model is
no flow depth, 0.5 inches of rainfall and 0.5 inches of infiltration as shown is Table 7.

Table 7. SUMMARY.OUT File for the Green-Ampt Verification

Claim 16: SCS Curve Number method infiltration is
accurately computed and volume is conserved
Substantiation of Claim 16. The implementation of the SCS runoff curve number (CN) loss method in the
FLO-2D model was funded by Pima County Regional Flood Control District. The rainfall loss is a function
of hydrologic soil type, land use and treatment, surface condition and antecedent moisture condition.
The SCS curve number method was developed from 24 hour hydrograph data on mild slope rural
watersheds in the eastern United States to predict rainfall runoff from ungaged watersheds. Runoff
curve numbers have been calibrated or estimated for a wide range of urban areas, agricultural lands and
semi-arid range lands. For large basins (especially semi-arid basins) which have unique or variable
infiltration characteristics such as channels, the method tends to over-predict runoff (Ponce, 1989). It
should be noted that the SCS curve number method removes water directly from the rainfall, so that if
there is no rainfall, there are no SCS method losses. If there is flooding on a surface without rainfall,
there are no transmission losses.
54

The SCS curve number parameters can be assigned graphically in the GDS to allow for spatially variable
rainfall runoff. Shape files can used to interpolate SCS-CNs from ground cover and soil attributes. The
SCS-CN method can be combined with the Green-Ampt infiltration method to compute both rainfallrunoff and overland flow transmission losses. For the combined loss routines, the SCS-CN method will
be applied to grid elements with rainfall during the model computational timestep and the Green-Ampt
method will compute infiltration for grid elements that do not receive rainfall during the timestep. This
enables transmission losses to be computed with Green-Ampt on alluvial fans and floodplains while the
SCS-CN is used to compute the rainfall loss in the watershed basin. The FLO-2D SCS method calculations
were verified for Pima County by JE Fuller Hydrology and Geomorphology and Stantec Consultants with
hand computations and simplified project applications. Application of a CN = 80 for the Diamond
Project with a total rainfall (P) of 3.1 inches results in a total volume of actual runoff Q:
Q = (P - 0.2 S)2/ (P + 0.8 S) = (3.1 - 0.2 * 2.5)2/(3.1 + 0.8 * 2.5) = 1.33 inches
where:
S = 1000/CN - 10 = 2.5
which in Table 8 below is the equivalent of the total point rainfall minus the infiltrated and abstracted
water 3.1 inches - 1.78 inches = 1.32 inches. Applying this value to the volume conservation in Table 5,
there are 10374 grid elements 300 ft square:
Volume of Runoff = 1.32 inches/12 inches/ft * 10374 * 300 * 300 ft2/43560 ft2/acre = 2,357.73 acre-ft
The sum of the floodplain storage plus the floodplain outflow = 2,316.01 + 41.66 = 2,357.67 acre-ft as
shown in Table 8.

Table 8. Validation of the SCS Curve Number Computations

55

Claim 17: FLO-2D can be used to predict alluvial fan flooding.

Substantiation of Claim 17. In August 2010 JE Fuller Hydrology and Geomorphology completed a report
for the FCDMC entitled: "Refinement of Methodology: Alluvial Fan Flood Hazard Identification and
Mitigation Methods Study." The scope of this study was to update and refine the FCDMC current
Piedmont Flood Hazard Assessment Manual (PFHAM) methodology, to identify engineering procedures
to quantify flood hazards on alluvial fan landforms, to recommend hazard mitigation measures, and to
refine landform definitions used in the PFHAM. Based on the analysis of four evaluation fans (two fans
shown in Figures 32 and 33), it was concluded that "(f)or the PFHAM study, the FLO-2D model was
selected as the best available model, a finding which is consistent with the findings of other agencies...".
From the hydrologic modeling analyses it was concluded that "...FLO-2D is preferred over HEC-1 for
modeling fans and alluvial plains..." and from the hydraulic modeling analyses "...FLO-2D modeling is
preferred for modeling fans and alluvial plains."

Figure 32. FLO-2D Model Predicted Depth and Velocities for White Tanks Fan 36 (JE Fuller, 2010)

56

Figure 33. FLO-2D Mode Predicted Depth and Velocities for White Tanks Fan 36
The PFHAM report and methodology was submitted to a panel of experts ("Blue Ribbon Panel) for peer
review which met on June 2-3, 2010. The Blue Ribbon Panel consisted of experts from a variety of
engineering, scientific, and regulatory disciplines associated with alluvial fan flood hazard assessment.
The Blue Ribbon Panel concluded that the methodology was "...reasonable, defensible, and scientifically
sound..." and that two-dimensional modeling was "... strongly recommended for alluvial fan flood
hazard assessment" including flooding attenuation. It was also concluded that the proposed hazard
assessment methodology using the FLO-2D Mapper program flood hazard delineation method of low,
moderate and high hazard was acceptable.
3.4.3 Summary of the Hydrology Component
The hydrologic component in FLO-2D is critical to accurate flood hazard delineation because the area of
inundation is primarily a function of both the flood inflow hydrographs plus effective rainfall on the
project area. Rainfall can occur directly on ponded areas, rivers, streets, buildings and infiltration losses
can be computed for any surface. The rainfall runoff from urban areas or watersheds can be routed
over the grid system to the water courses and then routed downstream in the channel component all in
one model simulation. The combined FLO-2D hydrologic and hydraulic flood routing is a significant
advancement over the traditional 1-D lumped parameter unit hydrograph method watershed approach.
The FCDMC (2010) stated that the District was "...having great success with the application of the FLO2D hydrologic and hydraulic modeling in support of mapping and managing flood hazard areas within
Maricopa County." "...(T)his year FLO-2D is also being used for areas within our ADMP study areas to
57

model watersheds that would normally be modeled with HEC-1. FCDMC would not be using it as the
primary modeling tool on this year's ADMP's if it were not a valuable and appropriate tool." The PFHAM
report by JE Fuller (2010) recommended that the "...engineering tool for hydrologic modeling of active
alluvial fans and alluvial plains in Maricopa County is FLO-2D".
3.5 Component Applicability
The FLO-2D model can be used to simulate the complete hydrologic cycle including rainfall, infiltration,
evaporation, groundwater-surface water exchange and flood routing. Model components used in the
hydrologic and hydraulic simulations include channel and street routing, urban flooding, alluvial fan
flooding, sediment transport, mud and debris flows, levee confinement, levee overtopping and levee
and dam breach. Each flood project is unique in its combined utilization of these components. These
combinations lead to an infinite range of potential variability in a given model application and not all of
these components are required for every flood hazard delineation project. For FEMA Flood Insurance
Studies (FIS), the flooding is generally of short duration so components such as evaporation,
groundwater-surface water exchange are not required. In addition, for FIS studies sediment transport
modeling is not required. As such, these particular components along with the mud and debris flow
component and the dam breach erosion component are not discussed in this document. The FLO-2D
model does not simulate the following flood conditions:

Two-dimensional channel flow. The channel component simulates one-dimensional depth

averaged flow.

Pressure flow is not simulated directly, but it can be simulated by a third party program such as
a storm drain model and applied as a hydraulic structure rating table.

Levee confinement and levee overtopping are the remaining FLO-2D components left for discussion.
Claim 18: Levee confinement is accurately simulated by
FLO-2D in eight potential flow directions.
Substantiation of Claim 18. Levee assignment occurs along the grid element boundary in any of the 8
potential flow directions. The user assigns the levee blocked direction and the crest elevation. As long
as the crest elevation is greater than the water surface elevation, the levee will stop flow from being
exchanged between grid elements. Figure 34 displays the Middle Rio Grande floodplain flooding
confined by the levee system on both sides of the river. This project is one of the FLO-2D example
projects and is used as a benchmark test for model revisions.

58

Figure 34. Rio Grande Overbank Flooding Confined by Levees (red outline in Maxplot)

Claim 19: Levee overtopping discharge is accurately

computed in the FLO-2D model.
Substantiation of Claim 19. The FLO-2D model uses a broad crested weir equation to compute the levee
overflow. The weir coefficient for levee overtopping is hardwired in the model at 2.85. The levee length
is a function of the grid element width (times the width reduction factor WRF-value). To validate the
levee overflow, a simple model was created to force a discharge of 1,000 cfs over the levee in one grid
element flow direction. A levee was assigned in a straight line across the floodplain with a crest
elevation 30 ft or more above the floodplain elevation (red line in Figure 35). One flow direction in one
levee element was assigned a crest elevation of only about 2 ft above the floodplain. The inflow to the
grid system is a steady discharge of 1,000 cfs. At the end of a 4 hour simulation, the discharge past the
levee was approximately 1,000 cfs. Reviewing the final water surface elevation, the head H on the levee
was 6.58 ft (levee crest = 4,705.00 and the final water surface elevation = 4,711.58). The discharge Q at
hour 4.0 was computed as:
Q = CLH1.5 = 2.85 * (50 ft * 0.41421)*6.581.5 = 996.24 cfs
This predicted discharge is reported exactly in the LEVEE.OUT file (Table 9).

59

Figure 35. Simple Levee Example Project with 1,000 cfs Steady Discharge

Table 9. Reported Levee Overtopping Discharge at Hour 4.0

60

3.6 Accuracy of Computational Results

Model accuracy is a defined as the degree to which the numerical solution approaches the true solution.
Accuracy depends on the resolution of the data base (including inflow hydrographs, rainfall, topography,
cross sections, etc.), model assumptions and limitations, and convergence of the solution scheme. For a
1-D model, an exact analytical solution is possible. Finite difference solutions are only approximate, so
for a complex river system model, an exact solution is not possible.
FLO-2D is a comprehensive hydrodynamic model with many detailed hydrologic and hydraulic
components. The solution of the full dynamic wave equation is based on a Newton-Raphson
approximation of a central difference numerical scheme that is subject to very small timesteps to insure
solution convergence of 0.01. The accuracy of the hydraulic computations are based on predictions of
velocity and depth plus or minus 0.01 (fps or ft). The accounting of the volume conservation is typically
on the order of millionths of one percent.

Claim 20: The FLO-2D code is free of

volume conservation errors.
Substantiation of Claim 20. The FLO-2D model reports on volume conservation errors as previously
discussed in the Substantiation of Claim 1. Over the past 22 years, all of the accounting code errors for
volume conservation have been gradually eliminated.

Claim 21: Variation of results between model versions is

not a indication of computational inaccuracy.
Substantiation of Claim 21. Different versions of the FLO-2D model can yield slightly different results in
complex projects simulations because of variation in the computational timestep related to changes in
the numerical stability criteria. The results from simple case studies with known analog solutions will
still be replicated exactly, but complex models simulated with different versions may have slight
variations in results that arise from improved numerical stability methods or enhancements to urban
detail components. For example, the change in emphasis from the DEPTOL and WAVEMAX stability
parameters to the Courant criteria has resulted in a smoother and faster model with larger timesteps
and less variation in hydraulic results.
A simple example of the variability in results can be shown in the Monroe project example. This was a
Corps of Engineers alluvial fan flood hazard delineation project (Figure 36). The Version 2007 results
were obtained with the DEPTOL and WAVEMAX stability controls and the Version 2009 model was run
with the Courant criteria as the sole numerical stability control of the timestep. The Version 2009
simulation run time was only about 50% of that of the Version 2007 model. The differences in the
predicted maximum depths are displayed in Figure 37. All the differences in the predicted maximum
depths are 0.20 ft or less. The predicted maximum velocities are compared in Table 10 for Versions
2007 and 2009.
61

Figure 36. Corps' Monroe Alluvial Fan Project with a Defined Channel Using Version 2007

Figure 37. Corps' Monroe Alluvial Fan Project Difference in Maximum Depth
(all the differences in maximum depth are 0.2 ft or less)

62

Maximum Channel Velocity

Listed in Descending Order (top 24)
Version 2007
NODE

MAXIMUM VELOCITY
FPS OR MPS)

2946
2922
2968
2985
2998
2793
2753
2830
2863
2894
1172
2666
2619
1414
2567
1011
1253
931
1092
1334
369
289
2455
2394

14.93
14.93
13.81
12.06
11.19
11.16
11.16
10.31
9.82
9.72
8.98
8.83
8.83
8.79
8.77
8.62
8.59
8.41
8.37
8.36
8.33
8.33
8.28
8.28

TIME OF OCCURRENCE
(HRS)

2.59
2.59
3.02
2.50
2.31
3.04
3.04
3.04
3.03
3.03
3.48
3.06
3.06
4.23
3.13
3.47
3.47
3.19
3.03
2.71
3.83
3.83
3.11
3.11

Table 10. Difference in Maximum Velocity Between Version 2007 and Version 2009 for the Corps'
Monroe Project
When reviewing the Monroe project along with the Flume Case Study and the Camarillo Hills Drain, it is
demonstrated that project results can vary with model enhancements to improve speed and stability
while still converging to known analog solution results.
A source of model version differences are changes or enhancements to the model detail components.
This may include changes in variables, new components or adding spatially variable components. A
complete list of the model revisions between Version 2007 and Version 2009 are presented in the
Appendix. Some recent model revisions include:

Courant stability parameter previously hardwired at 1.0 is now a user assigned variable.
Rainfall runoff from buildings that are completely blocked.
Spatially variable limiting Froude numbers.

The sheer volume of computations during a flood simulation can also be a source of noise error
generation and slight variability in results. Not all of the variables and arrays in the FLO-2D model are
coded as double precision numbers. Real numbers only track 8 significant digits. When real numbers
are employed instead of double precision numbers, small differences between large numbers can lead
to variation in hydraulic results with changes in timestep magnitude.

63

3.7 Benchmark Testing

Benchmark testing has two purposes: 1) To verify the accuracy of model updates and new releases; 2)
To maintain the consistency of individual components. Individual benchmark tests have been
established for the FLO-2D Program to test the following components:
a.
b.
c.
d.
e.
f.
g.
h.
i.
j.

Overland flow
Channel flow
Channel/overland flow exchange
Street flow
Buildings (ARF/WRF values)
Hydraulic structures
Rainfall/Infiltration
Levees and levee overflow
Levee and dam breach
Sediment transport

A batch file program has been created that runs all the benchmark tests sequentially. Each test is
automatically compared against a known or previous set of results and a message is displayed indicating
that either the test passed or failed for a specific component. The benchmark tests are run to verify that
the model revision produces the anticipated results. All benchmark tests are used to verify that any
code revisions have either negligible effect on the model output, generate differences within acceptable
limits, or meet anticipated variation in the results as in the case of model enhancements. The FLO-2D
benchmark testing program is designed to provide model consistency and reliability. Baseline projects
are also used in the testing program to check component interaction. The baseline projects are actual
flood projects that are considerably more complex than the individual component benchmark tests and
encompass a broad range of flow conditions and component interaction.
It is the intent of FLO-2D Software, Inc. to be transparent with its modeling practices. When a modeling
issue is identified, the workflow of the FLO-2D staff is as follows (Table 11):
Event/Action
1.
2.
3.
4.
5.
6.

Outcome

Issue is reported to FLO-2D support center.

Data files and other supporting information are compiled.
Appropriate code is identified, revised and compiled.
Benchmark tests are conducted.
If a Benchmark tests fail, further revisions are necessary.
New executable is posted to the website with revisions document.

Support log updated.

Support log updated.
Support log updated.
Document testing in support log.
Update web site.

Table 11. FLO-2D Model Revisions Reporting Procedure

64

FLO-2D model revision and update since June 2010 posted on the FLO-2D website (www.flo-2d.com).
Future model revisions will be tagged with a version and build number identified with a numeric code in
the format:
VV-YR.MM.BB
where:
-VV is a two digit number indicating the version number (e.g. 11 for 2011)
-YR is a two digit number of the year that major new components or routines were implemented
(e.g. 12 for 2012);
-MM is a two digit number of the month that major new components or routines were implemented
(e.g. 06 for June);
-BB is a two digit number that identifies the executable build. A change in the build number
represents a significant bug fix or model revision. Build numbers are not changed for typos,
minor revisions or enhancements.
For example, 11-12.04.02 indicates that the 2th build or compiled revised executable of Version 2011
was released to the public and posted on the website in April of 2012. The model writes this tag to the
SUMMARY.OUT file and other places. A FLO-2D Model Revisions document is updated every time there
is a bug fix or enhancement in the model or processor programs. In the document, each update release
to the public will have build tag number listed to identify the model revisions with that build. Finally in
all new versions, the model build number will be displayed on the FLO-2D model runtime introductory
message to the screen, on the screen graphics, in the About dialog box in the GDS, in the CONT.DAT file
line one and in SUMMARY.OUT file.

65

IV. Summary of Validation Studies

This chapter is organized into a series of case studies. They are presented in the order that they were
addressed in the report to provide more information on the validation tests.
4.1 Case Study 1. Overland Flow-Flume with Analytical Solution
The Case 1 flume study was developed by a consulting firm (Exponent) in 1999 to verify if the FLO-2D
results were accurate. This particular flume was one of a series of similar models involving flumes or
open prismatic channels of varying sizes. This model was 60 elements long by 4 elements wide, each
element being 200 ft x 200 ft. Other flumes were 20 ft x 20 ft up to 2,000 ft square. The slope was
0.003 and a uniform n-value was set at 0.05. Given a steady discharge of 8,000 cfs spread across the 4
inflow nodes (2,000 cfs per inflow element), the velocities and depths can be computed with Manning's
equation as steady and uniform flow in the lower portion of the flume. Like a prototype flume, the flow
has to accelerate to a steady flow condition since the inflow elements have no convective acceleration
component in the downslope direction. There is also no local acceleration term in the full dynamic wave
momentum equation due to the steady flow condition and for this prismatic channel case, the
momentum equation collapses to the kinematic wave equation. The unknown hydraulic variables of
velocity and depth can be computed using Manning's equation and continuity. The FLO-2D predicted
final steady, uniform flow depth and velocity are 6.82 ft and 5.86 fps respectively as shown in Figures 4
and 5 which is the exact analog solution as follows:
Q = VA = 1.486/n R2/3Sf1/2 A
where:

A = flow area = width * depth = 200 ft * flow depth d

V = depth average velocity
n = 0.05
Sf = friction slope = So = bed slope using the kinematic wave equation = 0.003
R = hydraulic radius flow depth d for a wide shallow channel
Q = discharge 8,000 cfs

Substituting:
Q = 8,000 cfs = (1.486/0.05) *d2/3 * (0.003)1/2 * 200 ft * d
d = (8,000/325.57)3/5 = 6.83 ft
V = Q/A = 8,000 cfs /(6.83 ft*200 ft) = 5.86 fps
This case study validated that the FLO-2D model computes overland flow accurately. Variations of this
example has been used repeatedly by consultants and studies to challenge the basis of the eight flow
direction FLO-2D model. In fact, physical model flumes studies have been undertaken in universities to
test the FLO-2D flume results.

66

4.2 Case Study 2. Overland Flow - Truckee River

Tetra Tech, Inc. was contracted by the US Army Corps of Engineers, Sacramento District, to conduct a
flood hazard delineation project for the Truckee River through Reno, Nevada. This project was
supported by the City of Reno and Washoe County Flood Control District. The FLO-2D model was
selected for this study because most of the flood volume in the lower reach of the study area was on the
floodplain. The focus of this study was a January 1997 flood that was documented with gage data, high
water marks, pictures and video. This data base was supplemented with over 100 post flood channel
cross section surveys. An existing HEC-2 model was updated and used to calibrate in-channel flow nvalues.
The key to the FLO-2D overbank flood analysis was the relationship between the volume of water in the
channel and the volume of water on the Truckee Meadows floodplain area. Downstream of this area,
all the flow in the river and on the floodplain was forced over the geologic control (bedrock sill) and into
a narrow canyon (Figure 38). This broad floodplain has historically flooded in response to the backwater
at the canyon entrance. Just upstream of the canyon a USGS gage (New Vista) was located. Upstream
of the Truckee Meadows area was another gage (Reno gage) which essentially captured all the flow
entering the Meadows reach. Between the two gages, the river conveyance capacity was limited to
about 10,000 cfs. The 1997 flood peak discharge was approximately 20,000 cfs. Complicating the flood
replication was the fact that the travel time associated with the overbank flooding through the Truckee
Meadows floodplain area was on the order of 6 hours whereas the channel flow was less than 0.5 hours.
To replicate the downstream New Vista gage stage for the 1997 flood, it was necessary to accurately
route the flow in the channel, the overland flow over the huge meadow and the return flow to the
channel upstream of the gage. Following the successful replication of the 1997 flood, the model was
applied to other historical flood events that were less well documented. The ability of the model to
accurately replicate the lower gage stage as shown in Figures 9 and 10 verified that three components of
the model functioned correctly in concert:

Channel routing

Unconfined overland flood routing

Channel-floodplain exchange (both overbank flow and return flow to the channel)

It is noteworthy that this project marked the end of the use of the channel power regression
relationships for the channel cross sections. If the actual surveyed cross section data were used in the
model instead of the power regression relationship, the predicted stage at the New Vista gage would be
an almost perfect match. Only power regression relationship were used to represent the cross sections
when the model was initially developed for this project.

67

Reno Gage

New Vista Gage

Figure 38. Sparks Truckee Meadows Area General Peak Flow Path (Google Earth, 2011)

4.3 Case Study 3. City of Camarillo, California FIS Study

FEMA issued a Preliminary Flood Insurance Study (FIS) reporting Digital Flood Insurance Rate Maps
(DFIRM) for Ventura County and the City of Camarillo on September 16, 2005. In February 2006, the
City appealed and FEMA in April 2006 agreed with the technical assessment of the Citys appeal and
requested additional analysis. In January 2008, the City filed a detailed two dimensional FLO-2D model
floodplain study (27,979 - 100 ft elements) in support of their appeal. The revised floodplain boundaries
and elevations were incorporated into the DFIRM with a date of January 20, 2010. After the DFIRMs
went into effect, it became apparent that it was difficult to determine whether flood insurance was
required in certain areas. The City came to the conclusion that a more refined hydraulic model should
be prepared to better depict the 100-year flood for flood insurance and requested that 10 ft grid
elements be used in the FLO-2D model. It was presumed that a higher resolution grid system was
required to more accurately define the shallow flow flooding in the densely populated urban areas.
The new FLO-2D FIS Study 10 ft model encompassed 1,093,307 grid elements, over 5,000 buildings,
almost 2,400 channel elements, spatially variable n-value assignment (including the streets), privacy
walls, and numerous hydraulic structures to simulate the culverts and bridges (Figures 39 - 41).

68

Figure 39. City of Camarillo 10 ft FLO-2D Grid System with the Building Shape File

Figure 40. Detail of the City of Camarillo 10 ft FLO-2D Grid System with LiDAR Data

Figure 41. Spatially Variable n-value Shapefile

69

There were several areas that had limited predicted flooding in the 10 ft model compared to the 100 ft
model based on the higher resolution topography, streets, and the assignment of building area/width
reduction factors. In many areas, the flood inundation was confined to the streets. The area between
Lewis Road and Ponderosa Drive depicts the refined model resolution and the importance of flood
conveyance in the streets (Figure 42).

Figure 42. Flooding Differences Between the Models (100 ft model left and 10 ft model right)
Some of the predicted subdivision flooding in the 100 ft model was not predicted in the higher
resolution 10 ft model (Figure 43).

Figure 43. Flooding along Ponderosa Drive (100 ft model left and 10 ft model right)
The refined model had better resolution of grid element elevation along the highway crest. In the 100
ft model, flooding crosses Highway 101 (Figure 44), but the flood is contained by highway topography in
the 10 ft model (Figure 45). The interpolation of lower elevations on each side of the highway reduced
the highway element elevations in the 100 ft model. The 10 ft model predicts that the flooding would
be confined to elements north of the east bound lanes.

70

Figure 44. 100 ft Model - Flooding across Highway 101.

Figure 45. 10 ft Model - Flooding is confined by Highway 101.

The higher resolution 10 ft FLO-2D model more accurately assessed the shallow urban flooding for the
City of Camarillo 100-year FIS DFIRM maps. The 10 ft grid system represented the internal street
topography and predicted that much of the flooding would be confined to the streets. Using the area
and width reduction factors to limit building flood storage and simulation flow obstruction forced more
of the predicted flooding into the streets. With more of the predicted flow in the streets, the floodwave
processes faster through the urban area. The City of Camarillo supported the DFIRM map submittal to
FEMA.
4.4 Case Study 4. Middle Rio Grande 2005 Spring Runoff Monitoring Project
The FLO-2D Middle Rio Grande (MRG) model has been significantly expanded since the its first
application in 1996. Over the years, the initial 500 ft grid model was replaced with a 250 ft model which
was then calibrated with the Corps 2005 study: Overbank Monitoring of the 2005 Spring Release from
Cochiti Dam. The overbank monitoring study included aerial photography, video, cross section surveys
and water surface elevation measurements to establish baseline data associated with maximum Cochiti
Dam releases, which was used to spatially calibrate the FLO-2D predicted overbank flooding.
On June 1, 2005, the Corps of Engineers took aerial photography (color infrared orthophotography) of
the Middle Rio Grande during spring high flow releases from Cochiti Dam. The flows were measured at
6,900 cfs at the gaging station below Cochiti. The floodplain inundation in these aerial photos between
Cochiti Dam and Elephant Butte Reservoir was digitized. The digitized mapping represented river
surface and floodplain inundation and included features such as: main channel, islands, sand bars,
inundation floodplain and abandoned channels.

71

The digital terrain model (DTM) data base was compiled from six different mapping efforts by the Corps
and Bureau of Reclamation over six years. Each DTM data set represented a specific reach of the Middle
Rio Grande. The DTM data sets were provided in various formats and had different reference elevation
datum. All the data sets were compiled and converted to a consistent datum using the New Mexico
State Plane (Central) North American Datum (NAD) 1988 horizontal and North American Vertical Datum
(NAVD) 1988 vertical reference. The resolution of the DTM data varies by reach. The resolution of the
original DTM data sets was adequate to generate 2 ft contour maps.
Over 480 cross sections had been surveyed throughout the Middle Rio Grande (MRG) from Cochiti Dam
to Elephant Butte Reservoir. Most of the cross sections were surveyed in conjunction with the Bureau
of Reclamations river maintenance program. These cross sections were distributed and interpolated to
the 3,241 channel elements in the model. Levees constrain the potential flooding through most of the
MRG valley and the levees on both sides of the river are incorporated into the model. In the MRG
model, the inflow hydrograph was assigned to a channel element representing the Cochiti Dam outlet
works and outflow nodes were assigned to grid elements in the headwaters of Elephant Butte Reservoir.
The FLO-2D MRG model was used to predict the area of inundation, water surface elevation and
discharge anywhere in the MRG system.
FLO-2D model calibration was accomplished by reach. The primary data base that was used for the
model calibration was the Corps 2005 overbank monitoring study. The various USGS gage data were
used to replicate the flow hydrographs. The model discharge routing (compared with gage
hydrographs) and model spatially variable flood inundation (compared with infrared color photography)
were reviewed with each calibration run. Water surface elevations were measured at specific times at
various cross section locations. FLO-2D predicted water surface elevations were written to a file along
with the measured water surface elevations at the simulated survey time (12 noon, June 1, 2005).
Model calibration was a balance between the hydrograph calibration, water surface calibration and
matching the area of inundation.
Two validation runs were made for the period from April 3-11, 2004 (Figure 46) and May 5-22, 2004
(Figure 47) from San Acacia to San Marcial for discharges in the range of 1,500 cfs to 3,000 cfs. No
adjustments to the data files were made for these additional simulations. For the April 2004
hydrograph, FLO-2D under-predicts the reported discharge at the USGS San Marcial gage and for the
lower-flow, May 2004 hydrograph, the model over-predicts the reported discharge at the San Marcial
gage. This demonstrates the variability of the gage-discharge relationship and frequent mobile bed
shifts at the San Marcial gage. Using the identical model but different inflows based on the San Acacia
gage, the reported discharge is about 20 percent different from the predicted discharge for a portion of
the hydrograph, but high in one case and low in the other. In both hydrographs, however, the predicted
and reported discharges converge near the end of the hydrograph. This epitomizes the difficulty that
flood modelers have with field data bases associated with mobile bed river systems. Frequent gage
calibration discharge measurements are required to reset the reference stage. Once a flood routing
model has been calibrated with gage data, it generally will produce more consistent and accurate
discharge predictions than the gage measurements on the sand bed river.

72

Figure 46. FLO-2D Predicted vs. USGS Gage Discharge at San Marcial April 3-11, 2004
00

Figure 47. FLO-2D Predicted vs. USGS Gage Discharge at San Marcial May 5-22, 2004

4.5 Case Study 5. California Aqueduct Operational Design Flows

This project was one of a series of projects for the Corps of Engineers Sacramento District, the California
Department of Water Resources (DWR) and the California Department of Justice regarding the 1995
Arroyo Pasajero flooding. The arroyo flooding from the City of Coalinga to the California Aqueduct
destroyed the arroyo I-5 Highway Bridge. A flood detention basin protecting the California Aqueduct
filled and was breached discharging flood waters into the Aqueduct. The discharge in a 10 mile reach of
the California Aqueduct near Huron, California (Figures 48 and 49) was simulated with the FLO-2D model
using 96 channel elements in a 500 ft grid system. The aqueduct is difficult to simulate with a numerical
73

model because of the very mild slope (5.7 x 10-5), high flow depth and low Froude numbers. In addition,
the bedslope is variable because of land subsidence associated with agricultural groundwater pumping
(Figures 50). The variable bed slope altered the design discharge operational conditions.

Arroyo Pasajero

California Aqueduct

Figure 48. Aerial Image of the FLO-2D Reach of the California Aqueduct Near Huron
(from Google Earth, 2011)

Figure 49. California Aqueduct Near Huron (from Kewaneh, Google Earth, 2011)

74

Figure 50. California Aqueduct Bedslope Near Huron

To prepare the FLO-2D model for the simulation of the aqueduct flooding, the aqueduct model was
calibrated with the design flows. The original design discharge for this concrete lined trapezoidal
channel was approximately 8,150 cfs. The roughness (Mannings n-values) for most of the 10 mile
reach ranged from 0.020 to 0.024 due to subsidence (slope variability). The model predicts consistent
discharge throughout the reach. The predicted velocities and flow depths are uniform and replicate the
hydraulic design conditions. The Corps and DWR used the FLO-2D model to address the operation of the
aqueduct gates during selected return period floods.
4.6 Case Study 6. Green River FLO-2D Model from Flaming Gorge Dam to the Colorado Confluence
As part of the U.S. Fish and Wildlife Upper Colorado River Endangered Fish Species Recovery
Implementation Program, a FLO-2D model of the Green River system from Flaming Gorge Dam to the
Colorado River confluence (over 412 miles) was developed. The project had a number of purposes
related to the endangered fish species migration, spawning and drift cycles. Cross section surveys were
collected throughout the Green River system to observed changes in channel morphology over a long
period. The FLO-2D model was developed to predict river flow response to Flaming Gorge Dam Releases
in conjunction with spring runoff from the major tributaries. The timing of peak flows, opening of
backwater habitat and predicting bottomlands overbank flooding were analyzed with the model.
One of the project objectives was to determine the impact of the Flaming Gorge Power Plant diurnal
releases on backwater habitat in Canyonlands National Park just upstream of the Colorado River
confluence. The Dam power releases had a daily variation between 2,000 and 4,500 cfs and if this
discharge range persisted downstream, it could open backwater habitat and then dewater it within 24
hours, forcing endangered larval fish back into the river flow. The endemic fish species, particularly the
Colorado Pike Minnow and the Razorback Sucker, lay eggs in the upstream reaches of the Green River
system and when the larval fish hatch, they begin a downstream drift cycle. When the larval fish reach
75

downstream flooded bottomlands and backwater areas, they quickly grow in the warmer nutrient rich
water. If they are flushed out of this habitat back into the river, fish drift into Lake Powell downstream
of the Colorado River confluence and are preyed upon by nonendemic lake fish species. To assess the
variability on backwater habitat from the power plant releases, the FLO-2D model simulated the diurnal
power generation hydrograph shown in Figures 52-54. The location of the Jensen gage is shown in
Figure 51.

Ouray Gage

Jensen Gage
Colorado River
Confluence
End of Uinta Valley

Figure 51. Green River Bed Slope from Flaming Gorge Dam to the Colorado River Confluence
While the integrity of the Flaming Gorge power release spikes is sustained through the steep Green
River Canyon in Dinosaur National Park to the Jensen Gage (Figure 24 repeated below), the spikes are
largely attenuated by the middle of the mild slope Uinta Valley at the Ouray gage (Figure 52) and they
are essentially melded into one hydrograph crown by the end of Uinta Valley as shown in Figure 53. By
the Colorado River Confluence, the two spike releases are a single hump as shown Figure 54.

Figure 24. FLO-2D Predicted vs. Measured Discharge, Jensen USGS Gage for the Flaming Gorge Releases
76

Figure 52. Flaming Gorge Power Plant Spike Release Hydrograph at the Ouray Gage

Figure 53. Flaming Gorge Power Plant Spike Release Hydrograph at the end of Uinta Valley

Figure 54. Flaming Gorge Power Plant Spike Release Hydrograph at the Colorado River Confluence
77

V. References
Cunge, J.A., F.M. Holly Jr., and A. Verwey, 1980. Practical Aspects of Computational River Hydraulics,
Pittman Advanced Publishing Program, London, UK.
Fletcher, C.A.J., 1990. Computational Techniques for Fluid Dynamics, Volume I, 2nd ed., Springer-Velag,
New York.
Flood Control District of Maricopa County, 2010. Letter to Jim O'Brien Re: "The Use of FLO-2D for
Hydrologic Modeling in Maricopa County", Phoenix, AZ.
Green, W.H. and G.A. Ampt, 1911. "Studies on soil physics, part I: The flow of air and water through
soils," J. of Agriculture Science.
International Association of Hydro-Environment Engineering and Research, 1994. "Guidelines for
Documenting the Validity of Computational Modelling Software", Madrid, Spain, www.iahr.com.
James, M.L., G.M. Smith, and J.C. Wolford, 1985. Applied Numerical Methods for Digital
Computation, Harper and Row, New York, N.Y., Third Edition.
Jin, M. and D.L. Fread, 1997. Dynamic flood routing with explicit and implicit numerical solution
schemes, J. of Hyd. Eng., ASCE, 123(3), 166-173.
JE Fuller Hydrology and Geomorphology, 2010. " Refinement of Methodology: Alluvial Fan Flood Hazard
Identification & Mitigation Methods," Final Report submitted to the Flood Control District of
Maricopa County, Phoenix, AZ.
Michael Baker, Jr., 2001. "A Review of the FLO-2D Model for Riverine Applications," Letter Report
prepare for FEMA, by the Alexandria, VA office.
Miller, J.J., 1996. "Comparison of Models to Mitigate Flood Hazard to Transportation Alignments on
Alluvial Fans," Masters Thesis, Dept. of Geoscience, Univ. of Nevada, Las Vegas, NV.
Nalesso, M., 2009. "Integrated Surface-Groundwater Modeling in Wetlands with Improved Methods to
Simulate Vegetative Resistance to Flow," Ph.D. Dissertation, Florida International University, Miami,
FL
Ponce, V.M. and F.D. Theurer, 1982. Accuracy Criteria in Diffusion Routing, J. of Hyd. Eng., ASCE,
108(6), 747-757.
Ponce, S.M., 1989. Engineering Hydrology, Prentice Hall, Englewood Cliffs, New Jersey.

78

Appendix
U.S. Army Corps of Engineers Original Two Letters of Certification
Letters of Endorsement from:
Flood Control District of Maricopa County
Mohave County Flood Control District
Cochise County Flood Control District
Email Endorsements from:
California Department of Water Resources
U.S. Corps of Engineers, Sacramento District
Truck River Flood Project
List of FLO-2D Version 2009 Model Revisions

79

Original Corps Letters of Certification

80

81

82

83

Letters of Endorsement

84

85

86

87

88

Email Endorsements
U. S. Army Corps of Engineers, Sacramento District
Senior engineering staff in the Hydraulic Design Section, Sacramento District, USACE, have
reviewed the document FLO-2D Model Validation for Version 2009 and Up, dated June 2011.
Based on our review of this document, as well as 15 years of successful application of FLO-2D
on numerous Sacramento District projects, we believe that FLO-2D v. 2009 is technically sound
for the purpose of floodplain mapping in support of FEMA Flood Insurance Studies.
Lea Adams, P.E.
Chief, Hydraulic Design Section
Sacramento District
US Army Corps of Engineers
1325 J St
Sacramento CA 95814

California Department of Water Resources

The Floodplain Evaluation Branch of the California Department of Water Resources (DWR),
Division of Flood Management, has reviewed the FLO-2D Model Validation (Version 2009 and
up) prepared for FEMA, dated June 2011, as prepared by FLO-2D Software, Inc. Based on this
review, we endorse the findings presented therein and support the recommendation that the
FLO-2D Flood Routing Model (Version 2009 and later) be included on FEMAs list of approved
hydraulic models for Flood Insurance Studies.
David J.W. Wheeldon, Chief
Floodplain Evaluation Branch
Division of Flood Management
California Department of Water Resources
Truckee River Flood Project, Reno, Nevada
I have reviewed the document FLO-2D Model Validation for Version 2009 and Up, dated June
2011. Based on my review of this document, I believe that FLO-2D v. 2009 is technically sound
for the purpose of floodplain mapping in support of FEMA Flood Insurance Studies. The Truckee
River Flood Project has been using the FLO-2D model for over 11-years. The Truckee River
Flood Project has worked with the Sacramento District of the US Army Corps of Engineers on
planning and evaluating numerous project alternatives for the project.
The Truckee River Flood Project uses FLO-2D to estimate flood flows that could occur in the
Reno, Sparks and Truckee Meadows region due to emergency releases from and potential
89

breaches of the Dams upstream of Reno-Sparks. These inundation maps are used by the
Regional Emergency Operations Center in running table top exercises of various flood
emergencies that could happen in the region. The flood project also monitors flood elevations
in critical flood pool areas to evaluate the adverse flood impacts that could happen if fill is
placed in these critical areas. Local ordinances require mitigation of floodplain storage volume
that may be displaced by a project.
Paul D. Urban, P.E. , Project Manager
Truckee River Flood Project
Reno, Nevada

90

91

92

Revisions, Enhancements and Bug Fixes

to the FLO-2D Model and Processor Programs
Since October 1, 2009
Updated 5/02/11 (Most recent revisions are shown in blue.)
FLO-2D Model Version 2009.06
Revisions and bug fixes in the FLO.EXE program include (most recent at the bottom):
1. Revised the license code for the registry file (10/9/09).
2. Edited dam and levee breach routine for the pipe breach roof collapse, channel slump collapse
and the sediment transport equation (10/9/09).
3. Edited a metric format statement to include more digits for levee overtopping discharge
(10/25/09).
4. Revised the levee overtopping discharge broad crested weir coefficient to 1.44 for the metric
option (10/25/09).
5. Fixed the format for a plot variable format (11/5/09).
6. Fixed the graphical display of the interior channel elements at runtime for channels with more
than one segment (11/14/09).
7. Edited two format statements for larger integer grid numbers (11/14/09).
8. The initialization of some sediment transport metric parameters for MPM-Smart was not
complete (11/14/09).
9. Revised the reservoir water surface elevation assignment for values less than sea level
(11/15/09).
10. Fixed the index for the interior channel element arrays for the animation output file (11/15/09).
11. RAIN.EXE may be missing from the C:\Program Files\FLO-2D folder. If it is download the
RAIN.EXE from the update page and replace it (11/18/09).
12. The reformatting of the channel data for writing to the HYCHAN.DAT in metric did not have
sufficient digits for the bed shear stress (12/1/09).
13. A variable initialization error was identified and fixed for the channel when bank elevations are
assigned in CHAN.DAT but are not necessary (12/5/09).
14. More variable initialization errors. These errors are the result of switching from Lahey to Intel
Fortran for Version 2009 to support multiple processor programming. Intel does not
automatically assign a zero to non-initialized variables like Lahey Fortran. These errors were
encountered when writing the output files. One error was related to channel and mudflow. The
other error was writing the DEPTHDUR variable to the output file (12/13/09).
15. Fixed volume conservation errors for rain on channel outflow nodes and rain on streets in
metric. Improved plot of channel flow depth at runtime. Fixed the check box for the volume
conservation reporting in the summary dialog box at the end of the simulation. More variable
initialization errors were fixed in the output data reporting (12/22/09).
16. A variable DSECINCREASE was not set to zero (1/13/10).

93

17. For flow upstream in a channel an adjustment to the routing scheme was implemented to
increase the velocity. The coefficient for diffusive wave velocity was revised when the
momentum term shifted direction. (2/2/10).
18. The graphics display at runtime was fixed so that the interior channel elements display the
channel flow depth (2/2/10).
19. For channel routing, edited the channel read subroutine to stop when the end of the channel is
reach for displaying the interior channel elements (2/15/10).
20. Corrected the RGRIDTOT = 0 variable initialization for metric. This was already done for English
units (5/8/10).
21. Fixed a bug for metric mudflow with a channel. The stability criteria was being by-passed for low
flows based on 1.0 m instead of 1.0 ft in metric (6/5/10). This fix was released in BUILD No. 0910.06.01. The build number is listed at the end of the SUMMARY.OUT file and is being for the
first time with this bug fix.
22. The numerical solution of the momentum equation was expanded for the sign convention of
convective acceleration term for upstream channel flow in the case of tidal conditions
(7/14/10). BUILD No. 09-10.07.01.
23. The Courant number stability criteria which was previously hardwired and set to 1.0 was
redesigned to be an optional user variable as line 2 in TOLER.DAT. The default value is now 0.6
and the user can assign a value in the suggested range of 0.3 to 1.0 (7/14/10). BUILD No. 0910.07.01
24. Binary files of the output data are now generated for the last completed output interval
(multiple of TOUT in CONT.DAT). This means that the model can be restarted with the last
completed output interval for every simulation regardless of what cause the model to stop (i.e.
successful completion of the model, inadvertent computer termination, or user interrupted
event) (7/14/10). BUILD No. 09.10.07.01.
25. The HYCROSS.OUT had incorrect times listed when streets were included in the some but not all
of the prescribed floodplain cross sections (7/14/10). BUILD No. 09.10.07.01.
26. The infiltration porosity parameter POROS has been adjusted to accommodate the Flood
Control District of Maricopa County drainage manuals that specifies that the soil moisture
deficiency parameter DTHETA represent the volumetric soil moisture deficit which includes the
porosity. If POROS is assigned 0.0, then DTHETA includes the porosity and the DTHETA ranges
from 0.0 to 0.5. If DTHETA represents only the soil moisture deficit, set a reasonable soil
porosity in the range from 0.35 to 0.45 (11/15/10). BUILD No. 09-10.11.02.
27. The floodplain limiting Froude number is now spatially variable and is assigned in an optional
data file FPFROUDE.DAT. The values in this file will supersede the global assignment of the
limiting Froude number FROUDL in CONT.DAT. The format of the file is simply: F(line
character) Grid element number limiting Froude number. These values can be graphically
assigned in the GDS (11/15/10). BUILD No. 09-10.11.02
28. Fixed the binary file status (open) for the hydraulic structure data when continuing a previous
simulation (11/20/10). BUILD No. 09-10.11.02.
29. Expanded the hydraulic structure routine to allow multiple culverts to discharge to one culvert
(11/20/10). BUILD No. 09-10.11.02.

94

30. Corrected code for channel mudflow with hydraulic structures. This coded error was due to an
improper placement of an ENDIF statement that resulted in a fatal Fortran error that would not
allow the model to run (12/21/10). BUILD No. 09-10.12.02.
31. There was a metric conversion error for a Breach output parameter that would cause a fatal
error (1/24/11). BUILD No. 09-11.01.03.
32. Improved the depression storage initial routing for rainfall by decrementing the timestep for
negative flow depths instead of redistributing the flow. This was important for ARF reduced grid
elements (1/28/11). BUILD No. 09-11.01.03.
33. Fixed an array index for reading multiple inflow sediment supply size fraction groups (1/28/11).
BUILD No. 09-11.01.03.
34. Added code to identify and eliminate negative grid element storage when very small rainfall
begins to accumulate and runoff. This may occur when the flow depth first starts to exceed the
TOL value using large timesteps on grid elements with ARF reduced area. The volume
conservation error was minor (almost negligible), but the error was eliminated with this revision
(1/30/11). BUILD No. 09-11.01.03.
35. There was a bug that was introduced for writing channel hydraulic structure discharge to the
temporary file in the latest build. This did not affect any of the model computations, only the
writing of the structure discharge to the HYDROSTRUCT.OUT file. The discharge was reset to
zero before the final results were written to file (2/3/11). Build No. 09-11.02.03.
36. Writing the channel Froude number to the HYCHAN.OUT file was corrupted by failure to
initialize the variable prior to writing to file (2/7/11). Build No. 09-11.02.03.
37. In the channel variable regression relationships for computing the channel top width above the
assigned exceedance value, two errors were corrected. These errors were essentially reporting
errors that did not affect the actual flood routing because the top width is only used to assign
the right bank element (2/15/11). Build No. 09-11.02.03.
38. The run time of the model in SUMMARY.OUT was not being reported correctly. The runtime
was computed based on the CPU runtime. This was appropriate before parallel processors were
introduced. The runtime computation was revised to be based on the actual start and end of
the computation sequence in the model (2/21/11). Build No. 09-11.02.03 .
39. The floodway routine was improved for flow exchange between the floodway and floodway
fringe modifying the assignment for the floodway element as they filled with water (2/26/11).
Build No. 09-11.02.03.
40. The Karim Kennedy equation had a metric conversion error for the sediment D50 size (3/9/11).
Build No. 09-11.03.03.
41. The RAINCELL spatially variable rainfall assignment was re-initialized to enable the grid element
data to appear in random order in the RAINCELL.DAT file (4/6/11). Build No. 09-11.04.04.
42. The TIMETOPEAK output file was triggered by the BREACH time and if there was no dam or
levee breach simulation, the time was reported incorrectly (5/3/11). Build No. 09-11.05.04.
43. Expanded the cross section write format in the BASE.OUT file for larger discharges (5/4/11).
Build No. 09-11.05.04.

95

Revisions and bug fixes in the Processor Programs include (most recent at the bottom):
1. MAXPLOT The Version 2007 executable should be replaced with the 2009 in the FLO-2D folder.
The Version 2007 executable was accidentally installed in the model subdirectory. (10/25/2009).
2. GDS Removed the C:\TEMP\FLO-2D folder address. This is no longer needed (11/6/09).
3. MAPPER Rename the Mapper_2009.CHM to Manual_Mapper_2009.CHM so the Mapper
program can read the help files (11\6\09).
4. FLOENVIR Fixed the graphical display of the interior channel elements for more than one
channel segment (11/14/09).
5. MAXPLOT Fixed the graphical display of the interior channel elements for more than one
channel segment (11/14/09).
6. GDS Added warning messages for possible missing outflow nodes and for the channel
interpolation cross sections (11/18/09).
7. GDS Fixed the Breach Switch in the Levee Editor dialog box. The switch was required to be
turned on to save the levee data files even though a breach was not being simulated
(12/02/09).
8. To import AutoCAD DXF files into the GDS, the MO24rt.exe file may need to be reinstalled with
new control parameters: (/CFGHIJKM). This can be accomplished manually by performing the
following:
In Windows Explorer, locate the file C:\Program Files\FLO-2D\Mo24rt.exe. In the DOS command
window or from the Run command, execute the Mo24rt.exe file as follows:
C:\Program Files\FLO-2D\Mo24rt.exe/CFGHIJKM
9.
10.
11.
12.
13.

This will create all the files in the C:\Program Files\Common\ESRI folder (12/2/09).
MAXPLOT Fixed the channel flow depth display of non-bank interior elements (12/22/09).
FLO2DINTER.DLL Update for the GDS DLL library. Replace this file in the Windows\System32
folder (1/7/10).
FLOENVIR The display of the channel interior elements was fixed for channels with more than
one segment (2/2/10).
Mapper Some minor display changes and typos were fixed (2/2/10).
GDS A number of bug fixes and enhancements were made including (2/2/10):
New warning message for channel cross section interpolation.
Eliminated the GDS writing zeros in the INFLOW.DAT file when no hydrographs are
entered.
New warning message if RAIN.DAT is missing for RAIN.EXE and updates rainfall data
when all the *.DAT files are saved.
Deleted NOFLOCS outside the channel after channel realignment.
Levee duplicate assignments to the same boundary warning message.
Fixed the channel cross section plots to join the station lines.
Spelling errors were corrected.

14. GDS Setting the Total Rainfall in the RAIN dialog box was fixed so that the enter value was save
when assigning the rainfall distribution (5/8/10).
15. GDS Saving the global infiltration was corrected. The infiltration file INFIL.DAT was being
created without user initiation. Now INFILT.DAT is only generated when the user assigns either

96

16.
17.

18.
19.
20.
21.
22.
23.
24.
25.

26.
27.
28.
29.
30.

31.

global or individual parameters. The default global parameters are automatically assigned when
the user double clicks on an individual grid element to assign infiltration (5/8/10).
PROFILES The interpolation routine was adjusted for metric cross sections (6/15/10).
GDS A number of enhancements were incorporated for mega DTM data files including
(7/14/10):
Reading very large mega DTM data files exceeding 15 million points.
Reading new ASCII format DTM data files.
Reading multiple DTM data elevations files.
Null assignment of grid elements without DTM data.
Options for interpolation grid elements with null elevation assignment.
GDS Levee crest elevation can now be assigned from 3-D polylines data (7/14/10).
GDS Revised to read, assign and write optional Courant number in the TOLER.DAT (7/14/10).
GDS Expanded the HEC-RAS channel data interface extensively to enable geo-reference HECRAS channels to be automatically aligned an assigned cross sections.
rb_exe.exe channel right bank element assignment increased channel element numbers to 8
places, download and put this file in the FLO-2D folder (8/2/10).
rbcheck.exe channel right bank element check write to file, increased channel element
numbers to 8 places, download and put this file in the FLO-2D folder (8/2/10).
GDS Floodplain limiting Froude numbers are now spatially variable and can be assigned in the
GDS. The GDS was also modified to enable user defined Courant numbers (8/8/10).
PROFILES and HYDROG have been updated to handle grid systems greater than one million grid
elements (8/8/10).
GDS right bank element assignment rb_exe.exe program. The number of right bank extension
was improved for rectangular and trapezoidal channels. Download and put this file in the FLO2D folder (2/7/11).
HYDROG Expanded the cross section column width to handle grid element numbers greater
than 1,000,000 (2/14/11).
MAXPLOT Improved the file compare routine for plotting the differences between two output
files. Both graphical and computation improvements were made (2/15/11).
HYDROG Fixed a read format error introduced by a line change in the SUMMARY.OUT file
(3/2/11).
GDS Fixed an error in the levee profile tool (4/20/11).
FLOENVIR and MAXPLOT Fixed the levee plot graphics for flow directions for the case where
there was more than 200 consecutive grid elements in the first row or column that all had the
same coordinate value of either x or y (4/25/11).
GDS -- Fixed an error in the street realign tool. Default spatial street width was assigned a 1
instead of a 0. (5/2/11)

Revisions and bug fixes in the GUI program include (most recent at the bottom):
1. The version number was correct to match the FLO-2D model version number 2009.06 (5/8/10).

97