VA
F.P 600/7
3C-048
Tennessee
Valley
Authority
Office of
Natural
Resources
United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Cincinnati OH 45268
EPA-600/7-80-048
March 1980
Research and Development
User's Guide to
TVA-HYSIM
A Hydrologic
Program for
Quantifying Land-Use
Change Effects
_'.._*'.*-A-'
Interagency
Energy/Environment
R&D Program
Report
-------
EPA-600/70-80-048
November 1980
USER'S GUIDE TO TVA-HYSIM
A HYDROLOGIC PROGRAM FOR QUANTIFYING
LAND-USE CHANGE EFFECTS
by
Roger P. Betson, Jerad Bales, and Harold E. Pratt
Office of Natural Resources
Division of Water Resources
Tennessee Valley Authority
Norris, Tennessee 37828
IAG No. D9-E721-DS
Project Officer
Eugene F. Harris
Energy Pollution Control Division
Industrial Environmental Research Laboratory
Cincinnati, Ohio 45268
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report was prepared by the Tennessee Valley Authority and has
been reviewed by the Office of Energy, Minerals, and Industry, U.S. Environ-
mental Protection Agency, and approved for publication. Approval does not
signify that the contents necessarily reflect the views and policies of the
Tennessee Valley Authority or the U.S. Environmental Protection Agency, nor
does any mention of trade names or commercial products constitute endorse-
ment or recommendation for use.
THE TENNESSEE VALLEY AUTHORITY MAKES NO REPRESENTATION OF ANY KIND
WHATSOEVER, INCLUDING, BUT NOT LIMITED TO, representation or warranties,
expressed or implied, or MERCHANTABILITY, FITNESS FOR USE OR PURPOSE,
accuracy or completeness of processes, procedures, designs, definitions,
instructions, information, or functioning of these programs and related
material; TVA further expressly disclaims any knowledge of purpose for which
these programs may be utilized or its applicability for such use, nor shall
the fact of making it available constitute any such representation,
warranty, or knowledge, nor does TVA assume any liability, responsibility or
obligation arising from the use or malfunctioning of these computer programs
or related materials.
-------
FOREWORD
When energy and material resources are extracted, processed,
converted, and used, the related pollution impacts on our environment and
even on our health often require that new and increasingly efficient pollu-
tion control methods be used. The Industrial Environmental Research
Laboratory-Cincinnati (lERL-Ci) assists in developing and demonstrating new
and improved methodologies that will meet these needs both efficiently and
economically.
This report is a user's guide for a computer package designed to
quantify the hydrologic effects of land-use change. These programs have
been developed to permit their use at an interactive computer terminal, and
to streamline the output to the types of information used in land-use plan-
ning applications. This program package and user's guide should be of
interest to planners and consulting engineers in such applications as deter-
mining the probable hydrologic consequences of surface coal mining on the
hydrologic balance as required under Public Law 95-87. For further informa-
tion contact the Oil Shale and Energy Mining Branch of the Energy Pollution
Control Division.
David G. Stephan
Director
Industrial Environmental Research Laboratory
Cincinnati
m
-------
ABSTRACT
TVA-HYSIM is a computer package containing complex hydrologic
models specifically designed for ease of application in land-use planning
studies. TVA-HYSIM contains models capable of continuous hydrologic simula-
tion, as well as a rainfall generator and an erosion component.
This user's guide outlines the information required to operate the
programs and how this information is obtained, shows examples of input and
output, and provides examples of job controls needed to operate the program.
Model components are described in sufficient detail so that changes to the
algorithms may be made if so desired.
TVA-HYSIM is not adapted to handling dynamic land-use conditions,
but rather is designed to be used as a planning tool so that the end effects
of the land-use change can be evaluated before the change occurs. Thus in a
typical land-use change evaluation the model package would first be used to
simulate hydrology under present land-use conditions and then used to simu-
late the post land-use change hydrology. Some strategies for using TVA-
HYSIM to determine the effects of land-use change on the hydrologic balance
are offered.
This user's guide is submitted by the Tennessee Valley Authority,
Division of Water Resources, in a partial fulfillment of the terms under
Interagency Agreement No. D9 E721-DS with the U.S. Environmental Protection
Agency. Work was completed as of February 1, 1980.
IV
-------
CONTENTS
Forward
Abstract
Figures
Tables . .
List of Abbreviations and Symbols.
I. Introduction 1
A. Overview 1
B. Model Description 2
C. Region of Applicability 3
II. Program Input/Output Description 4
A. General Description 4
B. Interactive Input Information by Questions. ... 8
C. Optional Output 12
D. Example Storm Hydrograph Output 12
III. Basin Characteristics 16
A. Determination of Basin Characteristics 16
1. Land Cover Measures 16
2. Topographic Characteristics 16
3. Soil Associated Characteristics 17
4. Carbonate Rock Associated Measures 18
5. Other Land-Use Measures 19
6. Sediment Associated Measures 19
7. Convolution Interval (DT) 20
8. Storm Selection Threshold Measures 21
9. Meteorological Measures 22
B. The Watersheds Used to Develop the Regionalized
Relationships 24
C. Range of Values Used in Developing Regionalized
Relationships 24
IV. User Strategies 33
A. The Types of Simulations That May be Obtained . . 33
B. Are the Basin Characteristics "Correct"? 34
C. Validation/Verification 36
D. What if the Regionalized Relationships Are
Not Applicable? 37
E. Diagnostics 39
-------
CONTENTS (continued)
V. Description of TVA-HYSIM Components 40
A. Introduction 40
B. Stochastic Rainfall Generator Component 42
C. Continuous Daily Streamflow Model Component ... 45
1. Interception Storage 45
2. Storm Runoff Volumes (a) Impervious Areas . . 47
3. Storm Runoff Volumes (b) Pervious Areas ... 47
4. Ground Water Runoff Volumes 48
5. Dormant Season Recharge 48
6. Soil A Horizon Moisture Storage Capacity. . . 48
7. Potential Runoff Volume Losses 49
(a) Bypass Seepage
8. Potential Runoff Volume Losses 49
(b) Transmission Losses
9. Potential Runoff Volume Losses 50
(c) Pervious Area Runoff Losses
10. Evapotranspiration 50
11. Runoff Routing 51
12. Regionalized Model Parameter Prediction
Equations 51
D. Storm Hydrograph Model Component 52
1. Precipitation Excess Distribution 52
2. Storm Burst Definition 54
3. The Unit Hydrograph 55
4. Regionalized Unit Hydrograph Prediction
Relationships 55
E. Suspended Sediment Model Component 60
1. Impervious Area Dust and Dirt 60
2. Pervious Area Storm Sediment 61
3. Sediment Routing 62
VI. Computer Requirements 64
A. General Computer Requirements 64
B. RUNOFF Input Data File 67
C. TSO Commands for Interactive Runs 67
D. Job Control Language (JCL) for Batch Runs .... 75
References 78
Appendix
English to Metric Conversion Factors 81
-------
FIGURES
Number Page
1 Interactive Input 5
2 Interactive Output Using Model TVA-HYSIM ... 7
3 Simulated Storm Hydrograph 13
4 Range of Values - Standard Deviation Cube
Root Monthly Rainfall 25
5 Schematic of Watershed Model 41
6 Continuous Daily Streamflow Model Schematic . 46
7 Double Triangle Unit Hydrograph and Lag Time
Definition 56
8 Interactive Procedure Flowchart 65
9 Batch Procedure Flowchart 66
10 CLIST for Interactive Computer Runs 72
11 Job Control Language for Batch Computer
Runs 76
TABLES
Number Page
1 Calibration Watersheds for Model
TVA-HYSIM 26
2 Basin Characteristic Range Used in
Regionalized Relationships 32
3 Daily Rainfall Transitional Probability
Coefficients 42
4 TVA Continuous Daily Streamflow Model
Regionalized Parameter Prediction
Equations 53
Regionalized Equations for Predicting
Coefficients in Equations for TL, UP and T2.
59
-------
TABLES (continued)
Number Page
6 Program RUNOFF Subroutines and
Descriptions 68
7 Program RUNOFF Functions and
Descriptions 71
-------
LIST OF ABBREVIATIONS AND SYMBOLS
Because of the use of FORTRAN nomenclature, the difference between
abbreviations and symbols becomes difficult to define. Therefore, all
abbreviations and symbols are listed alphabetically. The location or
equation in which these variables, parameters, or coefficients are first
encountered are indicated.
Symbol
AA Used in unit hydrograph definition (eq. V-46)
ACCIA Daily rate of dust and dirt accumulation (Section II.B.ll)
AHORD Soil A horizon depth parameter (Section V.C.6)
ARF Accumulated storm rainfall (eq. V-38)
AWC Available water holding capacity of soil (Section III.A.3)
AW Parameter in Continuous Daily Streamflow Model (eq. V-17)
B Parameter in Continuous Daily Streamflow Model (eq. V-17)
BB Used in unit hydrograph definition (eq. V-46)
BFQ Baseflow discharge (eq. V-74)
BGWR Beginning value for groundwater reservoir (eq. V-37)
BHORP Soil B horizon permeability parameter (Section V.C.6)
B NPE Duration of a burst of precipitation excess (Section II.D)
B PE Burst precipitation excess (Section II.D)
B RF Burst rainfall (Section II.D)
B ROI Burst runoff intensity (Section II.D)
BSMI Beginning soil moisture reservoir value (eq. V-36)
C Cover term in Universal Soil Loss Equation (Section II.B.10)
-------
CC Used in unit hydrograph definition (eq. V-46)
CFS Cubic feet per second (Section II.A)
CN SCS curve number based on soils and land-use (Section II.D)
CN-PE Curve number used in determining precipitation excess
distribution (Section II.D)
CO Cutoff used to define storm rainfall burst (eq. V-42)
CONC Concentration (eq. V-74)
CSLOPE Channel slope (eq. III-l)
D Ratio of median sediment grain size to one micron (Section II.D)
DA Drainage area in square miles (eq. III-4)
DD Drainage density (eq. III.2)
AE Increment of elevation along main channel (eq. III-l)
DK Constant in evapotranspiration equation (eq. V-23)
AL Increment of distance along main channel (eq. III-l)
DLF Deep loss measure (Section II.B.8)
DS Parameter in Continuous Daily Streamflow Model (eq. V-17)
DT Convolution interval (Section II.B.12)
DUR Storm duration (eq. V-39)
FOR Percent forest/100, plus one (eq. V-30)
GI Growth index, used in evapotranspiration adjustment (eq. V-23)
GRO Ground water runoff (eq. V-26)
GROK Ground water recession coefficient (eq. V-26)
GROKW Winter GROK coefficient (Section V.C.ll)
GROKS Summer GROK coefficient (Section V.C.ll)
GWDOR Ground water dormant season recharge (Section V.C.5)
GWK Parameter in Continuous Daily Streamflow model (eq. V-19)
GWL Ground water loss due to bypass seepage (eq. V-20)
X
-------
GWR Ground water reservoir volume (eq. V-17)
GWV Ground water volume to be allocated (eq. V-19)
H Maximum difference in watershed elevation (eq. III-4)
HTFLUX A measure used to increase evaporation of intercepted
moisture (Section II.B.13)
I Interception (eqs. V-13)
IN Runoff in watershed inches (eq. IV-1)
ITPS Factor used to delay peak sediment concentration (Section II.D)
IUSG Instantaneous unit sediment graph (eq. V-72)
IUSGI Impervious area unit sediment graph (eq. V-73)
JD Julian day (eq. V-5)
K Soil erodability term in Universal Soil Loss Equation
(Section II.B.10)
L Main channel length (eq. III-7)
LAT Latitude (eq. III-9)
LC Length in miles across the watershed of the contour representing
25, 50, or 75 percent of the watershed height (eq. III-4)
LENGTH Watershed length term (eq. III-5)
LI Length of grid lines used in calculating DD (eq. III-2)
LOSS Long-term average annual evapotranspiration (eq. V-23)
LS Length-slope term in Universal Soil Loss Equation (Section
II.B.10)
LT Literature values (Table 2)
LULL Interval between bursts during storm rainfall (eq. V-43)
M Constant in dust and dirt washoff equation (eq. V-68)
MAXRF Maximum rainfall threshold for defining storms (Section II.B.15)
MINRF Minimum rainfall threshold for defining storms (Section II.B.15)
MINRO Minimum runoff threshold for defining storms (Section II.B.15)
xi
-------
N Number of grid intersections (eq. III-2)
P Conservation practice term in Universal Soil Loss Equation
(Section II.B.10)
PCM A measure of percent mined (Table 5)
PD Transitional probability for dry (eq. V-3)
PD/D Transitional probability of dry given dry (eq. V-3)
PD/W Transitonal probability of dry given wet (eq. V-2)
PE Precipitation excess (Figure 3)
PE. Precipitation excess from impervious areas (eq. V-16)
PE Precipitation excess from pervious areas (eq. V-18)
PEIN Normalized precipitation excess intensity (eq. V-47)
PET Long-term monthly potential evapotranspiration (eq. V-23)
PERM Soil permeability (Table 5)
PHI Constant loss term used in determining PE (Section II.D)
PKARST Parameter affecting storm runoff in karst terrain (Section II.B.8)
PR Transitional probability (generic) (eq. V-l)
PPM Suspended sediment concentration in parts per million (Section
II.A)
PPM1 Suspended sediment concentration at one cfs/mi2 (eq. V-74)
PW/W Transitional probability of wet given wet (eq. V-2)
QP Storm hydrograph peak discharge (eq. V-71)
R Dust and dirt daily removal rate (Section II.B.ll)
ra Coefficient in transitional probability equation (eq. V-l)
rb Coefficient in transitional probability equation (eq. V-l)
RF Rainfall, used to refer to monthly or storm rainfall (eq. III-9)
RF Mean monthly rainfall (eq. 111-10)
RFD Daily rainfall (eq. V-4)
XII
-------
RF5 Five-minute rainfall (eq. V-10)
RFH Hourly rainfall (eq. V-7)
RF. Available moisture after interception and precipitation excess
are removed (eq. V-19)
RF Daily rainfall less interception (eq. V-16)
RF2 Rainfall for storm greater than one inch (eq. V-12)
RI Retention index (eq. V-17)
RO Storm runoff (precipitation excess) (eq. III-9)
S Maximum potential retention (eq. V-38)
SB Coefficient in time lag-PEIN equation (eq. V-48)
SC Coefficient in time lag-UP equation (eq. V-50)
SD Standard deviation (eq. III-ll)
SE Coefficient in time lag-UP equation (eq. V-50)
SEDDAY Accumulated dust and dirt (eq. V-66)
SF Coefficient in time lag-T2 equation (eq. V-51)
SG Coefficient in time lag-T2 equation (eq. V-51)
SHAPE Watershed shape (Table 5)
SI Seasonal index (eq. V-17)
SINU Sinuosity (eq. V-39)
SLOPE Watershed slope in percent (eq. III-4)
SMR Soil moisture reservoir (eq. V-17)
SRO Storm runoff (eq. V-25)
SROK Storm runoff routing constant (eq. V-25)
SURES Storm runoff reservoir (eq. V-25)
S|J Coefficient in time lag-PEIN equation (eq. V-48)
Tl Time to initial unit hydrograph peak (Figure 7)
T2 Time to inflection point on unit hydrograph (Figure 7)
xiii
-------
T3 Time base of unit hydrograph (Figure 7)
tl Average duration of one-inch storm (eq. V-ll)
t2 Average duration of storm larger than one-inch (eq. V-12)
TC Time of concentration (eq. III-7)
TDSRO Parameter indicating runoff that occurs on day of rain
(eq. V-25)
T50 Time to half the area under unit hydrograph (Figure 7)
TL Time lag (Section II.D)
TIP Transmission loss parameter (eq. V-21)
TONS Suspended sediment load in tons (eq. V-71)
UGMOD Coefficients which can be used to modify the TL-PEIN relation
(Section IV.D.2)
UP Peak of the unit hydrograph (Figure 7)
UR Inflection point on unit hydrograph (Figure 7)
USLE Universal Soil Loss Equation (Section III.A.6)
USLEP Product of 95 and terms in the USLE equation (Section II.D)
USLEXP Exponent in Modified USLE equation (Section II.D))
WASH Dust and dirt washed off during storm (eq. V-68)
WB Term relating QP, the maximum PE. and the time to peak
(eq. V-72) J
Y Long term annual runoff/rainfall (eq. V-27)
YP Random number from uniform probability distribution (eq. V-4)
(j micron
XIV
-------
SECTION I
INTRODUCTION
A. OVERVIEW
The Tennessee Valley Authority Hydrologic Simulation Model (TVA-
HYSIM) is a computer package containing complex models specifically designed
for ease of application in land-use planning studies. TVA-HYSIM contains
models capable of continuous hydrologic simulation as well as a rainfall
generator and an erosion component. This package is formulated to provide
the level of output normally required for land-use planning studies and yet
be very simple to operate (no learning period). The approach is to use an
"interactive" input technique, wherein the computer prompts the user for the
information.
An earlier package has many additional options including programs
for determining optimized model parameters by adjusting the models to
observed data. (This more complex model version is also available from the
authors.) Because of the many options, however, a learning period is
required before this earlier program can be used with ease.
This user's guide briefly describes the models in the program
package TVA-HYSIM, outlines the information required to operate the programs
and how this information is determined, and shows examples of the input/
output. Also, some strategies for using the models are offered. A later
section describes the model components in sufficient depth so that changes
to the program may be made by the user, if necessary. And finally, examples
of the job controls needed to operate the computer package are provided.
English units have been used deliberately throughout this report.
Because model algorithms, particularly the regionalized parameter prediction
equations, were developed based on measures in English units, it would be
virtually impossible to incorporate metric units into the report or into the
computer program. A table of English to metric conversions is included as
an Appendix for convenience.
The information provided in this user's guide may appear to be
excessive for a model designed for ease of application. Certainly not all
of this information is needed to operate the model package. The extended
documentation included in this guide goes well beyond the information needed
to simply operate the model. This additional information is provided to
simplify other aspects of model application including: job controls, deter-
mining basin characteristics, and modifying model components, if necessary.
-------
B. MODEL DESCRIPTION
TVA-HYSIM provides users a relatively easily applied tool for
quantifying the effects of land-use change on the hydrologic balance, yet
the model package is capable of handling the complex geologic-hydrologic
system. Briefly, TVA-HYSIM consists of several linked components. A rain-
fall generator drives the package and supplies rainfall to a continuous
daily streamflow model. Hourly or shorter duration rainfall is supplied by
the rainfall generator to the storm hydrograph model for those storms
selected for simulation. Runoff volumes are calculated in the daily stream-
flow model. Storm suspended sediment loads and concentrations may also be
simulated in conjunction with the storm hydrographs.
TVA-HYSIM contains "regionalized" model components. Regionalizing
a model involves adjusting the model to data at locations where adequate
data are available to obtain optimized values for each model parameter. In a
second step, each optimized parameter is related to some combination of
basin characteristics in the upstream watershed and/or climatological
measures. Once these relationships are developed, the model may be applied
at ungaged locations since the model parameters can be predicted from basin
characteristics and appropriate climatological measures.
The advantage of regionalization is obviousthe model may be
applied at locations where gaged data are unavailable. A regionalized model
cannot, however, be applied blindly to all situations. For instance: (1)
the model must be used with caution at locations outside of or under condi-
tions differing from those used in regionalizing the model; (2) judgment
must be exercised to ensure that reasonable results are obtained since
validation data are unavailable and; (3) if measures of basin character-
istics are mis-scaled because a different computation technique is used,
because an error was made in measurement; or because a bad estimate of a
characteristic is made in a case where needed information (such as a soils
map) is lacking, simulation results may be unrepresentative but the effect
probably will not be obvious. It is hoped that this user's guide will
contain enough information about TVA-HYSIM so that these caveats do not
become problems.
The basic set of models from which TVA-HYSIM was formulated has
been under development for about 15 years. This development began with a
project in the rural Upper Bear Creek watershed located in northwest Alabama
(TVA, 1973a). Urban capabilities and algorithms to handle the effects of
carbonate rock terrain on hydrology were developed at a project located in
Knoxville, Tennessee (Betson, 1976). The model has recently been adapted to
handle surface mining in an ongoing project (Betson, 1979b, Bales, 1979,
Barr, 1979). Section V in this report describes in some detail the newer
features of the model components not documented elsewhere. (An EPA Report
to be issued in late 1981 will document model development, calibration, and
validation in greater detail.)
TVA-HYSIM may be used for a variety of land-use evaluations. There
are provisions for differentiating among hardwood and conifer forest types,
several agricultural cover classes, and unvegetated conditions. Impervious
-------
areas (urban conditions) can be considered as can the effects of surface
mining. The model has provisions for handling the effects of the under-
drainage that influences storm runoff in areas of soluble carbonate rock
(Betson, 1977). And finally, suspended sediment loads can be considered as
washoff of dust and dirt from impervious surfaces and as erosion from
pervious areas.
C. REGION OF APPLICABILITY
The regionalized relationships incorporated in TVA-HYSIM were
derived using hydrologic data collected in the Tennessee Valley and sur-
rounding area. This region includes six physiographic provinces: Blue
Ridge, Valley and Ridge, Cumberland Plateau, Highland Rim, Central Basin,
and Mississippi Embayment (Fenneman, 1938). These provinces include a
considerable range of physical watershed, soil, topographic and meteor-
ological conditions. The model has been calibrated using data from water-
sheds in this region receiving an average of from 38 to 77 inches of rain-
fall per year.
The region where this model has been calibrated is roughly bounded
between latitudes 34°N and 37°N and between longitudes 82°W and 89°W. To
the extent that conditions are similar to those within this region, the
model will apply elsewhere. Its applicability, however, should be verified.
In more northern climates where snow and snow melt runoff are significant
considerations, changes to model components will become necessary to ade-
quately model the snow-snow melt phenomenon continuously. Approaches are
suggested in Section IV. The model is not adapted to arid-climate
hydrology. A list of watersheds used in calibrating the models, which was
part of the regionalization process, is included in Section III of this
report to provide an idea of the range of conditions that was included.
-------
SECTION II
PROGRAM INPUT/OUTPUT DESCRIPTION
A. GENERAL DESCRIPTION
The input information required to operate TVA-HYSIM is input to
the program interactively. Experience has shown that the typical input
formatting which utilizes poorly defined FORTRAN "type" variables requires a
learning period before the procedure can be used with confidence, and even
then this approach is error-prone. The interactive free-form mode used in
TVA-HYSIM was designed to be used at a time-sharing terminal so that the
user simply responds to data prompts, or requests, made by the computer.
This system is very simple to use since the computer essentially defines the
format for the user. The same data input sequence is also used for optional
batch runs that may be made with TVA-HYSIM although, of course, there is no
direct interaction between computer and user per se.
Figure 1 shows the interactive input required to make a three-year
simulation of continuous daily streamflow and selected large storms. The
watershed used for this simulation, Cane Branch near Parkers Lake, Kentucky,
was the mined watershed (10.5%) in the strip mine effects study reported by
Musser et al, (1970). (Although there are actually no impervious areas in
the watershed, a two percent impervious area was added to the land cover so
that all input questions and all possible options could be shown in this
example.) The set-up shown in Figure 1 resulted in the three-year sequence
of simulation of hydrologic information for which the selected information
is summarized in Figure 2. Shown in Figure 2 are simulations for user
selected large storms in each year which include the rainfall and runoff in
inches, peak discharge in cubic feet per second (cfs), suspended sediment
load in tons, and average and maximum suspended sediment concentration in
parts per million (ppm). The annual simulated rainfall, runoff, and minimum
one-day discharge are shown for each year, as is the total sediment load for
those storms simulated. At the end of the output the average annual rain-
fall, runoff (in inches), and suspended sediment load (in tons) for the
entire simulation period are shown.
Although most of the questions shown in Figure 1 are straight-
forward, each will be explained in sequence, and the information or charac-
teristics requested will be defined. Section III explains in more detail
how these characteristics are measured and where the necessary information
is obtained.
This example run was set up so that all possible questions would
appear. In most runs, however, not all of the questions will necessarily
-------
FIGURE 1
Interactive Input
Input Data Values As Called For In Each Line With A Comma Or Blank Between Each Value.
I. What is the name & location of the watershed?
Cane Branch Hr. Parkers Lake, KY.
2. Do you want a batch printout stored? (Yes or No)
Yes
3. Is sediment to be simulated? (Yes or No)
Yes
4. What is the drainage area & land cover percentages:
Hardwood, conifer, pasture, small arain, row crop, impervious, unvegetated? (8 Values)
?
0.67 71 14 2.5 0 0 2 10.5
5. Are special directly connected impervious areas to be considered? (Yes or No)
Yes
6. What is the percent? (1 Value)
?
1
7. What are watershed characteristics values: Slope, shape, drainage density,
curve no., sinuosity, % mined, soil perm., available water holding capacity? (8 Values)
?
194 2.04 11.4 50 0.14 10.5 2.98 5.28
8. What are the values for bypass seepage (DLF) & karst areas (PKARST)? (2 Values)
7
0 0
9. What is the average sediment concentration at 1 CFS/SM in PPM? (1 Value)
7
10
10. What are the USLE factors: Soil-erodibility (K), slope length & gradient (LS),
cropping management (C), & erosion control practice (P)? (4 Values)
?
0.3 5.0 0.0126 1.0
II. What are the sediment associated coefficients: Grain size ratio (D), daily impervious
area removal rate (R), & daily accumulation rate (ACCIA)? (3 Values)
?
1 0.08 1.0
12. What is the convolution time interval? (Hours & minutes with a period between) (1 Value)
?
i.o
-------
FIGURE I.(CONTINUED)
13. What are the long-term or expected mean rainfall, and runoff or average latitude, &
htflux? (4 Values)
?
47.2 0 36.9 3.0
14. Number of years to be simulated and beginning year (2 Values)
?
3 01
15. What lower limits are to be used in defining a storm to be simulated? (MINRF & MINRO,
& MAXRF) (3 Values)
7
1.5 0.2 2.5
16. What are the 12 values for the mean of the cube roots observed monthly rainfall
in water year sequence*? (12 Values)
?
1.315 1.49 1.54 1.57 1.545 1.70 1.57 1.59 1.61 1.62 1.45 1.46
17. What are the 12 values for the standard deviation of observed monthly rainfall in water
year sequence? (12 Values)
?
0.38 0.27 0.28 0.28 0.28 0.28 0.27 0.25 0.26 0.30 0.30 0.31
18. What are the 12 values for the mean monthly potential evapo-transpiration in water
year sequence? (12 Values)
?
2.72 1.44 0.88 0.95 1.44 2.89 4.24 5.34 5.86 5.97 5.29 4.22
IHN002I STOP 99
19. Are you ready to run the program? (Yes or No)
Yes
-------
FIGURE 2
Interactive Output Using TVA-HYSIM
Cane Branch Nr. Parkers Lake, KY.
Storm Data RF(IN) RO(IN) MAX Q(CFS) SS(TONS) AVG PPM MAX PPM
I/ 8/ 1 2.11 0.7478 18.92 102.70 2829. 8328.
2/ ]/ 1 1.85 0.6950 14.28 82.04 2431. 3602.
6/ 4/ 1 3.19 0.5226 19.36 83.19 3279. 5486.
Annual RF = 48.77 IN. & annual RO = 16.001 IN for year 19 1
Minimum one-day flow for 19 1 -- 0.0 CFS
$ The total sediment for the above storms is 267.93 tons
Storm Data RF(IN) RO(IN) MAX Q(CFS) SS(TONS) AVG PPM MAX PPM
12/23/ 1
1/24/ 2
2/24/ 2
3/1 9/ 2
5/30/ 2
1.72
2.71
1.53
1.64
4.95
0.3396
1.3697
0.7424
0.8720
1.9841
6.50
32.72
94.86
19.08
92.66
34.47
194.34
253.33
109.70
434.71
2091.
2922.
7029.
2591.
4513.
5073.
3816.
11749.
3889.
7956.
Annual RF = 48.49 IN. & annual RO = 22.031 IN for year 19 2
Minimum one-day flow for 19 2 -- 0.01 CFS
$ The total sediment for the above storms is 1026.55 tons
Storm Data RF(IN) RO(IN) MAX Q(CFS) SS(TONS) AVG PPM MAX PPM
I/ 5/ 3
1/23/ 3
11 4/ 3
7/15/ 3
1.55
1.77
3.24
2.80
0.5164
0.9329
0.1531
0.3740
9.34
29.22
5.29
16.10
53.24
146.46
17.81
60.16
2123.
3233.
2397.
3314.
4418.
4765.
584?.
5780.
Annual RF = 49.81 IN. & annual RO = 17.664 IN for year 19 3
Minimum one-day flow for 19 3 -- 0.00 CFS
$ The total sediment for the above storms is 277.68 tons
The following are the simulated averages for 3 years
RF = 49.02 RO = 18.59 SED = 524.05
-------
appear. Some questions are contingent on the response to a previous
question or a measure previously entered.
The first item at the top of Figure 1 involves initiating the
CLIST (see Section VI) which begins the program. HYSIM is the name of the
CLIST.
B. INTERACTIVE INPUT INFORMATION BY QUESTIONS
1. What is the name and location of the watershed? A descriptor not
to exceed 80 characters.
2. Do you want a batch printout stored? (Yes or no). A "yes" causes
detailed storm by storm simulated hydrographs and continuous daily
streamflow model simulation information to be stored (at increased
run-time) for subsequent printout. The storm hydrograph informa-
tion may be obtained at the end of the interactive printout (see
last paragraph this sub-section). Information from the continuous
daily streamflow model and/or the storm hydrograph model may be
obtained in either a separate job step when the program is run
from a terminal or using the optional batch run (see Section
II.C).
3. Is sediment to be simulated? (Yes or no) If "no", questions 9,
10, and 11 will be deleted.
4. What is the drainage area and land cover percentages: hardwood,
conifer, pasture, small grain, row crop, impervious, and unvege-
tated? (8 values) The drainage area contributing to streamflow
(surface and/or ground water) is in square miles and the land
cover values are in percent (100% must be accounted for).
5. Are special directly connected impervious areas to be considered?
(Yes or no) This question is contingent on entering a non-zero
value for impervious areas in question 4. A "yes" results in
question 6.
6. What is the percent? (1 value) Special directly connected areas
are impervious areas that drain directly into a stream. This
feature should not be used when the impervious area is much
greater than about 17 percent.
7. What are the watershed characteristics for slope, shape, drain-
age density, curve no., sinuousity, % mined, soil perm., avail-
able water holding capacity? (8 values) These are watershed
characteristics typically obtained from topographic and soils
maps.
a) Slope A weighted measure of main-channel slope in feet per
mile.
-------
b) Shape Dimensionless measure defined as the squared length of
the main channel in one-mile chords (with fractional value
included) divided by the drainage area.
c) Drainage Density As measured on a 1/24,000 scale topographic-
map (7.5 minute quadrangle map) in miles/mi2.
d) Curve No. The Soil Conservation Service curve number averaged
Across the watershed (SCS 1972, 1975).
e) Sinuosity The ratio of the length of the main channel to the
channel length measured in one-mile chords, minus one.
f) % Mined A measure of the percent of the watershed that is
mined, where this mining has a substantial affect on the
timing of the storm hydrograph. See Section III for informa-
tion on the range of values used in deriving the regionized
relationships.
g. Soil Perm. Soil permeability averaged across the watershed in
inches per hour.
h. Available Water Holding Capacity As determined from published
soils information and averaged across the watershed in inches.
What are the values for bypass seepage (DLF) and karst areajs
(PKARST)? (2 values) These are measures typically used only in
soluble carbonate rock areas where drainage within the rock system
begins to dominate the streamflow. See Section III for hints on
how these measures may be estimated.
a. Bypass Seepage (DLF) This is a ratio index of the amount of
potential ground water runoff expected to bypass the location
on a stream where simulations are being made (DLF=1.0, no
ground water runoff occurs; DLF=0, no potential ground water
runoff is lost to deep seepage).
b. Karst Areas (PKARST) This is a ratio index of the amount of
potential storm runoff from pervious areas expected to enter
the soil carbonate rock system, except during very large
storms when the capacity of the soil to hold this additional
moisture may be exceeded. (PKARST=1.0, storm runoff occurs
only during very large storms; PKARST=0, all potential storm
runoff occurs).
What is the average sediment concentration at 1 cfs/sm in PPM?
(1 value) This is an estimate of the suspended sediment concen-
tration in parts per million that would occur at a discharge of
one cubic foot per second per square mile and is used with base-
flow in the model to provide a base concentration for sediment
simulations when storm hydrographs are printed. Any reasonable
value will suffice as the value does not appreciably affect the
-------
sediment load predictions (i.e., 10 ppm). This question appears
only when question 3 is answered "yes".
10. What are the USLE factors: soil credibility (K), slope length
and gradient (LS), cropping management (C) , and erosion control
practice (P)? (4 values) These are measures in the Universal
Soil Loss Equation averaged across the watershed. This question
appears only when the answer to question 3 is "yes". See Section
III.A.6 for a detailed description of these parameters.
11. What are the sediment associated coefficients: grain size ratio,
daily impervious area removal rate (R), and daily accumulation
rate (ACCIA)? (3 values) These terms are associated with the
suspended sediment simulations. This question appears only when
the answer to question 3 is "yes." See Sections III.A.6 and V.E
for details on how these terms are determined.
a. Grain Size Ratio This is the ratio of the expected median
suspended sediment grain size to one micron.
b. Daily Impervious Area Removal Rate (R) This is a measure of
the fraction of dust and dirt that accumulates on impervious
surfaces during days of no rain, which is then removed due to
the effects of wind and traffic.
c. Accumulation Rate (ACCIA) ACCIA is the daily accumulation
rate of dust and dirt on impervious surfaces in pounds per
acre. Normally this and the companion removal rate R are used
only for urban situations where the impervious areas are more
extensive. Values of zero may be used where the percent of
impervious area is relatively small.
12. What is the convolution time interval? (Hours and minutes with a
period between) (1 value) Convolution is the process of multiply-
ing a distribution of precipitation excess by a unit hydrograph to
obtain streamflow. The convolution interval should be about 1/4
of the time to peak of the hydrograph. See Haan and Barfield
(1978) or any hydrology text for a discussion of the unit hydro-
graph. The convolution interval, DT, must be a five-minute
increment evenly divisible into an hour (i.e., 5, 10, 15, 20, and
30 minutes) or an integer multiple of an hour. DT is expressed in
hours and minutes with a decimal between (i.e., for 20 minutes DT
= 0.20). DT is converted to hours and fractions of an hour
internally in the program.
13. What are the long-term or expected mean rainfall, runoff, or
average latitude, and HTFLUX? (4 values).
a. Mean Rainfall The estimated long-term average annual water-
shed rainfall in inches.
10
-------
b. Mean Runoff The estimated expected annual runoff in inches.
Where the latitude relationship applies (see next measure) a
value of zero may be used. At other locations, the mean
annual runoff is either estimated from records from nearby
applicable streamgages or by subracting estimated long-term
evapo-transpirative losses from average rainfall.
c. Average Latitude The average latitude in degrees for the
watershed is used to estimate long-term average annual evapo-
transpiration in inches (loss) in the model with a relation-
ship (loss=65.5 - latitude) generally applicable at locations
noted in Section I.C. At locations where this relationship is
not applicable, a value of zero should be used for this
measure and the expected long-term runoff estimated.
d. HTFLUX This is a ratio measure to allow intercepted rainfall
to evaporate at a rate faster than the normal rate of trans-
piration (see Betson, 1979a) . A recommended value of 3.0 will
allow intercepted water to evaporate at three times the rate
of transpiration. (This feature is more important in forested
areas.)
14. Number of years to be simulated and beginning year (2 values)
The choices are arbitrary. For simulations that involve flood
flow frequency analyses, 25 years are usually run.
15. What lower limits are to be used in defining a storm to be simu-
lated? (MINRF and MINRO arid MAXRF) ~~(3 values) These a^re
controls which will, if desired, limit the number of storm hydro-
graphs that will be simulated. Any one day or two day rainfall
(with corresponding runoff) greater than these values will cause a
storm hydrograph to be simulated. In long-term simulations, the
goal is to set lower limits so that only the three to four storm
hydrographs with the largest peaks are simulated per year. This
decreases run time and associated costs as well as limiting the
output to the desired information.
a. Minimum Rainfall (MINRF) Each storm with rainfall equal to or
exceeding MINRF and simulated runoff equal to or greater than
MINRO will have a storm hydrograph simulated.
b. Minimum Runoff (MIWRO) This minimum value of storm runoff is
used to screen low runoff storms when the purpose of the run
is peak flow analysis.
c. Maximum Rainfall (MAXRF) All storms with rainfall equal to or
exceeding this threshold will have a storm hydrograph simu-
lated regardless of the storm runoff volume.
16. What are the 12 values for the mean of the cube roots of observed
monthly rainfall in water year sequence? (12 values) The cube
root of monthly rainfall at a representative raingage is summed by
11
-------
months for a recommended 25-year record and the monthly average
determined. (October is the first month in the water year.)
17. What are the 12 values for the standard deviation of observed
monthly rainfall in water year sequence? (12 values) The
standard deviations of the array of cube root of monthly rainfall
values determined for question 16 are supplied here.
18. What are the 12 values for mean monthly potential evapo-
transpiration in water year sequence? (12 values) Mean evapora-
tion values from a published land pan or as computed using
standard equations that are based on meterological data are
supplied here.
19. Are you ready to run the program? (yes or no) A "yes" begins the
program. A "no" will cause an exit from the TVA-HYSIM program.
This question allows the user to exit and begin again if input
errors are detected.
If the answer to question 2 was "yes," another question will be asked fol-
lowing printout of the storm summaries (see Figure 2) as to whether a list-
ing of the storm hydrographs is desired. A "yes" to this question will
begin the sequential printing of all storm hydrographs.
C. OPTIONAL OUTPUT
TVA-HYSIM has several output options. The interactive terminal
output which contains much of the essential information that is simulated is
shown in Figure 2. If the response to question 2 was "yes," much more of
the simulated information is stored. The next sub-section describes the
detailed storm hydrograph printout that may be obtained at the terminal.
These hydrographs are stored on a disk file and may also be printed with a
user-supplied utility program that will print a normal file. Similarly,
water year simulations of daily rainfall and streamflow are stored on the
disk file but this information may be obtained only with a utility program.
Section VI explains how this is accomplished in more detail.
A batch run of the program is another option. (Batch runs are
made by reading data into the computer by cards and receiving output at a
line printer, as opposed to an interactive run in which data is keyed direct-
ly into the computer via a teletype terminal.) The data set-up for batch
runs is similar to that used in the interactive runs; but, of course, all of
the printout may be obtained in a single step. Section VI provides an
example of a job control set-up for a batch run.
D. EXAMPLE STORM HYDROGRAPH OUTPUT
Figure 2 shows an example output containing simulated storm values
for the short-form printout option. Figure 3 shows an example storm hydro-
graph simulation that may be obtained using either the interactive approach
12
-------
in
a
UJ O
a. o
z
r-
0*
r*.
*
t-~
u
Q
»
u<
^
^
z
Q
b^
CO
a:
UJ
*
_j
taJ
O
I
X
a:
ta
o
a:
o
>~
x
X ~
a:
O -V
CO
O *
oJ
t
< U.
-J O
X X
to o
t
°°
t
1
a:
O
X
ID
UJ >-
-1 ic
.-I
> UJ
UJ *
Jj J
to
SO CO
uj or
Z UJ
z -x
i^l j;
^- <
X
Jat
^
2
x
CJ
at
;Q
Al
o
z o o uj
CO) 0 X '0 '
X a: o u. AJ r»-
O < a: » 0
0 * . j*
z x a - z
< u »
09 -*
Z 0 (*» O *M 0
o ru o o o o
or &
09
J) 0 « X 0
co o o O'O si o x*
c_> o o o o-
£ in
*
00 f^ ,O
0 0 0 0 -«
« * * o
X O* O'O 0 o -1 "-
ac * *" *
3 O ^*'JTr
in -» o &
3 -M 3
0 O 0 * 0 C»
» 0 «
* *
UJ O *
> Jl O O O -* 4)
O: OH) r- 9 vD
X *OO X30 3 9
^» * a:* no
>< i-i"3O J C> C> O^ Di»
* O
»~ 0 0
UJtn Hf\lO~ «9fl
x «> aj o * o o
LH 3;OO *O^ O "-
tn o
>< _j *^ in
o * ex x -
-* o o
t» «
*
o
3t Ml"\ Oo COST
O~* ** » i « Kl K- fl O
ui in rv o
-J QC or
< O O C» -^ (oJ Jj
^9 5TM OC> *- X
UJO 00 00 (J »
(OX* * ^ r^ C>
cj- x < «-« <0
CO X CT
(O cj fti to m «
- 00 -"f^l t-f^JCt-
oc AJ in ° * " uj
»- UJ O * o CO UJ
LJX- oo oo a: X
« O 7 CO UJ <
a: _J > _j w » x
< en ~* _j uj uj o <
r
Q O r»- t i * ^ --* ^- ^-« Kt
UJUJ*-O - ;>
_J or i- u. _j
3 ui 3 o z a: .u
«-*< z < r> «-
co * *« a: x *-
O Uf
O X 0
u. o *>-
a: i^ » cr-
9
00 0 2T -
(J £
O*
9
o o- -.
a; o >-» i/» ^*
* X 0 CO
no x z
o o
*-*
o
in so
9 CO
t- " O (0
:> o
1C
x m
CO -0
rj in
~t O
tr o
UJ
X
X
j\ «* O
fU sO ««. O
O i>| i O
a: * or o cj o
3O 3 CJ
0 *«
a: o
(V 35 «
r^ r« o
X » X o o
oo :D
o
o
!- O
t- 0
* o
=» a: o
o »
to m o -*
CJ O «t
>-« Kl I ^ op
t ( O <*«!
CO M O
a: ac x
ij UJ »-
U i
< 0 0
r cr
X
UJ < DO
Ul O <
X ox
< 0
ae o »
z :D r-
_j
r ui
x t-
O O 9
p^ Z O O O
* * *
x
H>
uj a:
*
< a;
o
z X
0 CL
to =!
z o:
*J O
o
. cc
-i ~ °
u ^
o X
0
2
i or
^ o
^
3t O
O |jj
^ ^
2 i s
>^ CO
o --
jj ro
-t u, LU
^ o or
XX ^
a: O
to o
*- XX U.
t « * *
1
I
1
(t O
o +
X 0
*- t> oc
Z> oo Q
* X
OK O O O
UJ >- 000
_f
< o o o
> UJ ooo
^ * + *
UJ < O O O O
UJ _J UJ ^O «* "-
CO » ^> CO O
Z UJ O T-> "~t f\l
Z 1C LU «
uJ a: a: 3 o o
>- < Q.
^
uJ O
Z Ul
T>
x or
O uj
Z (O O O O
4 $ * *
ac. o o o o
D
t->
<4J oj -* 'U "O
Z X
*
(J t-
13
-------
c ac x ac
K a.
ac.
x. ac ae ae or
a. x
0000000000000000000000000000000000000000000000003000000000
UJ
oooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
r m « - so O- 3 -»
14
-------
or a batch run. Shown on each storm hydrograph simulated are the simulated
storm date along with the random number used to generate the storm rainfall
(see Section V.B). The word "simulated" at the left margin just under the
title will always appear to remind the user that the storm rainfall and
runoff values have been simulated rather than observed. A list of the input
watershed characteristics that are used in the storm hydrograph component is
provided. Next, the total storm rainfall and its distribution (in convolu-
tion time intervals-DT) are tabulated followed by a tabulation of the pre-
cipitation excess and its distribution.
For each rainfall burst (see Section V.D.2 for burst definition),
predicted hourly parameters of the unit hydrograph are shown. On the same
line characteristics of the storm burst are given and include: time lag
(TL), weighted precipitation excess intensity (B ROI), rainfall (B RF),
precipitation excess (B PE) , and precipitation excess duration (B NPE).
(See Section V.D. for further definitions.) Occasionally, as in this
example, the sum of the burst rainfall volumes does not equal the total
rainfall. This is due to internal round-off in the computer.
The next line presents the unit hydrograph parameters in integer
DT units along with the loss term PHI, the SCS curve number used in determin-
ing the precipitation excess distribution, and DT. This curve number,
CN-PE, is based on a relationship among storm rainfall, runoff, and PHI (see
eq. V-38) and differs from the watershed average curve number, CN, that was
input. DT is printed out in hours and fractions of an hour, as opposed to
the way it is read in.
If the sediment simulation option is exercised, the storm hydro-
graph output will contain information relevant to this option including: a
computed product of constants in the Universal Soil Loss Equation multiplied
by 95, USLEP; the exponent on the runoff energy term in the modified
Universal Soil Loss Equation (0.56), USLEXP; a term to delay the peak sedi-
ment concentration, ITPS; the grain size ratio, D; the impervious area
washoff term, R; the dust and dirt accumulation rate, ACCIA; and the simu-
lated storm sediment load from impervious and pervious areas (in tons).
The last segment of the storm hydrograph output is a print-plot
which has a column for observed flows, predicted flows, and the difference
between the observed and predicted flows, all in cfs per DT unit. (This is
a general print-plot routine used in other programs--in TVA-HYSIM the
observed and the error columns will always be zeros.) These predicted
runoff values (usually beginning and ending with baseflow) are plotted at a
scale shown at the top of the print-plot such that the maximum predicted
value is full-scale. Also shown on the right margin of the print-plot are
the rainfall (less interception), the incremental suspended sediment load in
tons and the suspended sediment concentration. The total sediment load is
printed at the bottom of the print-plot.
15
-------
SECTION III
BASIN CHARACTERISTICS
A. DETERMINATION OF BASIN CHARACTERISTICS
Because TVA-HYSIM contains regionalized models, the basin charac-
teristics must be determined in the same manner as were those character-
istics that were used in developing the regionalized relationships. This
section describes how each of the characteristics is measured.
1. Land Cover Measures
These measures, listed in question 4, Section II, can best be
obtained by direct observation or from aerial photographs. For larger
watersheds, the overall percent forest may be estimated from current topo-
graphic maps which depict forest areas, and an estimate of the remaining
agricultural cover distribution estimated from county information in a
current USDA Census of Agriculture (assuming the distribution in the water-
shed corresponds with that in the county).
2. Topographic Characteristics
These characteristics are determined using a 7-1/2 minute quad-
rangle map, 1/24,000 scale.
a. Drainage Area If the drainage area is not published, the topo-
graphic divide of the watershed is depicted on the map and the
area obtained using a planimeter or by counting intersections
within the divide on a grid overlay. Known non-contributing areas
such as sinkholes should not be included. (See sub-section A.4.a
of this section. )
b. Channel Slope is a weighted measure calculated from the point of
simulation to the watershed divide along the main channel.
Channel slope is calculated in feet per mile. The number of miles
between each contour crossing of the main channel upstream of the
simulation point is tabulated (AL). The associated difference in
elevation (AE) is then tabulated in feet (AE should usually be
constant). The weighted channel slope can then be calculated as
CSLOPE = (IAL/I(AL/VAE/AL))2 III-l
16
-------
c. Sinuosity This is an index measure of the sinuosity of the main
channel. The measure is defined as the ratio of the actual length
of the main channel to the main channel length as measured in
one-mile chords (with fractional value included), minus 1.0.
d. Drainage Density A grid intersection method is used to compute
drainage density. The grid size in map scale miles should be
approximately equal to the square root of (drainage area in square
miles/75). This should give about 100 grid intersection points,
regardless of the drainage area size. First, all channels and/or
courses that might carry water are delineated on a 1:24,000 scale
contour map and extended as indicated by contour crenulations ("V"
shaped cusps). Next, the total number (ZN) of intersections of
the channel system with the grids are tallied. Then the total
length (ZL1) of all grid lines within the watershed (in miles) is
determined. Drainage density is calculated as
DD=1.571 IN/ILI III-2
e. Shape This is a dimensionless measure of the shape of the basin.
It is defined as the squared length of the main channel measured
in one-mile chords (with the fractional value included) divided by
the drainage area in square miles.
3. Soil Associated Characteristics
Soils information is essential for operating TVA-HYSIM. An SCS
county soil survey is available for many counties. Where unavailable, a
soils association map is often available. Usually, the local Soil Conserva-
tion Service county representative can provide the necessary soils informa-
tion. The calculation of these characteristics begins with a determination
of the percent of the watershed in each soil type. This information is best
obtained from soil maps using a grid-intersection counting method. Informa-
tion on the hydraulic properties of the various soil types are contained in
the more recent County Soil Survey reports, or may be obtained from standard
soils series descriptions which can usually be obtained from a county SCS
representative (also, see Section IV.B).
a. Soil Permeability Permeability in inches per hour is usually
expressed in the soils series descriptions as a range of values.
A separate range of values may be given for each depth horizon.
The average of the range of permeability values for each depth
horizon is used. These permeabilities are weighted by the incre-
mental horizon depth to determine an average permeability value
for each soil type. Only the depths of each soil type provided in
the descriptions, usually to about six feet maximum, are used.
Where hard-pans or other confining layers occur, only values above
this horizon are considered. A watershed average value is deter-
mined by weighting the permeability of each soil by the portion of
the watershed it occupies.
17
-------
Available Water Holding Capacity (AWC) Water holding capacity
(usually expressed in inch/inch) is typically given in the soil
series descriptions as a range for each of a number of soil
horizons. The average value of the range is multiplied by the
incremental depth it represents and these amounts summed to deter-
mine an AWC for each soil. A weighted watershed average value is
then computed.
Curve Number (CN) This is the SCS curve number (SCS 1972, 1975)
which combines the percent of the watershed in A, B, C, D, hydro-
logic classes of soils with land cover. See the SCS references or
Haan and Barfield (1978) for procedures for computing CN. Care
must be taken to ensure that the soil series description corres-
ponds with the assigned hydrologic class (see Section IV.B). In
practice, the land cover occupying each soil type is seldom known.
Therefore, an assumption can be made that forest occupies the
poorer soils (class D) while row crops occupy the better soils
(class A or B). Each land cover is allocated to a logical hydro-
logic soil group and the corresponding CN for that cover-
hydrologic group is determined from the tables. A watershed
average CN is then determined.
4. Carbonate Rock Associated Measures
These measures apply in areas underlain by carbonate rock. A
geologic map must be available in those areas where carbonate rock is
suspected in order to determine rock type. Determination of the following
measures requires some knowledge of geohydrology.
a. Modifications to Drainage Area In soluble carbonate rock areas,
the geologic divide may not correspond with the topographic
divide. Springs may bring in water from areas outside the topo-
graphic divide and water may be lost from within the topographic
divide through sinkholes. To the extent possible, adjustments to
the drainage area should be made where there is supporting
evidence such as sinkholes.
b. Bypass Seepage (DLF) In the absence of any hydrologic data this
measure is difficult to estimate. It can be approximated as the
fraction of the watershed area underlain by very soluble carbonate
rock. This measure primarily affects the yield of potential
ground water runoff (see Section V.C.7).
c. Karst Areas (PKARST) This measure controls the yield of potential
storm runoff. Values are assigned to each rock type underlying
the watershed as follows: very soluble carbonate rock,
PKARST=1.0; moderately soluble carbonate rocks, PKARST=0.5; all
other rocks PKARST=0.0. Judgment values between these numbers can
be used. An average value for the watershed is determined based
upon the percent of the watershed each rock type occupies (see
Section V.C.9).
18
-------
5. Other Land-Use Measures
a. Percent Mined The measure applies only to a surface mined area
that has a significant effect on the storm hydrograph. For
example, if the land is returned back-to-contour and revegetated,
the value used here would be zero. It is important that the value
used for this measure not exceed the range of values used during
regionalization (0-24 percent). (See the last sub-section in this
section.)
b. Directly Connected Impervious Areas In the algorithms used in the
daily flow model, an impervious area amounting to less than 17
percent of the watershed area does not affect the volume of storm
runoff (see Section V.C.2.a). The assumption is that when the
impervious area is less than 17 percent, the runoff from these
impervious surfaces drains into adjacent pervious areas. However,
occasionally there are impervious surfaces, such as roads parallel-
ing a stream, that drain directly into the stream. These areas
should be considered even though the total impervious area is less
than 17 percent. If a value is provided for the directly connected
measure, that value will be used as the impervious area contribut-
ing to storm runoff, regardless of the impervious area measure
used for land cover.
6. Sediment Associated Measures
There are provisions for simulating suspended sediment from both
impervious areas (dust and dirt) and pervious areas (erosion) within a
watershed. The model formulations are from published sources so that
estimates of the input measures can be made without field measurements.
a. USLE Terms These are the terms in the modified Universal £k>il
Loss Equation (Williams, 1975) determined for the pervious
portions of a watershed. The Modified Universal Soil Loss
Equation (without the runoff energy term) is expressed by:
USLEP=95 LSKCP I1I-3
Terms required as model input include:
LS is the length slope term for the watershed as determined using
the method of Williams and Berndt (1976),
K is the soil erodability factor for each soil averaged across the
watershed,
C is the cover term averaged for the watershed, and
P is the conservation practice term,
The length-slope term, LS, for a watershed is determined in the
following manner.
SLOPE = 25H(LC25+LC50+LC75) III-4
5280 DA
19
-------
where: SLOPE = slope in percent,
H = maximum difference in watershed elevation in
feet,
LC = length in miles across the watershed of the
contour representing 25, 50, or 75 percent
of H, and
DA = drainage area in square miles.
The length term, LENGTH, is calculated as
LENGTH = .264/DD III-5
where LENGTH = length in feet and
DD = drainage density.
The length-slope term then becomes
T ]rWPTH 0*5
LS = ( ^0 * ) (0.065+0.0454SLOPE+0.0065(SLOPE)2) III-6
12.. o
The text by Haan and Barfield (1978) summarizes methods for deter-
mining the other terms.
b. Grain Size Ratio (D) If a grain size distribution is available,
the ratio of the median grain size to one micron is used. Other-
wise, in the absence of additional information assume a value of
one.
c. Daily Impervious Area Removal Rate (R) The removal rate R is the
fraction of accumulated dust and dirt removed daily by wind,
traffic, and sweeping. A value of 0.08 is recommended if sediment
from impervious areas is to be simulated.
e. Accumulation Rate (ACCIA) This is the daily accumulation rate for
dust and dirt on impervious surfaces in pounds per acre. (See
Donigan and Crawford, 1976, and Metcalf and Eddy, 1971, for values
for several cities.) Values typically range from 1 to 10 Ib/ac or
more. Since the yield of dust and dirt from impervious surfaces
is usually low relative to the sediment from pervious surfaces,
the R and ACCIA terms can be assumed to be zero if the impervious
portion of a watershed is small (<10-15 percent).
7. Convolution Interval (DT)
Although selection of a convolution interval (DT) is not critical,
the use of a DT that is too small can significantly add to the computer run
time, while one that is too long will generally result in poor simulations
and may cause computer diagnostics. The text by Haan and Barfield (1978,
p81ff) presents a number of equations for predicting unit hydrograph time
parameters.
20
-------
In general, the following relationship adapted from Kirpich (1940)
may be used to give an initial estimate of the time of concentration, which
is the time of travel from the hydraulically most distant point in the basin
to the watershed outlet.
TC=2.6 L°-77(L/H)°-385 III-7
where: TC is the time of concentration in hours,
L is the main channel length in miles,
H is the difference in elevation between the basin outlet
and the hydraulically most distant point in the
watershed, in feet.
Using TC as an estimate of the time to peak and recognizing that there
should be at least three DT units prior to the peak, DT may be estimated in
hours as
DT=TC/3 III-8
DT would then be rounded to 5, 10, 15, 20, or 30 minutes, or an integer
multiple of one hour. DT is converted to hours or fractions of an hour
internally in the program. In locations where significant mining or urban-
ization has occured, or will occur, it may be necessary to reduce DT some-
what. Conversely, where there is an extensive forest cover and/or permeable
soils, the DT estimated by these equations will be too short. Following an
initial simulation, the adequacy of the estimated DT may be determined from
a printout of storm hydrographs as shown in Figure 3 and adjustments may be
made if necessary. If, for instance, the time to peak is consistently less
than three DT units, DT should be shortened.
8. Storm Selection Threshold Measures
The threshold measures allow the user to control the number of
storm hydrographs which are simulated per year of continuous simulation.
These measures should be selected so that 3 to 4 storm hydrographs per year
are simulated when flow frequencies are to be determined. When the model is
used to simulate continuous storm hydrographs (for example, if a year of
sediment loads are simulated) these limits would be set much lower.
a. Minimum rainfall (MINRF) The value for a larger storm that
typically falls during the winter (high runoff) season should be
selected.
b. Minimum Runoff (MINRO) This threshold is used to eliminate storms
that occur during the summer when runoff is so low as to preclude
a high peak discharge.
c. Maximum Rainfall (MAXRF) This threshold forces hydrographs for
all of the storms equal to or larger than this threshold value to
be simulated regardless of the associated runoff. This option can
be important: when urban areas are involved.
21
-------
To simulate three to four storms per year in an area where annual
rainfall averages between 45 and 55 inches, a first trial set of values
might be MINRF=1.5, MINRO=0.2, and MAXRF=2.5 inches.
9. Meterological Measures
Although the required meterological inputs are seemingly rather
gross measures, the simulations are quite sensitive to these values being
representative. The long-term average rainfall and runoff (or loss.)
estimates are used in the regionalized relationships for predicting param-
eters in the continuous daily streamflow model and thus affect the alloca-
tion of all precipitation among streamflow, soil moisture recharge, and
evapotranspiration. The measure of long-term monthly evaporation, on the
other hand, need only have a reasonable distribution among the 12 months as
a volumetric correction is made within the model so that the annual total
corresponds with the read-in loss determined from latitude (see Section
II.B.lS.cj or from rainfall minus runoff. In contrast, the statistical
measures of rainfall are critical since generated precipitation drives the
entire system.
a. Long-term Average Rainfall This information can best be obtained
from published NOAA-National Weather Service records or comparable
sources that determine long-term normal or mean rainfall for
reporting stations. The raingage selected should be nearby and
representative. Records from stations less than 25 years old
should be adjusted to a long-term mean at a longer-record station.
The station selected to give the long-term mean should also be
used in determining the monthly statistics. The long-term mean
annual rainfall need not correspond exactly with that at the
long-term station, however, if a better estimate can be made from
another source (for example, if there is a long-term rainfall
isohyetal map available which indicates a different value). Data
should be compared from several stations in the area to ensure
that a representative value is being used.
b. Long-term Average Runoff (or Latitude) Only one of these two
measures is needed. Within the region of applicability (Section
I) latitude should normally be used unless there is evidence to
indicate that the long-term runoff computed using this method will
be in error. Rearranging terms in the equation presented in
Section II.B.13.C, results in the following equation for predict-
ing long-term runoff (RO) from long-term rainfall (RF) and
latitude (LAT).
RO=RP-65.5+LAT III-9
When a value for LAT is entered, RO should be entered as zero.
At locations outside the region of model applicability the long-
term runoff should be entered (this can be determined from RF
minus an expected loss). In this case a zero should be entered
for average latitude.
22
-------
c. HTFLUX This measure is a ratio of the rate of evaporation of
intercepted rainfall to the normal rate of transpiration. A value
of 3.0 should be used in the absence of better data. This allows
intercepted water to evaporate at three times the rate of trans-
piration. The HTFLUX measure becomes more important on forested
watersheds.
d. Mean of the Cube Root of Monthly Rainfall This information must
be obtained from representative long-term raingage records (at
least some 25 years in duration). The monthly and annual simu-
lated rainfall averages will be very nearly the same as those of
the station used, so this station must be representative. If,
however, very unusual rainfall patterns occurred during a month or
two, some adjustment to these data may be justified. For example,
if the average for a particular month differed considerably from
the published normal, because of unusual rainfall during a year or
two, some adjustment should be made. These adjustments should
seldom be necessary for very long-term stations.
These values are determined by taking the cube root of all monthly
rainfall values and then determining a mean for each month. As an
option, the mean of the cube root of monthly rainfall may be
estimated from published station monthly mean or normal data with
the following equation:
3VRF = 0.935(RF)°'3436 111-10
where RF is the long-term mean monthly rainfall for a given month.
This equation is valid for monthly rainfall within the range of
about 2.5 inches to 9 inches. This monthly information is input
in water year sequence; i.e., October is the first month.
e. Standard Deviation of Cube Root of Monthly Rainfall (SDj The
monthly rainfall simulations are quite sensitive to the values
used for these standard deviations. These data must be obtained
from a representative nearby long-term raingage record. In order
to understand how these numbers combine to simulate rainfall, a
simple computation can be made. In a 20-year simulation, on the
average, there will be one monthly rainfall simulated outside the
range determined by the following equation:
RF monthly = (3V'RF ± 2SD)3 III-11
Thus for a typical October with a long-term mean of the cube root
of monthly rainfall (3A/RF) equal to 1.3 and assuming an SD of .29
(lowest observed), at least one month of October rainfall will be
outside the range of 6.7-0.4 inches for a 20-year simulation
period. If, on the other hand, the maximum observed SD is used,
the range becomes 10.4-. 07 inches. This points up the need for
care when selecting a representative long-term gage.
23
-------
The computed standard deviations should be smoothed somewhat as
the calculated values are quite sensitive to unusual monthly rain-
fall values, even in a 25-year record. This smoothing should
result in a fairly consistent pattern from month to month. Figure
4 shows an average pattern and the extreme values obtained analyz-
ing data from 34 raingage locations across the Tennessee Valley
area and into Kentucky where rainfall averaged from 39 to 59
inches annually.
Actual calculation of the standard deviation of the cube root of
monthly rainfall involves the following:
1. Determine the cube root of the rainfall for a given month for each
year in the period of record and compute the mean value (beginning
with October).
2. Determine the standard deviation of these cube root values for the
given month.
3. Repeat for other 11 months
4. Smooth standard deviation values as necessary
B. THE WATERSHEDS USED TO DEVELOP THE REGIONALIZED RELATIONSHIPS
Table 1 lists the watersheds for which hydrologic data were used
to obtain optimum values for the parameters of the continuous daily stream-
flow model and the storm hydrograph model. These optimized values were used
subsequently to develop the regionalized relationships. Shown in the table
are the gage type, the latitude and longitude of the watershed, the physio-
graphic province involved, the percent impervious or mined, percent forest,
the annual rainfall and the model with which the data were used.
C. RANGE OF VALUES USED IN DEVELOPING REGIONALIZED RELATIONSHIPS
Listed in Table 2 are those measures used in the regionalized
relationships to predict model parameters. Some of these measures are used
in standardized formulations (such as terms in the Universal Soil Loss
Equation) and consequently are not regionalized measures, per se. Acceptable
values may be obtained from the literature for these measures, therefore,
these terms are noted with an "LT". Where upper and lower bounds for the
measures are given, it is important that computed measures stay within these
limits or the regionalized relationships may not apply. Section IV will
provide some clues on how to handle situations where measures exceed these
limits.
The characteristics listed in Table 2 generally follow the se-
quence in which they were presented in the Section II description of the
interactive input information. A relative measure of the importance of
keeping measures within the stated bounds is also shown in Table 2. The
24
-------
ro
a
«+
E
o
cr
o
o
o
a>
.Q
3
O
c
o
(D
Q
-o
D
C
O
I
I
<*
O
o>
c
o
O)
3
a>
'NOI1VIA3Q QdVONViS
25
-------
13
0)
CO
£>
cj
co
cj
CO
CJ
CJ
CJ
CO
CJ
CO
-^
cj
CO
CJ
W
c 1 i 1 W
CO CO 01
d +< .d
d d u
d -rl C!
< <0 K-l
«
P +»
C H
0)
H
>_3
W
Q
0
s
OH
0
Ci-l
CO
p
co
PH
w
H
^
O
I i
H
P^
PQ
t 1
CJ
LO ^D LO ^-
LO LO LO LO
CM CM LO ai
CM ^f CM CN
/_x
1-1 CM
4-> O ^
d
OJ C T3
cj co I)
l-i ,0 C
0) 1-1 -H
P-l E3 S
o o o o
CD
1 U U
O -H d s-~*
r-l 43 -r-i r 1
W &, > s_/
>> C3 O
PJ w P-l
P^ P"! P^ P-^
w H S £3
00 /-v
CO CO CN
S CJ -H
rH 1-1 5
CO < *_/
1-1
o
. '*~r
00 00
rt ai
0 T3
1-1 '^
^
-1-1 00
CO CU
i-^ 'O
* /
00 1-^ CM
O I-H 00 C^
...
CM r~- o '-*
r^. r*- \^ *o
....
r r^ i . t >
00 00 00 00
en en co co
....
^d*
co
LO
oo
CO
oo
CM
00
l-i
CJ
^*^
1-1 CM
CO 1
CU CJ
PQ CO
N.^
1-1
CU )-3
ft <
ft
£D
l-i
CJ
^-^
1-4 CO
CU CJ
PQ CO
^^
^l
CO CM
CU 1
PQ CO
^s
1-1
CU 1-5
ft<
ft
t3
CU
a
c
^_)
d
o
CJ
^»/
26
-------
0)
co
:=>
CO
en
U
CO
CO
c_>
u
o
CO
CO
i 1
CO
^
d
d
<3^
i t
CO
4-1
d
H
CO
PS
_,_,
d
0)
u
1-4
W
0)
X!
U
d
i i
4-1
CO
0)
S-4
O
-* o
LO LO
o o
O 00
\O
LO
LO
t--
>« m o
X) 1-1 S-l
PH 00 P^
CM
pq
55
co
CO
PQ
;=>
o
P-I
CJ
OH
LO
eu
CM
CO
CH
CJ
a
eq
00 ^-.
CO CO CM
d cu -H
H 1-1 S
CO *3S "'
s-t
n
I
CN
O
O
O
LO
I
LO
CO
00
00
o
o-
CN
CO
oo
CO
LO
CM
LO
oo oo
d cu
o -o
LO
oo
oo
00
LO
OO
CM
00
oo
00
03
00
LO
00
sO
00
LO
00
LO
00
4-1
CO 0)
1-1 "O
CD CU
oo pi
CO >
O H
<
CN
LO
CO
(H
C^
LO
CO
CO
o
CO
!=>
CM
vO
CO
CO
O
CO
JD
sO
LO
CO
\ CO
< o
> CO
t-H
vd
CO
CO
o
CO
r-~
LO
CO
CO
o
CO
JD
i i
^D
CO
CO
O
CO
E3
O
^
CO
CO
o
CO
J3
OO
LO
CO
CO
o
co
CO
OJ
CQ
1-1
0)
ft
ft
£3 I-?
CO
CO
S-l
0)
I-I
PH
1-1
OJ
1 1
1 1
H
^
4-1
i 1
CO
£5
H
V
CO
4-1
1-1
cO
ft
CO
£^
H
. %
i-i d
C_) O
d
00 co
d , c^
H CD
1-1 I 3
ft
co
**^
CD
^_l
CJ
CD
d
CO
£^
f-H
r.
1 1
1 1
H
|-H
^
"TD
CO
XI
CO
M
cu
0)
1-1
CJ
CO
d
S
O
^
0?
55
H
~
XI
CO
^^
^
S-l
o
[T |
t*^
CU
d
CO
0
J^H
E-t
rv
>>
4-)
4-1
r-j
i 1
CJ
S-l
cu
>
H
PH
>.
1-1
O
g
W
55
H
*s
00
1-1
£3
Xi
S-l
CO
3:
^SH
CD
CD
1-1
CJ
1-1
cu
4-1
4J
I-I
cq
^
H
s
d
H
M
w
^3
<
a
CO
^
0)
CD
l-i
C_J
p3
O
O
u
CJ
CO
£^j
H
*.
ft
CO
W
13
d
H
3:
13
r i
O
'O
CD
(^
H
^_)
d
o
o
>. /
27
-------
cu
co
S3
CO
CO
CO
co
CO
o
CO
co
CO
CO
CO
CJ
co
CJ
r-l ii co
CO CO CU
|w *W p£3
ecu
(3 -H (3
<; co ii
o
\o
IT)
oo
LO
r-.
O-
> CO O
43 SH SH
OH OOOn
O
CO
Oi
O
CM
U
OH
U
CJ
OH
CJ
H
O
p-l
CJ
S
o
CM
O-i
CJ
S
o
O-i
CJ
S
LO
O4
CJ
W
0)
00 *~~
CO co CM
a cu !-
H SH E
CO < '-
SH
O
O
oo
Cs)
LO
00
CM
O
00
o
LO
vO
00 00
a cu
O T3
O
00
00
00
LO
00
00
00
o-
OO
00
LO
o-
00
v±
00
-P OO
0) CUl
oo pJ
(3 >J
e> HI
e>
LO
CO
CO
o
CO
vO
CO
CO
CO
CO
co
o
co
£5
CO
CO
o
CO
CO
vO
CO
CO
Jc O
H
CO
.
-;<
H
CO
. CO
;< o
CO
CO
^o
CO
--. en
Jc O
E-H CO
S3 S3
CO
vO
CO
~-» co
-Jc O
E-H CO
S3 S3
C^
vO
CO
CO
a
CO
S3
so
LO
CO
co
o
CO
S3
g
r&4 o^d
T3
cu
<-J
CO
SH
CU
<0
S
cu cj
CU -i-l
SH h^
CJ
00
4-< -H
r-i eq
cu
co
M H
CU
0) >
SH CU
CJ r-l
CO
T3 ^3
U ^j
Ji co
(-1 0
0
H
cu
rH
M i 1
CU -H
0) >
SH to
CJ to
O
>< CJ
^
cq
CJ >
CO O
SH O
pq S
t3
(3 CU
O CU
4J SH
r-l O
CU
W
43 iz;
CJ H
a
S-l CO
pq e
cu o
o
CJ
a
CO
s-i iz;
PQ H
a >
O co
co S
S-l SH
CU O
*o Jz;
a
u iz;
a H
CO
SH ~
Pq co
S
a SH
CU O
SH
0
u iz;
a H
CO
SH
pq co
S
to SH
rH O
28
-------
TABLE 1 (continued)
Physio -
Drainage graphic Percent Annual
ong. Area Province Urban or Percent Rainfall Used
deg.) (mi2) (1) Mined (2) Forest Inches (3)
^ ^
4-) 00
B) CJ
13 rO
*-*"
CO
CJ
O
O
rH
O
00
CM
OO
OO
CO
.
\O
CO
H
55
H
*\
^
4J
H
j«s cj
CJ
cu d
S-l O
CJ CO
d
43 43
CO O
d >->
S-l
OH-
IO
CM
00
o
CM
CM
i i
oo
CM
vQ
CO
CO
o
CO
^
CJ 55
OJ H
CJ -
CO
O -H
i-H S-l
B! S-l
4-1 O
4-1 55
3
PQ
CJ
00
OO
0
I 1
LO
00
o
00
1 I
vO
OO
CO
o
CO
:=>
CO
rH
Bj
W 1-^*1
H
v>
^4 w
0) T3
H
^3
r^d "
CJ d
CJ O
S-l -H
CJ 4->
Bj
CU 4->
CJ CO
J
4J d
CO O
CJ CO
43 rH
CJ -H
ES
t 1
co co
0\1 \O
LO O-
oo >
H H
r2^ J^
J*i H H
CJ
S-l T3 CJ CJ
CJ O CJ rH
O SH i 1
CJ S CJ -H
r-« 1 (-H fcjl
pJ I^H F^ t*N
4-1 00 4-) O
w d s-i d
CJ W 3 NH
43 0
CJ PH
CO
CJ
CM
00
CM
vO
i i
O
o-
00
o
^o
CO
<£
>
H
55
H
»s
^ qj
CJ r-l
(D rH
r-! *H
CJ >
X
~O 0
s-i d
43
H
CO
LO
1 1
LO
o
rH
00
o
^o
CO
H
55
H
«v
ri^ CJ
CJ rH
CJ i-l
S-l -H
CJ >
X
4-> 0
co d
S-l &
H
PH
co
,__!
rH
H
W
rj
O
H
4-)
B!
4-1
d
B)
r-l
0-1
CO
OO
OO
CM
H
55
H
#s
CJ
I 1
1 1
H
>
X
o
(3
N
T3
CJ
3
d
r-(
4->
d
o
u
^^
29
-------
co
co
13
CD
d
d
H
4_)
d
o
o
-
M
pq
H
Annual
nt Rainfall Used
st Inches (3
cu CD
O SH
to 0
CU I-M
PH
S^ CS
4-1 O ^
d
CU d T3
u co CD
SH Xi d
CD S-l -H
PH C3 2
00
CO
in
i i
0
CU
1 U O
O -H d
H XI -H .-N
CQ ft > iH
>> CO O ^-^
Xi SH !^
PH 600H
60
co CO .' s
d CU CM
H SH -H
co < E
Q
60 60
d cu
O T3
i-l -~ '
4-1 60
CO CU
1-3 rO
1 '
«
pq
i i
in
.
T 1
m
CM
00
^
m
CO
cu oil
60 pj <
to R >
O Hi H
CJ
55
u -
d s-i
CO 0)
CD
(-{
CO
S-l
cu
4-1
CO
^2
SH- 4-1
pq co
CD
S-l U
CU -H
AS CU
S-i }_j
CO
PH
CJ
o
o
T-H
o
PH
PQ
>Hf
O
"^
r-H
xj-
OO
^
m
CO
co
CO
!=)
ly* r .
r H C^
CD
CD ~
S-l CO
cj d
H
O CO
CJ rH
H PH
H O
CJ CJ
H
rH
A! rH
pr | CJJ
H
.
55
CM
CO
00
0
fV|
pq
o>
»
o
rH
m
CM
00
i 1
m
CO
CO
o
CO
p
0
55
AS >
CU CD
CU (0
SH O
CJ S-l
d
pO OJ
03 Q-i
j_l
CJ
In
*
XI 0
cj 55
d
co ~
S-l CD
pq o
XI
d w
O CU
4-1 CO
rH O
o
pq
co
-^ co
u
o\ f-.
vo m
o
O CTl
r-H O-\
0 0
pq pq
O> "3"
0>
O^ i~~i
r- o
CN CO
00 00
>
S_i 55 AJ 4-1
CJ H CU -H
CD cj
d s-i
co -H cj d
HO O
d -H d >>
M d CS to
!=> rH cq
o
55 55
TJ
d
CO
iH
S-l
CU
&
CJ
II
PH
CJ
> -,
d
H
co
CO
CQ
cu
,H
i 1
H
>
XI
co
CO
55
II
PQ
55
» «y
H
^,
d
<
H
ft
ft
H
CO
co
H CD
CO 60
CO T3
H -H
2 PH
II ^
CD
pi i i
pr* rH
CO
^> >
OO
CO II
ON
TH PH
^ ' ^
d *
co d
S co
cu cu
d 4-1
ti ^
CU i-l
r-y ;. f^ |
S~~\
1 t
^^
X ^
,2
CD
d
H
S
S-l
O
/_N
1 1
^^
d
CO
Xi
S-l
d
4-1
d
cu
U
^J
CU
ft
cu
SH
C8
TD
CD
4-1
CO
H
>-q
/ v
CM
i 1
CD
T3 CU
O CU
S co
co
XI CU
ft d
co d
S-l CU
60H
0
T3 0
^ 4J
S-l CO
0 S-l
4-i cu
CO >
H
II d
co -;<
CU
13
o
e
^
o
4-1
e
CO
cu
SH
4-1
CO
K*^
rH
H
co co
T3 i 1
CD
CO T3
d o
0 S
d
d xl
H 4-1
4-1 O
d Xl
O
0 II
II co
^v^,
CJ CJ
.' \
CO
v-^
-------
critical index ranges from 1 for measures that must be kept nearly within
the range used in regionalization, to 3 for measures where this range is
relatively unimportant.
31
-------
TABLE 2
BASIN CHARACTERISTIC RANGE USED IN REGIONALIZED RELATIONSHIPS
Characteristic
Drainage area (sq. miles)
% Hardwood
% Conifer
% Pasture
% Small grain
% Row crop
% Impervious
% Unvegetated
% Directly connected impervious areas
Slope (ft/mi)
Shape
Drainage density (mi/sq.mi.)
Curve number
Sinuosity
% mined
Soil permeability (in/hr)
AWC (inches)
Bypass seepage
PKARST
Sediment cone, at 1 csm*
K Univ. Soil Loss Equation term.
LS "
p IT 11 11 If
p II II II IT
Grain size ratio
Daily imp. area removal rate
Daily D&D accumulation rate
Convolution time DT (hours)
Mean rainfall (inches)
Mean runoff (inches)
Latitude (degrees)
HTFLUX-
Lower
Value
0.2
0
0
0
0
0
0
0
0
10.9
1.2
9.4
35
0.015
0
1.2
2.6
0
0
10
LT
LT
LT
LT
1
LT
LT
0.083
38
10
34.3
3
Upper
Value
175
100
100
100
100
100
45
86
5
1170
7.3
17
82
0.56
24
8.1
11.7
1.0
1.0
10
LT
LT
LT
LT
2
LT
LT
2
77
50
36.9
3
Critical
£
Index
2
3
3
3
3
3
2
3
3
1
1
1
1
1
1
1
1
1
1
3
-
-
-
-
2
-
-
1
2
2
2
2
^default or standard value used
LT=literature values
a-l=measures should be kept within stated bounds
2=measures should be kept near stated bounds
3=measures should be reasonable
32
-------
SECTION IV
USER STRATEGIES
A. THE TYPES OF SIMULATIONS THAT MAY BE OBTAINED
Figure 2 shows the basic output that is obtained from the inter-
active version of TVA-HYSIM. The basic output consists of storm rainfall,
runoff, and peak discharge, and if the sediment option is exercised, the
suspended sediment load, and average and maximum sediment concentration for
each storm simulated. The annual rainfall, runoff, minimum one-day flow,
and sediment load is given for each year simulated and the average annual
rainfall, runoff, and sediment load are shown at the end of the simulation.
The units for rainfall and runoff are watershed-inches (equivalent depth
over the contributing drainage area). Sediment loads are in tons. Peak
discharge and minimum flow are in cubic feet per second.
If a flood frequency is to be evaluated, the storm threshold con-
trols (Section II.B.15) should be set so that on the average three to four
storms per year are simulated. If these thresholds are set too low many
storms will be simulated, the computer cost will be high, and nothing will
be gained from the simulation of additional storms; but if the thresholds
are too high, no storms may be simulated during one or more years. The
thresholds should be set with the MINKF corresponding to a storm size that
typically occurs several times during a high runoff season. MAXRF should be
set at a reasonable high storm or daily, rainfall value. The threshold
MINRO can be set to any reasonable value that will eliminate those storms
with high rainfall that occur during the low runoff season but do not
produce significant peak flows.
Water yield can be determined from the average annual rainfall and
runoff shown at the end of the printout (Figure 2). If only water yield is
to be analyzed, the rainfall-runoff threshold values should be set high so
that no storm hydrographs are simulated.
Minimum one-day flows are determined for each year during the
course of the simulations and are shown on the interactive printout. If
minimum flow simulations for other durations are needed, they may be
obtained by answering "yes" to question 2 (Section II) which causes a
detailed printout to be stored. When running the program from a time-
sharing terminal, the detailed printout of the continuous mean daily dis-
charge simulations and the storm hydrograph simulations will be stored on a
disk file. This information may be obtained in a second job step using a
user-supplied utility program that will print a normal file to a line
printer.
33
-------
When the program is run from cards (batch), the job controls can
be used to obtain the equivalent interactive printout along with the contin-
uous daily streamflow model and storm hydrograph model simulations. Section
VI has an example job control set-up showing how this batch run might be
accomplished. When this printout is obtained, the continuous daily stream-
flow information will contain identification information and parameter
values. Following this, there will be two tables (matrices) for each year
simulated. The first table is the simulated daily rainfall for the water
year; the second table is the simulated daily runoff. These simulated
runoff values are in watershed-inches printed in scientific notation. The
minimum simulated flows for any desired duration can be determined from this
latter table and converted to conventional units with the following equa-
tion:
CFS=DA-IN/0.0372 IV-1
where: CFS is cubic feet per second (per day),
IN is the average daily flow in watershed-inches, and
DA is the drainage area in mi2.
Suspended sediment simulations are tied to the storm hydrograph
simulations. Obviously, as more storms are simulated per year, the total
annual sediment load that is computed will increase (although typically most
of the sediment will be associated with the larger storms). Therefore, if
annual sediment loads are the consideration, the rainfall-runoff thresholds
should be set very low so that most storms are simulated. If this is done,
however, only a few years should be simulated to keep the computer run-time
reasonable.
TVA-HYSIM is not adapted to handling dynamic land-use conditions.
For example, there are no provisions for handling the changing land-use that
occurs in urban areas or the during-mining phase of surface mining. This
model package is designed to be used as a planning tool so that the end
effects of the land-use change can be evaluated before the change occurs.
Thus, in a typical land-use change evaluation the model package would be
used first to simulate hydrology under present land-use conditions (or any
other baseline condition) and then used to simulate the post land-use change
hydrology. Used in this manner, the model package could, for example, be
used to determine many of the probable hydrologic consequences of surface
mining as required under PL95-87 (30CFR 780.21C).*
B. ARE THE BASIN CHARACTERISTICS "CORRECT"?
There are a number of basin and climatological measures needed to
operate TVA-HYSIM. Section III described at some length how these charac-
teristics are to be measured to help assure that they are determined in the
same way that they were for the regionalized relationships. Nevertheless,
problems in determining these characteristics will occur. Probably, the
most serious problems will center around the soils-associated measures
(Section III.A.3).
^Surface Coal Mining and Reclamation Operations-Permanent Regulatory
Program, March 1979.
34
-------
The various soil measures are essential to operating TVA-HYSIM.
This model cannot be used where soils information is unavailable. Soil-
association maps can be used with caution at locations where detailed SCS
soils survey maps, or equivalent, are unavailable. However, even where
detailed county soil surveys are available, there are limits to the applica-
bility of certain of the measures at a given site. (1) The hydrologic
measures (available water holding capacity and permeability) published in
newer County Soil Survey reports or in established series descriptions are
typically based upon measurements made at a particular location. How
applicable these measures may be at another site many miles away with the
same soil series is unknown. The use of range averages for these measures
as a computational device does not circumvent the fact that at a particular
site the actual values could be at either side of the range, or beyond it.
(2) The calculation of curve number, CN, is highly dependent upon the hydro-
logic soil group classification which is based upon minimum infiltration
rates (SCS, 1975). Because of changes in soil classifications, the
published hydrologic information for a particular soil may not adequately
reflect the characteristics of that soil in a particular county. Therefore,
it is important that published descriptions in the County Soil Survey
describe an infiltration rate corresponding with the hydrologic soil group.
(For instance, the relatively low CN used in the example in Section II,
Figure 1, results from an adjustment to account for high infiltration
capacities.) If they do not agree, adjust the hydrologic soil group. (3)
Where hard-pans, plow-pans or fragipans exist which significantly inhibit
the downward movement of water, the soil moisture associated measures should
be determined only down to the impeding horizon. (4) Although some soils
are deep with bedrock at depths of 10 feet or more, the soil associated
measures are computed only as deep as the typical profile measures are given
(usually about six feet maximum).
Most of any problems encountered in simulating sediment will be
associated with determining the cover term "C" provided the approach of
Williams and Berndt (1976) is used to compute the length-slope term. The
simulation of suspended sediment loads is highly sensitive to the value used
for the cover term. Yet, for many cover conditions published values for "C"
may vary by a factor of from two to ten. The simulated loads will vary
proportionately. This means that, to a large degree, the reasonability of
the pervious area sediment simulations will depend upon the estimate used
for the cover term. In watersheds with mixed land-use, these estimates are
most critical for disturbed areas and agricultural lands.
And finally, as pointed out in Section III.A. 9. e, even when as
much as 25 years of monthly rainfall data are used to determine the standard
deviation of the cube root of monthly rainfall, some adjustment to these
calculated standard deviation values may be necessary. If the standard
deviations (SD) are too large, inordinately large and small monthly rainfall
values will be simulated occasionally. If the SD values are too small the
simulated monthly values will vary little from the mean monthly values. If
unusual monthly rainfall amounts have occurred in the observed data, the
computed standard deviations will probably be too large or too small. A
somewhat smoothed seasonal distribution of the standard deviations should be
used and be patterned after the average relationship shown in Figure 4.
35
-------
C. VALIDATION/VERIFICATION
In the absence of data for validation, simulation results using a
hydrologic model cannot be accepted with complete confidence. Even at
locations where a regionalized model is presumed to apply, verification of
the results is necessary. There are too many possible sources of error that
range from mistakes in measuring characteristics, to poorly estimated
measures, to using unrepresentative published measures, to simply making
mistakes in entering information into the computer. At locations beyond the
region of applicability for the models, validation becomes a necessity.
Perhaps the safest approach to validating the model package is to
simulate at a watershed where rainfall, streamflow and, with luck, suspended
sediment data are being, or have been, collected. The range and means of
the simulated annual rainfall, runoff and sediment loads can be compared
with the observed data. Similarly, the optional printout can be obtained
and monthly and daily rainfall extremes can be compared with observed data
for reasonability. The simulated large-storm peak discharges can be
compared with the published data. Through this process, and by changing
some of the "soft" or estimated measures described in the previous section,
the user can begin to develop confidence in the model and an appreciation of
the ^art" involved in determining some of these measures.
Where appropriate validation data are simply unavailable, then the
results should still be verified for reasonability. The following steps
should help assure that the results are at least reasonable.
1. Be sure that the simulated average annual rainfall, runoff, and
sediment load, shown at the bottom of Figure 2, are reasonable.
The average annual rainfall, for longer simulations, should corres-
pond well with the read-in value in question 13, Section II. The
mean runoff should also be fairly close to the read-in runoff or
rainfall minus the computed loss as obtained from values supplied
for that question. The average annual sediment load and peak
concentrations should correspond with values measured on similar
streams in the vicinity or be within the range of values obtained
from applicable published sources such as Dawdy (1967).
2. If large-storm hydrographs are simulated for a number of years,
the flood frequency relationship should be checked against predic-
tions obtained from published sources such as the statewide flood
frequency reports of the U.S. Geological Survey. While complete
agreement should not be expected, the two approaches should
predict 100-year floods, for example, that are reasonably similar.
Site factors, such as land-use, that might not be accounted for in
the published relationships should be considered when making the
comparisons.
3. The large-storm simulated rainfall and runoff values obtained from
the interactive printout (Figure 2) should be reasonable. If too
many unusually large storm rainfall values are found, the daily
transitional probabilities (Section V.B) may have to be adjusted
36
-------
or the standard deviation of the monthly rainfall cube roots may
have to be reduced somewhat. The optional simulated daily rainfall
printout (see Section IV.A, paragraph on minimum flows) is helpful
in these evaluations.
4. Storm hydrograph simulations (obtained when the answer to question
2, Section II, is "yes") should be checked to be sure the hourly
(or shorter) rainfall distributions are intuitively reasonable.
Considering the total rainfall involved and the season, the dura-
tions should be reasonable and not too "bursty". If a number of
storms appear to have unreasonable intensities, adjustments may be
needed in the hourly transitional probabilities.
5. From the storm hydrograph simulation (Figure 3) check to be sure
the value used for DT (question 12, Section II) is less than the
shortest computed unit hydrograph time to peak (Tl). Also, check
the computed time-lag values (TL) to be sure they are reasonable.
Time-lag can be estimated from standard procedures (for example:
See Haau and Barfield, 1978). The computed time-lag shown on the
printout should be reasonable considering the watershed character-
istics in general and the burst runoff intensity (B ROI). Un-
reasonable time-lag values could indicate an error in measuring
one or more of the unit-hydrograph associated watershed character-
istics, that one or more characteristics are beyond the range of
values shown in Table 2, or that the model is being used beyond
the applicable "region".
D. WHAT IF THE REGIONALIZED RELATIONSHIPS ARE NOT APPLICABLE?
There are probably three reasons why changes to the model might be
necessary: snow is a dominant consideration in the hydrology of the area; a
land-use that was not considered in the regionalization is encountered; the
model package is being used in an area where the regionalized relationships
are inapplicable. Each will be considered and approaches suggested for
handling the problem.
1. Where snow melt is an occasional occurrence, the effect on the
hydrology can be ignored. This model has no explicit provisions
for handling snow melt, as such. In areas where snow melt is
important and cannot be ignored, the transitional probabilities
(Section V.B) should be adjusted to reflect the snow melt pattern
(PW/W probably increased during months of snow melt to provide
more apparent days of "rain".) If a snow pack typically develops
during certain winter months, the average monthly rainfall pattern
used should reflect the snow melt regime rather than the water
equivalent of the snow buildup (with appropriate adjustments in
the standard deviation of the monthly cube root of rainfall). In
areas where snow packs dominate the hydrology, the previous adjust-
ments may not work well and major changes to the rainfall
generator may be necessary.
37
-------
2. If a land-use is encountered that has not been considered in the
regionalized relationships (for example, a raountaintop removal-
head of hollow fill surface mining operation) and the regionalized
relationships do not work well, modifications to the unit hydro-
graph regionalized relationships may be needed. The regionalized
relationships for predicting the unit hydrograph model parameters
can be modified rather simply in the model. Arriving at what
modifications are necessary, however, is somewhat complicated and
requires analysis of storm hydrograph data. Section V.D.A
describes the regionalized relationships used to predict the unit
hydrograph parameters and notes that there is a vector of modifi-
cation constants defaulted to 1.0 in a data definition statement
in subroutine SCHAR1. The various equations for predicting coef-
ficients in the time-lag precipitation intensity equation can be
modified by redefining this modification variable named UGMOD.
The recommended method for determining values for the vector of
UGMOD modification variables involves adjusting the unit hydro-
graph model to observed storm hydrograph data using an analytical
storm hydrograph model (obtainable from the authors). When this
is done for a range of storms, the runoff intensity for each storm
is plotted versus the time lag determined from the analysis
program and the value obtained using the existing regionalized
relationships. The regionalized relationship is then modified
with an appropriate UGMOD variable so that the slope of the rela-
tionship best fits the time-lag values obtained from analysis.
This is a rather complex process and should be attempted only
after the analysis of a large number of storms.
3. When validation testing with observed data indicates that the
regionalized relationships may not be applicable, there is a
recommended hierachy for considering modifications (the component
models are described in Section V). First, be sure that the
stochastic rainfall generator is simulating reasonable monthly
volumes since the monthly simulations are easiest to compare with
observed data. Then check the daily rainfall volumes, their
seasonal distribution, and the hourly rainfall volumes for storm
periods for reasonability. The rainfall generator is the most
easily modified component and is critical to simulations with all
other components.
Second, compare the annual and monthly streamflow simulations from
the continuous daily streamflow model with available data and
check the values for reasonability. Simulations with this model
are sensitive to the estimates of long-term mean precipitation and
runoff (question 13 Section II) and the measure of soil available
water holding capacity. If these measures are incorrect, the
simulations will be affected. Also, if the runoff loss terms (DLF
and PKARST question 8, Section II) are incorrect, the simulations
will be affected. The continuous daily streamflow model has been
adjusted to data from watersheds where the average annual runoff
ranged from about 10 inches to about 50 inches. The model cannot
handle arid-climate hydrology, however, because of some of the
38
-------
basic formulations (see Section V.C.3). If all of these pre-
cautions are considered and there is still a need to modify the
regionalized relationships, an analytic version of the model
(obtained from the authors.) must be adjusted to daily rainfall-
runoff data to obtain optimized parameter values for a number of
watersheds. Appropriate changes can then be made to the region-
alized relationships (Table 4, Section V).
Finally, check the storm hydrograph model simulations. Be sure
the basin characteristics are measured well and that none are
outside of the ranges shown in Table 2. If a critical measure is
outside this range, verify the measure. Try a computer run with
the measure set on one of the tabular limits. If all of the
measures appear to be correct and consistent with the methods of
measuring them described in Section III, and the hydrographs
appear to have consistent error, then modifications to the model
may be made as described in the previous paragraph.
The authors will be interested in any major modifications that are
made to TVA-HYSIM and the results obtained.
E. DIAGNOSTICS
Diagnostics, either from the operating system or from programmed
messages indicating invalid conditions, may occur during the course of the
simulations. It is impractical to attempt to describe them. Generally,
however, because of the straightforward information input procedures used in
this program, diagnostics usually indicate input data or procedure errors.
39
-------
SECTION V
DESCRIPTION OF TVA-HYSIM COMPONENTS
A. INTRODUCTION
TVA-HYSIM consists essentially of four linked components. Figure
5 shows a schematic of the model package. It is driven by a stochastic
rainfall generator that simulates a daily rainfall distribution for the
continuous daily streamflow model. Storm runoff (one or two day) amounts
identified in the continuous daily streamflow model, along with the corres-
ponding storm rainfall, are passed to the storm hydrograph model. Here, an
hourly (or shorter period) storm rainfall distribution is computed by the
rainfall generator. This distribution is used as the basis for computing a
storm hydrograph. When an urban sediment option is exercised, the continu-
ous daily streamflow model also determines the washoff of dust and dirt from
impervious surfaces for storm periods and transfers this information to the
suspended sediment model component. Suspended sediment loads from pervious
and impervious areas are distributed in the suspended sediment model using
precipitation excess from the storm hydrograph model component.
The output block on Figure 5 termed TVA-HYSIM indicates the output
that is illustrated in Figure 2. The optional storm hydrograph output shown
in Figure 5 is illustrated in Figure 3. There are no provisions in TVA-
HYSIM for obtaining at a time sharing terminal the daily runoff option shown
on Figure 5 which consists of lengthy tables of annual daily rainfall and
runoff. These latter simulations may be obtained in a separate job step or
when the program is run batch as explained in Sections II and VI.
This section describes each of the four model components. These
descriptions are presented in sufficient detail so that if changes in
component model algorithms become necessary for other regions or conditions,
these changes may be more easily made.
In the interest of keeping this user's guide easy to follow by
avoiding duplication, certain of the variables and coefficients are renamed
and therefore do not correspond with their counterparts in the FORTRAN
listings. The changes are slight and should not create problems in identify-
ing the algorithms in the listing. A number of subroutines are mentioned in
this section to identify the location of algorithms. These subroutines are
described in the next section in Table 6.
40
-------
RAINFALL
GENERATOR
STORM
HYDROGRAPH
MODEL
DAILY
STREAMFLOW
MODEL
STORM TOTAL
STORM
HYDROGRAPHS
f
SEDIMENT
(OPTION)
DAILY
RUNOFF(RO)
(OPTION)
TVAHYSIM
ANNUAL RF^RO
STORM EVENT:
PEAK DISCHARGE
SEDIMENT LOADS
Figure 5 : Schematic of Watershed Model
41
-------
B. STOCHASTIC RAINFALL GENERATOR COMPONENT
This rainfall generator uses a disaggregation approach. The
advantage of this approach (as opposed to the more conventional serially
dependent Markovian approach) lies in the ease with which the seasonal
characteristics and statistics of the rainfall are preserved. The rainfall
generator disaggregates generated monthly rainfall into: (1) a distribution
of daily rainfall using transitional probabilities; (2) the daily rainfall
into an hourly rainfall distribution; and (3) if shorter duration rainfall
is needed, the hourly rainfall into five-minute amounts which are then added
together to produce amounts for durations evenly divisible into an hour (5,
10, 15, 20 or 30 minutes).
Observed monthly rainfall is normalized using a cube root trans-
form. Random numbers from a uniform probability distribution (subroutine
URAND) are operated upon in subroutine GAUSS to generate values that pre-
serve the mean of the cube root of monthly rainfall and standard deviation
values that are read-in. These generated values are converted to monthly
rainfall in subroutine RFSIM. Next, the monthly rainfall is disaggregated
into daily rainfall using transitional probabilities of the following form:
D1?rb
PR = e-raRF V-l
where: PR is a transitional probability,
RF is the monthly (or daily or hourly) rainfall, and
ra, rb are coefficients.
The coefficients for two of the transitional probabilities are
defined in the program; the probability that a day will be dry given that
the previous day was dry (PD/D); and the probability of wet given wet
(PW/W). The following identities may be used to determine the remaining
transitional probabilities:
PD/W = 1 - PW/W V-2
PD = 1/[((1-PD/D)/PD/W) + 1] V-3
Table 3 includes transitional probabilities used in subroutine RFSIM to
determine a rain or no rain event for each day.
TABLE 3
DAILY RAINFALL TRANSITIONAL PROBABILITY COEFFICIENTS
Probability
PW/W
PD/D
Coefficient
ra
rb
ra
rb
Oct.
1.0
-0.06
0.1
0.60
Nov.
1.4
-0.30
0.14
0.40
Month
Dec. -Aug.
1.36
-0.38
0.18
0.31
Sept.
1.4
-0.30
0.14
0.40
42
-------
Rainfall is allocated to days of rainfall using the following
Weibull distribution to convert values from a normalized probability distri-
bution function (subroutine URAND) into rainfall:
RFD = (19.0 logU/YP)1'176 + O.D/100 V-4
where: RFD is the daily rainfall (unadjusted) in inches, and
YP is a random number from a uniform probability distribution.
An adjustment is next made to each day of rainfall by adding
together the unadjusted RFD's and multiplying each by the ratio of this
total to the predicted monthly rainfall previously determined. This adjust-
ment forces the distribution to total to the predicted monthly value.
Similarly, daily rainfall totals are disaggregated into hourly
amounts whenever storm rainfall simulations are made. Subroutine HOURF uses
a Fourier function to allow the probability of wet given wet (PW/W) to vary
across the year. The formulation for the transitional probabilities are:
PW/W = exp[-(.15-.03sin[(27t-JD/365)-0.8])RFD~'3] V-5
PD/D = exp[-0.041 RFD^] V-6
where: JD is the Julian day beginning with Oct. 1=1, and
RFD is the daily rainfall.
Seasonal phasing in equation V-5 is in radians and the -0.8 term
causes the highest PW/W (winter) to occur about February 15.
Rainfall is allocated to hours of rainfall using random numbers
drawn from a uniform probability distribution and transformed with the
following Weibull distribution function:
RFH = ((101og(l/YP))0'8 + 1.0)/100 V-7
where: RFH is hourly rainfall in inches.
An adjustment is next made to each value of hourly rainfall to
assure that it adds to the correct total predicted for the storm.
A final subroutine FIVMIN determines five-minute rainfall amounts
each hour there is rainfall (when DT is less than an hour) for use in the
storm hydrograph model. The two transitional probabilities used are:
PW/W = exp[-0.026RFH~°'65] V-8
PD/D = exp[-3.0RFH°'5] V-9
and the Weibull distribution is:
RF5 = (2.0 log(l/YP) + 1.0)/100 V-10
43
-------
where: RF5 is a five minute rainfall amount.
When the pre-set transitional probabilities and Weibull distri-
butions are used, the only information needed to operate this model
component are the means of the cube root of monthly rainfall and the corres-
ponding standard deviations. This is typically obtained from about 25-years
or more of monthly raingage data. Section III.A.9 explains in detail how
these data are obtained.
At locations where the pre-set probabilities (Table 3) do not
apply, the probabilities must be adjusted by changing data definition state-
ments in the appropriate subroutines. The basis for such adjustments must
be observed data. The basic transitional probability relationship, equation
V-l, may be used to help in revising the coefficients. For example, if
storm durations are too long or too short, estimate an expected duration for
a 1 inch rainfall such that about half the storms are longer and half are
shorter. The "ra" coefficient in the basic equation will equal:
ra = -ln(0.5)/tl V-ll
where: ra is a coefficient in the transitional probability equation,
and
tl is the average duration of a one-inch storm in hours.
The ra term can be calculated for both summer and winter and then used with
the Fourier relationship in equation V-5, or a single-season relationship
may be used.
Similarly, the average duration for a larger storm is estimated
and the "rb" term estimated from the following:
In(tl/t2)
In RF2 V-12
where: t2 is the average duration for a storm other than one-inch (for
example, a large storm) in hours, and
RF2 is the corresponding rainfall for the storm other than one
inch, in inches.
This approach may be used to estimate coefficients for either the
PW/W or PD/D transitional probabilities using observed data, in this case
hourly rainfall abstracts. Usually the hourly transitional probabilities
are most critical. Revisions to the Weibull distributions are less straight-
forward and if necessary can best be done by summarizing a relatively long-
term period of observed rainfall (daily, hourly, or five-minute) and compar-
ing with corresponding simulated values.
44
-------
C. CONTINUOUS DAILY STREAMFLOW MODEL COMPONENT
This model component performs the basic moisture accounting for
TVA-HYSIM. A daily time unit is used because of the ready availability of
daily rainfall and streamflow data for calibration and validation and
because this is about the longest time scale at which continuous moisture
accounting can be performed and the individual processes considered. The
model is fairly well documented (TVA, 1972; Betson, 1976). Figure 6 shows a
schematic of the model. This component model has features for handling
urban conditions (including the washoff of dust and dirt as described in the
subsequent section on sediment), the effects of under-drainage in soluble
carbonate rock terrain on streamflow, and a variety of land covers.
Although the information is not needed to operate TVA-HYSIM, model
algorithms are documented in this section so that should problems in operat-
ing the model be encountered, or should it become desirable to modify this
model component in other regions, the formulations and the subroutines in
which the algorithms occur can be more easily identified. The format of
this documentation loosely follows the schematic in Figure 6.
1. Interception Storage. This is the first of the model compartments and
is deterministic. Forest and non-forest interception is handled separately.
For forested areas the following interception equations for 20-year old
loblolly pine and hardwoods are based upon the work of Swank et al (1972)
and Helvey and Patric (1965), respectively:
Loblolly pine (20-years old) I = 0.02 + 0.12RF V-13
Hardwood (growing season) I = 0.04 + 0.06RF V-14
Hardwood (dormant season) I = 0.02 + 0.024RF V-15
where: I is the interception in inches, and
RF is the storm rainfall in inches.
For non-forested areas an interception capacity of 0.05 inch per
day is used. During winter periods when rainfall is sustained over many
days and evapotranspiration is low this interception compartment could
become unrealistically large. Therefore, there is a limit on the amount of
intercepted water that can be held during the dormant season and is as
follows: pine (0.3 inch), hardwood (0.08 inch), non-forest (0.05 inch).
During the growing seasons these limits are each increased by 0.12 inch.
Evaporation of intercepted rainfall occurs at a faster rate than
the computed evapotranspiration rate (described subsequently in Section
V.C.10). As described by Betson (1979, p. 58 ff), this additional evapora-
tion is necessary to account for the higher losses experienced in forests,
most noticeably in pine forests. The literature indicates that evaporation
of intercepted water occurs at a rate some three times the transpiration
rate (Stewart 1977; Singh and Szeicz, 1979). The variable HTFLUX (read in
question 13, Section II) is the ratio at which the evaporation of inter-
cepted water occurs in relation to the computed transpiration rate. The
interception calculations are all made in subroutine MODEL.
45
-------
o
O)
o
CO
i_
3
O>
MCTUIMV3H.LS
46
-------
2. Storm Runoff Volumes (a) Impervious Areas. An algorithm developed by
Miller and Veissman (1972) is used for predicting daily storm runoff (precip-
itation excess) from urban areas. The algorithm, in subroutine MODEL,
determines daily precipitation excess based upon the portion of a watershed
which is in excess of 17 percent impervious:
where:
PE. = 1.165 RF
(IMPERVIOUS - 17)/100
and (IMPERVIOUS - 17) ^ 0
V-16
PE . is precipitation excess in inches from impervious
surfaces ,
RF is daily rainfall less interception in inches, and
r IMPERVIOUS is the percent of the watershed with
surfaces .
impervious
The algorithm is based upon the observation (by Miller and
Viessman, 1972) that when less than 17 percent of a watershed is impervious
the water tends to drain onto pervious surfaces and the "urban effect" is
insignificant. This 17 percent minimum does not apply if there are areas
directly connected to a stream through sewers or culverts. Thus, if a value
is entered in response to interactive question 6 (Section II) that value
will be used to compute impervious area precipitation excess. Question 5
should be answered "yes"; however, only when the total impervious area is
less than about 17 percent and there are directly connected areas.
3. Storm Runoff Volumes (b) Pervious Areas. The algorithm in subroutine
MODEL for determining daily precipitation excess from pervious areas is an
adaptation of a rational model presented by Betson et al (1969). The
algorithm is based upon the assumption that the yield of precipitation
excess is proportional to the amount of moisture stored in the system.
where:
RI = AW + (DS - AW) SI) e
-B(SMR + GWR)
PE =
P
(RF
RI2)
0-5 _
RI
V-17
V-18
RI is a retention index in inches,
AW is a model parameter associated with winter storms,
DS is a model parameter associated with summer storms,
B is a model parameter associated with runoff volumes in
1.0/inches,
SI is a phenologic index that equals one in summer and zero
in winter, with interpolated values for spring and fall,
SMR is the moisture in inches stored in the soil moisture
reservoir in inches,
GWR is potential runoff stored in the ground water reservoir
in inches, and
PE is the pervious area precipitation excess in inches.
47
-------
The phenologic index SI is used in several of the model compart-
ments to allow seasonal variation. It is determined in subroutine SEASON
and the beginning of each season is defined as: Fall = 15 (October 15);
Winter = 92; Spring = 183; Summer = 212. Should it become necessary to
alter this phasing, a change must be made in program DATSET (which controls
the interactive input). The season is defined in a write statement (FORMAT
190) just prior to the question which asks for the mean of the cube root of
monthly rainfall.
4. Ground Water Runoff Volumes. After interception and precipitation
excess have been removed the residual storm rainfall becomes a potential for
ground water runoff. These computations are made in subroutine MODEL. The
portion of the residual storm rainfall that will become ground water is
proportional to the yield of daily precipitation excess from pervious areas:
GWV = (PE /RF )GWK-RF. V-19
and GWV ^ RF.
i
where: GWV is a volume to be added to the ground water reservoir
(GWR) in inches,
RF is the daily rainfall less interception,
GWK is a model parameter, and
RF. is the available moisture after interception and precipita-
tion excess have been removed.
5. Dormant Season Recharge. For watersheds with high soil water holding
capacities, such as in clay and loam soils, a recharge to the ground water
reservoir can occur as vegetation becomes dormant in the fall. During the
fall period, moisture held in the soil under tension by the vegetation is
released as the vegetation becomes dormant. This moisture is added to the
ground water. These accretions are taken from the soil moisture reservoir
in subroutine MODEL (when sufficient moisture is available) at a daily rate
according to the parameter GWDOR, in inches, and added to the ground water
reservoir. This feature has effects only minimum flow simulations when
using TVA-HYSIM.
6. Soil A Horizon Moisture Storage Capacity. Soils that have shallow A
horizons and/or have low permeability rates in the B horizon will experience
relatively large volumes of precipitation excess once the storage capacity
of the upper soil horizon is exceeded. This heavy runoff will occur even
though the total moisture stored in the system may be low, which according
to equations V-17 and V-18 should produce low volumes of precipitation
excess. Therefore, two parameters, the A horizon depth (AHORD), and the B
horizon permeability (BHORP) are defined which provide a limit above which
all excess moisture to be allocated to the soil moisture reservoir becomes
precipitation excess. The sum of the two parameters is this limit with
recovery storage capacity in AHORD occurring at a daily rate of BHORP. Both
parameters are preset to 1.5 inches in subroutine CHAR for TVA-HYSIM.
48
-------
7. Potential Runoff Volume Losses (a) Bypass Seepage. Losses of potential
runoff can occur for a variety of reasons. In areas with deep alluvial
soils, much of the ground water and some of the potential surface water
runoff may occur beneath the surface. In areas underlain by soluble
carbonate rocks this bypass drainage can become significant and must be
accounted for to keep the simulations reasonable.
Bypass losses are the losses of potential ground water runoff that
bypass the site where simulations are being made, for whatever reason.
These losses are often significant in carbonate rock terrain. These bypass
or deep losses are removed in subroutine MODEL directly from the potential
ground water runoff, in proportion to a parameter (DLF in question 8,
Section II) and are not considered further in the simulations:
GWL = GWV DLF V-20
where: GWL is a deep, or bypass loss, in inches,
GWV is the ground water volume computed in equation V-19, and
DLF is value read-in in question 8 (Section II).
When DLF equals zero, no losses occur (GWL = 0) and when DLF
equals one no ground water runoff occurs. In the absence of any site infor-
mation or data, DLF is difficult to determine and probably should be set to
zero. In carbonate rock areas it has been set equal to the fraction of the
watershed underlain by very soluble carbonate rocks.
8. Potential Runoff Volume Losses (b) Transmission Losses. Transmission
losses occur when potential precipitation excess originating from impervious
areas does not reach the simulation site. These losses are most pronounced
during smaller storms when precipitation excess originating from driveways,
roofs, etc., infiltrates into lawns, pervious surfaces or dry tributary
channels. These losses decrease as the storm size increases until at some
value of precipitation, all of the potential precipitation excess will reach
the simulation site. The impervious area precipitation excess, as calcu-
lated in eq. V-16, is adjusted for transmission losses using the following
algorithm:
PE. = (PE./TLP) PE. = PE.2/TLP V-21
and (PE./TLP) g 1.0
where: PE. is impervious area precipitation excess in inches, and
TLP is a transmission loss parameter preset to 1.0 inch.
This adjustment occurs in subroutine MODEL where the parameter is
redimensioned to remove the effect of the size of impervious area. The
value for parameter TLP is preset, to 1.0 inch in program DATSET for applica-
tions with TVA-HYSIM. Transmission losses are not lost from the system but
become part of the residual rainfall used in the pervious area storm runoff
volume calculations.
49
-------
9. Potential Runoff Volume Losses (c) Pervious Area Runoff Losses. In
carbonate rock areas, in particular, the potential precipitation excess that
should be realized from pervious areas based on the physical watershed
characteristics and antecedent moisture conditions, often does not material-
ize. This potential runoff may be all or partially lost into sinkholes
through pervious channel seeps or through rapid percolation into near-
surface channels in the soluble carbonate rock. It is important to
recognize this phenomenon, where it exists, since it can affect all of the
simulations. Where these pervious area losses occur, not only must storm
precipitation excess be reduced, but this loss must be cascaded along with
the remaining residual rainfall into subsequent compartments, or reservoirs,
since it must be accounted for in the moisture budgeting. The algorithm
used to modify PE (as calculated in eq. V-18) to account for losses due to
soluble carbonate rock systems is:
PE = PE (1-PKARST) V-22
P P
where: PE is the pervious area precipitation excess from equation
V-18 in inches, and
PKARST is a read-in value from question 8 (Section II).
PKARST is defined so that a zero value results in no loss while a
value of one results in no precipitation excess (unless the A horizon limit
described in Section V.C.6 is exceeded). Values for PKARST are estimated
based upon the fraction of the watershed area underlain by soluble carbonate
rock as described in Section III. A. 4. These adjustments are made in sub-
routine MODEL.
10. Evapotranspiration . Monthly values of evapotranspiration are used in
the model. Adjustments are made to the mean monthly evapotranspiration
index read-in in question 18 (Section II) based on seasonal growth index
(GI) relationships. The total evapotranspiration thus computed is then
forced to total either the long-term loss predicted from the latitude
(question 13, Section II) or optionally from long-term mean rainfall minus
expected runoff. The algorithm used is:
12
LOSS = DK I (PET. GI.) V-23
where: LOSS is the long-term annual evapotranspiration in inches,
DK is a factor used to equate the accumulated products of
PET-GI to the expected long-term loss,
PET is long-term monthly potential evapotranspiration as
measured, for example, by a land pan, in inches, and
GI is a crop growth index - a ratio of monthly evapotranspira-
tion to potential evapotranspiration.
This algorithm forces the measure of potential evapotranspiration
used to have a monthly distribution based upon cover present in the water-
shed, and for the annual total to be equal to the expected long-term annual
loss. The monthly growth index values (Holtan and Lopez, 1973) are con-
50
-------
tained in data definition statements in subroutine ETYEAR where the monthly
values of evapotranspiration are calculated. Computed daily evapotranspira-
tion first depletes the intercepted rainfall (see Section V.C.I) and then
the soil moisture reservoir.
An adjustment to the expected long-term annual evaporation (LOSS)
is made when unvegetated areas are present. This adjustment is patterned
after a relationship devised by Douglass and Swank (1972) which related an
increase in runoff to the reduction in forest stand:
LOSS = LOSS + 1.39 - 0.13 (UNVEGETATED) V-24
and: UNVEGETATED § 10.7 percent
where: UNVEGETATED is the percent of the watershed unvegetated.
11. Runoff Routing. Although not relevant to the operation of TVA-HYSIM in
which storm precipitation computed by the continuous streamflow model is
routed in the storm hydrograph model, the continuous daily streamflow model
also routes runoff on a daily basis. Daily streamflow simulations can,
therefore, be obtained as explained in Section IV.A.
Precipitation excess originating from impervious areas (PE.) is
assumed to become streamflow on the day of the rain. Precipitation excess
originating from pervious surfaces (PE ) is routed on a daily basis as
follows:
SRO. = TDSRO-(PE ). + SUKES. (1-SROK) V-25
where: SRO is routed precipitation excess in inches,
TDSRO is a parameter indicating that portion of the PE that
becomes runoff on the day of the rain,
SURES is the storm runoff reservoir in inches, and
SROK is a storm runoff recession parameter.
On a day of rainfall, that portion of the precipitation excess
that does not become runoff (1-TDSRO) is allocated to the reservoir SURES,
where it runs off on subsequent days. No distinction is made between
surface runoff and interflow.
Ground water runoff (GRO) is routed daily from the ground water
reservoir (GWR) using a recession constant (GROK):
GRO = GWR (1-GROK) V-26
There are provisions in the model for separate summer and winter
ground water recession constants, GROKS and GROKW, respectively. All rout-
ing is done in subroutine MODEL.
12. Regionalized Model Parameter Prediction Equations Table 1 lists the
watersheds involved in developing regionalized relationships for the contin-
uous daily streamflow model (total of 28 watersheds). An analytic version
51
-------
of the model was first adjusted to three or four years of continuous daily
rainfall and streamflow data using an optimization technique (TVA, 1972).
Next, relationships between the optimized model parameters and watershed
characteristics were developed. These relationships are shown in Table 4.
(All necessary coefficients for deterministic algorithms were presented
earlier and certain of the model quasi-parameters that must be estimated
were described.)
Three additional parameters have values internally set in the
program: AHORD (1.5 inches), BHORP (1.5 inches) and TLP (1.0 inch) as
explained in previous sections.
The model parameters are predicted in subroutine CHAR using these
regionalized relationships. The last two equations in Table 4 are used to
predict initial values for the soil moisture reservoir (BSMI) and the ground
water reservoir (BGWR), respectively. The model begins calculations on July
1 with a three-month lead-in used to further adjust these reservoirs before
simulations begin on October 1.
D. STORM HYDROGRAPH MODEL COMPONENT
Storm rainfall and runoff are determined in the continuous daily
streamflow model. Storm periods (one or two days) above threshold criteria
(question 15, Section II) are identified in subroutine MODEL and the basic
storm hydrograph model component, subroutine STORM, is called. Here, the
total storm rainfall and precipitation excess determined by the continuous
daily streamflow model is distributed into shorter time intervals, the unit
hydrograph parameters are predicted, the storm precipitation excess and unit
hydrograph are convoluted, and the storm sediment computations are made.
The description of this component is organized roughly in the sequence in
which these computations are performed. (The sediment-associated calcula-
tions are described in sub-section E.)
1. Precipitation Excess Distribution For each storm identified in the
continuous daily streamflow model for storm hydrograph simulation, the total
rainfall is distributed into hourly (subroutine HOURF) or shorter time
periods (subroutine FIVMIN) depending on the convolution interval, DT, read
in question 12 (Section II). A precipitation excess distribution is
computed using this rainfall distribution and a modification of the Soil
Conservation Service (SCS 1972, 1975) method using a constant loss parameter
PHI. This SCS distribution technique reduces to:
SRO. = (ARF. - 0.2S)2/(ARF. + 0.8S) V-38
where: SRO. is the accumulated storm precipitation excess at time j
in inches,
ARF. is the accumulated storm rainfall in inches, and
S is the maximum potential retention which is related to a
SCS curve number, CN-PE, which is defined as:
CN-PE = 1000/(10+S).
52
-------
TABLE 4
TVA CONTINUOUS DAILY STREAMFLOW MODEL
Parameter
AW
DS
B
GWK
GWDOR
TDSRO
SROK
GROKS
GROKW
'IONALIZED PARAMETER PREDICTION EQUATIONS
Prediction Equation
AW = 9000
DS = 3400
B = 8.2
GWK =0.28
GWDOR = 0.
5 68
Y g 200
3 92
. Y g 200
3 3
Y V-29
- AWC (FOR)
00003 AWC2'6
Equation No.
V-27
V-28
V-30
V-31
TDSRO = 0.75 - 0.31 (log DA) -
0.069 (log DA)2
arid TDSRO g 1.0
SROK=e-1-6l/DA
0.2
^ 1.0
GROKS =1-4.70 AWC~2'485 g 1.0
-1 48
GROKW = 1 - 0.78 AWC ^ 1.0
V-32
V-33
V-34
V-35
BSMI =1.04 AWC - 0.0343 AWC2
BGWR = 0.0023 AWC2'9
V-36
V-37
where: Y is long-term annual yield (runoff/rainfall) (question 13,
Section II),
AWC is the available water holding capacity of the soil in
inches (question 7, Section II),
FOR is a measure of the portion of watershed area covered by
forest [(hardwood + conifer)/100] + 1.0 (question 4,
Section II), and
DA is the drainage area in mi2 (question 4, Section II).
53
-------
The SCS technique is then modified to increase losses during lulls
in multi-burst storms with a constant loss variable PHI. This variable is
subtracted from each RF. value prior to accumulation to obtain a new ARF.
PHI is allowed to vary from storm to storm and is calculated in subroutine
SCHAR1 using the following equation:
PHI = 0.0567 - 0.0003 FOR - 0.0971 SINU - 0.015 PE
+ 0.0332 PKARST + 0.0314 (ARF - PE)
+ 0.024 (ARF/DUR) V-39
where: FOR is the fraction of the watershed covered by forest plus
one,
SINU is the sinuousity measure read-in in question 7 (Section
ID,
PE is the storm precipitation excess,
PKARST is the measure read-in in question 8 (Section II), and
DUR is the storm duration in hours.
The revised ARF calculation is made as follows:
n
ARF. = I RF. - PHI V-40
J j = l J
subject to PHI § RF.
The distribution of precipitation excess over time is determined
from equation V-38 by differencing:
PE. = SRO. - SRO. n V-41
J J J-l
2. Storm Burst Definition The unit hydrograph, far from being a fixed
characteristic as visualized by Sherman (1932), is highly variable from
storm to storm and from burst to burst within storms, particularly for
smaller watersheds. As will be described in the next two sub-sections, the
unit hydrograph function used in the storm hydrograph component of this
model varies according to the precipitation excess distribution. When storm
rainfall is intermittent, as is common in the summer, a point is reached
(depending upon the size of the watershed involved) where these intermittent
precipitation excess distributions must be subdivided into bursts which can
be individually characterized more meaningfully for the unit hydrograph
relationships. Two definitions are needed, a lower cutoff and a defined
lull. The lower cutoff is needed to cover the situation where rainfall is
sustained but low for a period such that an insignificant amount of
sustained precipitation excess is calculated. The cutoff is a limit of
precipitation excess (for each convolution interval DT) below which a lull
may be defined:
54
-------
CO = 0.01 SRO DT V-42
subject to CO g 0.01 inch.
A lull is the minimum duration during which PE. must be below
the computed CO in order for a burst to be designated. Bursts are calcu-
lated in subroutine STORM where a lull is defined in convolution interval
units, as
LULL = (2.5/DT) DA°'2 V-43
where: DA is the drainage area in mi2.
There are no constraints on the number of bursts that may be
identified during a storm.
3. The Unit Hydrograph TVA-HYSIM uses a unit hydrograph to distribute the
precipitation excess across the storm hydrograph. This unit hydrograph is
represented with the double-triangle function depicted in Figure 7. The
rationale for this functional representation was presented by TVA (1973b)
and Betson (1976). In essence, it was developed from the concept of partial-
watershed area contributions to storm runoff.
The symbols used in Figure 7 are:
UP is the peak ordinate of the initial response triangle,
generally the unit hydrograph peak, in inches per hour,
Tl is the time of the initial peak UP in hours,
UR is the ordinate of the recession inflection point (infre-
quently the maximum peak) in inches per hour,
T2 is the time to the recession inflection point in hours,
T3 is the time base of the unit hydrograph in hours, and
TL is time-lag defined as the time from the occurrence of
\ DT unit to the centroid of the unit hydrograph.
The five parameters UP, UR, Tl, T2, and T3 completely describe the
double-triangle unit hydrograph.
4. Regionalized Unit Hydrograph Prediction Relationships A time-lag con-
cept is used as the basis for predicting the unit hydrograph parameters.
The definition of time-lag employed shown on Figure 7 is that developed by
Overton (1970) which is essentially the time between the occurrence of 50
percent of the precipitation excess volume and 50 percent of the storm
hydrograph volume.
55
-------
w TIME, hours
T50
TIME, hours
Figure 7: Double Triangle Unit
Hydrograph and Lag Time Definition
56
-------
In unit hydrograph terms, if 50 percent of the unit area occurs
before Tl (see Figure 7), the time to 50 percent of the unit area T50 is:
T50 = (Tl/UP)^ V-44
If T50 is greater than T2 the relationship is:
T50 = T3 - [(T3-T2)/UR]^ V-45
And, if T50 lies between Tl and T2, as in the usual case, the
relationship is
T50 = [UP-(UP2-[1-M]-[BB/CC])^]-[CC/BB] + Tl V-46
where: AA = UP-T1,
BB = UP-UR, and
CC = T2-T1
Using an adaptation of the method proposed by Troxler (1978),
Bales (1979) developed a measure of the intensity of the burst precipitation
excess distribution as:
PEIN = [ I (PE.2)/IPE]/(IPE/RF)
= J
n 2
= I (PE.) RF/(IPE)2 V-47
where: PEIN is a normalized precipitation excess intensity in inches
per hour,
PE . is the precipitation excess at time j ,
IPE is the total storm precipitation excess, and
RF is the total storm rainfall.
This measure of precipitation excess intensity, normalized by dividing by
the yield, provides a measure of intensity that considers antecedent condi-
tions (yield) .
Based on work by Overton (1967, 1968, and 1971), the normalized
precipitation excess intensity is used in the prediction of a lag-time
unique to each watershed for each storm. The equations are of the form
CD
TC = S(J (PEIN) V-48
and TL = TC/1.6 V-49
57
-------
where: TL is time lag in hours,
TC is time of concentration in hours, and
S|J and SB are coefficients to be predicted.
The predicted time-lag for the storm is next used in subroutine
SCHAR1 in relationships to predict the parameters UP and T2. The relation-
ship for UP is:
UP = SC-TLSE V-50
where: UP is the initial unit hydrograph peak in inches/hour, and
SC, SE are coefficients to be predicted.
Similarly, the relationship for T2 is:
T2 = SF-TLSG V-51
where: T2 is the time to recession inflection, in hours, and
SF, SG are coefficients to be predicted.
Based upon the fact that geomorphic thresholds govern channel
formation (Schumm, 1973), separate equations are used to predict the coef-
ficients in Equations V-48, V-50 and V-51 for small basins and for large
basins. Table 5 shows the equations used to predict these coefficients for
basins less than or equal to two square miles along with the equations used
for larger basins.
The basin characteristics in Table 5 are defined as follows:
DA = drainage area (question 4),
CN = SCS curve number (question 7),
PCM = an internally defined measure of percent mined
(question 7) defined as: (% mined/100) + 1.0,
PERM = soil permeability, inches per hour (question 7),
CSLOPE = channel slope, feet per mile (question 7),
DD = drainage density, mi/mi2 (question 7),
AWC = available water holding capacity, inches (question 7),
SHAPE = watershed shape (question 7),
FOR = an internally defined measure based upon information
in question 4:
FOR = [(HARDWOOD + CONIFER)/I00] + 1.0, and
SINU = a measure of sinuosity (question 7).
58
-------
co
H
55
W
[- _j
CLJ
HH
f^H
fi_l
w
0
CJ
O
^
i i
H
CJ
ti
Q
m c2
P-I
w
/₯3 f-\
M-J ^M'
<3 f1-!
H
00
£5
0
1 1
H
£-3
O*
w
Q
W
N
h- 1
t-1
55
O
HH
O
rVI
PH
CM
E-i
Q
S
^*
*.
PH
*v
(»_3
H
PH
CO
55
O
r 1
H
£D
0-
w
Jg^
I-H
d
o
H
4-1
CO
3
O1
W
u
r-t
4_J
cn
H
£-{
0)
CJ
co
cO
f*
CJ
d
rH
M
CO
PQ
d
o
4-)
d
CD
d
0
X
w
1
d
o
CJ
1
14-1
OJ
o
CJ
1
o
12!
J"~l
Jz
H~l
w
PH
0
PH
W
PH
»T*
CO
O
D£
<£
Q
£ 3
W
P-i
0
CO
o
5"J
PH
w
PH
*5~]
CJ
PH
25
CJ
o>
xHf
O
o
in
a^
r-i
o
m
CO
.
i
o
vj-
1 1
t
o
CM
0
00
1 1
o
CO
00 OS
m m
i i
M-* £*
00
CO
~*
o
m
CO
OS
o
O i i
in CM
I-H Ln
. .
O 1
CO
0)
1 1
r-j
E
OJ
i-l
CO
3
CJ1
to
O OO CM
* in r^
4-1 rH O
d i o
CO
-d
4-)
i-i r-
0) in
4-> O
cfl
£> £>
O^ O
vH/
CM
CM
1
O rH
o m
i d
CO
o
m
*x
oo
>H^
w
d
o
H
4_>
CO
3
&
W
d
rH -H
i 1
r-» T3
0)
1 W
3
D
CO
CO
4_j
d
^
M
H
B
o
CM
A
<£
Q
j-i
O
*4H
~T
CO
»
0)
1 1
PH
S
CO
CU
^
O
PH
00
co
-------
The equations for the four coefficients used to predict time-lag
(V-52 and V-53, V-58 and V-59) are shown in subroutine SCHARl with a
variable UGMOD(I), which is set to 1.0 in each case, in a data definition
statement. As explained in Section IV.D.2, this variable may be used to
modify the regionalized relationships, if necessary.
The equation for T3 is as follows for all drainage areas:
T3 = 6.026TL°-818DA0'113 V-64
To complete the solution of the double-triangle unit hydrograph Tl
and UR are determined solving relationships derived from the geometry of the
double triangle. Depending upon the value of T50 relative to Tl and T2,
equation V-44, V-45, or V-46 is used to either solve for Tl directly
(equation V-44) or UR directly (equation V-45) or Tl in terms of UR
(equation V-46). The following equation, which forces the unit hydrograph
to have an area of 1.0, is then used to solve for the remaining measure:
UR = [2-(UP«T2)]/(T3-Tl) V-65
All of the above calculations are performed in subroutine SCHARl.
Because of the way these regionalized relationships are formu-
lated, modifications for other locations (regions) and other land-use condi-
tions can be made rather easily should these regionalized relationships
prove to be inadequate. Probably all that would need to be revised are the
equations used to predict the coefficients in the time-lag equation V-48
(equations V-52, V-53, V-58, V-59). This can be done by changing appropri-
ate values for the variable UGMOD. These changes can be developed for a
particular land-use condition by analyzing several storm hydrographs to
develop a relationship between TL and PEIN. Revisions to the regionalized
relationships to handle a variety of land-uses is, of course, more involved.
Convolution of the predicted unit hydrograph with the precipi-
tation excess distribution to compute a storm hydrograph is accomplished in
subroutine STORM. The storm hydrograph is added to base flow computed from
the ground water predicted in the continuous daily streamflow model. Inter-
polations between the daily values of predicted ground water are made so
that simulated storm hydrograph, as portrayed in the print-plot shown in
Figure 3, do not usually begin or end at zero flow.
E. SUSPENDED SEDIMENT MODEL
TVA-HYSIM has provisions for simulating the washoff of dust and
dirt accumulated on impervious surfaces and the sediment eroded from
pervious areas. Standard formulations are used; thus there are no model
parameters, as such, in this model component.
1. Impervious Area Dust and Dirt. Dust and dirt accumulations on
impervious surfaces are accounted for in the continuous daily streamflow
60
-------
model component according to an algorithm presented by Donigan and Crawford
(1976).
SEDDAYt = SEDDAYt_1-(l-R) + ACCIA V-66
where: SEDDAY is the dust and dirt on impervious surfaces at time t
in pounds/acre,
R is the fraction removed daily due to traffic, wind, or
sweeping, and
ACCIA is the dust and dirt accumulation rate, Ib/ac/day.
This equation is used to accumulate dust and dirt every day there
is no rain. The accumulation reaches a limit of:
SEDDAYt = ACCIA/R V-67
which occurs in 1/R days. The limit is based upon the assumption that the
accumulated rate will diminish as the amount builds up simply due to the
effects of wind and traffic.
On days when rainfall occurs a portion of the accumulated dust and
dirt will wash off. The algorithm that is used was devised by Metcalf and
Eddy (1971, p. 178) for the EPA SWM model, wherein the pounds of pollutant
(WASH) washed off during time interval, dt, is proportional to the pounds on
the impervious surfaces (SEDDAY) at the beginning of the period.
^dlWASH) = -dCSEDDAY) = M . SEDMy v_6g
dt dt
which integrates to:
MI-
WASH = SEDDAY (1-e ) V-69
The relationship used in TVA-HYSIM is similar to that used in the
SWM model in that it is assumed that one-half inch of runoff (depth on the
impervious surfaces) in one day will wash away 90 percent of the dust and
dirt accumulated on impervious surfaces. This leads to the algorithm for
daily washoff during days of rain. (Time, by definition, in this case is
one since the time unit is one-day.)
SEDDAY - WASH = SEDDAY (l-e~4'6'PEi*t) V-70
This accounting is performed in subroutine MODEL. The washoff
load is transferred to subroutine SEDMEN where it is routed across a storm
hydrograph.
2. Pervious Area Storm Sediment. The approach used to simulate suspended
sediment loads from pervious areas is the modified Universal Soil Loss
Equation devised by Williams (1975).
TONS = 95 (53.33DA-PE -QP)'56 LSKCP V-71
61
-------
where: TONS is the storm suspended sediment load in tons,
DA is the drainage area in mi2 (the product 53.33DA converts
PE in inches to acre-feet),
PE Is the pervious area precipitation excess volume in
watershed inches,
QP is the storm hydrograph peak discharge in cfs,
LS is a length-slope factor,
K is a soil erodibility factor,
C is a crop management factor, and
P is an erosion control practice factor.
Equation V-71 is used to calculate the suspended sediment load
from pervious areas for the storm. These calculations are performed in sub-
routine SEDMEN. The latter four terms in equation V-71 are read-in with
question 10 (Section II). The LS term is usually defined only for small
fields. When determined for the entire watershed the approach proposed by
Williams and Berndt (1976) should be used.
3. Sediment Routing Separate approaches are used for routing the dust and
dirt washed off impervious surfaces and the pervious area sediment load.
The washed off dust and dirt is distributed across the impervious area
precipitation excess vector PE. using equation V-70, which has the effect of
allocating much of the washoif early in the storm (the well-known first-
flush effect), yet in proportion to the magnitude of the precipitation
excess occurring in each time interval. Routing of the impervious-area
washoff load across the storm hydrograph is accomplished in subroutine
SEDMEN using an adaptation of Williams' (1978) instantaneous unit sediment
graph (IUSG):
IUSG = (DT/ITPS) g'WB'DT'VD v_72
where: DT is time in DT units, question 12 (Section II),
ITPS is a factor used to delay the peak of the IUSG (defaulted
to 1.0 in TVA-HYSIM),
WB is an internal variable determined from a relationship
among the maximum precipitation excess value, the peak
discharge, and the time to peak, and
D is a ratio of mean suspended sediment grain size to one-
micron, |J, question 11 (Section II).
The adaptation of equation V-72 for impervious areas involves the
assumption that 90 percent of the washoff will occur by the time T3, the
base of the storm unit hydrograph. This assumption redefines WB and yields
the following algorithm for an impervious area unit sediment graph (IUSGI).
IUSGI = (DT/ITPS)e-2-3°3DT^D/T3 V-73
This impervious area unit sediment graph is multiplied by the
storm runoff unit hydrograph and the product forced to unity. The resultant
impervious area unit sediment graph is then convoluted with the computed
incremental dust and dirt loads to determine a distribution of dust and dirt
washoff across the storm hydrograph.
62
-------
The pervious area IUSG, determined using equation V-72, is multi-
plied by the storm runoff unit hydrograph (and the product forced to unity)
to provide a pervious area unit sediment graph. This pervioizs area unit
sediment graph is then convoluted with the incremental pervious area precipi-
tation excess values squared. The sum of these convoluted values is then
divided by the sediment load computed using equation V-71 to determine a
ratio which when multiplied by the distribution will provide a correct
total. A rational development for this technique is presented by Williams
(1978).
Baseflow sediment is also determined. Using the average concen-
tration read-in in question 9 (Section II) corresponding with a discharge of
one cubic foot per second per square mile (PPM1), loads are computed for
ground water simulated by the continuous daily streamflow model. The
algorithm used in subroutine STORM to compute baseflow sediment concentra-
tions is:
CONC = PPM1 (BFQ/DA)1'5 V-74
where: CONC is the suspended sediment concentration associated with
baseflow in mg/1,
PPM1 is a read in concentration for one cfs/mi2, question 9,
(Section II),
BFQ is the simulated baseflow (ground water) discharge in cfs,
and
DA is the drainage area in mi2.
The suspended sediment loads determined for each DT interval
(baseflow plus impervious area washoff plus pervious area loads) are shown
to the right of the storm hydrograph print-plot (see Figure 3). The corres-
ponding concentrations are also shown. The interactive output shown in
Figure 2 tabulates the total pervious and impervious area sediment load.
(The baseflow load during storm periods is negligible.) The average storm
concentration and the maximum simulated concentration are also shown for
each storm.
63
-------
SECTION VI
COMPUTER REQUIREMENTS
A. GENERAL COMPUTER REQUIREMENTS
Because program TVA-HYSIM is interactive, because the input data
process is simplified, and because the interactions among model components
necessitates use of storage files, a number of input/output devices are
required. The job control statements needed for every computer system on
which the program TVA-HYSIM might be used are impossible to anticipate.
However, to assist users, the job control statements used to operate the
model on the TVA system are provided. These statements are provided as a
guide for use in adapting TVA-HYSIM to other systems. This section also
describes the general computer requirements for the model.
As noted in the introduction, TVA-HYSIM is a version of a complex
model that has a number of options unavailable in TVA-HYSIM. These options
were omitted because they have limited application in land-use planning
studies and they do make the model more complicated to use. In the descrip-
tions of some of the material in this chapter, a few of these unavailable
options are encountered.
The TVA-HYSIM is composed of two programs which are written in
FORTRAN IV. The programs are compiled on the TVA system using IBM Gl or H
Extended compilers. They are being run on TVA Amdahl 470/V6-II computer
with a MVS JES3 IBM batch system and IBM OS/VS2 TSO interactive system.
Figures 8 and 9 are flowcharts showing how the programs fit together for
interactive and batch runs, respectively.
The first program (DATSET) drives the interactive feature. DATSET
requests the input data and creates the formatted data file for input to the
second program (RUNOFF). Program DATSET accepts list-directed read state-
ments (free formatted) which the Gl and H extended compilers will handle.
On TVA's computer system DATSET uses 50K bytes of storage and \ CPU second.
The program variables are in single precision. This program uses the fol-
lowing data set reference numbers which specify the input/output devices:
5 - Reads input from terminal or cards,
6 - Outputs to terminal or printer,
10 - Outputs to RUNOFF input data file.
The second program (RUNOFF) is a complex program which performs
all of the hydrological simulations. RUNOFF receives input data in
formatted form from the data file created by the program DATSET. On TVA's
64
-------
o
JZ
o
?
2
u_
o
o
CL
(U
o
o
00
L.
65
-------
o
_c
o
0>
u
o
CD
0>
0>
k_
3
66
-------
computer system, RUNOFF uses 230K bytes of storage and about \\ CPU second
for each year simulated (3 to 4 storms per year). The variables in this
program are in double precision. Table 6 gives a description of all sub-
routines in program RUNOFF. Table 7 describes each function used. These
functions must be provided by the user or be included in the computer system
software.
Program RUNOFF uses the following data set reference numbers for
specifying input/output devices:
1 - Outputs the storm hydrograph model results to printer or disk
file.
2 - A temporary working file to store the storm hydrograph input data.
This file is fixed length, 80 bytes per record. About 16K bytes
are needed for this file.
3 - Outputs the daily flow model results to the printer or disk file.
4 - Outputs the observed and simulated discharge from the daily and
storm models to a tape for plotter use later (option not available
in TVA-HYSIM).
5 - Reads input data from the RUNOFF input data file or from cards.
6 - Outputs interactive output to the terminal or printer, also error
messages are written to this data set reference number.
7 - Punches simulated daily streamflow values to cards for use with
other programs, (option not available in TVA-HYSIM).
8 - A temporary working file. This file consists of fixed length, 24
bytes per record. About 48 bytes are needed.
B. RUNOFF INPUT DATA FILE
On the TVA system, an on-line disk is used to store the RUNOFF
input data file created by DATSET. This file is a partitioned data set with
input data for a single watershed being a member of the file. A sequential
file can be used on disk or tape. The file has the following attributes:
(1) record format (RECFM) of Fixed length and Blocked (FB), (2) record
length (LRECL) of 90 bytes, and (3) block size (BLKSIZE) of about a % of a
track on a 3330 disk.
C. TSO COMMANDS FOR INTERACTIVE RUNS
This section describes the TSO command procedure (CLIST) used to
run the interactive form of HYSIM. Figure 10 is the CLIST used in TVA to
run HYSIM. Due to the various types of interactive systems not all can be
described. The TVA CLIST can be used, however, as a guide to set up the
CLIST for another computer system. The following describes lines of the
CLIST that may be need more explanation.
CLIST
Line # Description
20 This suppresses TSO informational mes-
67
-------
TABLE 6
PROGRAM RUNOFF SUBROUTINES AND DESCRIPTIONS
Subroutine
Main
MODEL
CHAR
SEASON
ETYEAR
PRINT
QUDAT
RFSIM
GAUSS
URAND
SMODEL
Description
Driver for the subroutines. In addi-
tion, it reads in operational controls
for a variety of options not available
in TVA-HYSIM.
Main subroutine which operates the
continuous daily streamflow model
(DFM).*
Calculates daily flow model parameters
from basin characteristics (DFM).
Calculates the daily season variable
for the four seasons (DFM).
Calculates daily evapotranspiration
from land cover and GI curves (DFM).
Prints optional daily rainfall and
simulated flow by months (DFM).
Simulates daily water quality infor-
mation (option not available in TVA-
HYSIM) .
Generates daily rainfall (DRFG) from
the mean of the cube roots of ob-
served monthly rainfall and the
standard deviation of this distribu-
tion.
Generates a normally distributed
variable from the mean and standard
deviation of cube root of monthly
rainfall (DRFG).
Generates random numbers with a
uniform probability distribution
(DRFG, HRFG, FRFG).
Drives the storm hydrograph model
(SHM) when used by itself (option not
available in TVA-HYSIM).
^Denotes the model component where subroutine is usedsee end of table.
68
-------
TABLE 6 (continued)
STORMD
SCHAR1
STORM
RUNOFF
HOURF
FIVMIN
INTERP
STATS
PLOT
SEDMEN
DATE
MOV
Reads observed storm hydrograph data
from cards and stores on data set
reference number 2 (option not avail-
able in TVA-HYSIM). Portions of this
subroutine are used in TVA-HYSIM for
procedural operations (SHM).
Reads storm hydrograph watershed
characteristics from data reference
number 2, calculates storm hydrograph
parameters from basin and storm charac-
teristics (SHM).
Main component of storm hydrograph
model (SHM).
Determines precipitation excess distri-
bution (SHM).
Generates hourly rainfall (HRFG).
Generates five-minute rainfall (FRFG).
Interpolates hourly rainfall for
shorter time periods (SHM) (option not
available in TVA-HYSIM).
Calculates statistics indicating
agreement between simulations and
observed data when provided (option
not available in TVA-HYSIM) (DFM,
SHM).
Provides a print-plot of a year of
daily flow (not available in TVA-
HYSIM) or a storm hydrograph (DFM,
SHM).
Calculates storm suspended sediment
loads (SSM).
A dummy subroutine to replace a TVA
system subroutine which obtains the
date from an internal computer clock.
A subroutine to initialize arrays to
zero. This replaces a software sub-
routine on the TVA system.
69
-------
TABLE 6 (continued)
CORE A TVA system subroutine to modify the
format of variable data. This
replaces a software subroutine on the
TVA system.
DFM = Continuous daily streamflow model
DRFG = Rainfall generator, daily
HRFG = Rainfall generator, hourly
FRFG = Rainfall generator, five-minute
SHM = Storm hydrograph model
SSM = Suspended sediment model
70
-------
TABLE 7
PROGRAM RUNOFF FUNCTIONS AND DESCRIPTIONS
FORTRAN software
Functions
ALOG
DABS
DATAH
DEXP
DFLOAT
DLOG10
DMAX1
DMOD
DSQRT
EXP
FLOAT
IDINT
IFIX
MOD
SIN
SNGL
Descriptions
Natural logarithm in single precision.
Absolute value in double precision.
Arc tangent in double precision.
Exponential in double precision.
Converts from integer single precision
to real double precision.
Common logarithm in double precision.
Maximum value in double precision.
Same as MOD except in double pre-
cision.
Square root in double precision.
Exponential in single precision.
Converts from integer single precision
to real single precision.
Truncation in double precision.
Converts from real single precision to
integer single precision.
Integer modular arithmetic in single
precision (used to abstract fractional
part of numbers).
Sine in single precision.
Obtains most significant part of a
real double precision number and
returns to a real single precision
number.
71
-------
FIGURE 10
CLIST for Interactive Computer Runs
/*CLIST FOR TVA-HYSIM */ 010
CONTROL NOMSG 020
/* 040
/* RUN PROGRAM "DATSET" TO SET UP A DATA SET FOR PROGRAM "RUNOFF" */ 130
/* 140
FREE ATTR(PDSOUT,PDSIN,RO,TEM) 145
ATTR PDSOUT OUTPUT 150
FREE F(FT05F001, FT10F001) 160
ALLOC DA(*) F(FT05F001) 170
ALLOC DA('H.HP325246.SSAM.DATA(TEMPNAME)') F(FTIOFOOI) + 130
USING(PDSOUT) 185
/* 190
CALL 'H.HP326246.PROGRAM.LOAD(DATSET)' 200
/* 210
WRITE ARE YOU READY TO RUN THE PROGRAM?(YES OR NO) 220
READ &ANS 230
IF &ANS NE YES THEN GOTO END 240
/* 250
/* RUN PROGRAM "RUNOFF" */ 260
/* 270
ATTR PDSIN INPUT 280
FREE F(FT01F001,FT02F001,FT03F001,FT05F001,FT10F001) 290
DEL TEMP.SPACE 300
DEL '$HDAI04.PRINTDAY.DATA' 310
DEL '$HDAI04.PRINTSTO.DATA' 320
ALLOC DA('H.HP326246.SSAM.DATA(TEMPNAME)') F(FT05F001) USING(PDSIN) 330
ATTR R9 BLKSIZE(133) LRECL(133) RECFM(F,B,A) 340
ALLOC DA('$HDAI04.PRINTDAY.DATA') F(FT03F001) NEW SPACE(9900,500) + 350
BLOCK(133) USING(RO) 360
ALLOC DA('$HDAI04.PRINTSTO.DATA') F(FTOIFOOI) MEW SPACE(59900,500) + 370
BLOCK(133) USIMG(RO) ~ 380
ATTR TEM BLDSIZE(SO) LRECL(80) RECFM(F.B) 390
ALLOC DA (TEMP. SPACE) F'(FT02F001) NEW SPACE(2,1) BLOCK(80) USING(TEM) 400
ALLOC DUMMY F(FT04F001) 410
/* 420
CALL 'H.HP326246.PROGRAM.LOAD(RUNOFF9)' 430
SET <CC=&LASTCC 435
/* 440
DEL TEMP.SPACE 470
/* 480
IF <CC NE 99 THEN GOTO END 490
WRITE DO YOU WISH TO PRINT THE STORM HYDROGRAPHS?(YES OR NO) 500
READ &ANS 510
IF &ANS NE YES THEN GOTO END 520
L '$HDAI04.PRINTSTO.DATA' NONUM 530
/* 540
END: STOP 550
72
-------
sages from commands or statement in
the CLIST.
145 Releases attribute names PDSOUT,
PDSIN, RO, and TEM.
150 Assigns the file attribute of OUTPUT
to PDSOUT. OUTPUT specifies that the
data set is to be used as output only
to the program DATSET.
160 Releases (frees) files allocated to
the data set reference numbers 5 & 10.
170 Allocates the terminal to data set
reference number 5 using attributes
assigned to PDSOUT.
180 Allocates RUNOFF input data file to
data set reference number 10.
200 Calls program DATSET to run the
program.
220 Writes to terminal "ARE YOU READY TO
RUN THE PROGRAM? (YES OR NO)".
230 Reads the response to the previous
question. If YES, then the operations
proceed on through the CLTST to call
program RUNOFF. If NO, then go to line
550 to end the CLIST.
280 Assigns the file attribute of INPUT to
PDSIN. INPUT specifies that the data
set is to be used as input only to the
program RUNOFF.
290 Frees files allocated to data set
reference numbers 1, 2, 3, 5, & 10.
300-400 In these statements, three files are
deleted then re-created. Typically
the files could be created one time
and used repeatedly. However, on the
TVA system, deleting and re-creating
these files seems to work better.
300-320 Delete three files TEMP.SPACE,
$HDAI04.PRINTDAY.DATA, and $HDAI04.
PRINTSTO.DATA.
73
-------
330 Allocates RUNOFF input data file to
data set reference number 5 using the
attribute assigned to PDSIN.
340 Assigns to RO the attributes of block
size (133) record length (133), and
record format of fixed (F) and blocked
(B) and American National Standards
Institute control characters (A).
350-360 Creates the file "$HDAI04.PRINTDAY.
DATA" and allocates it to data set
reference number 3 using the
attributes RO.
370-380 Creates the file "$HDAI04.PRINTSTO.
DATA" and allocates it to data set
reference number 1 using the
attributes assigned to RO.
390 Assigns to TEM the attributes of block
size (80), record length (80), and
record format of fixed (F) and
block(B).
400 Creates the file "TEMP. SPACE" and
allocates it to data set reference
number 2 using the attributes assigned
to TEM.
410 Allocates data set reference number 4
to a dummy file since it is not used.
430 Calls program RUNOFF to run the
program.
435 Assigns the last return code from
program RUNOFF to the variable <CC.
&LASTCC is a built-in function command
procedure to obtain the last return
code.
470 Delete file TEMP.SPACE.
490 This statement checks the return code
from program RUNOFF to see if it
equals 99. If it is not equal to 99,
the operations proceed to line 550 and
stops the CLIST. If it is equal to 99,
then the operations proceed on through
the CLIST.
74
-------
500 Writes to terminal "DO YOU WISH TO
PRINT THE STORM HYDROGRAPHS? (YES OR
NO)".
510 Reads the response input to the
previous question. If YES, then the
operations proceed on through the
CLIST to print the storm hydrographs.
If NO, then the operations proceed to
line 550 to stop the CLIST.
530 Prints the storm hydrographs which
have been stored on file "$HDAI04.
PRINTSTO.DATA" by program RUNOFF.
550 Stops the CLIST.
D. JOB CONTROL LANGUAGE (JCL) FOR BATCH RUNS
This section will describe the JCL used to run TVA-HYSIM in a
batch mode in TVA. Figure 11 is a listing of the JCL used in TVA. This can
be used as a guide to setting up JCL for a different computer system. The
following will describe lines of the JCL that may need additional explana-
tion:
Line // Description
50 The JOBLIB card. It provides the file
name for the partition data set where
the programs are stored in load mode
form.
70 Calls program DATSET to run the
program.
100 Allocates the printer to data set
reference number 6.
110 Allocates RUNOFF input data file to
data set reference number 10.
130 Allocates the card input to data set
reference number 5.
140 These are the responses to the inter-
active questions in Figure 1, Section
II. These response are punched on
cards with a comma or blank between
each value. This amounts to 19 cards
or less.
75
-------
FIGURE 11
Job Control Language for Batch Computer
//NXYSIM JOB 326285,HPRATT.ENGR.LAB,MSCLEVEL=1,CLASS=K
//* VERSION DATE: 12/14/79.
//* PROCEDURE TO RUM "HYSIM" BATCH.
//*
//JOBLIB DD DISP=SHR,DSN=H.HP325133.LIB
//*
//DATSET EXEC PGM=DATSET
//*
//*
//FT06F001 DD SYSOUT=A
//FT10F001 DD DISP=SHR,
//DSN=H.HP326246.SSAM.DATA(TEMPNAME)
//FT05F001 DD *
//*
***(Data cards with the responses to interative questions)***
//*
//RUNOFF EXEC PGM=RUNOFF9,REGION=230K,
// TIME=3
//FT01F001 DD SYSOUT=A,
//DCB=(RECFM=FBA,LRECL=133,BLKSIZE=3059)
//FT02F001 DD UNIT=SYSPL,DISP=NEW,
// SPACE'=(80,(100,100)),DSN=&&STORM,
// DCB=(RECFM=FBA,LRECL=80,BLKSIZE=400)
//FT03F001 DD SYSOUT-A,
// DCB=(RECFM=FBA,LRECL=133,BLKSIZE=3059)
//FT04F001 DD DUMMY
//FT05F001 DD DISP=SHR,
// DSN=H.HP326246.SSAM.DATA(TEMPNAME)
//FT06F001 DD SYSOUT=A
RUN PROGRAM "DATSET" TO
SET UP DATA FOR PROGRAM
"RUNOFF".
--PRINT QUESTIONS AT PRINTER
CREATED INPUT DATA FOR
PROGRAM "RUNOFF".
--RUN PROGRAM "RUNOFF" TO
SIMULATE FLOW.
PRINT STORM HYDROGRAPH
RESULTS.
TEMPORARY STORAGE FILE FOR
STORM HYDROGRAPH MODEL
INPUT DATA.
--PRINT DAILY FLOW RESULTS.
--PLOT DATA FILF IF SET UP.
--INPUT DATA TO "RUNOFF"
PROGRAM.
PRINT SHORT FORM & ERRORS.
010
020
030
040
050
060
070
080
090
100
no
120
130
135
140
150
160
170
180
190
200
210
220
230
240
250
260
270
280
76
-------
160 Calls program RUNOFF to run the
program.
180-190 Allocates data set reference number 1,
the storm hydrograph results, to the
printer.
200-220 Allocates data set reference number 2
to a temporary storage file for storm
hydrograph model input data.
230-240 Allocates data set reference number 3,
the daily flow results, to the
printer.
250 Allocates data set reference number 4
to a dummy file since it is not used.
260-270 Allocates RUNOFF input data file to
data set reference number 5.
280 Allocates data set reference number 6,
the interactive output, to the
printer.
77
-------
REFERENCES
Bales, J. , 1979, "TVA Strip Mine Assessment Model: Hydrologic Component,"
Symposium on Surface Mining Hydrology, Sedimentology and Reclamation, Uni-
versity of Kentucky, December 4-7.
Barr, W. C. , 1979, "TVA Strip Mine Aquatic Assessment Model: Aquatic Biota
Modules," Symposium on Surface Mining Hydrology, Sedimentology and Reclama-
tion, University of Kentucky, December 4-7.
Betson, R. P., R. L. Tucker, and F. M. Haller, 1969, "Using Analytic Methods
to Develop a Surface-Runoff Model," Water Resources Research V5(l).
Betson, R. P., 1976, "Urban Hydrology - A Systems Study in Knoxville, Ten-
nessee," Tennessee Valley Authority, Water Systems Development Branch, P.O.
Drawer E, Norris, Tennessee 37828.
Betson, R. P., 1977, "The Hydrology of Karst Urban Areas," In Hydrologic
Problems in Karst Regions, Published by Western Kentucky University, Bowling
Green, Kentucky (Dilamarter, R. R. and S. C. Csallany, eds.).
Betson, R. P., 1979a, "The Effects of Clear Cutting Practices on Upper Bear
Creek, Alabama, Watersheds," Tennessee Valley Authority, Water Systems
Development Branch, Norris, Tennessee 37828, Report No. WR28-1-550-101, 100
p..
Betson, R. P., 1979b, "Overview of TVA Strip Mine Aquatic Impact Assessment
Model," Presented at Winter Meeting ASAE held in New Orleans, December
11-14.
Dawdy, D. R. , 1967, "Knowledge of Sediment in Urban Environments," ASCE
Journal Hydraulics Division HY6 (Nov.).
Donigan, A. S., and N. H. Crawford, 1976, "Modeling Nonpoint Pollution from
the Land Surface," EPA Environmental Research Laboratory, Athens, Georgia,
EPA-600/3-76-083, 292 p.
Douglass, J. E., and W. T. Swank, 1972, "Streamflow Modification Through
Management of Eastern Forests," U.S. Forest Service Research Paper SE94,
S.E. Forest Experiment Station, Asheville, NC, 15 p.
Fenneman, N. M., 1938, "Physiography of Eastern United States," McGraw Hill,
New York.
78
-------
Haan, C. T. , and Barfield, B. J. , 1978, "Hydrology and Sedimentology of
Surface Mined Lands," Office of Continuing Education, University of
Kentucky, Lexington, Kentucky 40506, 286 p.
Helvey, J. D. , and J. H. Patric, 1965, "Canopy and Litter Interception of
Rainfall by Hardwoods of Eastern United States," Water Resources Research,
1:193, pp. 193-206.
Holtan, H. N., and N. C. Lopez, 1973, "USDA HL-73 Revised Model of Watershed
Hydrology," Plant Physiology Institute Report No. 1, USDA Agricultural
Research Service Hydrograph Laboratory, Beltsville, Maryland.
Kirpich, P. Z., 1940, "Time of Concentration of Small Agricultural Water-
sheds," Civil Engineering 10(6).
Metcalf and Eddy, 1971, "Storm Water Management Model - Vol. 1, Final
Report," EPA Water Pollution Control Series 11024DOC07/71, 352 p.
Miller, C. R. , and W. Viessman, Jr., 1972, "Runoff Volumes from Small Urban
Watersheds," Water Resources Research V8 No. 2.
Musser, J. J., C. R. Collier, R. J. Pickering, and others, 1970, "Hydrologic
Influences of Strip Mining - Chapters A-C," U.S. Geological Survey Profes-
sional Paper No. 427, U.S. Government Printing Office, 0-383-348.
Overton, D. E., 1967, "Analytical Simulation of Watershed Hydrographs from
Rainfall," Proc. Intern. Hydrol. Symp., Fort Collins, CO, September 6-8,
1967, pp. 9-17.
Overton, D. E., 1968, "A Least-Squares Hydrograph Analysis of Complex Storms
on Small Agricultural Watersheds," Water Resources Research, Vol. 4, No. 5,
pp. 955-963.
Overton, D. E., 1970, "Route or Convolute," Water Resources Research, V6(l),
pp. 43-52.
Overton, D. E., 1971, "Estimation of Surface Water Lag Time from Kinematic
Wave Equations," Water Resources Research, Vol. 7, No. 3, pp. 428-440.
Schumm, S. A., 1973, "Geomorphic Thresholds and Complex Response of Drainage
Systems," in Fluvial Geomorphology, Proceedings of the Fourth Annual Geo-
morphology Symposia Series, September 27-28, 1973, Published by State Uni-
versity of New York, Binghamton, New York.
Sherman, L. K. , 1932, "Streamflow from Rainfall by the Unit-Graph Method,"
Eng. News-Rec. Vol. 108, pp. 501-505, April 7.
Singh, B., and G. Szeicz, 3979, "The Effect of Intercepted Rainfall on the
Water Balance of a Hardwood Forest," Water Resources Research V15(l), pp.
131-138.
79
-------
Soil Conservation Service, 1972, "National Engineering Handbook," Section IV
Hydrology, Part 1 - Watershed Planning, Washington, DC.
Soil Conservation Service, 1975, "Urban Hydrology for Small Watersheds,"
Technical Release No. 55.
Stewart, J. B. , 1977, "Evaporation from the Wet Canopy of a Pine Forest,"
Water Resources Research V13(6J, pp. 915-921.
Swank, W. T. , N. B. Gobel, and J. D. Helvey, 1972, "Interception Loss in
Loblolly Pine Stands of the South Carolina Piedmont," Journal of Soil and
Water Conservation (27), pp. 160-164.
Tennessee Valley Authority, 1972, "Upper Bear Creek Project - A Continuous
Daily Streamflow Model," Division of Water Management, Research Paper No. 8,
Knoxville, Tennessee, 99 p.
Tennessee Valley Authority, 1973a, "Summary Report on the Upper Bear Creek
Experimental Project," Knoxville, Tennessee.
Tennessee Valley Authority, 1973b, "Storm Hydrographs Using a Double-
Triangle Model," Division of Water Management, Research Paper No. 9, Knox-
ville, Tennessee, 111 p.
Troxler, W. L., 1978, "A Stormwater Simulation Model for the Tennessee
Valley," MS Thesis, Department of Civil Engineering, University of Tennes-
see, Knoxville, Tennessee.
Williams, J. R., 1975, "Sediment Yield Prediction with Universal Equation
Using Runoff Energy Factor," Proceedings of the Sediment-Yield Workshop,
Oxford, Mississippi, U.S. Dept. of Agr., ARS-S-40. pp. 244-252.
Williams, J. R. , 1978, "A Sediment Graph Model Based on an Instantaneous
Unit Sediment Graph," Water Resources Research V14(4), pp. 659-664.
Williams, J. R. , and Berndt, H. D., 1976, "Determining the Universal Soil
Loss Equation's Length-Slope Factor for Watersheds," in Soil Erosion: Pre-
diction and Control, Soil Conservation Society of America, Ankeny, Iowa, pp.
217-225.
80
-------
APPENDIX
ENGLISH TO METRIC CONVERSION FACTORS
English Unit
inches (in)
feet (ft)
miles (mi)
square feet (ft2)
acres (ac)
square miles (mi2)
cubic feet (ft3)
pounds (Ib)
tons
pounds/acre (#/ac)
tons/sq mi (Tons/mi2)
ft3/second/mi2
acre-feet (ac-ft)
Multiplied By
2.54
0.305
1.61
0.093
0.405
2.59
0.0283
0.454
907.2
1.120
3.5
0.0109
1223.
Converts To
centimeters (cm)
meters (m)
kilometers (km)
square meters (m2)
hectares (ha)
square kilometers (km2)
cubic meters (m3)
kilograms (kg)
kilogram (kg)
kilogram/hectare (kg/ha)
kilogram/hectare (kg/ha)
m3/second/km2
cubic meters (m3)
81
-------
TECHNICAL REPORT DATA !
(Please read Instructions on the reverse before completing)
1. REPORT NO. 2.
EPA-600/70-80-048
4. TITLE AND SUBTITLE
"User's Guide to TVA-HYSIM"
A Hydro logic Program for Quantifying Land-Use Change
Effects
7. AUTHOR(S)
Roger P. Betson, Jerad Bales, and Harold E. Pratt
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Tennessee Valley Authority
Water Systems Development Branch
P.O. Drawer E
Norris, Tennessee 37828
12. SPONSORING AGENCY NAME AND ADDRESS
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cinncinnati, Ohio 45268
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
November 1980
6. PERFORMING ORGANIZATION CODE
1
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO. .
EPA-IAG-D9-E721-DS
13. TYPE OF REPORT AND PERIOD COVERED ;
Final: 6/75-11/80
14. SPONSORING AGENCY CODE
EPA-ORD j
15. SUPPLEMENTARY NOTES
This project is part of the EPA planned and corrdinated Federal Interagency Energy/
Environment R&D Program
16. ABSTRACT
TVA-HYSIM is a computer package containing complex hydrologic models specifically
designed for ease of application in land-use planning studies.
This user's guide outlines the information required to operate the programs and
how this information is obtained, shows examples of input and output, and provides
examples of job controls needed to operate the program. Model components are de-
scribed in sufficient detail so that changes to the algorithms may be made if so
desired.
TVA-HYSIM is not adapted to handling dynamic land-use conditions, but rather is
designed to be used as a planning tool so that the end effects of the land-use
change can be evaluated before the change occurs. Thus in a typical land-use
change evaluation, the model package would first be used to simulate hydrology
under present land-use conditions and then used to simulate the post land-use
\ change hydrology. Some strategies for using TVA-HYSIM to determine the effects
of land-use change on the hydrologic balance are offered. ;
j
17. KEY WORDS AND DOCUMENT ANALYSIS f
a. DESCRIPTORS
Hydrology
Hydrologic Models
Land-Use Change
Surface Mining
18. DISTRIBUTION STATEMENT
Unlimited
b. IDENTIFIERS/OPEN ENDED TERMS
Water Quality
Flooding
Sedimentation
Computer Program
19. SECURITY CLASS (This Report)
Unclassified
20. SECURITY CLASS (This page)
Unclassified
c. COS AT I Field/Group
B
|
i
21. NO. OF PAGES
96 !
22. PRICE
1
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
------- |