United States
Environmental Protection
Agency
Environmental Research
Laboratory
Athens GA 30601
Research and Development
EPA-600/S3-83-069 Nov. 1983
Project Summary
Application of Hydrologic
Simulation Program-FORTRAN
(HSPF) in Iowa Agricultural
Watersheds
Anthony S. Donigian, Jr., John C. Imhoff, Brian R. Bicknell,
James L. Baker, Douglas A. Haith, and Michael F. Walter
The Hydrologic Simulation Program-
FORTRAN (HSPF) was applied to two
watersheds in Iowa to test the model's
capability in evaluating the effects of
agricultural best management practices
(BMPs) on water quality. The project
first involved refining HSPF to incorporate
a pesticide risk assessment methodology.
Before the application of HSPF to the
Four Mile Creek Watershed, an evaluation
of the sensitivity of model parameters
for BM Ps was done and a parameter esti-
mation manual was prepared. The model
was calibrated and verified in the
watershed. The water quality effective-
ness of alternative BMP scenarios was
then evaluated using the parameter
estimation manual. To evaluate the
ease of model scale-up and extrapolation
to a larger basin, the model was then
applied to the Iowa River Basin above
Coralville Reservoir.
These applications of HSPF in water-
sheds of different size in Iowa demon-
strated that HSPF is a flexible and
realistic means of approximating the
impacts of candidate BMPs on water
quality in small watersheds and large
river basins. Moreover, risk assessment
procedures can be used with the
simulated chemical concentration time
series produced by the model to
evaluate the impacts on selected
aquatic organisms.
True verification of any model's
ability to simulate the effects of BMPs,
however, must await the availability of
both pre- and post-BMP implementation
data. Until such data have beeV) collected,
model? such as HSPF can still be used
to approximate the effects of BMPs on
stream water quality,7and sensitivity
analyses of various model parameters
can assist the BMP evaluation and
planning process.
These applications provide one of the
first systematic attempts to combine a
detailed simulation of agricultural
runoff and soil processes, which calcu-
late surface and subsurface pollutant
transport to receiving waters, with sub-
sequent simulation of instream trans-
port and transformations. The result is a
comprehensive simulation of water-
shed hydrology and water quality.
This Project Summary was developed
by EPA's Environmental Research
Laboratory. Athens. GA, to announce
key findings of the research project that
is fully documented in three separate
reports (see Project Report ordering
information at back).
Background
HSPF is a comprehensive program for
modeling sediment, pesticides, nutrients,
and other water quality constituents in
runoff from urban, agricultural, and other
lands. The model allows detailed simula-
tion of stream hydraulics, water quality
processes, pesticide and nutrient be-
havior in soil, and sediment-contaminant
transport. Extensive data handling and
analysis procedures that support and
complement the simulation capabilities
are included.
-------
A unique feature of HSPF is the incor-
poration of runoff models—the Agri-
cultural Runoff Management (ARM)
Model and the Nonpoint Source (NFS)
Model and stream transport and fate
models-in a systematic framework. Sim-
ply put, the model uses information on
meteorology, land use, and agronomic
characteristics to simulate a time history
of the quantity and quality of the runoff.
Flow rate, sediment load, and nutrient
and pesticide concentrations are pre-
dicted. The model then takes these re-
sults and information about the stream
channels in the watershed and simulates
the processes that occur in these
streams. This produces a time history of
water quantity and quality at any point in
the watershed. HSPF can be applied to a
wide range of water resource problems.
The key attribute that makes it widely ap-
plicable is its ability to simulate the con-
tinuous behavior of time-varying physical
processes and to provide statistical sum-
maries of the results.
The work reported here was performed
as part of a comprehensive field evalua-
tion program in two Iowa watersheds
(Figure 1). The purposes of the program
were to test and demonstrate (1) the ca-
pability of agricultural BMPs to achieve
water quality goals, and (2) the applicabil-
ity of water quality planning tools (such as
HSPF) to the BMP evaluation and selec-
tion process. Program results are de-
scribed in "HSPF Parameter Adjustments
to Evaluate the Effects of Best Manage-
ment Practices," "Modeling Water Quali-
ty and the Effects of Agricultural Best
Management Practices in Four Mile
Creek, Iowa," and "Preliminary Applica-
tion of HSPF to the Iowa River Basin to
Model Water Quality and the Effects of
Agricultural Best Management Prac-
tices."
HSPF Parameter
Adjustments for
Evaluating BMPs
A key problem in applying most
models, especially those with an empirical
basis, is how to adjust model parameters
to represent the effects of a specific
practice, or combination of practices, that
comprise a BMP or system of BMPs.
"HSPF Parameter Adjustments to Evaluate
the Effects of Agricultural Best Manage-
ment Practices" qualitatively assesses
the effects of selected agricultural
practices on runoff, erosion, and chemical
processes, and quantifies the associated
adjustments to model parameters based
on the current state of science. Although
the specific parameter changes are
Iowa
Cedar Rivers
I Z>> Basin
Coralville
Reservoir!? lowa
City
Figure 1. Location of Four Mile Creek and Iowa-Cedar River Basin Sites for lowa Field
Evaluation Program.
particular to the HSPF and ARM models,
the information presented is generally
applicable and should be pertinent to
many models having similar representa-
tions of the relevant processes.
In summary, the report discusses the
technical aspects of conventional practices
and candidate BMPs as a basis for
predicting BMP impacts on relevant
agricultural runoff processes in lowa
watersheds. Guidelines and recommen-
dations are included for adjusting all
major parameters whose effects can be
reasonably quantified. Deficiencies in
the current state of knowledge of BMP
effects and recommendations for future
research and model improvements are
presented and discussed.
The approach to assessing the manner
in which HSPF parameters should be
adjusted to represent the effects of
candidate BMPs involved six steps:
• Define conventional agricultural
practices for lowa watersheds.
• Select and define candidate BMPs.
• Evaluate qualitatively the impact of
each candidate BMP on agricultural
runoff processes (quantity and quality)
-------
relative to conventional practices.
• Identify HSPF model parameters that
control or affect specific runoff
processes (quantity and quality).
• Evaluate qualitatively the relative
effect of each candidate BMP on the
HSPF model parameters identified
above.
• Quantify the effects of the candidate
BMPs on the identified model param-
eters based on available data, current
literature, and, lacking all else, best
judgment or experience.
The qualitative assessments provided the
basis for quantifying the changes in
model parameters.
The qualitative assessment of how the
candidate BMPs could be represented by
the indicated changes in HSPF parameters
relative to those for conventional practices
is presented in the report. The candidate
BMPs are divided into nonstructural,
structural, and input management prac-
tices.
Finally, the report provides a discussion
of the major HSPF runoff, sediment and
chemical parameters in terms of adjust-
ments or changes needed to simulate the
effects of the candidate BMPs.
Modeling BMP Effects-
Four Mile Creek
"Modeling Water Quality and the
Effects of Agricultural Best Management
Practices in Four Mile Creek, Iowa"
discusses the calibration and verification
of HSPF in a relatively small watershed
(approximately 20 sq. mi.). Such efforts
are needed to evaluate model capabilities
and develop sufficient confidence so that
the model can be used for BMP analyses.
Hydrologic calibration involves iterative
adjustments in selected parameter
values based on comparison of observed
and simulated runoff volumes and storm
hydrographs. The model representation
and calibrated parameters are then
verified by performing the same compari-
sons on an independent data set, i.e.,
runoff volumes and hydrographs that
were not used in the calibration. In the
Four Mile Creek watershed, observed
runoff data for calibration and verification
were available for three watershed sites
and three small field sites. Approximately
7 years of daily streamflow data (and
selected hydrographs) were available for
the watershed sites, and 2 1/2 years of
runoff data were available from the field
sites. A summary of the annual runoff
and daily flow statistics for calibration
and verification at the Traer gage, a
watershed site on Four Mile Creek is
given in the report. The calibration and
verification results of modeling the
watershed hydrology compare reasonably
well with the observed values.
The calibration and verification results
of modeling alachlor concentrations are
shown in Figure 2. The relative timing of
applications and the first significant
storm event have a major impact on the
resulting chemical runoff. Although the
alachlor simulations are generally higher
than observed values, the concentrations
and loads are generally within the range
of the observed values. Oftentimes,
major discrepancies can be explained by
the absence of, or large errors contained
in, sampling data during storm events.
The simulations of alachlor edge-of-
field loadings and stream concentrations
showed that approximately 30% of the
alachlor reaching the stream did not
reach the watershed outlet at Traer. This
Calibration
is likely due to adsorption onto sediment
particles, resulting deposition in the bed,
and decay of the compound in the
channel bed.
Another method of viewing calibration
and verification results is to determine
whether decisions on risk or exposure to
aquatic organisms would be different if
the analyst based his decision on the
simulated or the observed chemical
information. Figure 3 demonstrates how
the frequency (percent of time) of acute,
chronic, and sub-lethal conditions might
be determined for a particular stream
given a time series of chemical concentra-
tions. Using the risk assessment meth-
odology integrated into HSPF, the ob-
served and simulated chemical concen-
trations were analyzed to determine the
percent of time conditions within each
region shown in Figure 3 would exist. The
results of this analysis using a hypothetical
•c
\
Oi
c
•3
Figure 2. Calibration and verification results for solution alachlor loading at Traer, IA.
3
-------
a
Sublethal Region
Duration
Figure 3. Lethality analysis of chemical concentration data.
organism are shown in Table 1. A
hypothetical organism was used because
all the values observed for alachlor
concentrations were considerably lower
than any of the maximum acceptable
toxicant concentration values for all
species of fish found in Four Mile Creek.
The table shows that the observed and
simulated values agree quite closely
indicating that the ultimate decision on
aquatic risk and exposure would be the
same whether the observed or simulated
values were used.
In summary, the instream pesticide
simulation results show that reasonably
good simulations can be obtained by
using small site parameter values for
alachlor in conjunction with hydrologic
parameters calibrated on the entire
watershed and a reasonably good instream
hydraulic and sediment transport simula-
tion. In effect, the work has shown that an
adequate simulation of the hydrologic
transport components on a watershed
basis provides a viable means of simulat-
ing chemical transport and aquatic risk.
The nutrient simulation followed the
same general procedures as the pesticide
simulation. The nutrient parameters
calibrated on the small sites were
extrapolated to the entire watershed and
used in conjunction with the watershed
hydrology and stream hydraulics to
simulate nutrient concentrations and
loadings at the watershed outlet. Unlike
the pesticide simulation, nutrients are
contributed from all land within the
watershed whether or not the land
received fertilizer applications.
The daily concentrations and loads
from nitrate, ammonia, and chloride
generally indicate that the loads are
simulated considerably better than
concentration values, but that the overall
simulation is reasonably good. The large
differences between the simulated and
recorded values for ammonia and phos-
phate at the watershed outlet are likely
due to the lack of simulation data on in-
stream sediment and sediment-nutrient
interactions. HSPF did not simulate
sediment-nutrient interactions as was
done for pesticides.
Although the results were viewed as
preliminary, it was concluded that the
model is capable of representing the
overall watershed system behavior in
terms of nutrient simulation. Additional
calibration efforts and the suggested
model enhancements would likely increase
significantly the agreement of simulated
and observed values and thus improve
the preliminary results presented here.
As a demonstration of the model's
utility, a BMP scenario was evaluated
using the parameter values reported in
the previous study. Using the assumptions
and associated changes in parameter
values for the BMP scenario described in
the report, a comparison can be made of
the simulated base condition versus the
BMPscenario. Table 2 shows the average
percent reductions in runoff (8.3%),
sediment (41.7%), alachlor (for solution
and sediment, 32.5% and 68.2%, respec-
tively), and nutrients (varies byform)tobe
expected from the BMP.
Finally, Table 3 shows the results of the
simulation of the BMP scenario in terms
of reductions in the fraction of time when
acute and lethal conditions exist, following
the lethality analysis methodology men-
tioned previously. The reductions indicate
an 8% reduction in the percentage of time
when pesticides exceeded the MATC
levels in the watershed. Although the
values listed here are specific to the
conditions under which this BMP scenario
was simulated, the overall methodology
and analysis indicates how the procedures
can be used to evaluate the effects of
BMP scenarios on the resulting risk of
exposure of aquatic organisms to chemi-
cals.
Although this study demonstrated the
overall utility of the HSPF model for
evaluating BMPs, numerous weaknesses
and difficulties were identified. Needed
improvements include a better capability
to predict pesticide decay in the field
under various environmental conditions;
better understanding of nutrient param-
eters, of tile drainage and snowmelt
effects and of field-to-stream sediment
delivery; and a better basis for assumptions
about agronomic practices.
Table 1. Lethality Analysis for Alachlor in Four Mile Creek for Hypothetical Organism
Global Exceedance
(% of time)
1978
Calibration
DBS SIM
Acute Region
Above MATC Value
Sublethal Region
(below MATC)
0.5
4.4
95.6
3.3
6.6
93.4
1976
Verification
OBS SIM
0.5
3.3
96.7
1.9
8.1
91.9
1976-78
OBS SIM
0.2
1.5
98.5
1.2
3.2
96.8
MA TC - Maximum Acceptable Toxicant Concentration.
(0.003 mg/l used above).
OBS - Observed values.
SIM - Simulated values.
4
-------
Table 2. Comparison of Base Conditions with BMP Scenario on Four Mile Creek
BMP Scenario % Difference
Base
Conditions
Runoff (mm)
76
77
78
Total
Sediment (tonne/ha)
76
77
78
Total
Alachlor (kg/ha)
Solution:
76
77
78
Total
Sediment:
76
77
78
Total
69.9
48.8
161.8
280.5
0.437
0.042
1.381
1.860
0.00166
0.00009
0.00212
0.00387
0.00012
0
0.00010
0.00022
60.7
43.7
152.9
257.3
0.175
0.022
0.888
1.085
0.00084
0.00005
O.OO168
0.00261
0.00003
0
0.00004
0.00007
-13.2
-10.4
- 5.5
- 8.3
-60.0
-47.6
-35.7
-41.7
-49.4
0
-20.7
-32.5
-69.2
0
-60.0
-68.2
Nutrients fkg/ha): 1 1/77 - 10/78
N03
Cl
NH3-Sol
*NH3-Sed
*PO4-Sol
*PO4-Sed
26.2
34.0
2.18
0.00045
1.69
0.00029
20.2
30.1
0.69
0.00019
1.14
0.00018
-22.9
-11.5
-68.3
-57.8
-32.5
-37.9
- Edge of Stream Loadings.
Table 3. Lethality Analysis of BMP Scenario for Alachlor in Four Mile Creek for Hypothetical
Organism
Global Exceedance
(% of time)
Base Conditions
Acute Region
Above MA TC Value
MAX
1.2
3.2
AVER
0.4
2.5
BMP Scenario
MAX
1.2
2.9
AVER
0.2
2.3
% Change
MAX
0.0
-9.4
AVER
-50.
-8.0
Sublethal Region
(below MA TC)
96.8
97.5
97.1
97.7
+0.3
+0.2
MA TC - Maximum Acceptable Toxicant Concentration
(0.003 mg/l used above).
MAX - Daily maximum concentrations.
A VER - Daily average concentrations.
HSPF Demonstration -
Iowa River Basin
"Preliminary Application of HSPF to the
Iowa River Basin to Model Water Quality
and the Effects of Agricultural Best
Management Practices" describes a
basin-scale model application that com-
bines the detailed simulation of agricul-
tural runoff and soil processes, surface
and subsurface pollutant transport to
receiving waters, and subsequent simula-
tion of instream transport and trans-
formations. The result is a comprehensive
simulation of river basin water quality.
Comparisons of water quality resulting
from conventional agronomic practices
and BMPs provide a basis for determining
the net effects and associated benefits of
BMP implementation. Furthermore,
using simulated concentrations of pesti-
cides and other toxic pollutants in
conjunction with lethality-duration in-
formation, the frequency of acute and
chronic toxic conditions can be determined
to assess the risk to aquatic life of
proposed practices.
The investigation of the Iowa River
Basin (approximately 7240 sq. km. reported
here is an extension or scale-up of the
modeling study in the Four Mile Creek
watershed (approximately 52 sq. km.)
described previously. The objective was
to extrapolate the methodology developed
on Four Mile Creek to the Iowa River
Basin to demonstrate its applicability and
functionality on a large river basin. This
demonstration must be viewed as prelimi-
nary due to the limited data and project
resources.
The report documents the procedures
and assumptions used in applying HSPF
to the Iowa River Basin. The simulation
results presented are indicative of the
type of information produced by the
model. The specific results and compari-
sons with observed data should not be
used as a final* determination of the
accuracy or reliability of HSPF, however,
because of the preliminary nature of this
work. Moreover, the model results
should not be used as a basisfor planning
decisions on agricultural nonpoint pollu-
tion and BMPs in the study area without
additional calibration efforts and re-
evaluation of the underlying assumptions
on which this demonstration rests. This
investigation was directed to the assess-
ment of operational problems of modeling
chemical fate and transport at the river
basin scale.
The development of a simulation plan
involved four steps: characterize the area
with regard to meteorologic conditions,
soils, topography, land use, pollutant
sources, etc.; segment the basin to define
areas of homogenous hydrologic response;
evaluate streamflow and water quality
data to devise a modeling and calibration
scheme; and ascertain the relative
importance of various pollutant sources.
Prior to actual HSPF application to the
entire basin, a limited calibration of
hydrology and water quality was con-
ducted to demonstrate sufficient agree-
ment between model results and available
data so that the model could then be used
for BMP analysis and evaluation.
For the actual demonstration of HSPF,
a BMP scenario similar to that exercised
on the Four Mile Creek watershed, was
evaluated on the Iowa River Basin, i.e.,
conservation tillage plus contouring.
Table 4 shows a comparison of loadings
in the Iowa River at Marengo for the 5-
year simulated base conditions and BMP
evaluations. Average percentage reduc-
-------
Table 4.
Comparison of Loadings in the Iowa River at Marengo for Base Conditions and BMP
Simulations
Year
Base
BMP
% Difference
RUNOFF (mm)
SEDIMENT
(tonnes/ha)
SOLN. ALACHLOR
(kg/ha)
SED. ALACHLOR
(kg/ha)
NITRATE N
(kg/ha)
AMMONIA N
(kg/ha)
1974
1975
1976
1977
1978
A verage
1974
1975
1976
1977
1978
Average
1974
1975
1976
1977
1978
A verage
1974
1975
1976
1977
1978
Average
1974
1975
1976
1977
1978
A verage
1974
1975
1976
1977
1978
Average
183.0
124.0
80.0
47.8
299.0
147.0
3.91
0.88
0.56
0.019
5.69
2.21
0.0278
0.0026
0.0008
0.00
0.0068
0.0076
0.0032
0.0002
0.00
0.00
0.0007
0.0008
31.0
14.9
9.5
4.9
18.5
15.8
0.48
0.57
0.53
0.37
0.91
0.57
170.0
116.0
73.9
42.4
280.0
136.0
2.62
0.47
0.12
0.012
5.49
1.74
0.0219
0.0017
0.0004
0.00
0.0048
0.0058
0.0020
0.0001
0.00
0.00
0.0004
0.0005
29.8
9.5
6.2
3.0
13.1
12.3
0.41
0.30
0.20
0.09
0.46
0.29
-7.1
-6.4
-7.6
-11.3
-6.4
-7.5
-33.0
-47.0
-79.0
-37.0
-3.5
-21.0
-21.0
-35.0
-50.0
.
-29.0
-24.0
-38.0
-50.0
-
.
-43.0
-38.0
-3.9
-36.0
-35.0
-39.0
-29.0
-22.0
-15.0
-47.0
-62.0
-76.0
-49.0
-49.0
lions are approximated for runoff (7.5%),
sediment (21.0%), alachlor (24% and
38% for solution and sediment, respec-
tively), nitrate (22%) and ammonia (49%).
The sediment loss reductions are
somewhat less than expected. This is
likely due to the fact that a significant
portion of the total sediment loss is
derived from the channel system itself,
which would not be significantly affected
by the BMPs. The zero reductions in
alachlor occurred during years of extreme
drought. The reductions for nitrate and
ammonia were lowest in the first year of
the simulation period due to the same
initial nutrient storages in the soil for
both the base conditions and the BMP.
As in the Four Mile Creek application,
one of the possible uses of continuous
modeling of chemical fate and transport
is to evaluate the risk of exposure of
aquatic organisms to various magnitudes
and durations of chemical concentrations.
1978 were not sufficient to reduce the
concentrations below the 30 ppb thresh-
old assumed in this risk analysis. Overall,
the reductions indicate a 23% reduction
in the percent of time when lethal condi-
tions occurred in the watershed.
Specific conclusions from this study
are:
1. HSPF can be used to model the flow,
sediment, and water quality from
large agricultural river basins.
2. Many model parameters, primarily
these related to hydrology and sedi-
ment, are calibration dependent.
3. Meteorologic data for different por-
tions of the basin are a critical
component.
4. With only minimal calibration effort,
simulation of flow frequencies on the
main stem of the Iowa River was fair
to good in this study.
5. Sediment simulation was judged to
be fair to poor due to insufficient
calibration, lack of data, and model
deficiencies.
6. Simulated pesticide (i.e., alachlor)
loadings and concentrations were in
the expected range, asgenerally were
the simulated nitrate-nitrogen and
ammonia-nitrogen concentrations.
Several needed improvements to HSPF
were identified in this study:
1. Accommodation of nutrient and
chemical inputs with precipitation.
2. Simulation of both ionized and un-
ionized forms of ammonia and sedi-
ment-ammonia interactions.
3. Better definition of bed water quality
and sediment processes, possibly
Table 5 shows the reductions in the with a layered representation.
fraction of time when acute and lethal 4. Use of output of nutrient transforma-
conditions exist under the simulated tions as aids in calibration and
BMP scenario. It is interesting to note analysis of model results.
that although the BMP scenario provided
substantial reductions in the peak
concentrations (ranging from 9% to 41 %),
the absolute reductions in 1974 and
Table 5. Lethality Analysis of BMP Scenario for Alachlor in the Iowa River at Marengo, Iowa
Base Conditions
Global Exceedance
(% of time)
BMP Scenario
i Change
Acute Region
Above MATC Value
Sublethal Region
(below MATC)
0.49
3.50
96.50
0.49
2.68
97.32
0
-23.4
- 0.8
MA TC - Maximum Acceptable Toxicant Concentration
(0.003 mg/l used above).
-------
AnthonyS. Donigian. Jr., John C. Imhoff, andBrianR. Bicknellare with Anderson-
Nichols and Co.. Palo Alto. CA 94304; James L. Baker is with Iowa State
University, Ames, IA 500 J1; Douglas A. Haith and Michael F. Walter are with
Cornell University, Ithaca, NY 14853.
T. O. Barnwell, Jr., is the EPA Project Officer (see below).
This Project Summary covers three reports, entitled:
"HSPF Parameter Adjustments to Evaluate the Effects of Agricultural Best
Management Practices,"(Order No. PB 83-247 171; Cost: $13.00)
"Modeling Water Quality and the Effects of Agricultural Best Management
Practices in Four Mile Creek, Iowa," (Order No. PB 83-250 183; Cost: $ 13.00)
"Preliminary Application of HSPF to the Iowa River Basin to Model Water
Quality and the Effects of Agricultural Best Management Practices," (Order
No. PB 83-250 399; Cost: $13.001
The above reports are available only from: (costs subject to change)
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Research Laboratory
U.S. Environmental Protection Agency
College Station Road
Athens, GA 30613
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
U.S. POSTAGE
PAID
Cincinnati, Ohi
Permit No. G3!
Official Business
Penalty for Private Use $300
U.S. GOVERNMENT PRINTING OFFICE: 1983-759-102/0794
------- |