Section 7
Model Development for Dutchman Lake
v 7-11
FINAL REPORT
concentrations obtained from GWLF output. The monthly values were summed over
the water year for input to BATHTUB. To obtain flow in units of volume per time, the
depth of flow was multiplied by the drainage area and divided by one year. To obtain
phosphorus concentrations, the nutrient mass was divided by the volume of flow.
7.4 Model Calibration and Verification
The GWLF model was calibrated prior to BATHTUB calibration. The GWLF model
for the Dutchman Lake Watershed was calibrated to flow data, as tributary phosphorus
concentrations were not available. Nutrient concentrations entered into the GWLF
model were calibrated based on response occurring in the BATHTUB model.
Therefore, the nutrient block of the GWLF model and the BATHTUB model were
calibrated together to reach agreement with observed data in Dutchman Lake.
7.4.1 GWLF Calibration
The GWLF model must run from April to March to coincide with the soil erosion
cycle, which is when the majority of erosion occurs in agricultural areas. GWLF does
not retain erodible sediment between model years, so the model year must begin after
the previous year's sediment has been washed off. The model assumes that the soil
erosion cycle begins with spring runoff events in April and that erodible soil for the
year has been washed off by the end of winter for the cycle to begin again the
following April. GWLF generates monthly outputs including precipitation, flow,
runoff and nutrient mass per watershed, and annual outputs including precipitation,
flow, runoff, and nutrient mass per land use. These outputs are part of the input for the
BATHTUB model.
In-stream nutrient data was not available for model calibration, so GWLF was only
calibrated to flow. The monthly average flow output from GWLF was compared to the
monthly average streamflow calculated from USGS gage 03612000 with the drainage
area ratio method presented in Section 5.1.3. The model flow was calibrated visually
through the recession constant and seepage constant. Visual calibration is a subjective
approach to model calibration in which the modeler varies inputs to determine the
parameter combination that looks like the best fit to the observed data (Chapra 1997).
According to the GWLF manual, an acceptable range for the recession constant is 0.01
to 0.2. No range suggestions are provided for the seepage constant. Figure 7-4 (at the
end of this section) shows the comparison between the two flows for subbasin 1 of
Dutchman Lake. The GWLF model for Dutchman Lake was visually calibrated with a
resulting recession constant of 0.2 and a seepage constant of 0.1 in each subbasin.
Once calibrated, the model output data could properly be included as BATHTUB
inputs. The GWLF model was not validated as flow was calibrated by visually
comparing 16 years of observed flow.
Although in-stream nutrient concentrations are not available for the tributaries to
Dutchman Lake, Clean Lakes Studies have been conducted by Illinois EPA on various
Illinois lake watersheds, which do provide in-stream nutrient data for lake tributaries
including dissolved and total phosphorus. The dissolved and total phosphorus
