HSPF Snow Modeling Parameters in an Urban Setting: Evaluation Using NWS Coop and SNODAS Snow Data and Re-Calibration Using PEST

An evaluation and provisional re-calibration of a set of regional Hydrologic Simulation Program — FORTRAN (HSPF) snow modeling parameter values used by the U.S. Army Corps of Engineers in the Chicago, Illinois, region were carried out for two small gaged urban watersheds in the region using data from water years 1996–2007. The simulated snow fluxes (snowfall, sublimation, and melt) and storage (snowpack) were evaluated by comparison with daily observed snowfall and snowpack-depth data from two National Weather Service (NWS) Cooperative (Coop) observer stations and output from NWS-NOHRSC's Snow Data Assimilation System (SNODAS). HSPF snow parameter sensitivities and provisional re-calibration were performed using the Model Independent Parameter Estimation Routine (PEST) calibration program. It was observed that for the period of available SNODAS data (water years 2004–7), using the original snow parameter values, the HSPF-simulated cumulative snowfall snow-water equivalent (SWE) exceeded both the corresponding SNODAS and NWS Coop values by 20–40%. The HSPF-simulated cumulative snowmelt exceeded the corresponding SNODAS value by a similar amount. However, the HSPF-simulated cumulative snowpack depth values were significantly smaller than both the NWS Coop and SNODAS values, suggesting that the HSPF-simulated snowpack has too small a residence time. In support of this observation, the HSPF-simulated discharge in the winter months in these watersheds tends to be too early, with simulated flows too large in mid-winter and too small in the early spring compared to observed flows. This result shows the potential usefulness, despite their limitations, of the combined NWS Coop and SNODAS data for guiding the simulation of snow-related processes. In addition, the provisional re-calibration of the snow parameter values for these watersheds using PEST was able to improve this discharge timing problem, thus showing PEST's potential for providing improved simulation of HSPF snow processes.