Calibration of BASINS HSP-F in Support of a Watershed Approach to CSO Long Term Control Planning

Abstract
BASINS (Better Assessment Science Integrating point and Nonpoint Sources) HSP-F was used as part of the watershed approach in developing the City of Buffalo Long Term Control Plan for CSO abatement. More specifically, the focus of this chapter is the Buffalo River watershed and 33 CSOs that discharge to the lower Buffalo River. Estimated sediment, bacteria, and metals loads from the watershed and CSOs were compared for design storms of different frequencies and on an annual basis. Loadings of these contaminants from the upper watershed exceeded the CSO loadings. It was essential to have a calibrated BASINS HSP-F model to make the watershed loading estimates. This chapter reviews the procedure and results of the model calibration effort, including some of the challenges to estimating both runoff quantity and sediment transport.
The Buffalo River watershed is 446 square miles (1,155 km2) in area and has a USGS gauge station on each of the three major tributaries, Cazenovia Creek, Cayuga Creek, and Buffalo Creek. The BASINS HSP-F, version 2, was calibrated on an annual basis, using daily mean flow for a dry year (1995), a wet year (1992), and an “average” year (1990). The r2 between observed and modeled daily mean flow for the three gauge sites (0.56-0.73) and the Nash Sutcliffe coefficient (0.06-0.50) tended to be better for the “average” year. The model was most sensitive to changes in the parameters UZSN and LZSN (upper and lower zone nominal soil moisture storage, respectively).
Larger errors in model estimates for the years 1990, 1992, and 1995 frequently were traced to the rainfall data used to drive the runoff simulations. The rainfall record used in the calibration was for only one gauge site, the Buffalo Airport. Observation of weather radar and qualitative notation from field personnel indicated that considerable spatial variability occurred in rainfall patterns for the area. Validation runs were conducted for the year 2000 using only the Buffalo Airport rainfall data and spatially averaged rainfall data that also included two other rain gauges within the watershed. The validation run with the spatially averaged rainfall data had a higher r2 and Nash Sutcliffe coefficient as compared to the validation run with the Buffalo Airport data alone.
A lengthy time series of observed suspended solids data was not available to calibrate the sediment erosion and transport component of the model. However, turbidity data measured continuously at various sites along the Buffalo River in 2000 were available and relationships between suspended solids and turbidity were developed using least squares regression. For the purpose of model calibration the daily mean turbidity values were run through the appropriate regression equation to construct a suspended solids time series. Results of the calibration run for an example river reach that represents Cazenovia Creek, near the city line, and for the lower Buffalo River, near the Ohio St. bridge were evaluated in detail. Visually, the observed and modeled sediment time series for the two reaches corresponded, although quantitatively, the r2 was low (in the range of 0.29). The model did a better job of representing erosion and transportation from the upper part of the watershed and had greater difficulty in representing the sediment deposition processes within the more hydraulically complex dredged channel (Ohio St. bridge site).
This paper is only available in PDF Format: View full text PDF