Improving HEC-HMS Streamflow Prediction Through Localized Bias Correction of Stage IV Radar Rainfall: A Case Study of the Elm Fork Trinity River Basin During Tropical Storm Hermine

Abstract

Precipitation is the principal cause of runoff production and the source of much uncertainty in hydrologic modeling. While National Weather Service (NWS) Stage IV multisensor precipitation data have good spatial and temporal resolutions, localized biases could still be present in watershed-scale hydrologic modeling. Objectives: This study evaluates whether localization of Mean Field Bias (MFB) correction for NWS Stage IV precipitation can improve streamflow prediction in Elm Fork Trinity River Basin under the influence of Tropical Storm Hermine from September 5 to September 9, 2010. Methods: Rainfall hourly measurements from three Automated Surface Observing System (ASOS) stations were used for estimation of MFB correction factor. Three Hydrologic Engineering Center Hydrologic Modeling System (HEC-HMS) simulations were performed for (1) spatially interpolated rainfall data from gauging network, (2) raw Stage IV precipitation data, and (3) MFB-corrected Stage IV precipitation data with the same basin settings. Results: The correction reduced precipitation PBIAS to about 0% at all three ASOS stations. At Gainesville Airport station, RMSE of precipitation was reduced from 6.77 mm to 5.88 mm. NSE of simulated streamflow was increased from 0.723 to 0.882, and the peak discharge was decreased from 11,210 ft³/s to 8,574 ft³/s (versus 6,560 ft³/s observed peak flow). Conclusion: Localized MFB correction for precipitation improved hydrologic modeling for this storm event using only three rain gauges without changing HEC-HMS basin settings.