Impacts of Radar Indicated Rainfall on Distributed Rainfall-Runoff Modeling

Abstract

A case study is presented in which the impact of using spatially distributed, radar indicated, rainfall for lumped-distributed rainfall runoff modeling is demonstrated. The calibrated model was used to prioritize siting of runoff detention devices to minimize flood flows and sediment erosion in North Fish Creek, Bayfield County, WI. Modeling for the study demonstrated the importance of using bias adjusted, radar indicated rainfall for calibration. A HEC-HMS model with 97 sub-basins with 1km2 average area was calibrated to a full summer of continuous simulation that included two floods of historic magnitude. The model employs explicit soil moisture accounting, kinematic wave overland flow, and Muskingum-Cunge channel routing. The Preistly-Taylor technique was used to estimate evapotranspiration with solar radiation, temperature, and a crop coefficient to track soil moisture losses in continuous simulation. A suite of distributed parameters were developed, using satellite derived, land cover, SSURGO soils data, checked with field sampling, and digital elevation data interpolated from 1:24K topographic maps. Radar rainfall estimates were derived using NEXRAD level III data at a spatial resolution of roughly 500m. These data were smoothed and sampled to a 10m grid in order to better represent rainfall gradients. It was also bias corrected to match the cumulative rainfall observed by a USGS rain gage at the modeled watershed outlet. To examine the impact of using spatially and temporally distributed rainfall, three scenarios are presented and compared. First, the radar indicated precipitation for each of the 97 sub-basins is applied. Second, the mean radar indicated precipitation is applied to the entire watershed, using the time distribution from the rain gage. Finally, the rain gage indicated rainfall is applied uniformly over the watershed. The scenarios are compared, based upon the measured hydrographs at the stream gage.