Abstract
Information regarding topographic, meteorologic, geologic, and geomorphic characteristics is increasingly available in spatially-explicit digital formats. This information, along with increased computing capabilities (such as faster computers, GIS, etc.), provides the ability to efficiently characterize these spatial features for specific areas. Of interest is whether the inclusion of new watershed characteristics may improve our ability to predict extreme hydrologic events, with particular emphasis on low streamflow prediction. Lumped models of these hydrologic processes (such as regional regression equations) often produce estimators with unacceptably large errors. Using a continuous digital elevation model (DEM) of the conterminous United States, watershed boundaries were developed for 1573 streamflow gauges of the USGS's Hydro-Climatic Data Network (HCDN). The historic discharge records at these sites are considered to be of the highest quality available in the United States. Using the watershed boundaries, numerous digital grids were employed to create a new database of watershed characteristics. The digital grids utilized include a 40-year monthly time series of PRISM's orographically weighted precipitation and temperature grids, the USGS's MUID (STATSGO) geology grids, and the original DEM from which many topographic estimators can be developed. Having grids in digital format allows the efficient processing of the grids, the ability to easily derive a variety of statistics from the grids, and provides more reproducible results than those obtained by manual processing. Specifically, the study investigates the inclusion of new parameters in lowflow regional regression models, and the changes in these models as they are compared across regions. Preliminary results indicate that regional regression models of low streamflow quantiles, which traditionally have very high model errors, could be improved in some regions by including topographic, climatic, and hydrogeologic statistics, though in many regions the models performed poorly.