| At ungauged river sites a regional regression technique is often used to estimate the magnitude and frequency of floods and low streamflows, which are critical to effective management of water resources. In a region centered on eastern Tennessee and western North Carolina, flood and low streamflow regional regression models were constructed with input variables derived from 1,478 raster datasets. Of interest is whether models are improved by using raster datasets with finer horizontal resolution and newly employed raster datasets. The results of this study indicate that the model performance is not significantly impacted by the horizontal resolution of raster datasets. The performance of flood and low streamflow models was significantly improved with inclusion of hydrologic and hydrogeologic variables, respectively. A custom GIS application was developed to efficiently calculate descriptive statistics with a large number of raster datasets. The developed GIS application provides a powerful tool for various environmental studies. |
