Comparison of input precipitation sources in streamflow forecasting
Avant, Brian Keith
MetadataShow full item record
Hydrologic modeling requires meteorologic and landscape data as inputs for streamflow forecasting and prediction. While many remote areas lack adequate ground-based data for model calibration and use, remotely sensed (e.g., satellite) data can provide the information needed for global applications. Sources of remotely sensed meteorologic data include TRMM, GPM, and GRACE, as well as reanalysis data, such as MERRA and GLDAS. This study investigates the accuracy of hydrologic modeling when satellite data are used in place of ground-based meteorological data. A spatially distributed streamflow model, Hydrologic Simulation Program-FORTRAN (HSPF), is used to model six watersheds (2,315 to 11,400 km$^2$) in the Coastal Plain and Piedmont physiographic provinces of the Southeastern United States. Observed streamflow is compared with modeled streamflow using meteorologic data from both ground-based and satellite sources. The fitting error and bias were computed for each input product/method scenario using BASSET, a baseflow separation program developed for this study. TRMM precipitation and MERRA potential evapotranspiration data are shown to be suitable substitutes for ground-based meteorological data for the watersheds examined. TRMM/MERRA coupled inputs actually perform better (i.e., provide a better fit to observed data) than ground-based inputs for the largest watershed modeled.