A Hydrologic Model Calibration Exercise for Regional Climate Change Impact Assessment of the Conterminous U.S

Numerous studies have investigated the hydrological impacts of climate change in the U.S. using projections from multiple general circulation models downscaled by means of regional climate models, statistical methods, and hydrologic models. Most of these studies focused on a small number of local watersheds without consideration to larger-scale regional climate change impacts, or utilized macro-scale hydrologic models with coarser spatial resolution that are insufficient to characterize the delicate surface hydrology. To improve the results of regional hydro-climate impact assessment, there is a need for better spatial coverage as well as resolution of hydrologic models. The main challenge has been the availability of a comprehensive set of higher resolution calibrated physical parameters. Focusing on the need of regional hydro-climate impact assessment, a data-intensive hydrologic model calibration exercise is performed for over 2000 USGS hydrologic Subbasins (HUC8) in the conterminous U.S. at the resolution of 1/24th degree (~4km). Both USGS WaterWatch monthly runoff and NWIS daily gage observation are used to calibrate the baseline variable infiltration capacity (VIC) hydrologic model. Several statistical matrices are used to evaluate the model performance at each HUC8, including the Pearson correlation coefficient (R), root mean square error (RMSE), Nash-Sutcliffe model efficiency coefficient (NSEC), bias (B) and the percent bias (PB). The overall results show that the physical models simulate closely the observed values with about sixty four percent of the HUC8s having an average NSEC of 0.95 The model performance was vastly better in wet region basins than they were in arid region. The current baseline VIC model can hardly be improved in arid and desert regions (covering about twenty percent of the HUC8s) where the NSEC values are below zero. Overall, the new 4-km model implementation for the conterminous U.S. shows promising improvement over the ones evaluated at 12-km resolution.