Impacts of different precipitation data sources on water budgets

Radar and rain-gauge precipitation datasets are used to explore the impacts of different precipitation data sources on water budgets simulated by the Three-Layer Variable Infiltration Capacity (VIC-3L) land surface model over the watershed of the Illinois River at Watts, Oklahoma. Hourly grid-based NEXRAD (Next Generation Radar) Stage III precipitation data (approximately 4 X 4 km(2)) were aggregated in time and space to daily precipitation at 1/8 degree and compared to the University of Washington (UW) daily precipitation data, which were gridded at 1/8 degree based on rain-gauge precipitation data. Comparisons of the temporal cumulative precipitation magnitudes indicate that the NEXRAD precipitation is smaller than the UW data and the rain gauge measurements. Hyetographs obtained from the NEXRAD data in general appear to be narrower with higher peaks. In addition, the NEXRAD data are better in capturing the precipitation spatial distributions than the UW data. Investigations of water fluxes based on simulations forced by the two types of precipitation datasets suggest that realistic streamflow simulations, compared to the observed daily streamflow at the outlet of the Illinois River at Watts, can be obtained if model parameters are calibrated. Compared to soil moisture of the total zone, runoff and evapotranspiration are more sensitive to the temporal and spatial distributions of precipitation, with runoff being most sensitive. Also, runoff and evapotranspiration obtained based on the NEXRAD precipitation data show more spatial heterogeneities than those obtained by using the UW precipitation data. Moreover, the magnitudes of the components of the water budget can be strongly impacted by the VIC-3L calibrated parameters, but their spatial distribution patterns may not be significantly affected. (C) 2004 Elsevier B.V. All rights reserved.