Semi-distributed modelling of basin hydrology with radar and gauged precipitation

Even though gauged rainfall data generally provide accurate depth measurements, sparsely spaced, gauging stations cannot effectively account for the spatial variability of precipitation at basin scale. On the other hand, radar data such as the WSR-88D stage III radar rainfall data can generally capture the spatial variability of rainfall fields, but tends to underestimate rainfall depth of stratiform storms, or both convective and stratiform storms if a storm is of low intensity. To take advantage of both the strength of radar data (mapping accurate spatial variability of rainfall) and that of gauge data (accurate depth measurements), the two data sets were merged together by the Statistical Objective Analysis (SOA) scheme. The event-based hydrologic experiments using a semi-distributed, physics-based hydrologic model (distributed physically based hydrologic model using remote sensing, DPHM-RS) revealed that WSR-88D Stage Ell radar rainfall data simulated more accurate runoff hydrographs than gauged data for convective storms but less accurate runoff hydrograph for stratiform storms, because radars measured slightly more rainfall than gauges for convective storms, but substantially less rainfall for stratiform storms. However, after merging WSR-88D stage III radar data with gauge data by SOA, the radar's underestimation of stratiform storm depth decreased substantially, but the adjustment could be counter productive for convective storms. Results show that rainfall spatial variability, depths, and hydrologic model resolution play a major role on the accuracy of simulated runoff volumes and peak flows. Copyright (c) 2006 John Wiley & Sons, Ltd.