Role of satellite and reanalysis precipitation products in streamflow and sediment modeling over a typical alpine and gorge region in Southwest China

Satellite and reanalysis precipitation products, as new and complementary data sources, are attractive for hydrometeorological applications, especially in data-sparse areas. This study evaluates the accuracy of two satellite precipitation products (TMPA 3B42V7 and PERSIANN-CDR) and one reanalysis precipitation product (NCEP-CFSR) against gauge precipitation observations with four statistical indices over the upstream of the Lancang River Basin (ULRB), Southwest China. The reliability and applicability of these precipitation products as inputs to a hydrological model (Soil and Water Assessment Tool, SWAT) for streamflow and sediment simulations are also assessed. Furthermore, we compare the spatial plots of extreme water yield (99 percentiles) and suspended sediment yield (99 percentiles) driven by the four precipitation sources, and investigate the spatial and temporal variability of water yield and suspended sediment yield over the ULRB. Results show that for direct comparisons with gauge precipitation observations, monthly TMPA 3B42V7 precipitation product performs the best at the basin scale with the smallest error and bias, and the highest correlation, followed by NCEP-CFSR, and PERSIANN-CDR. For modeling-based indirect inference, TMPA 3B42V7 presents great capability for streamflow and sediment simulations in the SWAT model on a monthly time step at the basin outlet, and PERSIANN-CDR also performs well. NCEP-CFSR shows acceptable skills in modeling sediment but unacceptable skills in modeling streamflow. Extreme water yield presents moderate spatial variability over the ULRB while extreme suspended sediment yield presents strong spatial variability. Water yield of this basin shows a decreasing trend during 1998-2008 while there is no obvious trend in suspended sediment yield in this period. (C) 2019 Elsevier B.V. All rights reserved.