Evaluation of Remotely Sensed Precipitation and Its Performance for Streamflow Simulations in Basins of the Southeast Tibetan Plateau

Abstract

our satellite-based precipitation products [TMPA real time (T-rt), its gauge-adjusted version (T-adj), Climate Prediction Center (CPC) morphing technique (CMORPH) real time (C-rt), and its gauge-adjusted version (C-adj)] were evaluated by a gauge-based synthesis dataset. Further, these products along with the CMORPH gauge?satellite blended version (C-ga), which is virtually C-adj in precipitation ungauged regions and is controlled by gauge analysis over regions of a dense station network, were intercompared with daily streamflow predicted by the distributed vegetation interface processes (VIP) model in the Lhasa and Gongbo basins of the southeast Tibetan Plateau. Results show these satellite-based products perform better in summer than in other seasons. Relative to the gauge-based synthesis dataset, for areal precipitation of the Lhasa basin from 2007 to 2010, biases of C-rt and T-rt are ?10.49% and 157.88%, respectively. Biases of C-adj and T-adj are 3.42% and 24.12%, respectively. The C-rt bias is underestimation of the volume of observed rainfall correctly detected and overestimation of the volume of falsely alarmed rainfall, while T-rt bias comes from overestimation of the volume of observed rainfall correctly detected. Simulation efficiencies of stream discharges driven by T-adj and C-adj are better than those by T-rt and C-rt, which are consistent with the accuracies of these products. With benchmarked model parameters using the gauge-based dataset, C-adj presents well for simulation, while T-adj needs parameter recalibration to achieve good skills. Compared to T-adj and C-adj, better simulation could be obtained by C-ga in precipitation-gauged regions.