资源环境科技发展态势分析平台

Evaluation of satellite-based quantitative precipitation estimates (QPEs) with reliable and independent ground-based measurements is important for both product developers and users. Here, we present a comprehensive evaluation on 3 high-resolution QPEs, namely, the Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS), the latest non-real-time post-processing version of Tropical Rainfall Measuring Mission (TRMM 3B42 V7), and the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Record (PERSIANN-CDR), in 3 basins with different climates in China. The accuracy of 3 QPEs in reproducing the spatial extent of daily and monthly precipitation (PR) as well as extreme PR indices was evaluated. Two simulation scenarios were utilized to evaluate the efficiency of hydrologic events forecasting of these 3 QPEs quantitatively. The results indicated that the 3 QPEs generally show high accuracy in estimating monthly PR in 3 basins, among which TRMM 3B42 V7 performs best (coefficient of determination R² < 0.94) followed by CHIRPS (R² < 0.91). However, all QPEs tend to overestimate daily PR of 3 basins, resulting in low accuracy at the daily scale (R² < 0.35). For estimation of the extreme indices, the 3 QPEs show large differences in the spatio-temporal accuracy, but all with better performance in humid (R² < 0.86) than arid (R² < 0.7) basins. Similarly, all 3 QPEs show better performance in simulating streamflow in humid than arid basins. TRMM 3B42 V7 (Nash-Sutcliffe coefficient of efficiency, NSCE < 0.96) and CHIRPS (NSCE < 0.9) perform better in simulating streamflow in humid basins than PERSIANN-CDR (NSCE < 0.88), while PERSIANN-CDR performs best in arid basins (NSCE<0.67). However, 3 QPEs mostly underestimate peak flow and overestimate soil moisture in all basins, suggesting that the necessity of improving hydrologic efficiency for all of them.