资源环境科技发展态势分析平台

Satellite precipitation estimates (SPEs) provide important alternative precipitation sources for various applications especially for regions where in situ observations are limited or unavailable, like central Asia. In this study, eight SPEs based on four different algorithms, namely, the Tropical Rainfall Measuring Mission Multisatellite Precipitation Analysis 3B42, Climate Prediction Center morphing technique (CMORPH), Global Satellite Mapping of Precipitation (GSMaP), and Precipitation Estimation from Remotely Sensed Imagery Using Artificial Neural Networks are evaluated by using an improved evaluation system over central Asia with respect to their performance in capturing precipitation occurrence and magnitude. Both satellite-only and gauge-corrected versions are assessed against gauge-gridded reference from June 2001 to May 2006. Main results show that all SPEs have difficulties in accurately estimating mountainous precipitation with great overestimation/underestimation in both winter and summer. In winter, CMORPH products fail to capture events over ice-/snow-covered region. In summer, large overestimations dominated by positive hit bias and missed precipitation are found for all products in northern central Asia. Interestingly, 3B42 and CMORPH products show great false alarm percentages (up to 90%) over lake region, which is more significant in summer than in winter. Significant elevation-dependent errors exist in all products, especially for the high-altitude regions (>3,000 m) with missed error and hit error being the two leading errors. Satellite-only products have large systematic and random errors, while the gauge-corrected products demonstrate significant improvements in reducing random errors. Generally, the gauge-corrected GSMaP performs better than others with good skills in reducing various errors.