Ground validation and error decomposition for six state-of-the-art satellite precipitation products over mainland China

Evaluating the quality of satellite precipitation products (SPPs) is crucially important for SPPs' applications and improvements. This study aims to comprehensively evaluate the accuracy of six widely used SPPs (GSMaP, IMERG, TMPA, CMORPH, PERSIANN, and CHIRPS) from 2000 to 2017 over mainland China using 2372 rain gauges. The performance of SPPs at different spatio-temporal scales and altitudes are explored. The probability density function (PDF) at different precipitation intensities is also considered. In addition, special attention has been paid to analyze the error components of SPPs from different sources. Results demonstrate that: (1) GSMaP outperforms other SPPs with good statistical metrics, followed by IMERG. PERSIANN and CHIRPS have more space for improvement. All SPPs perform well in summer while worse in winter. Meanwhile, the accuracy of all SPPs varies region by region, which is better in southeast coastal basins than in northwest inland basins. (2) The overall performance in low-altitude regions is superior to high-altitude regions. The KGE (or RMSE) shows a significant negative (or positive) correlation with altitude, whereas POD, FAR, and CSI exhibit a slight dependence on altitude. (3) TMPA can well reproduce the PDF of different precipitation intensities. All SPPs' PDF is more consistent with gauges for moderate (5–10 mm/d) and heavy precipitation (10–50 mm/d), while deviates from gauges for tiny/light (0.1–5 mm/d) and violate precipitation (>50 mm/d). (4) The error components of each SPP could be larger than total errors, and the false bias is the major error source. GSMaP achieves a better performance with low hit and missed bias, whereas CHIRPS behaves poorly with large hit, missed and false bias. On the whole, the study demonstrates the error characteristics and accuracy performance of six mainstream SPPs over mainland China, which is expected to provide valuable information for hydrometeorology applications.