Performance of satellite-based and reanalysis precipitation products under multi-temporal scales and extreme weather in mainland China

Gridded precipitation products are efficient supplements to surface meteorological gauges and thus a comprehensive evaluation of such products is essential. This study systematically evaluated the performance of 4 gridded precipitation datasets with hourly scale in mainland China, with ground gauge precipitation as a benchmark. In particular, we assessed the performance of two satellite-based precipitation estimations (SPEs): Integrated Multi-satellite Retrievals for GPM (IMERG) and Global Satellite Mapping of Precipitation (GSMaP), and two reanalysis precipitation estimations (RPEs): Modern-Era Retrospective analysis for Research and Applications, version 2 (MERRA) and Climate Forecast System Reanalysis (CFSR). The results reveal that all the products have better performance in low-latitude areas, and have worse performance in the high-latitude mountainous regions. SPE products have similar features of temporal-spatial performance and diurnal cycle estimation, and so have RPE products. The overall summer SPE accuracy surpasses that of all RPE products, and vice versa in winter. The error decomposition of each product linked the summer bias to excessive false events of light precipitation. In winter, the SPE error is generally a result of missed and false events, while that of the RPEs originates from false precipitation. The product accuracy increases with the time scale, and the daily MERRA accuracy exceeds that of other products. However, due to the challenge in representing convective precipitation, the accuracy of the diurnal cycle and peak hour is lower than that of the SPEs, and representing weaker amplitudes compared to the gauge precipitation. This consequently results in RPEs’ poorer performance in extreme heavy precipitation compared to the SPE products. The results of this study are useful for both developers and users of gridded precipitation products.