Impact of the crucial geographic and climatic factors on the input source errors of GPM-based global satellite precipitation estimates

• Published in: Journal of hydrology (Amsterdam) 2019-08, Vol.575, p.1-16
• Main Authors: Chen, Hanqing, Yong, Bin, Gourley, Jonathan J., Liu, Jiufu, Ren, Liliang, Wang, Weiguang, Hong, Yang, Zhang, Jianyun
• Format: Article
• Language: English
• Subjects: Climate; Elevation; GPM-GSMaP; Mainland China; Season
• ISSN: 0022-1694; 1879-2707
• DOI: 10.1016/j.jhydrol.2019.05.020

•We investigated the error features of the latest GPM-GSMaP estimates by separating the input sources.•We found that the input sources exhibit larger biases in the semi-arid and arid regions.•The retrievals of the input sources are affected by topography in different extents.•The performance of imagers generally outperforms that of sounders.•GMI exhibits inadequacies in detection capability and has large biases in the winter and spring months. The impact of crucial geographic and climatic factors on the input source errors of integrated multi-satellite precipitation estimates is an important but still unclear issue for both algorithm developers and data users. This study primarily focused on the impacts of the twelve input sources used in the latest Global Satellite Mapping of Precipitation for Global Precipitation Measurement (GPM-GSMaP) for different climatic regions, elevations, and seasons over mainland China. Our evaluation results show that the error features of the input sources from several passive microwave and infrared sensors are related to the accuracy of GPM-GSMaP precipitation estimates. The input sources show larger hits, misses and false biases in the semi-arid and arid regions, for which the false bias was particularly significant. As for the seasonality, the input data sources exhibit a better performance in summer and have relatively lower hits and higher biases in winter. We also found that precipitation retrievals of the input sources are affected by topography in different extents. In terms of passive microwave sensors, the conical-scanning imagers generally outperform the cross-track-scanning sounders, but the sounders-based precipitation estimates have relatively better detection capability. As one of the core sensors for GPM, the microwave imager GMI reveals inadequacies in representing areal rainfall patterns and has relatively large biases especially in the winter and spring seasons, suggesting that the current GMI algorithm might need to be further improved.