Multi-scale evaluation of six high-resolution satellite monthly rainfall estimates over a humid region in China with dense rain gauges

Abstract

This study evaluates and compares the performance of six high-resolution monthly satellite rainfall estimates (SREs), which include TRMM 3B43V6, TRMM 3B42RTV6, CMORPH, GSMaP MWR+, GSMaP MVK+, and PERSIANN, with dense ground rain gauges located in Ganjiang River Basin. The performance was evaluated at multiple spatial scales: the 0.25° × 0.25° grid, sub-catchment, and the whole basin. It was observed that 3B43V6 generally performed well and was able to capture the ground benchmark rainfall with slight overestimation, whereas all of the other SREs suffered large underestimation in the study area. Among the five pure satellite-derived products, 3B42RTV6 and CMORPH performed better, whereas PERSIANN performed the worst. All of the SREs except 3B43V6 showed a strong seasonal signature with much better performance in the wet season than in the dry season. The results also indicate that SREs performed better in the southeast and central regions, whereas poor performance was observed in the western mountains and in the northern plains. Furthermore, the spatial patterns of SREs errors are influenced mainly by the local terrain. The performance of SREs improved when the spatial scale was increased, whereas the performance reduced when the temporal scale was increased from month to year.