资源环境科技发展态势分析平台

This study evaluates the performance of latest version 5B (V5B) Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (IMERG) Final Run products during a 60-year return extreme precipitation storm on 7 May 2017 over southern China with gauge observations as the reference dataset. Version 4 (V4) Global Satellite Mapping of Precipitation (GSMaP) products and quantitative precipitation estimates derived from a local ground-based S-band dual polarization weather radar (radar, hereafter) were used for parallel comparisons. The satellite-only products (IMERGUncal and GSMaP_MVIO and gauge-corrected products (IMERGCaI, GSMaP_Gauge) were selected for this study. The results showed that: 1) GSMaP_MVK, IMERGUncal, GSMaP_Gauge and IMERGCaI generally capture the spatio-temporal patterns of storm-accumulated rainfall with correlation coefficient (CC) values about 0.76, 0.70, 0.68 and 0.72, respectively, while radar was well correlated with gauge measurement (CC about 0.94); 2) The GSMaP_Gauge (-19.38%), IMERGCal (-40.23%), GSMaP_MVK (-57.12%), IMERGUncal (-58.77%) satellite-based precipitation products all underestimated the storm-accumulated precipitation, while ground radar overestimated by 27.48%; 3) Both IMERGCal and IMERGUncal outperformed their GSMaP counterparts in capturing the time-series with much higher CC (0.50 vs. -0.21, 0.51 vs. 0.17); 4) Among the satellite-based QPE products, when the rainfall rates are < 5 mm/h, IMERGCal performs best with highest CSI. The V4 GSMaP standard products and latest V5B IMERG final run products still had resolution and accuracy limitations in estimating extreme precipitation for this type of warm sector rainfall storm. These findings will help developers of rainfall retrieval products to understand the errors and also help improve modeling of hydrological extremes.