Evaluation of Monitoring Capability of GPM-Series Precipitation Products for Beijing "23·7" Heavy Rainfall Event

This study aimed to evaluate the monitoring capability of remote sensing precipitation products associated with global precipitation measurement (GPM), specifically the GSMaP and IMERG products during heavy rainfall events. By using the Beijing "23·7" heavy rainfall event as a case study, based on 3-hourly observation data from ground stations within the study area, both near-real-time remote sensing precipitation products (GSMaP_NRT, IMERG_ER, and IMERG_LR) and post-processed remote sensing precipitation products (GSMaP_MVK, GSMaP_Gauge, and IMERG_FR) were comprehensively evaluated and analyzed. Precipitation statistical metrics (CC, RMSE, P-BIAS, and KGE) and precipitation detection metrics (POD, FAR, CSI, and HSS) were used to systematically evaluate the products' monitoring capability for heavy rainfall events across different spatial and temporal scales. Additionally, the temporal variation of the precipitation products was thoroughly analyzed based on the spatial scale of the study area. The results indicate that IMERG_FR shows the best overall precipitation monitoring capability (P-BIAS = -0.41, KGE = 0.54). The GSMaP_NRT and GSMaP_MVK demonstrate strong performance in characterizing the distribution of precipitation, while the IMERG series performs better in metrics with more concentrated distribution (0.48 < KGE < 0.54, 0.59 < HSS < 0.66). Near-real-time products generally overestimate extreme precipitation, while post-processing correction introduces additional instability while correcting the bias. Both types of products are affected by topographic factors and perform poorly in areas with elevation transitions (KGE < 0.5). Additionally, both types of products exhibit temporal advances or delays during extreme precipitation events.