Performance evaluation of IMERG products based on the extremely heavy rainstorm event (2021) once in a thousand years in Henan, China

Abstract

Henan, a province of central China, experienced rainfall at the level of rare extremely heavy rainstorms once in a millennium during the period from July 17th to July 23rd, 2021. Based on ground-based rainfall data from 116 gauges, the performance of Global Rainfall Measurement (GPM) Integrated Multi-satellite Retrievals for GPM (IMERG) products (IMERG-Early run (IMERG-E), IMERG-Late run (IMERG-L), and IMERG-Final run (IMERG-F)) were evaluated from the seven-day rainfall change process, the hourly rainfall change process on July 20th and the six traditional indexes. Results indicate that the overall performance of IMERG-F is the best, relative bias (RB) is −0.03 and correlation coefficient (CC) is 0.51. IMERG-L (RB, −0.09; CC, 0.49) is very close to IMERG-F More outstandingly, in the three days of 19th, 20th and 21st, RB of IMERG-L per day is −0.01, 0.02, and − 0.13, respectively. IMERG-L shows a greater ability to capture extreme short-duration heavy rainfall in the peak period compared to the other two products. Since weak rainfall overestimation coefficient (WROC) is >0.5, it can be concluded that RB of excessively overestimated gauges are >100% mainly determined by the overestimation during weak rainfall; Because of overestimation coefficient (OC) and weak rainfall coefficient (WRC), there is an obvious error trend that gauges with weak rainfall (≤ 1 mm/h) are easily overestimated and gauges with non-weak rainfall (> 1 mm/h) are easily underestimated; IMERG-F has higher entire accuracy and reduces the error trend because of the large number of PMW estimates.