Event‐Based Evaluation of the GPM Multisatellite Merged Precipitation Product From 2014 to 2018 Over China: Methods and Results

Abstract

AbstractEvaluations of satellite precipitation products have been routinely conducted, often by calculating statistical indicators on fixed time intervals (e.g., hourly and daily). However, precipitation is actually a discontinuous variable in the form of events. Event‐based evaluations could provide additional information to those based on fixed‐interval statistics but have rarely been done. To address this issue, the Integrated MultisatellitE Retrievals for Global Precipitation Measurement (GPM) (IMERG) V06B product is evaluated with hourly rain gauge data from approximately 50,000 stations in China. We first assess the feasibility of using a satellite product to delineate precipitation event properties and then track satellite errors with the evolution of precipitation processes. Our results show a strong positive relationship between event duration and mean intensity in IMERG, which is not obvious in the gauge data. IMERG presents weaker peak‐to‐mean intensity ratio and later peak occurrence time than compared to the gauge data. Higher false alarm and lower miss proportions exist near the beginning of an event in IMERG, and this pattern reverses near the end of the event. IMERG underestimates the precipitation rate at the beginning of an event but overestimates later. The IMERG bias in event properties revealed here has important implication on its applications like flood forecasting and storm identification, and the life cycle‐dependent performance of IMERG we found indicates the need of event dynamics and stage information in satellite precipitation retrieval processes, which is not in current algorithms. The event‐based evaluation carried out here can be applied to other satellite precipitation products as well.