Inspection of IMERG precipitation estimates during Typhoon Cempaka using a new methodology for quantifying and evaluating bias

Abstract

Bias analysis is a significant component in the performance evaluation of satellite precipitation products (SPPs), and the relative bias (RB) is usually adopted as a key metric in all relative studies. However, when calculating the RB, severe overestimation (positive RB) and underestimation (negative RB) with the opposite sign can be offset through summation when a small temporal scale changes to a large temporal scale, and the final result will ignore the actual situation, which weakens the validity and reliability of bias evaluation. To solve the above problem, the typhoon presenting capability (TPC), which consists of two metrics, the presenting capability indicator (PCI) and the bias type discriminant (BTD), was developed as a new methodology for error evaluation during tropical cyclone events. In this study, by applying the TPC, two quasi-real-time Integrated multi-satellite retrievals for global precipitation measurement (IMERG) products (i.e., the early run and late run) acquired during Typhoon Cempaka in July 2021 were evaluated. The main conclusions are as follows: (1) underestimation of the precipitation by the two IMERG products tended to occur at higher hourly rainfall intensities (>2 mm at gauge scale and 100 mm at regional scale), while overestimation occurred at lower intensities. (2) The false-alarm rate (FAR) and RB values were significantly higher in the Pearl River Delta (PRD) than in other regions of the study area. (3) Through calculation of the TPC, a positive and negative offset phenomenon was identified over the region around Zhanjiang City in western Guangdong. By introducing the TPC for performance evaluation of IMERG products during Typhoon Cempaka, it was concluded that TPC is an advanced replacement of the RB for disclosing bias situations when adopting IMERG to monitor and reconstruct typhoon processes.