Comparison of GPM IMERG, TMPA 3B42 and PERSIANN-CDR satellite precipitation products over Malaysia

Abstract The launch of the Global Precipitation Measurement (GPM) mission has prompted the assessment of the newly released satellite precipitation products (SPPs) in different parts of the world. This study performed an initial comparison of three GPM IMERG products (IMERG_E, IMERG_L and IMERG_F) with its predecessor, the TMPA 3B42 and 3B42RT products, and a long-term PERSIANN-CDR product over Malaysia. The performance of six SPPs was evaluated using 501 precipitation gauges from 12 March 2014 to 29 February 2016. The annual, seasonal, monthly and daily precipitation measurements were validated using three widely used statistical metrics (CC, RMSE and RB). The precipitation detection capability (POD, FAR and CSI), probability density function (PDF) and the 2014–2015 flood event analysis were also considered in this assessment. The results show that all the SPPs perform well in annual and monthly precipitation measurements. The spatial variability of the total annual precipitation in 2015 is well captured by all six SPPs, with high precipitation amount in southern East Malaysia, and low precipitation amount in the middle part of Peninsular Malaysia. In contrast, all the SPPs show moderate correlation at daily precipitation estimations, with better performance during the northeast monsoon season. The performance of all the SPPs is better in eastern Peninsular Malaysia, but poorer in northern Peninsular Malaysia. All the SPPs have good precipitation detection ability, except the PERSIANN-CDR. All the SPPs underestimate the light (0–1 mm/day) and violent (> 50 mm/day) precipitation classes, but overestimate moderate and heavy (1–50 mm/day) precipitation classes. The IMERG is shown to have a better capability in detecting light precipitation (0–1 mm/day) compared to the other SPPs. The PERSIANN-CDR has the worst performance in capturing all the precipitation classes, with significant underestimation of light precipitation (0–1 mm/day) class and overestimation of moderate and heavy precipitation classes. The IMERG near-real time products with finer temporal and spatial resolutions can be regarded as a reliable precipitation source in studying the 2014–2015 flood event in Malaysia.