Comparative analysis of bias correction methods for IMERG V06 precipitation products: case study in Guangxi, China

Abstract

ABSTRACT This study investigates the effectiveness of three bias correction methods: linear scaling (LS), local intensity scaling (LOCI), and quantile mapping (QM) in correcting the bias of Integrated Multi-SatellitE Retrieval for Global Precipitation Measurement (IMERG) V06 products (IMERG-Early run (IMERG-E), IMERG-Late run (IMERG-L) and IMERG-Final run (IMERG-F)) at different spatial and temporal scales in Guangxi, China. The results reveal that the LS and LOCI are effective in reducing the relative bias (RB) and root mean square error (RMSE) of IMERG products at the regional and monthly scales, while the QM tends to increase the overestimation or underestimation of precipitation. The QM outperforms the other two methods in improving the correlation coefficient (CC) of IMERG products at the monthly and daily scales, especially for stations with low correlation before correction. None of the methods can effectively correct the extreme values of IMERG products at a daily scale, and the correction efficacy of the three methods is limited for IMERG-F, which has a high initial accuracy. The QM has an advantage in enhancing the correlation between IMERG and gauge data. LS and LOCI have great potential in correcting the bias of IMERG near-real-time products.