Evaluation and Comparison of Satellite Rainfall Products in the Black Volta Basin

Abstract

This study evaluated the performance of five satellite rainfall products (CHIRPS, PERSIANN, TRMM, RFE, and ARC) in the Black Volta Basin (BVB) using four performance evaluation methods: pairwise statistics, categorical statistics, rainfall intensity distribution, and extreme rainfall indices. In all, 21 rainfall stations distributed across the BVB with daily data spanning from 1981 to 2010 were used in the study. A high linear relationship was observed between observed and satellite rainfall data at decadal and monthly time scales as compared to weak relationship at the daily and annual time scales. The rainfall amount was least underestimated by CHIRPS at all the time scales. CHIRPS, PERSIANN and RFE performed well with the least deviation (BIAS ≤ 10%) from the observed rainfall amount at all time scales. Considering the high correlation coefficient and good NSE at decadal, monthly, and annual time scales, rainfall in the BVB is best represented by CHIRPS, followed by PERSIANN, RFE, ARC, and TRMM in that order. Though the probability of correctly detecting rainfall events is high (POD = 0.57–0.94), the satellite products were not able to adequately detect rainfall events in the basin at the daily time scale. The TRMM product was better in reproducing a very high rainfall amount (R ≥ 5 mm/day) in the basin as compared to CHIRPS, PERSIANN, RFE, and ARC. Extreme rainfall indices (R20, R99p, CWD and SDII) in the study basin were best represented by CHIRPS. Generally, precipitation in the BVB is best represented by CHIRPS, followed by PERSIANN, TRMM, RFE, and ARC in that order.