Validating CHIRPS-based satellite precipitation estimates in Northeast Brazil

Satellite-based rainfall is an alternative source of information for regions such as the Northeast Brazil (NEB) where there are large areas that do not have ground observation stations. In this study, the monthly rainfall derived from the satellite-based rainfall product, Climate Hazards Group InfraRed Precipitation with Stations (CHIRPS v.2), is compared with observation from 21 ground stations in the NEB, for the period 1981-2013. Various metrics based on pairwise comparison were applied to evaluate its performance in estimating rainfall amount and rain detection capability. Results show that the CHIRPS data correlate well with observations for all stations (r = 0.94), but tend to overestimate low and underestimate high rainfall values (>100 mm/month). Although CHIRPS achieves better results during the wet season (March to May; bias = -4.60%), its ability for the rain detection is poor (probability = 0.44). The best global performance was noted in the Cerrado biome (r = 0.93; rain detection = 0.53), but fails to detect rain in other biomes (r = 0.92; rain detection = 0.16). The study concluded that the CHIRPS v.2 dataset can be a useful substitute for rain-gauge precipitation data outside the semiarid NEB. (C) 2016 Elsevier Ltd. All rights reserved.