A comprehensive assessment of precipitation products: Temporal and spatial analyses over terrestrial biomes in Northeastern Brazil

Abstract

Precise rainfall estimations at spatio-temporal resolution are vital for numerous applications, including agricultural monitoring, hydrometeorology, and water resources management. Northeast Brazil (NEB) is vulnerable to extreme interannual climate variability and precipitation is closely linked to the livelihoods of its citizens and the regional health of the economy. Satellite rain data is an alternative source of information for regions such as NEB. Thus, this study aimed to evaluate the performance of monthly rainfall estimations derived of latest versions from five precipitation products: (i) CHIRPS, (ii) ERA5-Land, (iii) TerraClimate, (iv) TRMM, and (v) IMERG in the NEB and comparing them to the observed precipitation for the period 2001–2019. The results showed that the satellite-based products were more regularly correlated with the observed precipitation than the model-based products. CHIRPS, IMERG and, TRMM showed an average Pearson's correlation coefficient (r) value above 0.93. All products tended to present lower relative error in the rainy months; TerraClimate gave the highest average value of root-mean-square error (RMSE) for NEB (41.61 mm/month). The best global performance was observed in the Amazon biome (r ∼ 0.90), and the poorest in the Atlantic Forest biome, which tended to underestimate the precipitation. In spite of the results in the Atlantic Forest, this study demonstrated that products of precipitation can be a useful complement to rainfall data for the NEB. The comprehensive assessment of precipitation products reported here will serve as a reference for managers and researchers to select the ideal product for its intended application and associated uncertainties.