Capability of satellite data to estimate observed precipitation in southeastern South America

Abstract

Precipitation is a fundamental component of the water cycle. Satellite-derived precipitation estimates with high spatial resolution and daily to subdaily temporal resolution become very important in regions with a limited ground-based measurement network, such as southeastern South America (SESA). This study evaluates the performance of four state-of-the-art satellite products, including IMERG V.06 Final Run, PERSIANN, PERSIANN CCS-CDR and PDIR-NOW in representing observed precipitation over SESA during the 2001–2020 period. The ability of each product to represent observed annual and seasonal precipitation patterns was assessed. Statistical and categorical evaluation metrics were used to evaluate the performance of satellite precipitationestimates at monthly and daily timescales. Our results report that IMERG and CCS-CDR achieve the best performance in estimating observed precipitation patterns at annual and seasonal timescales. While all satellite products effectively identify autumn and spring precipitation patterns, they struggle to represent winter and summer patterns. Notably, all satellite precipitation productshave a better agreement with observed precipitation in wetter regions compared to drier regions, as indicated by the spatial distribution of continuous validation metrics. IMERG stands out as the most accurate product, reaching the highest correlation coefficients (0.75 < CC < 0.95) and Kling–Guptaefficiencies (0.65 < KGE < 0.85, rate as good to very good performance). Regarding categorical statistical metrics, IMERG correctly estimates the fraction of observed rainy days (POD > 0.7, CSI > 0.6) and shows the lowest fraction of estimated precipitation events that did not occur. PERSIANN, CCSCDR and PDIR-NOW exhibit lower performances, mainly in drier areas. Moreover, PERSIANN and PDIR-NOW tend to overestimate observed precipitation in almost the entire SESA region. We expect this validation study will provide greater reliability to satellite precipitation estimates, in order to provide an alternative that complement the scarce observed information available for decision-making in water management and agricultural planning.