Performance of PERSIANN satellite precipitation products in a complex-terrain environment: Rio de Janeiro, Brazil

Abstract This study evaluated the performance of the satellite-based precipitation products PDIR-NOW and PERSIANN-CCS, both part of the PERSIANN system, in representing monthly and seasonal mean precipitation over the municipality of Rio de Janeiro (RJM), Brazil, during the period 2003–2024. The estimates were compared with observational data from the Alerta Rio monitoring network, aiming to assess the ability of these products to reproduce the spatial and temporal variability of precipitation in a region strongly influenced by orography. The evaluation was conducted through analyses of the spatial distribution of mean precipitation, the temporal evolution of monthly accumulations, the spatial patterns of mean error, and statistical metrics summarized using Taylor diagrams. The results indicate that both PDIR-NOW and PERSIANN-CCS are able to capture the seasonal pattern of the RJM rainfall regime, with better performance during summer and the rainy season, when higher correlations with observations and lower error values were observed. However, systematic biases were identified, with a tendency toward precipitation underestimation by PERSIANN-CCS and overestimation by PDIR-NOW, varying according to the season. Transitional seasons exhibited contrasting performances, with relatively better results for PDIR during autumn and for CCS during spring, whereas winter emerged as the period of poorest performance for both products, characterized by generalized underestimation, low correlation, and larger discrepancies relative to observations. Consistently, the main limitations of the PERSIANN products were associated with their inability to adequately represent topographic effects, resulting in the omission of the highest precipitation cores over the RJM mountain ranges in all seasons. These findings highlight the need for regional adjustments to satellite-based precipitation algorithms and for integration with local observational data in order to improve precipitation representation in areas with complex terrain.