Unraveling divergences: disagreement between precipitation datasets and stations in tropical South America

Abstract Gridded precipitation datasets are widely used to monitor hydroclimatic trends, yet their accuracy in tropical South America remains uncertain. Here, we systematically compare trends from state-of-the-art precipitation datasets with station data across the region. Our results reveal substantial disagreements in both the magnitude and direction of trends, with datasets frequently failing to capture observed station trends. Trend accuracy varies considerably, with better agreement in more densely sampled areas like the Caatinga, while the Western and Northern Amazonia show lower accuracy. Moreover, we find limited agreement between datasets and stations regarding documented patterns, such as southern Amazonia drying during the dry season and Western Amazonia wetting in the wet season. An exception is the Southeastern Cerrado, where most datasets consistently capture significant drying trends in the wet season. Our findings highlight the need for a multi-dataset approach and an improved observational network to assess precipitation changes in tropical South America.