Globally gridded precipitation products (GGPPs) are commonly used in impact assessments as substitutes for weather station data, each with unique strengths and limitations. Reanalysis products are among the most widely used for driving impact models, evaluating climate models, or bias-correcting and downscaling model outputs to generate climate change projections. However, they are often outperformed in accuracy by other GGPPs, particularly in tropical regions, including areas of the Global South. Therefore, we assessed the utility and suitability of GGPPs for climate and health research by examining how differences and uncertainties in these products affect area-level precipitation estimates, often used in health studies when epidemiological data are linked to administrative units. We compared reanalysis (ERA5/-Land) with satellite-based (CHIRPS, PERSIANN-CDR) and interpolated gauge-based products (CRUTS, GPCC), each a viable candidate to serve as reference climatology in climate change impact assessments. We focused on seasonal patterns, disease-related bioclimatic variables, and climate change-relevant indices, such as the number of wet or dry periods. Our findings revealed substantial variation in the accuracy of local precipitation estimates across GGPPs, with differences in maximum pixel precipitation values exceeding 75% between ERA5-Land and CHIRPS. These differences in GGPPs translated into area-level precipitation and, consequently, in vector carrying capacity estimates, demonstrating their impact on health assessments. Our analysis focused on Brazil and Colombia, two diverse countries differing for example, in orography, climate, and size. Each product was evaluated against national station data. Our results indicate that estimating tropical precipitation is particularly challenging for reanalysis, while CHIRPS demonstrated the best overall performance.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)