Precipitation Data Products for South America
South America's precipitation monitoring relies on a combination of satellite-based estimates, atmospheric reanalyses, and gauge-interpolated datasets, each with distinct strengths and limitations. The principal global products used across the continent are CHIRPS, GPM IMERG, ERA5 and ERA5-Land, and NASA POWER, alongside regionally tailored products such as MERGE (Brazil) and PISCO (Peru). The choice of product depends strongly on the target region, climate regime, elevation, and intended application.
Satellite-gauge blended products combine the spatial coverage of satellite infrared and microwave observations with the accuracy of ground-based rain gauges. CHIRPS, developed for the Famine Early Warning Systems Network, is the most extensively validated product across South America and was designed to provide GPCC-like performance with lower latency.[^c1] It is the only product that consistently produces unbiased precipitation estimates for most catchments in large-scale Brazilian evaluations.[^c4] GPM IMERG, produced by NASA and JAXA, provides the highest temporal resolution at half-hourly intervals and underwent major algorithm revisions in version 07.[^c3] NASA POWER, developed for agroclimatological applications, draws its precipitation fields from the [[MERRA-2]] reanalysis and has no independent gauge adjustment, making it more reliable for temperature and radiation than for rainfall.
Atmospheric reanalyses like ERA5 and ERA5-Land provide physically consistent meteorological fields by assimilating historical observations into numerical weather prediction models. ERA5 offers substantial improvements over its predecessor, including a resolution increase to 31 km and an all-sky radiance assimilation approach that corrected long-standing precipitation biases.[^c2] However, reanalysis precipitation in the tropics remains challenging: ERA5-Land shows systematic overestimation in the Bolivian highlands and performs poorly for rainfall in Brazilian evaluations, with no added improvement from the finer-resolution land replay.
No single product is universally best across the entire South American continent. Performance varies with elevation — all products degrade above 2,000 m in the Andes — and with climate regime, as demonstrated by validation studies in Uruguay, Colombia, Argentina, and Brazil that consistently show the importance of region-specific product selection based on local topography and climate.[^c8][^c5] Gauge density is a critical factor: the sparse station networks in the Amazon basin and the high Andes limit both the training and the validation of precipitation products.[^c6]
The choice of precipitation product has direct consequences for downstream applications. A 2025 study comparing six global gridded products across Brazil and Colombia found substantial variation in local precipitation estimates, with maximum pixel differences exceeding 75% between ERA5-Land and CHIRPS.[^c9] Reanalysis products performed particularly poorly in tropical settings, while CHIRPS demonstrated the best overall performance.[^c10] These differences propagated into health-oriented calculations such as vector carrying capacity for disease transmission modeling, demonstrating that product choice has tangible effects beyond climate research.[^c11] Discrepancies between products persisted even after regridding to a common spatial resolution, indicating that area-level aggregation alone does not resolve inter-product differences. The same study found that ERA5-Land systematically produced more wet days and fewer consecutive dry days than CHIRPS and PERSIANN-CDR, particularly in the Amazon and the Andes, with consecutive dry days 30–60% lower than CHIRPS in some regions, indicating substantially different characterizations of drought risk.[^c12] The authors recommended that health impact studies assess the sensitivity of their findings to the choice of precipitation product and, where possible, use multiple products to characterize uncertainty, identifying CHIRPS as providing a good balance of spatial resolution, temporal coverage, and accuracy for health applications in Brazil and Colombia.[^c13]
Bias correction methods, particularly quantile mapping, are widely applied to improve the accuracy of gridded precipitation products at local scales across South America.[^c7] Downscaling approaches including geographically weighted regression and multi-scale GWR further refine IMERG and other coarse-resolution products for catchment-scale hydrological applications.
The choice of precipitation product also has significant implications for hydrological assessments. Multi-catchment, multi-dataset evaluations across 95 South American catchments have found that ERA5 departs substantially from observation-based estimates with systematic biases and physically implausible water-storage changes, raising concerns about its suitability for large-sample hydrological studies in the region.[^c14] Flood regime analyses show that ENSO strongly modulates flood probability across the continent: flood probabilities increase by more than 120% during El Niño in the La Plata Basin and during La Niña in the northern Amazon.[^c15] Streamflow extremes respond more strongly than precipitation, indicating that hydrological amplification of climate variability must be accounted for in flood risk assessments.[^c16] Under climate change projections, southern Brazil may face up to fivefold more frequent floods while the Amazon and Pantanal may see reduced flood occurrence.[^c17]