Comprehensive Global Assessment of 24 Gridded Precipitation Datasets Across 18,428 Catchments Using Hydrological Modeling
Published in Hydrology and Earth System Sciences, 2026
Recommended citation: Abbas, A., Yang, Y., Pan, M., Tramblay, Y., Shen, C., Ji, H., Gebrechorkos, S. H., Pappenberger, F., Pyo, J., Feng, D., Huffman, G., Nguyen, P., Massari, C., Brocca, L., Tan, J., & Beck, H. E. (2026). Comprehensive global assessment of 24 gridded precipitation datasets across 18 428 catchments using hydrological modeling. Hydrology and Earth System Sciences, 30(11), 3399-3423. https://doi.org/10.5194/hess-30-3399-2026
Dozens of datasets now estimate how much rain falls anywhere on Earth, built from satellites, weather models and ground-based gauges, yet anyone needing to pick one has had little solid basis for the choice. This study ran the largest such comparison to date: each of 24 datasets was fed in turn into the same rainfall-to-river-flow model across 18,428 river basins worldwide, and judged on how closely the flows it produced matched flows actually measured at gauges. MSWEP version 2.8 — which merges several sources and corrects them against ground gauges — came out best overall, ahead of the strongest satellite-only product, which pairs soil-moisture-derived rainfall with microwave readings, and the strongest weather-model reanalysis. Scores varied systematically between climates, so the study also shows which dataset to prefer in a given part of the world, not just which is best on average.