Multi-source Fusion Framework for Statistical Downscaling of Global Monthly Precipitation
This study introduces a global-scale framework for statistically downscaling monthly precipitation data to a high spatial resolution of 1 km for the period 2000–2024. We integrate satellite-derived, reanalysis-based, and in situ observational datasets using an ensemble fusion approach that leverages the strengths of multiple global products, including ERA5, CHELSA, and IMERG. Statistical downscaling methodology is implemented using ground-based meteorological station data to improve the representativeness of local precipitation patterns. The framework incorporates spatial predictors and temporal dynamics to transform coarse-resolution inputs into fine-scale monthly precipitation fields. The resulting dataset provides improved consistency and detail across diverse climatic regions and data-sparse environments. This high-resolution precipitation product is designed to support a range of applications, including hydrological modeling, drought and flood risk assessment, and climate change impact analysis. Overall, the proposed approach offers a scalable and replicable methodology for generating detailed precipitation estimates by harmonizing global datasets with in situ observations.
