Digital Twin Earth Hydrology – Precipitation: Harnessing the Strengths of Individual Products

In recent years, initiatives such as the European Union’s Green Deal and Data Strategy have promoted the creation of Digital Twins of the Earth (DTE). These virtual representations aim to integrate state-of-the-art advancements in Earth Observation (EO), modeling, artificial intelligence, and computational power. Their purpose is to enable the visualization, analysis, and prediction of both natural systems and human-related activities, ultimately supporting sustainability efforts and addressing the challenges of climate change. Within this framework, the European Space Agency (ESA) launched the DTE Hydrology project, with a specific emphasis on the water cycle, hydrological dynamics, and their practical uses.

As part of this project, high-resolution (1 km, daily) datasets for critical water cycle variables are generated to replicate hydrological behavior and understand its interactions with human systems. Among these, precipitation plays a central role due to its influence on agriculture, water resource planning, economic stability, and disaster risk reduction. However, ground-based observation networks are diminishing worldwide, and many regions lack sufficient station density for reliable monitoring. Satellite-derived precipitation estimates are therefore essential to fill both spatial and temporal data gaps in these areas.

To overcome the limitations of individual datasets, the DTE-Hydrology initiative synthesizes precipitation data from multiple EO satellite platforms and methods, combining them with reanalysis data to create a unified, enhanced product. Specifically, precipitation estimates from IMERG-Late Run, SM2RAIN ASCAT (H SAF), and ERA5 Land are downscaled at 1 km spatial resolution and subsequently merged. The downscaling process utilizes detailed spatial information from the CHELSA (Climatologies at High resolution for the Earth’s Land Surface Areas) dataset, while merging weights are calculated using the Triple Collocation method.

The final merged product was thoroughly validated and compared against a range of datasets—both coarse-resolution sources such as H SAF, IMERG-LR, ERA5, EOBS, PERSIANN, CHIRP, GSMAP, and fine-resolution datasets like EMO, INCA, SAIH, COMEPHORE, and MCM—demonstrating its strong reliability and performance.