This study introduces a self-supervised data assimilation model utilizing a novel loss function to generate global total water storage anomalies (TWSAs) at a 0.5° spatial resolution. The model integrates hydrological simulations with Gravity Recovery and Climate Experiment (GRACE) and GRACE Follow-On (GRACE-FO) satellite mission data. High-resolution information efficiency is demonstrated by accurately closing the water balance equation in small basins while maintaining large-scale accuracy from GRACE(-FO) measurements. The resulting product enhances local natural hazard monitoring and provides insights into the water cycle's response to natural and anthropogenic activities. The methodology is adaptable to other TWSA data sources and offers valuable contributions to the geoscience community and society.

Junyang Gou, Benedikt Soja