This study addresses the challenges of managing transboundary water resources characterized by data scarcity, focusing on the Indus Basin as a case study. Groundwater depletion in such regions poses a critical threat to food and water security, necessitating effective management strategies. We leverage Gravity Recovery and Climate Experiment (GRACE) satellite data, a valuable alternative in data-scarce environments to monitor terrestrial (TWS) and groundwater storage (GWS) changes. GRACE/GRACE-FO products have widely being used by researchers for studying terrestrial water cycle. However, the existing resolutions of GRACE/GRACE-FO products has limited the ability to identify local-scale changes in TWS and GWS across smaller catchments. The research downscaled GRACE data to 1km2 resolution using data-driven machine learning models and spatial downscaling techniques to explore local vulnerabilities of TWS and GWS dynamics associated with both anthropogenic and climatic changes across 20 sub-regions in the Indus Basin. Comparison with in-situ GWS from 2200 monitoring wells shows that downscaling of GRACE data significantly improves agreement with in-situ data, evidenced by higher correlation coefficients (0.60-0.95). The integration of GRACE into machine learning models enables the identification of local vulnerabilities. Hotspots with the highest TWS and GWS decline rate between 2002-2023 were in downstream sub-regions Dehli Doab (-437, -457 mm/year), BIST Doab (-348, -410 mm/year), Rajasthan (-266, -286 mm/year), and Ravi (-190, -278 mm/year), primarily linked to anthropogenic stressors. The findings underscore the potential of GRACE not only as a data source but as a powerful tool for actionable decision-making in the sustainable management of shared water resources in transboundary basins.