Accurate rainfall estimates at high spatial resolution are necessary for sustainable regional development in a semi-arid region, where agricultural and economic activities are largely dependent on the local rainfall. However, the scarcity of rain-gauge data and the inherent limitations of satellite derived high-resolution gridded rainfall datasets in overestimating light rainfall and underestimating heavy rainfall events emphasize the need for spatial downscaling of the coarse-scaled rain-gauge interpolated data that has shown relatively better accuracy in the estimation of heavier and lighter rainfall events in semi-arid regions. In this study, a Multi-scale Geographically Weighted Regression (MGWR) model based on an integrated downscaling and calibration framework was developed to generate a high-resolution (1 x 1 km) gridded rainfall dataset at monthly and annual time scales from the coarse-scaled (similar to 25 x 25 km) Indian Meteorological Data (IMD) using seven environmental variables (elevation, slope, wind speed, AET, Day-LST, Night-LST, and NDVI) retrieved from different open-source Earth Observation datasets in a semi-arid catchment in India. The MGWR-downscaled rainfall data (with residual correction) was calibrated with the rain-gauge data using the Geographical Ratio Analysis (GRA) and Geographical Difference Analysis (GDA) methods, and compared with the rain-gauge data of 2002, 2012, 2020, and 2002-2021 to evaluate the performance of these gridded-high resolution products in capturing the intra-annual variations under normal, dry, wet and average rainfall years, respectively. The GDA-calibrated monthly downscaled dataset with an r2 of 0.87, RMSE of 34.84 mm, and a bias of 0.12 demonstrated higher resemblance with the gauge data in comparison to the original-IMD gridded rainfall data (r2 = 0.65, 56.96 mm and 0.08) between 2002 and 2021. This was also supported from rainfall periodicity (estimated with Multi-Taper Method) and variability analysis, wherein GDA-calibrated downscaled dataset exhibited the highest likeness to the rain-gauge dataset. Groundwater Potential Zones (GWPZ) estimated from the original-IMD gridded rainfall data (similar to 25 x 25 km) had overestimated the 'High' and 'Very High' GWP zones and underestimated the 'Moderate' and 'Low' potential zones compared to the GDA-calibrated downscaled dataset. This indicates that the GDA-calibrated downscaled high-resolution rainfall data provides a more detailed representation of spatial rainfall variability, leading to improved delineation of the GWPZ. The approach outlined in this study is geographically scalable and can be used to bridge the high-resolution rainfall data gaps in other semi-arid regions for climate adaptation studies by improving estimation of water balance components, runoff, infiltration, and other key hydrological processes. (c) 2025 COSPAR. Published by Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar