Estimation of Groundwater Storage Variability over Major River Basins of Western India using Multi-Temporal Satellite Datasets and in-situ observations

Groundwater is a vital freshwater resource that sustains agriculture, domestic water supply and industrial growth, particularly in the arid and semi-arid regions around the globe experiencing low rainfall.  Excessive groundwater abstraction for sustained crop production and water demand for the growing population has led to overexploitation of the freshwater resource globally. Multiple studies have reported significant groundwater depletion in major aquifers around the globe. It is imperative to monitor the availability groundwater at diverse spatial and temporal scales accurately to reduce the uncertainty in groundwater supply and formulate policies for sustainable groundwater resource management.

A major leap in the estimation of groundwater resource emerged after the launch of Gravity Recovery and Climate Experiment Mission (GRACE) in 2002. Integration of Terrestrial Water Storage (TWS) estimates from GRACE/GRACE FO and water storage components (surface runoff, evapotranspiration and soil moisture) from global hydrological models are employed to monitor groundwater storage variability in major aquifers globally. However, uncertainties have been reported in the estimation of groundwater depletion, majorly due to the course resolution of GRACE TWS and model uncertainties at regional scales. Hence, it is important to identify the uncertainties in ancillary datasets employed to monitor groundwater variability from space.

In the present study, we evaluated long term trends in terrestrial and groundwater storage anomaly (2002-2023) over 4 major river basins of western India using three GRACE/GRACE FO mascon solutions – JPL, CSR and GSFC and computed groundwater storage anomalies using a combination of TWSA and Soil moisture estimates from multiple hydrological models – GLDAS Catchment Land Surface Model (CLSM), WaterGAP and ESA_CCI Combined data products. We further evaluated the performance of a global downscaled water storage anomaly product generated from self-supervised data assimilation. It was observed that the groundwater estimates from multi-temporal datasets showed uncertainties in long term trends when validated against in-situ groundwater observations. Further, heterogeneity in inter-basin groundwater variability was observed, indicating a need for estimation of water storage components at regional scale incorporating human intervention for accurate groundwater measurement. Our results highlight the importance of groundwater change estimation from earth observation and modelled hydrological components to insights into changing dynamics of groundwater in semi-arid river basins.