Long-term groundwater level (GWL) measurement is vital for effective policymaking and recharge estimation.A Machine Learning model is developed to mitigate data gaps and accurately predict maximum and minimum GWL.An Upsampling Model is trained to produce high-resolution GWLs using low-resolution GLDAS data as input.The approach successfully upscales GLDAS data, allowing high-resolution recharge estimations and proactive resource management.