• A variably saturated coupled poroelastic model is developed to simulate soil deformation caused by groundwater depletion within a phreatic aquifer system at a basin scale. • A sparse grid-based Bayesian framework is proposed to infer hydraulic conductivity and compressbility by integrating the poroelastic model with observed data. • Utilization of InSAR-derived displacement datasets significantly reduces the uncertainty associated with aquifer properties, facilitating a comprehensive understanding of hydrogeological processes in an over-exploited aquifer system in Spain.