Groundwater depletion causes a critical risk for the sustainability of urban and agricultural resilience in semi-arid regions such as Tangshan city. This study deployed an integrated framework that combines understandable machine learning (Shapley Additive exPlanations (SHAP), Geographically Weighted Regression (GWR), spatial autocorrelation (Local Indicators of Spatial Association or LISA), and scenario-based recharge forecasting to evaluate the spatial drivers and patterns of groundwater stress and project potential future risks. Using spatial groundwater table data from 2022 and key environmental and anthropogenic variables such as evapotranspiration (ET), population, temperature, precipitation, and land use and land cover changes, an XGBoost (Extreme Gradient Boosting) regression model was trained to capture nonlinear spatial patterns. SHAP analysis found that ET and population density are prominent contributors to groundwater depletion in agricultural and urban zones. To capture spatial heterogeneity, GWR was utilized to estimate localized coefficients and construct a Vulnerability and Resilience Index (VRI) from normalized coefficients and residuals. LISA validated vulnerability zones and revealed transitional stress regions. Future risks are also projected using Coupled Model Intercomparison Project Phase 6 (CMIP6) climate data and land-use data to run recharge modeling from 2023 to 2049 for both representative concentration pathway (RCP) 4.5 and RCP 8.5. Results show that RCP 8.5 demonstrates highly unstable recharge with frequent negative episodes (ET > P), while RCP 4.5 shows relatively stable patterns of groundwater table. Furthermore, coupled with urban and agricultural expansion, RCP 8.5 intensifies depletion risks. This combined framework provides analytical understandings of spatial driver patterns and scenario-based risk assessments under climate and land use change. The findings of the study recommend priority zones for intervention and underline the importance of adaptive, scenario-sensitive groundwater governance in semi-arid, urbanizing regions.