This study aimed to model the green and blue water footprint of tomato cultivation in Bushehr province using hybrid machine learning models (QXGBoost, QRF, XGBoost, and RF) with uncertainty analysis. Results revealed that blue water footprint accounted for the majority (80%) of water consumption, while green water footprint constituted only 20%, indicating heavy reliance on water resources and climate vulnerability. Correlation analysis showed precipitation (P) and effective precipitation (Peff) had the strongest positive impact on green water footprint, whereas temperature and evapotranspiration (ETc) showed negative effects. For blue water footprint, ETc and Tmax exhibited the highest positive correlation. The QXGBoost model demonstrated optimal performance with a determination coefficient (Rsq=0.95) and low error (RMSE=0.16). Uncertainty evaluation using UNEEC method confirmed QXGBoost's superiority with prediction interval coverage (PICP=0.93) and symmetric confidence intervals. The study proposes water footprint reduction strategies including improved irrigation (subsurface/smart systems), drought-resistant cultivars, and soil moisture management (mulching). Findings emphasize transitioning to sustainable farming systems through advanced modeling and continuous water resource monitoring.