Study region: Minab County, located in Hormozgan Province, southern Iran, is a low-lying alluvial basin characterized by arid to semi-arid climatic conditions and recurrent riverine and flash flooding. Due to its agricultural importance and hydrological position at the downstream convergence of major rivers, the region is highly vulnerable to extreme rainfall events and floodinduced damages. Study focus: This study develops a high-resolution flood susceptibility map by integrating multisource datasets, including Shuttle Radar Topography Mission (SRTM)-derived topographic indices, Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) rainfall variables, soil texture data, and Landsat-8 spectral indices, namely the Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Land Surface Water Index (LSWI), and Land Use/Land Cover (LULC). A total of 585 flood inventory points from the January 2022 flood event were used to train and test seven machine learning models: Extreme Gradient Boosting (XGBoost), Adaptive Boosting (AdaBoost), Categorical Boosting (CatBoost), Random Forest (RF), Model Tree (MT), K-Nearest Neighbors (KNN), and Support Vector Machine (SVM). Model performance was evaluated using statistical error metrics and Receiver Operating Characteristic (ROC) curve analysis. Two preliminary hazard maps derived from terrain-hydrological and spectral indices were integrated using a correlation-weighted approach to generate a final flood susceptibility map. New hydrological insights for the region: The results indicate that ensemble tree-based models outperformed non-ensemble approaches, with CatBoost and RF models providing the most balanced and generalizable performance (Area Under the Curve > 0.97), while XGBoost and AdaBoost models exhibited near-perfect statistical fit that may reflect overfitting tendencies. Shapley Additive Explanations (SHAP) analysis reveals that short-term precipitation prior to the flood event and surface moisture conditions are the dominant drivers of flood susceptibility, while elevation, slope gradient, and distance to river act as secondary controls. The final susceptibility map derived from the January 2022 extreme event shows that more than half of the study area falls within high to very high flood-risk classes, particularly along the Minab River and surrounding agricultural plains. The proposed integrative and explainable machine learning framework provides a robust and transferable methodology for flood hazard assessment in arid and semi-arid regions of southern Iran.