Interplay Between Climate Extreme Events, Land Use, And Water Quality: An Artificial Intelligence Multi-Risk Assessment Approach

River water quality is critical in maintaining ecosystem health, as it can directly influence biodiversity and access to clean water. However, the interaction between extreme climate events and human activities can lead to compounded effects that significantly alter water quality dynamics. The impacts of these combined factors are often complex, non-linear, and poorly understood, posing significant challenges for water resource management. Supervised machine learning and explainable artificial intelligence offer innovative tools to address these complexities. This study applied an integrated framework combining Random Forest (RF) Classifiers with SHapley Additive exPlanations (SHAP), to reveal the intricate relationships between land use, climate extremes, and their compounded effects on water quality at a high spatial resolution (867 elemental river basin), testing it in Veneto Region (northeastern Italy). The framework was applied to provide annual predictions of impacts on water quality elements to support the evaluation of ecological status according to the Water Framework Directive 2000/60/EC. The models have been applied on water quality data from 2010-2022, considering as predictors seasonal hot, dry and wet extreme climate hazard indicators, together with land use/cover metrics and territorial characteristics to represent specific river basins’ vulnerabilities. Three RF models were developed for physicochemical elements, specific pollutants, and biological alterations water quality indicators, and resulted in overall accuracies of 0.87, 0.81, and 0.85, respectively. The findings highlighted that temperature extremes acted as critical drivers, particularly when combined with droughts. Specific natural features (i.e. % of basin vegetated area, natural river typology, soil permeability) were identified as buffers against adverse impacts on water quality following extreme climate conditions. Conversely, anthropogenic land use intensified negative effects, especially when exceeding specific thresholds. The results confirm that the applied approach has the potential to aid decision-making by providing insights into multi-hazard-risks on water quality and highlighting the importance of holistic river basin management plans that prioritize nature-based solutions, ecosystem restoration, and strategic land use policies to strengthen climate resilience.