Water quality in rivers under anthropogenic pressures: from parametric assessment to contaminants transport modelling

This doctoral thesis examines the qualitative status of surface waters and assessment methodologies, aiming to update the state of the art, identify key knowledge gaps, and propose practical tools for addressing this relevant issue. The research is structured in two parts, both focusing on water quality degradation driven by anthropogenic activities. The first part introduces the Biochemical Quality Index (BQI), developed to identify river segments exposed to potentially polluting discharges. The BQI accounts for both the specific environmental pressures associated with different types of discharges and their cumulative effects along the receiving watercourse. Its validity was confirmed through comparison with Chemical Oxygen Demand (COD) from monitoring stations, demonstrating strong consistency. Results highlight the utility of the BQI as a semi-quantitative tool for estimating water quality trends in unmonitored rivers, identifying the most impacted river segment, and pinpointing the activities contributing most to degradation. Among the sources (Wastewater Treatment Plants, contaminated sites, IPPC-IED facilities, and Seveso plants) Seveso plants exerted the greatest impact due to the hazardous nature of the substances they handle, which pose severe risks to aquatic ecosystems. The second part investigates the potential effects of Seveso plant accidents triggered by floods, classified as Natech (natural hazard triggered technological) events. The study focuses on modelling the fate of petroleum products in freshwaters, noting that most oil spill models target marine settings, while fluvial models are limited, computationally intensive, or unsuitable for floodwaters. To address these limitations, the CAESARLIS-FLOOD model was implemented for simulating oil spills in floodplains, combining a robust physical foundation with reduced computational cost, user-friendliness, and open access. Benchmarking against other models showed good agreement in reproducing dispersion, concentration patterns, and spatial extent of contamination, while providing significantly faster performance. These results demonstrate its potential as predictive tool and rapid decision-support system for emergency response.