Bayesian Network application to assess Water Quality from a Water-Energy-Food Nexus perspective: A case study in the Upper Adige River Basin (Italy)

In the context of Water-Energy-Food (WEF) nexus security, it is imperative to place greater emphasis on the water quality dimension to ensure sustainable and resilient systems. While traditionally much focus has been placed on the availability of water, recently the quality of water emerged as a critical factor that limits its supply across various sectors, including agriculture, energy production, human and ecological needs. 

The impacts of global change—including climate change and land use intensification to meet socio-economic development needs—are reshaping water availability and quality in complex ways, influencing both the quantity of usable water and its suitability for specific purposes. Understanding these interconnections is vital for assessing the broader implications of clean water availability, as poor water quality can constrain sectoral efficiency and undermine ecosystem health. A spatial Bayesian Network (BN) model has been developed to predict the conjoined impacts of future climate change and land use trajectories on water chemistry in the Upper Adige River basin in Northern Italy. It allows to predict different water quality indicators (e.g. nutrient concentration, Dissolved Oxygen, temperature, pH, Total Suspended Solids) at the sub-catchment and seasonal scale and to classify their status (i.e. LIMeco Index) according to the Water Framework Directive 2000/60/EC. The model has been implemented using ARIES  (Artificial Intelligence for Environment and Sustainability), a Machine Reasoning platform for data and model integration. The model has been trained with historical water quality data from 2013-2022, considering as predictors specific indicators that serve as proxies for the different nexus sectors as well as external drivers (i.e. climate and land use). The strength of this work lies in enabling a spatial understanding of the drivers influencing water quality, allowing the identification of critical sources of pressures on water quality related to different economic sectors, and the spatial mapping of priority areas most affected by these pressures, as well as the prediction of the conjoined impacts of different scenarios (i.e. climate change,  land use change, anthropic stressors). The findings highlighted that diffuse sources attributable to agricultural activities, forest management, and the presence of highly urbanised areas play a greater role in influencing nutrient concentration than point sources and that while expected land use changes are quite significant in some basins, their impacts are moderated by hydroclimatic variables such as flow conditions and temperature, which vary considerably between seasons. By identifying hotspots of nutrient pollution and the key variables influencing water quality, the findings provide valuable tools for local authorities to implement measures and plans aimed at mitigating water quality deterioration. In the broader context of WEF nexus management, the results of this research underscore the importance of proactive water management strategies that account for the complex interactions between land use, climate, and water quality.