Empirical Modeling of Stream Nutrients for Countries without Robust Water Quality Monitoring Systems

Water quality models are useful tools to understand and mitigate eutrophication processes. However, gaining access to high-resolution data and fitting models to local conditions can interfere with their implementation. This paper analyzes whether it is possible to create a spatial model of nutrient water level at a local scale that is applicable in different geophysical and land-use conditions. The total nitrogen and phosphorus concentrations were modeled by integrating Geographical Information Systems, Remote Sensing, and Generalized Additive and Land-Use Changes Modeling. The research was based on two case studies, which included 204 drainage basins, with nutrient and limnological data collected during two seasons. The models performed well under local conditions, with small errors calculated from the independent samples. The recorded and predicted concentrations of nutrients indicated a significant risk of water eutrophication in both areas, showing the impact of agricultural intensification and population growth on water quality. The models are a contribution to the sustainable land-use planning process, which can help to prevent or promote land-use transformation and new practices in agricultural production and urban design. The ability to implement models using secondary information, which is easily collected at a low cost, is the most remarkable feature of this approach.