Estimación de la Calidad del Agua Mediante Modelización con Árboles de Regresión para la Cuenca del Río Yanuncay

Watersheds face numerous challenges that deteriorate water quality, such as industrial and domestic pollution, inappropriate land use and poor solid waste management. These problems are exacerbated by the effects of climate change, thus increasing the need to monitor and evaluate this vital resource for ecosystems and human communities. In this study, data-driven models with a limited set of parameters were used, using the National Sanitation Foundation Index (NSF WQI), to estimate water quality in the Yanuncay River basin. Data on physical, chemical and biological parameters of the basin were collected and equations derived from curve fits were used to calculate the NSF WQI. Models based on regression trees with a reduced number of parameters were developed and evaluated in terms of their performance as tools for estimating water quality. The results indicated that the models with the random forest regression algorithm, trained with three parameters, were the most suitable, highlighting the importance of fecal coliforms and nitrates as key indicators of pollution. In addition to showing consistent performance, with an R2 greater than 0.90. In conclusion, this study suggests that regression tree-based models can optimize monitoring requirements without compromising accuracy in estimating the NSF Water Quality Index.