Impacto del cambio climático en las principales fuentes de agua potable de la Provincia del Azuay

Abstract

This study analyzes the impact of climate change on potable water sources in Azuay Province, Ecuador, focusing on the Tomebamba River basin, a key water source for the city of Cuenca. Historical and satellite precipitation data were used, applying Lasso regression and neural networks to predict future precipitation trends and their impact on river flow. A downscaling approach was employed to refine global climate model projections to a local scale. Historical precipitation data (1960-1990) were validated using double mass curves, and Lasso regression helped identify the most relevant variables. Neural networks were used to model and forecast future precipitation, evaluating model quality with error metrics such as RMSE and the Nash-Sutcliffe efficiency coefficient. Results show a significant increase in precipitation variability, with more extreme peaks, suggesting a higher frequency of extreme events such as heavy rains and droughts. The Germany AWI-CM-1-1-MR climate model, refined through downscaling, was identified as the most suitable for the region. Projections indicate a decrease in the Tomebamba River's flow, which could affect the potable water supply. It is recommended to use this model for future studies and consider critical scenarios for better water resource management.