Long-Term Water Quality Analysis and Eutrophication Prediction Model for a Subtropical Reservoir in the face of Climate Changes

In recent years, eutrophication of the water in reservoirs has posed threats to water resources, human health, and the ecological environment. Therefore, it is crucial to understand the water quality of reservoirs and the impacts of future climate change on their catchment areas for effective reservoir management. This study focuses on the Techi Reservoir and conducts a comprehensive analysis of the spatiotemporal variations in water quality trends over the past 30 years. The research findings indicate that the reservoir's water quality is significantly influenced by local climate change. For instance, over the past 30 years, there has been an upward trend in water temperature (with a slope of 0.00022 °C/day), leading to a significant decrease in dissolved oxygen in the reservoir (decreasing by approximately 0.038 mg/L per year), while conductivity has shown a significant upward trend (increasing by approximately 1.6 μS/cm per year). From a spatial distribution perspective, this study combines the analysis of potential non-point source pollution in the Techi Reservoir catchment area with historical stream water quality monitoring data. The results indicate that suspended solids in the reservoir are mainly generated between the Siji-Lang River and Nan-Hu River monitoring stations, which converge at the sampling point of the Song-Mao River mouth. As for nutrients in the reservoir, they primarily come from the Jin-Yuan River, followed by the You-Sheng River and He-Huan River. To understand the relationship between water quality and algal growth and to address future algal bloom risks, this study also establishes a decision tree model for identifying eutrophication based on historical reservoir water quality data, and validate the developed prediction model by the k-fold cross validation method. This model utilizes water quality parameters such as dissolved oxygen, pH, suspended solids, total phosphorus, and chemical oxygen demand to classify eutrophication (i.e., chlorophyll-a concentration) in the reservoir, achieving an accuracy of over 77 %. In summary, through long-term trend analysis of water quality monitoring data, identification of pollution pathways and estimation of non-point source pollution in the catchment area, as well as the establishment of a decision tree model based on water quality parameters, this study proposes strategies and measures for reservoir water quality protection and catchment area management in response to climate change, providing valuable insights for management authorities to consider. © 2024, Taiwan Agricultural Engineers Society. All rights reserved.