Comparison of the effects of climate change and anthropogenic factors on river water quality with a random forest model

ABSTRACT When predicting future river water quality, the weather and pollution sources need to be considered together, which makes water quality prediction difficult. Climate change directly affects river flow, water temperature, and water quality, as do anthropogenic effects. It is necessary to understand changes in future water quality and be well prepared. In this study, a random forest model was applied as the machine learning method to construct a water quality prediction tool, which was used to project water quality to the year 2100 under climate change and pollution change scenarios. Two rivers in Taiwan were chosen as case studies. The Xindian River has good water quality and less pollution, whereas the Jishui River has poor water quality and substantial wastewater pollution. The results show that the random forest model can achieve high precision in both cases in terms of biochemical oxygen demand (BOD), ammonia nitrogen (NH3-N), and dissolved oxygen (DO). The water quality prediction under climate change in the Sixth Assessment Report (AR6) demonstrated that BOD and NH3-N worsened by nearly 10% in 2100, but the DO trend was stable and did not change substantially. Variable flow discharge increases aeration, and high temperature decreases oxygen dissolubility, causing the DO concentration to remain unchanged. However, when pollution factors are considered, increasing domestic wastewater treatment in the Xindian River can significantly improve water quality under climate change scenarios. Similarly, efforts to treat livestock wastewater in the Jishui River resulted in high water quality improvement. The results revealed that the control of pollution factors could positively affect water quality by 40%. The results of this study reveal that while climate change affects water quality, anthropogenic factors affect it much more, particularly wastewater treatment.