Water Supply-Demand Management under Climate Change

Tianyi Xu, Xiaosheng Qin

This study aimed to investigate optimal multiple reservoir operations and water demand management considering climate change impact. In this study, conditional density estimation network creation and evaluation (CaDENCE) method was used for downscaling precipitation, and support vector machine (SVM) was used for downscaling temperature. The Bayesian neural network (BNN) model was applied to simulate the monthly reservoir inflows, which was used as the input to the optimization model. A multi-reservoir system was used for methodology demonstration, where three reservoirs were delivering water to an urban area. Several water-saving measures including long-term and short-term measures were involved in the optimization model to mitigate water shortage problem. The model aimed to maximize the total revenue obtained from water release of three reservoirs subject to constraints of available water supply, demand of water users, and cost of water demand management. The optimal water release schemes and adoption of water-saving measures under current and future climate-change conditions were obtained. The results showed that the water releases would increase at spring and decrease at winter under HadCM3 A2 emission scenario compared to the current condition.