Abstract: Under the influence of global climate change and ever-increasing human activities, frequent droughts have imposed great impacts on social and economic developments. The study on drought-resistant strategies for staged reservoir operation can produce a remarkable effect on efficient utilization of water resources, drought-resistance, and disaster reduction. This paper examines how to determine an optimal drought-resistant scheduling strategy in multi-year available water supply combination scenarios based on the minimum risk Bayesian decision, using a bee colony algorithm and a reverse recursion method that is equipped with an objective function to maximize water supply guarantee rate and minimize water supply failure depth. In a case study of the Jinpen reservoir in the Heihe River basin, we make a comparative analysis of the drought-resistant strategy of its operation based on the Bayes joint reverse sequential recursion, against those based on conventional operation, traditional reverse sequential recursion, and optimized reverse sequential recursion. We find that the former can not only reduce the probability of extreme water shortage under extreme drought conditions, but also improve the water supply guarantee rate and reduce the water supply failure depth. This shows the Bayes decision-based operation strategy raises the capability of a reservoir against extreme droughts and lays a theoretical basis for drought-resistant operation.

Key words: staged drought operation, Bayes theory, bee colony algorithm, reverse order recursion, drought-resistant operation strategy