摘要
This paper presents a new methodology for developing operating rules for conjunctive use of surface and groundwater. Bayesian Networks-based operating rules are trained and verified using the results of a multi-objective optimization model. Reduction of pumping costs, improving the groundwater quality, water supply with acceptable quality and controlling the groundwater table fluctuations are considered as objective functions of the optimization model. In order to provide Pareto fonts among these conflicting objectives, the combination of MODFLOW and MT3D groundwater quantity and quality simulation models and Non-dominated Sorting Genetic Algorithm II (NSGA-II) are used. The best solutions on the Pareto fronts, which are selected using the Young and Nash bargaining theories, are used to train and verify Bayesian Networks (BNs). In real-time water allocation from surface and groundwater resources, the BNs-based rules can be used without any need to run the time consuming optimization and conflict resolution models. The proposed methodology is applied to the conjunctive use of water resources in the Tehran region, Iran. The results show that using the operating rules can improve the groundwater quality and control the groundwater table fluctuations in the study area.