Reservoir water quality planning based on a robust multi-criteria decision-making approach under deep uncertainty

Recently, uncertainties caused by climate change have been a major challenge in water resource management. In this study, to examine how robustness metrics influence final solution choices, a Multi-Criteria Decision-Making (MCDM) approach within a Multi-Objective Robust Decision-Making (MORDM) framework is integrated. In this framework, the LARS-WG Regional Climate Change Simulation model is applied to generate meteorological data, and AI-driven machine learning models (MLP, SVR, GRNN, and ANFIS) are used to simulate runoff. Also, the CE-QUAL-W2 water quality model and the NSGA-III multi-objective optimization algorithm are applied. For analyzing uncertainties and assessing future strategies, three climate change scenarios (SSP5-8.5, SSP2-4.5, and SSP1-2.6) are evaluated (2030-2034). In the following, to select optimal solutions and enhance resilience in reservoir management under high uncertainty, robustness metrics, such as regret and satisfaction, are incorporated into an MCDM model. In this study, to improve water quality and emphasize the importance of flexible strategies in water resource management, specific withdrawal strategies are implemented. Therefore, this research aims to enhance project strategies by integrating MCDM and MORDM and to offer a comprehensive, structured approach for water managers to develop strategies that address climate change impacts and protect reservoir water quality.