Multi-Objective Operation- Decision-Making-Risk Propagation Analysis for Cascade Reservoirs Affected by Future Streamflow Process Variations

Significant climate variations have decreased the stability of water resource systems, leading to multiple uncertainties in streamflow response, reservoir operation optimization, decision-making, and adaptive adjustments for water resource scheduling. Understanding the impact of climate change on reginal streamflow is necessary and crucial to identifying reservoir operation strategies and decision-making responses. In this study, we created an integrated systematic “uncertain streamflow responses”– “reservoir operation”– “optimization”– “decision-making risk analysis” chain. Three bias-corrected and downscaled general circulation models (GCMs) were used to analyze the inter-model uncertainties under three representative concentration pathways (RCPs). The streamflow responses and uncertainty in the future were determined using a distributed hydrological model and the fuzzy extension principle under predefined scenarios and uncertainty levels. Then, a stochastic simulation model and modified stochastic multi-criteria decision-making model were applied to identify the effects of climate change projections and streamflow responses on reservoir multi-objective operation and decision-making. Moreover, risk quantification indices were used to determine the uncertainty propagation and potential risks accumulated in the chain. We applied this framework to cascade reservoirs in the Qing River Basin. The results indicate that the mean annual streamflow projected using selected GCMs will increase, enhancing the hydropower response and weakening the ecological benefit response. The Pareto non-dominated solutions optimized based on the streamflow projections obtained using the GCMs (under the same RCP) and hydrological model are more distinct than those based on different RCPs and the same GCM. Moreover, a high emission scenario may increase the uncertainty of the streamflow projections and reservoir operation responses, which is consistent with the finding that the decision-making process becomes more variable and sensitive with increasing streamflow uncertainty. Finally, we identified the preferred solutions for reservoir operation under different uncertainties, the respective expected values, and the 95% confidence interval bands to enhance the adaptability of future reservoir operation.