Joint flood control operation functions for reservoir groups based on the concept of affiliated reservoirs

The middle and lower reaches of the Yellow River have a long history of frequent flooding. In recent years, climate change and intensified human activities have increased meteorological instability and disaster risk. Floods in this region are characterized by high peak discharges, large volumes, and a short forecast lead time. This study proposed the concept of “affiliated reservoirs for flood control” to represent the hierarchical coordination between primary and affiliated reservoirs for elucidating the complex operational relationships between mainstream and tributary reservoirs. Specifically, a set of operation solutions comprehensively superior to the current status quo was generated by implementing an NSGA-III multi-objective optimization model. The SAR(1)-Gaussian Copula-Monte Carlo Simulation model was applied to simulate multi-scenario flood hydrographs. A random forest regression model was trained with the optimized solution set as the samples to construct operation functions. The joint operation function significantly improved the three objectives compared with practical operation. The daily average excess storage capacities of the primary reservoirs and affiliated reservoirs were reduced by 3.53% and 332.06%, respectively, and the risk at flood control stations was reduced by 0.22%. This study achieves intelligent mapping from optimization results to operation rules, providing an interpretable and scalable technical pathway for functionalized operations in complex river basins. • Affiliated reservoirs and their joint flood operation functions were proposed. • Integrated multi-objective optimization and machine learning to extract operation functions. • Assessed the robustness of joint flood operations against hydrograph variability and forecast errors. • Implementation and validation of joint flood operations in a mid-lower Yellow River reservoir group.