Main content area

A novel method for deriving reservoir operating rules based on flood classification-aggregation-decomposition

Zhang, Jingwen, Li, Zejun, Wang, Xu, Lei, Xiaohui, Liu, Pan, Feng, Maoyuan, Khu, Soon-Thiam, Wang, Hao
Journal of hydrology 2019 v.568 pp. 722-734
algorithms, case studies, flood control, hydrology, models, water reservoirs, watersheds, China
The potential of aggregation-decomposition (AGDP) for flood control of mixed reservoir system remains unclear due to the flood variability and complex hydraulic connections. In this study, we develop flood classification-aggregation-decomposition (FAD) operating rules to adapt for flood variability and to account for the trade-offs between flood loss of the tributaries and that of the mainstream. Three steps are involved in the proposed FAD operating rules: (1) flood classification by projection pursuit (PP) coupled with genetic algorithm (GA); (2) definition of FAD operating rules using the AGDP model coupled with piecewise linear function based on flood classification; and (3) optimization of FAD operating rules by parameterization-simulation–optimization (PSO). A case study is performed with a large-scale mixed reservoir system in the Xijiang river basin (China) in three scenarios: conventional operation for the status quo year (CO-SQY), conventional operation for the perspective year 2030 (CO-2030) and FAD operation for the perspective year 2030 (FAD-2030). The results clearly show that flood classification enables reservoir operating rules to better adapt for flood variability, and the FAD operating rules in the scenario of FAD-2030 outperform the conventional operating rules in the scenarios of CO-SQY and CO-2030, as it considers flood variability and the trade-offs between flood loss of the tributaries and that of the mainstream.