Changes in hydrological processes associated with climate change may increase frequency of severe flood and drought events. More effective water resources management is needed to mitigate impacts of those severe water-related disasters. As reservoirs can play an important role in water resources management, sophisticated operation of reservoirs considering real-time hydrological predictions can improve the capacity of existing water resources management systems to deal with severe floods and droughts.
In order to provide reservoir managers with quantitative and science-based information on expected benefits and risks to introduce operational hydrological predictions into reservoir operation from the long-term viewpoint, a Monte Carlo-based method to analyze effectiveness and risks of integrated operation of a multi-purpose reservoir for flood management considering real-time ensemble hydrological predictions is developed in this study. Preliminary release operation, in which water stored in the reservoir is released just before a flood event considering real-time hydrological predictions to secure more empty volume in the reservoir for flood control, is considered as an integrated reservoir operation method in this study. The outline of the proposed method is described as follows. Firstly, a model for synthetic generation of ensemble hydrological predictions with designed error structures is developed. The mean error of ensemble mean predictions and the spread of ensemble predictions are considered as the parameters to represent prediction error structures. A Monte Carlo simulation of preliminary release operation of the target reservoir is then conducted considering a number of ensemble hydrological predictions with a designed error structure generated by use of the model developed in the previous process, in order to analyze the effectiveness of preliminary release operation in flood control as well as expected risks in water utilization due to reducing water level of the reservoir. The effects of preliminary release operation are thoroughly investigated by conducting the Monte Carlo simulations of reservoir operation changing error structures of hydrological predictions considered in the preliminary release operation.
The proposed method was applied to preliminary release operation of the Nagayasuguchi Reservoir in the Naka River basin in Japan, clarifying expected effectiveness of preliminary release operation in flood management as well as potential shortage in water storage recovery and its possibility after a flood event with respect to each parameter setting for error structures of ensemble hydrological predictions. This information can be used for calculation of the rational amount of compensation to be paid to stakeholders who have a right to use water of the reservoir for shortage in water recovery. The results also showed the criteria of prediction accuracy that is needed to conduct preliminary release operation safely while recovering storage water level for water utilization after the flood event with more than a certain probability.