Abstract:Real-time reservoirs operation can be combined with effective hydrological forecast information, which is of significant guide for the trade-offs between the flood control and utilizable benefit. However, the use of hydrological forecast information faces unavoidable uncertainties. Therefore, the flood risk control analysis should be considered in the field of the real-time reservoir operation. A dynamic control model of multi-reservoir flood limited water levels (FLWLs) during flood seasons based on the two-stage flood risk analysis method is proposed to solve the problem that the length and accuracy of forecast period among reservoirs in the system do not match with each other. First, the two-stage flood risk analysis method evaluates the uncertainty of the flood forecasting by dividing the operation horizon into the forecast lead-time and the beyond-forecast time period. The risk within the forecast lead-time induced by the streamflow uncertainty is estimated by counting the frequency of failure numbers among all scenarios with the help of the scenario-based forecasts. The risk beyond the forecast time period caused by the possible high-water level at the end of the forecast period is determined using reservoir flood routing with the design flood hydrographs. A real-time model of multi-reservoir systems is then established by considering the two-stage flood risk analysis method, and the objective of this model is to maximize the hydropower benefit. The results show that: 1) the proposed two-stage flood risk analysis method can take into account the potential risks within and beyond the forecast time period; 2) the proposed dynamic control of the FLWLs in multi-reservoir systems can increase the total hydropower generation in summer flood season for the five-reservoir system of the Hanjiang River Basin by 0.23 billion kW·h in 2010 year without increasing the flood risk, and the dynamic optimal decision-making process of the multi-reservoir system’s flood storage can be obtained.