Optimized operation of river reservoirs based on a pattern recognition method

In order to meet the increasing demand of water supply, the Chinese government has built a lot of reservoirs in the Yellow River basin, and the main purpose of these reservoirs is to supply water to accessible areas. Some of these reservoirs were built on rivers characterized by a rather high sediment concentration. Normally, the reservoirs retain water to designed levels in both flood and dry seasons, and begin to release water only when the inflowing discharge or sediment load exceeds critical values. The critical values of the inflow discharge and the sediment load were defined based on historical hydrologic data that they may not be able to adapt to the real-time incoming flow and sediment conditions. Inappropriate operations may cause waste of water or serious loss of the reservoir capacity. To deal with this issue, a model aimed at optimizing the operation of reservoirs based on the pattern recognition method was proposed in this study. The model divides the flood season into three stages: the early stage, the main stage and the late stage, and then predicts the probability of the flow and sediment conditions for the three stages. The model optimizes the control operation parameters of each stage according to predicted incoming flow and sediment conditions. This model was applied in the operation of Tingkou Reservoir in China. The results demonstrated that the optimized operation significantly improved the quality of water supply in terms of waste of water and loss of reservoir capacity.