Study on Optimization of the Operation of Dams Using Ensemble Prediction and a Distributed Rainfall-Runoff Model

Abstract

Recently, heavy rain by typhoon increases risk of disaster everywhere in Japan. There has been considerable interest to improve the flood control function of dams by prior releasing because an action plan was enacted to fully use the existing hydraulic structures to prevent flood disasters. Carrying out prior releasing has a risk in terms of water supply purpose, in other words, it may cause artificial drought. Therefore, it should be taken into consideration that releasing larger amount of water than rain gives water shortage. In the present study, we suggested the method of dam operation based on rainfall forecast including prior releasing considering the risk in terms of water supply purpose. Concretely, first, we investigated that it could estimate the accuracy of forecasted accumulated rainfall based on Global Spectral Model (GSM) by adding the information of spreads calculated by Weekly Ensemble Prediction System (WEPS) in the Yodo river basin. Second, accumulated rainfall based on GSM errors using gamma distribution was analyzed. Third, the method of dam operation based on rainfall forecast including prior releasing was applied to past examples and the effect was verified. As a result, peak discharge in Hirakata point was reduced than normal operation in case of rainfall prediction was accurate.Engineering