Flood control is one of the most important issues of reservoir operation. Rivers in island countries like Japan, Philippines and Indonesia that have smaller reservoirs than continental countries needs short term reservoir operation for flood control. In Japan, typhoons give dominant amount of water to reservoirs. Pre-release of water which makes effective use of the capacity of a reservoir requires the rainfall amount forecast (hyetograph). Therefore, weather forecast of typhoons is indispensable for flood control. Masuda and Oishi (2013) developed the reservoir control operation model using stochastic dynamic programming with one week ensemble weather forecast. One week ensemble forecast consists of 51 members, gives many kinds of weather variables including rainfall amount, has lead time of one week. In fact, frequency of issue daily is a problem for using one week ensemble weather forecast. In the present study, we give a solution for the problem. For giving higher frequency, we propose to use typhoon forecast which issues four times a day but does not include rainfall amount. By using a similarity index with past typhoon and latest ensemble forecast result, a method to give typhoon ensemble reasonable forecasted rainfall amount has been developed. Four times a day is higher frequency than once a day but required to be improved for practical use. For such purpose, we propose a kind of Bayesian method to evaluate the probability of each ensemble member. The proposed method gives the better rainfall amount forecast by evaluation hourly. In the conference, explanation of the methods in detail as well as some case studies with comparison are discussed.
Oishi, Satoru and Masuda, Hiroko, "Study On Optimization Of The Integrated Operation Of Dams Using Ensemble Prediction" (2014). CUNY Academic Works.