渇水時貯水池操作における長期確率流量情報の利用性に関する一考察

Abstract

Impacts of long-term stochastic inflow predictions (SIPs) and their uncertainties on reservoir operation for water supply under drought situations are analyzed and discussed in this study. Multiple sets of SIPs are pseudo-randomly generated with five-day resolution for three months arbitrarily changing the two kinds of prediction's uncertainty, namely reliability and discrimination, for a comprehensive analysis of the impact of the SIP. Monte Carlo simulations of long-term reservoir operation for water supply under drought situations are then conducted considering generated multiple SIPs with various uncertainties. Proposed analyzing method was applied to an assumed reservoir which data was derived from Sameura Reservoir in Japan, demonstrating expected impacts of SIPs and their uncertainties on the long-term reservoir operation, and giving a suggestion what type of uncertainty in SIP is more important in real-time reservoir operation for more effective drought management.