Real-Time Reservoir Operation for Drought Management Considering Operational Ensemble Predictions of Precipitation in Japan

Abstract

A method for real-time reservoir operation for drought management considering operational ensemble predictions of precipitation is developed in this study. Two operational ensemble predictions provided by Japan Meteorological Agency, namely One-month Ensemble Prediction (denoted by EPS1 hereafter) and One-week Ensemble Prediction (denoted by EPSW hereafter), are considered here. Ensemble prediction of basin precipitation for the coming 1 month is estimated from the grid point values of precipitation predictions of both EPS1 and EPSW, integrating them by employing the precipitation prediction from EPSW for the coming 8 days and that from EPS1 for the period from 9 days ahead to 1 month ahead. Ensemble streamflow prediction for the coming 1 month is then estimated from the ensemble prediction of basin precipitation by use of Hydro-BEAM (Hydrological River Basin Environment Assessment Model), a cell-grid type rainfall–runoff model. Daily water release from the target reservoir system for water supply is then optimized by use of dynamic programming (DP) approaches so as to minimize drought damage caused by the gap between water supply from the reservoir system and demand in the downstream in the optimization horizon considering ensemble inflow prediction. The proposed method was applied to the reservoir system in the Yoshino River in Japan, showing the effectiveness of the proposed method to introduce operation ensemble predictions of precipitation into reservoir operation for drought management.