Real-Time Reservoir Operation for Flood Management Considering Ensemble Streamflow Prediction and its Uncertainty

A real-time operation method of a multi-purpose reservoir for flood management considering an ensemble streamflow prediction (ESP) is investigated in this study. The ESP is derived by a distributed rainfall-runoff model from an operational ensemble prediction of precipitation. Japan Meteorological Agency’s One-week Ensemble Forecast of precipitation, which is provided every day and has 51 ensemble members of six-hour precipitations for the coming eight days, is employed here. ESPs with 51 members for the coming eight days are calculated from the ensemble predictions of basin precipitation by use of Hydrological River Basin Environment Assessment Model (Hydro-BEAM), a distributed rainfall-runoff model. Reservoir states such as release or storage are then estimated for each ensemble member of the streamflow predictions to decide the amount and the timing of preliminary release for flood protection. Chance and the expected amount of recovery in storage water at the end of the flood event are also estimated for each scenario of reservoir operation to estimate impact of the operation for flood control on that for water supply for the following period in order to help reservoir manager with making a decision on preliminary release considering the prediction and its uncertainty and variety. The presented method was applied to Nagayasuguchi Reservoir, a multi-purpose reservoir for flood and drought managements located in the Naka River basin in Japan, demonstrating the effectiveness and potential to provide useful information for real-time preliminary release operation of reservoirs.
Nohara Daisuke, Nishioka Yuya, Hori Tomoharu. Real-Time Reservoir Operation for Flood Management Considering Ensemble Streamflow Prediction and its Uncertainty. In: SimHydro 2014. New Trends in Simulation. 11-13 June 2014 Ecole Polytech’ Nice (France) 2014.