Document Type


Publication Date



Ensemble hydrological predictions are considered to be useful for robust reservoir operation as they include multiple possible hydrological scenarios in the future as well as dispersion of the predictions, from which the degree of prediction uncertainty can be estimated. Although operational ensemble hydrological predictions have been available in many regions, they have not yet been widely used in the actual reservoir management due to the difficulty in the handling such complex information. In order to facilitate effective utilization of ensemble hydrological predictions in the actual reservoir management, a real-time reservoir operation method for drought management is developed considering operational ensemble forecasts of precipitation in Japan. One-week and one-month ensemble predictions of precipitation (EPPs) with respectively 51 and 50 ensemble members provided by Japan Meteorological Agency are employed here. Firstly, an ensemble prediction of daily precipitation in the target basin is estimated for the coming one month from EPPs by using artificial neural networks (ANNs), which are developed so as to model the statistical relationships between predicted values and observed basin precipitations. An ensemble streamflow prediction is then estimated by use of Hydrological River Basin Environment Assessment Model (Hydro-BEAM), a distributed rainfall-runoff model. Reservoir operation for drought management is then optimized considering the prediction for the coming one month by using sampling stochastic dynamic programming, which can consider both the stochastic and time-series natures of the ensemble prediction, so as to minimize drought damage caused by deficit in release waters. Water release is conducted according to the optimized operation strategy, updating EPPs and ensemble streamflow prediction. The proposed method is applied to water supply operation of Sameura Reservoir in the Yoshino River basin in Japan, demonstrating effectiveness of considering operational EPPs as well as the effects of uncertainty contained in the predictions on performances of the reservoir operation.


Session R60, Reservoir Operations and Management I



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.