Abstract
We developed a regional ensemble prediction system to support advanced dam operation including predischarge of dam water for the purpose of reducing flood risk and effective water resource management, simultaneously. The targets were Ooigawa and upper Saigawa Basin. The prediction experiment was performed during July to September in 2015 to 2017. Their prediction accuracy was evaluated, and the effective use of probabilistic predictions were discussed. Higher percentile prediction could capture flood risk well, on the other hand, lower percentile prediction could escape from false alarms. For the dam operation with hydro power generation in the target basin, 25 to 50 percentile predictions would be useful to control false alarm ratio under 10%, with reasonable amount of missing ratio with 35%, simultaneously.
