Abstract
Using stochastic state space techniques, the authors developed a practical real-time stochastic flood prediction method for dam operation. Here we consider the series tanks model and the storage function model as the candidate for run-off model. In order to select a suitable stochastic model, predicted results are evaluated by a measure which is taken notice of accuracy in the rising part of a flood. The procedure of model set-up adopted here is applied for two dam basins: Tonoyama dam basin (294km2) and Tsunokawa dam basin (724km2). It is shown, by using the historical data, that the proposed method is practical and useful. The developed models are going to be utilized for real dam operation in near future.
