Accuracy Validation of Flood Forecasting System for River with Flood Adjustment Dam in the Middle Basin

Abstract

Sagami River is the first class river administered by Kanagawa prefecture, designated as a flood forecast river in 2006. It is essential to predict the water volume discharged by dam since water volume discharged by Shiroyama dam accounts for 75 to 95 percent of water volume during floods at the prediction point on the river. Artificial neural networks model, one of the correlation analysis methods, was employed to estimate water volumes of dam such as water volume discharge and flowing. In this study, predictability of online forecasting system of Typhoon No. 9 on September 2007 which rose river level to the critical water level was validated. The results are as follows, 1) Predicted water level form almost agreed with actual water level form. Especially, the rapid rising part of water form which is important for both flood control and evacuation activities was successfully predicted in advance. 2) It was suggested that water level prediction accuracy in lower basin can be improved by increasing accuracy of the upper reach flow volume through two different method, outflow analysis and correlation analysis.