Abstract
This paper discusses the application of neural network prediction method of the annual riverbed evolution by using the long tern observation data in the Shigenobu river and its area. The proposed neural network method takes into account of the check dam structure construction effect and time-lug effects of the check dam structure, precipitation and river water flow on the river bed evolution. The multiple regression analysis and its prediction method are also discussed to compare with the proposed method. The proposed prediction method shows good agreement with the observed data including several sequential years prediction attempts, and are better than the prediction by the multiple regression method.
