Author(s)

Toshiharu Kojiri

Abstract

As a heavy rainfall has caused the flood disaster, it becomes an important issue to predict the rainfall and run-off discharge in advance. New observation equipments such as raingauging radar or telemetric gauging station for precipitation have been provided with covering all area of Japan. In this paper, we will propose the realtime rainfall prediction through radar data by structuring neural network. To consider the time hysteresis and prediction residuals caused by meteorological disturbance, the artificial intelligence technology such as recurrent typed neural network and chaotic prediction method are introduced. llntroduction Flood inundation is one of the serious natural disasters not only in Japan but also in the world. The Ministry of Construction in Japan has provided more than twenty radar-observatio

Keywords