実時間洪水制御支援のための降雨予測知識の獲得と適用

Abstract

In this paper, we propose one of SOFT for real time flood control, which is local rainfall prediction knowledge and the knowledge acquisition method for predicting frontal rainfall using some suppositions and using physically based estimating method of three-dimensional wind field. In this knowledge acquisition method, we suppose the relationship between vertical wind caused by topography and convection causing heavy rain.We got the local frontal rainfall prediction knowledge for real time flood control by this method. We got the two main things about local rainfall prediction. The horizontal wind velocity over 5m/sec observed by AMeDAS may be the index of heavy rain. And we can specify the place where strong vertical wind often causes.We applied those knowledge to the flood prediction system. As a result, the system made better prediction than without the local rainfall prediction knowledge.