集中型流出モデルの構成と流出予測手法

Abstract

An on-line runoff prediction scheme is presented on the basis of filtering and prediction theorydeveloped by Kalman and others. A lumped runoff model to be the base of runoff prediction isnewly developed, and, by incorporating noise terms which are products of Markov noises and statevariables, the runoff model is transformed into a stochastic one. The resulting stochastic differentialequations are solved by an iterative difference method with the aid of a statistical linearization techni-que. In order to verify our scheme, it is applied to some historical events. The rainfall predictiondata which are required in the course of runoff prediction are reproduced by Monte Carlo methods.