Vegetation overgrowth along river networks in Japan has been a serious engineering problem for both flood protection and ecological conservation. For proper riparian management under the severe circumstances of the global climate change, it would be necessary to predict the vegetation dynamics for long-term duration. In this paper, we try to expand the stochastic process model, developed for predicting the vegetation dynamics in a river channel, to the whole river network by employing a stream ordering system, i.e., Shreve's link magnitude. In the model, the flood discharge was stochastically simulated using a shot noise process, whose model parameters are examined to relate the link magnitude of a river network. The results of Monte Carlo simulation for several cross sections in the Kako River confirm that the present model is able to predict the vegetation dynamics in the whole river network. Moreover, responses of the vegetated channels are stochastically evaluated in terms of the changes of river discharge and channel geomorphology.
