Tree vegetation overgrowth in river courses has become a serious engineering issue for riparian management in Japan. In order to properly manage the river courses for both flood protection and ecological conservation, it would be necessary to accurately predict the tree vegetation dynamics for long-term duration. In this study, we developed a stochastic model for predicting the vegetation dynamics with emphasis on the interaction with flood events. In the model, the flood discharge was stochastically simulated using a filtered poison process, one of the popularly used approaches in hydrological time-series generation. The model included effects of vegetation growth, loss by the flood impact, and infant tree invasion. The results of Monte Carlo simulation quantitatively indicated that a suitable equilibrium state of tree vegetation density for both flood protection and ecological conservation was achieved by a certain flood level, and it was degraded as the flood impact became smaller.
