高次元Fokker-Planck方程式を用いた不確実性を有する降雨流出解析に関する研究

抄録

Because of severe flood disasters like The Kanto-Tohoku heavy rainfall in September 2015 that occurred in recent years. The society recognized both structural countermeasures and non-structural countermeasures are necessary. On the other hand, unlike earthquakes and tsunami, there is usually enough time for residents to evacuate in a flood disaster if they are appropriately informed. Thus, the prediction of runoff is a critical index for evacuation. To make the prediction, it needs to consider the uncertainty of rainfall intensity and model parameters in the rainfall-runoff analysis. M.Hino had first introduced the Kalman filter in forecasting the rainfall-runoff process which considered the uncertainty of the process, since then methods such as Kalman filter, ensemble Kalman filter, particle filter, data assimilation, had been used to consider the uncertainty effects in the rainfall-runoff process. However, these methods are based on filtering theory and statistical methods, which cannot recognize the physical meaning of the uncertainty. The present study is based on the theory of high-dimensional Fokker-Planck equation, aimed at suggesting a new way of rainfall-runoff analysis which can not only consider the uncertainty in the system but also identify the physical meaning of these uncertainties