Author(s): Keisuke Yoshida; Shiro Maeno
Linked Author(s): Keisuke Yoshida
Keywords: Data assimilation; Flow resistance; Vegetation; Flood flow
Abstract: This paper describes a data assimilation technique for identifying flow resistance or the bed roughness factor because of vegetation establishment in river floodplains. A shallow-water model was used as the constraint condition for a variational approach of data assimilation, which enables us to develop the adjoint shallow model. The optimal value of the bed roughness coefficient was determined using a quasi-Newton method with adjoint variables. The technique was applied to the inverse estimation of the bed roughness coefficient because of lush vegetation established on floodplains in the Asahi River, Japan during a flooding event. Results showed that the coefficient is reasonably identified, compared with the observation data, even if the iterative estimation of data assimilation starts with an initial guess quite different from the optimal value.
Year: 2013