Author(s): Keisuke Yoshida; Shiro Maeno
Keywords: Data assimilation; flooding; flow resistance; Manning's roughness coefficient; vegetation
Abstract: This paper describes data assimilation for estimating bed roughness factor due to vegetation in natural rivers. An adjoint shallow-water model was developed for variational data assimilation. Optimal values for distributed bed roughness coefficients were determined using a quasi-Newtonian method with flow and adjoint variables. The data assimilation technique was applied to inverse estimation of the distributed coefficients in the lower reaches of the Asahi River in Japan during flooding in 2011. For practical use, the assimilated data comprised water levels measured along the river at regular time intervals. Results show that the magnitude of each coefficient distributed over targeted subregions was reasonably consistent with the vegetation species established in the corresponding subregion, considering characteristic values of vegetation related to the flow resistance.