Inverse estimation of time-series variation of bed roughness coefficients by using adjoint shallow-water model

Abstract

It is well known that the Manning roughness coefficient possibly varies during floods, depending on the bed conditions with or without sand waves and vegetations, and it is then difficult to identify the effective value of the coefficient by means of the on-site observation of the bed materials during normal stages. This paper proposed a practical method of an inverse estimation of the time-variant coefficient by using the flow data. The method is based on the optimal control with an adjoint shallow-water model. The fundamental twin experiments were carried out with the synthetic observation data of water level and depth-averaged velocity, in order to verify the method. The results showed that the time-variation can be accurately identified by using more observation data than the number of the unknown variables.