Estimation of spatially varying parameters in three-dimensional cohesive sediment transport models by assimilating remote sensing data

Abstract

Based on a three-dimensional cohesive sediment transport model with the adjoint data assimilation, the spatially varying parameters are estimated by assimilating satellite-retrieved suspended sediment concentrations (SSCs) in the Hangzhou Bay, China. To reduce the ill-posedness of the inverse problem, an independent point scheme is developed and a combined independent point and Tikhonov regularization scheme (CIPTRS) is presented. The CIPTRS is calibrated in ideal twin experiments and the results reveal that the CIPTRS can restrict the influence of ill-posedness and improve the accuracy of parameter estimation. In practical experiments, the spatially varying settling velocity is estimated by assimilating the satellite-retrieved SSCs using different strategies, and the best modeling results are obtained when the CIPTRS is used. To further improve the modeling results, the spatially varying settling velocity and initial conditions are estimated simultaneously using the CIPTRS. The data misfit between observed and simulated SSCs is largely decreased and the simulated SSCs can reproduce the spatial and temporal features of observed SSCs. The experimental results indicate that the adjoint method is a useful method to estimate the poorly known parameters in practical applications and the CIPTRS can effectively improve the results of data assimilation and parameter estimation.