A three-dimensional cohesive sediment transport model with data assimilation: Model development, sensitivity analysis and parameter estimation. (30th June 2018)

• Record Type: Journal Article
• Authors: Wang, Daosheng; Cao, Anzhou; Zhang, Jicai; Fan, Daidu; Liu, Yongzhi; Zhang, Yue
• Is Part Of: Estuarine, coastal and shelf science. Volume 206(2018)
• Page Start: 87
• Page End: 100
• Publication Date: 2018-06-30
• Subjects: Cohesive sediment -- Transport model -- Data assimilation -- Adjoint method -- Sensitivity analysis -- Parameter estimation; Estuarine oceanography -- Periodicals; Coasts -- Periodicals; Estuarine biology -- Periodicals; Seashore biology -- Periodicals; Coasts; Estuarine biology; Estuarine oceanography; Seashore biology; Periodicals; 551.461805
• DOI: 10.1016/j.ecss.2016.08.027
• Languages: English
• ISSNs: 0272-7714

Abstract

Abstract: Based on the theory of inverse problems, a three-dimensional sigma-coordinate cohesive sediment transport model with the adjoint data assimilation is developed. In this model, the physical processes of cohesive sediment transport, including deposition, erosion and advection-diffusion, are parameterized by corresponding model parameters. These parameters are usually poorly known and have traditionally been assigned empirically. By assimilating observations into the model, the model parameters can be estimated using the adjoint method; meanwhile, the data misfit between model results and observations can be decreased. The model developed in this work contains numerous parameters; therefore, it is necessary to investigate the parameter sensitivity of the model, which is assessed by calculating a relative sensitivity function and the gradient of the cost function with respect to each parameter. The results of parameter sensitivity analysis indicate that the model is sensitive to the initial conditions, inflow open boundary conditions, suspended sediment settling velocity and resuspension rate, while the model is insensitive to horizontal and vertical diffusivity coefficients. A detailed explanation of the pattern of sensitivity analysis is also given. In ideal twin experiments, constant parameters are estimated by assimilating 'pseudo' observations. The results show that the sensitive parameters are estimated more easily than the insensitive parameters. The conclusions of this work can provide guidance for the practical applications of this model to simulate sediment transport in the study area.