Automatic calibration and simultaneous estimation of bottom friction and open boundary conditions in a tidal model

Data assimilation for parameter estimation in the context of coastal models generally utilises traditional variational or stochastic methods. While these methods are well established and effective in coastal applications, they often rely on gradient information and may suffer from convergence to local minima in highly nonlinear systems with larger parameter sets. In contrast, global optimisation techniques have gained increasing attention due to their ability to explore complex, multimodal and high-dimensional parameter spaces without requiring derivative information. In this paper, we explored the efficiency of a global optimisation method called Particle Swarm Optimisation within a data assimilation framework for the simultaneous estimation of bottom friction and boundary condition control parameters in a tidal model. The results from the twin experiments show that the method is efficient for parameter estimation without the need to remove the less sensitive parameters through sensitivity analysis. The method is also compared to a variational method to confirm this applicability. The optimal parameters from the calibration have been used to validate the model predictions at independent observation points, improving the predictions even in non-assimilated areas. Furthermore, with the assimilation of real data, joint estimation of calibration parameters with spatially varying bottom friction and correction in the amplitude and phase of the predominant tidal constituents lead to more accurate predictions of water elevation from the model.