Identification of uncertain parameters in a tidal model of the north sea by SPSA

The adjoint method is usually used to identify uncertain parameters in large-scale shallow water flow models. This requires the implementation of the adjoint model, which is huge programming effort. The work presented in this paper is based on stochastic simultaneous perturbation of all model parameters to generate a search direction and avoids the implementation of the adjoint of the original nonlinear model. The SPSA based calibration approach has been implemented for the estimation of the depth values in a shallow sea model of the entire European continental shelf. The results demonstrate that the SPSA method give comparable results to steepest descent method with little computational cost.