Bathymetry Estimation Using Ensemble Adjustment Kalman Filter In The Numerical Simulation Of M₂ Tide

Tide is an important ocean dynamic process.An accurate ocean tide model is the foundation for the numerical simulation of ocean circulation.The numerical simulation results of ocean tide in coastal areas are strongly affected by bathymetry,which is one source of model bias for tidal simulation.Data assimilation can estimate the uncertain parameters in the numerical model while adjusting the state variables with observations to improve the simulation results through enhancing the numerical model.Bathymetry estimation experiments are performed by using Ensemble Adjustment Kalman Filter（EAKF）in this paper during the simulation of M₂constituent for the Bohai Sea and the Yellow Sea to the north of 35°N latitude.To explore the ability of EAKF in bathymetry estimation for linear and non-linear tide models,a two-dimensional linear shallow water model and the external mode of the Princeton Ocean Model with generalized coordinate system（POMgcs）are used for data assimilation experiments.The ideal data assimilation experiments of bathymetry estimation for the semi-diurnal tide simulation in a regular bay mouth area based on the linear shallow water model are implemented.Both the ideal and practical data assimilation experiments of bathymetry estimation for M₂constituent simulation based on the external mode of POMgcs in the Bohai Sea and the Yellow Sea to the north of 35°N latitude are performed as well.The results of the ideal data assimilation experiments in the regular bay mouth area show that the EAKF performed well for bathymetry estimation based on the linear tide model.The strategy of bathymetry parameters estimation can be used for reference to the data assimilation experiments of bathymetry estimation implemented with the external mode of POMgcs.The results of ideal data assimilation experiments in the Bohai Sea and the Yellow Sea to the north of 35°N latitude show that the EAKF method can still reduce the state error and retrieve the“truth”bathymetry when the non-linear tide model is employed.In the practical data assimilation experiments of NAO.99Jb and tide gauge data,comparing with the 34 tide gauges,the errors of model simulated amplitude and phase lag for M₂constituent are reduced by 40.18%and 49.19%,respectively,by use of the posterior estimate of the bathymetry.