Application Of Ensemble Assimilation In Ocean Model FVCOM

With the improvement of people’s demand for marine forecast,operational oceanography has been paid more and more attention by the state and society.As one of the three important components of operational oceanography,the Ocean dynamicss model has become an important tool for the study of the ocean system.Among many ocean models,FVCOM,with its unique advantages of unstructured-grid,finite-volume,free-surface and so on,has achieved good results especially in ocean simulation of complex coastline,estuary and coast.Although,because that the ocean model itself is only an approximation of the ocean system,there must be errors,so that the ocean process can’t be perfectly simulated by the model.In order to reduce the error caused by the initial field and parameterization and enable the final result to match the actual situation,the ocean data assimilation method is proposed.The assimilation algorithm is used to make the observation data to continuously correct the model,so as to provide more accurate initial field for the ocean model.Ensemble assimilation algorithm is an important branch of data assimilation algorithm,which has the characteristics of less computing resources and higher computing efficiency.In this paper,taking the south China sea as an example,the satellite altimeter data is applied to the establishment of the tidal model of the south China sea by using the ensemble optimal interpolation algorithm and ocean model FVCOM.The required tidal data was extracted from the jason-1 satellite altimeter data by the along track method,and used as the observation data for assimilation.The harmonic constants of four tidal constituents were selected as the open boundary conditions.The water deepth data originated from Etopo1,which is a global topographic data sets.And finally the conclusion was obtained by analyzing and comparing the tidal constituents of seven tide gauge stations in the south China sea before and after the assimilation,that the accuracy of the model after data assimilation was improved.The average error of m1 reduced 19 cm,and that of phase lag reduced about 7 °.At the same time the average error of m2 reduced about 2 cm,and that of phase lag reduced about 29 °.It can be seen that the effect of assimilation improved significantly.In addition,the improvement effect is not the same in different positions.According to the analysis,this is related to the location of satellite sampling points,the assimilation methods used and the model parameters.