Application Of Local Error Subspace Transform Kalman Filter Assimilate Sea Level Anomalies In High Resolution Ocean Model

The errors of ocean models mainly come from the parameterization of physical processes,numerical methods,resolution,and uncertainty of atmospheric forcing and boundary conditions.Data assimilation is the main way to reduce model errors and improve model performance.The assimilation method mainly combines observations and model results to give more realistic analysis values.Data assimilation provides more precise initial conditions for the model,thus increasing the short-term and medium-term predictability of the model.The ensemble assimilation method ESTKF(Error Subspace Transform Kalman filter)in PDAF(Parallel Data Assimilation Framework)assimilation framework is used to assimilate the sea surface anomaly SLA(Sea Level Anomaly)to 1/12 degree high resolution oceanic circulation model NEMO(Nucleus for European Modelling of the Ocean)through simple Assimilation Experiments and actual assimilation.Two important assimilation parameters,local radius and forgetting factor,were optimized by experiments.The conclusion shows that the influence of localization radius on the spatial distribution of the analysis results is obvious: if the localization radius is too large,the false correlation in the background error covariance matrix cannot be well filtered;if the localization radius is too small,the analysis will be too detailed to make the physical quantities field conform to reality.The selection of Forget factor has a significant impact on the assimilation effect.Through theoretical experiments,we know that if Forget factor(value 0 to 1)is selected properly,it can significantly improve the assimilation effect,but it is not the smaller the better.If the selection is smaller,the assimilation results will be too confident in the model,and the adjustment of the observation information to the model will be weakened.Therefore,we should pay more attention to the selection of this parameter in actual assimilation.In the actual assimilation experiment,the conclusion is consistent with that of the simple experiment.