Application Of Hybrid Data Assimilation Methods To Mesoscale Eddy Forecasting In The South China Sea

Data assimilation(DA)is the basis for numerical ocean prediction and reanalysis applications,providing reasonable initial conditions for global or regional numerical ocean prediction.The main assimilation methods currently used in ocean DA are optimal interpolation,three-dimensional variational,four-dimensional variational(4DVAR),ensemble Kalman filter(En KF),and particle filter(PF).PF,which is free from linear and Gaussian assumptions,has become a hot research topic in the field of DA.However,PF faces serious filter degradation problems in the application of high-dimensional systems,and hybrid DA methods based on PF and traditional methods have become effective solutions.The Localized Weighted Ensemble Kalman filter(LWEn KF)and Implicit Equal-Weights Variational Particle Smoother(IEWVPS)are two recent hybrid DA methods.The LWEn KF is a hybrid of PF and En KF,and the IEWVPS is a hybrid of PF and 4DVAR.Based on the Regional Ocean Model(ROMS),this paper designs and implements a test environment for the operational application of hybrid data assimilation methods.We conducted a study on the operational application of hybrid data assimilation methods in the northern South China Sea(SCS),and systematically tested the performance of LWEn KF and IEWVPS in practical applications.Considering both assimilation and mesoscale eddy forecasting effects and computational costs,the LWEn KF has the most potential for operational applications compared with the traditional 4DVAR and En KF assimilation methods.This paper further applies sea surface height from along-track satellite altimeters(AT-SSH),sea surface temperature based on satellite swath coordinates(S-SST)and insitu temperature and salt(T/S)profiles to the hybrid data assimilation test system and fully tests the performance of the LWEn KF in the northern SCS.The results of the study show that for the sea surface observations,the assimilation effect of the LWEn KF is comparable to that of the En KF and significantly better than that of the LPF.For the T/S profiles,the LWEn KF is significantly better than that of the En KF and LPF.For mesoscale eddies,the LWEn KF can effectively forecast their sea surface characteristics,and the forecast results can well represent the three-dimensional structure of the mesoscale eddies.