The Predictability Studies Of The Kuroshio Intrusion Into The South China Sea

The Kuroshio Intrusion(KI)into the South China Sea(SCS)is a key process that greatly modulates multi-scale dynamics in the northeastern SCS.Using the Regional Ocean Modeling System(ROMS)and the Conditional Nonlinear Optimal Perturbation(CNOP)approach,the predictability of the KI is studied from the perspectives of optimal precursor(OPR)and optimally growing initial error(OGIE).Our study found that the OPRs obtained by the CNOP approach are located in the southern part of the Luzon Strait and constrained in the upper 400 m.The OPRs exhibit the anticyclonic eddies(AEs)spatial pattern,with positive sea surface height(SSH)perturbations,positive temperature perturbations and clockwise-rotating velocity perturbations.After superimosimg the AE-like perturbation to the reference state,the perturbation grows rapidly and moves northwestward along the Kuroshio path.Simultaneously,this perturbation forces the Kuroshio to continuously move westward and trigger the occurrence of KI at last.The eddy-energetic analysis shows that the rapid growth of the AE-like perturbation is attributed to the barotropic instabilities.Additional examination with the HYCOM oceanic reanalysis and AVISO remote-sensing data further validate the importance of the AE-like perturbations in the KI triggering process.On the other hand,we obtained two types of OGIEs by the CNOP approach.Both types of OGIEs are also localized in the southern part of the Luzon Strait.The first type of OGIE,with negative SSH,negative temperature and anticlockwise-rotating velocity error,leads to an underestimate of the KI amplitude.The second type of OGIE,manifesting the features of AEs,is nearly opposite the first type,leading to an overestimate of the KI amplitude.The growth mechanism of the OGIEs is similar to that of the OPRs,which determines the high similarity between them.Finally,we identify the sensitive areas of KI used to implement targeted observations and evaluate the effects of targeted observations with observation system simulation experiment.The evaluation demonstrates that the implement of targeted observations in the sensitive areas could efficiently remove the initial errors and improve the accuracy of KI prediction.