Predictability and Error Growth Dynamics of the Kuroshio Extension State Transition Process in an Eddy-Resolving Regional Ocean Model

Based on an eddy-resolving Regional Ocean Modeling System (ROMS) and the conditional nonlinear optimal perturbation (CNOP) method, we set up a nonlinear optimization system to investigate the effects of the fastest-growing initial errors on the prediction of the Kuroshio Extension (KE) transition process from the stable to the unstable state. The results reveal that the larger values of the CNOP-type fastest-growing initial errors are mainly located upstream of the KE over the upper 500 m. Furthermore, the dynamic mechanism of the error growth is explored in terms of error movement, maintenance and development during the KE transition period. The results indicate that the errors are transported eastward from the south of Japan to the KE’s first quasi-stationary meander by linear advection. The nonlinear advection slows down the moving speed of these errors, leading to their long-term existence in the first quasi-stationary meander of the KE. In this situation, errors have enough time to absorb energy from the background field continuously through barotropic energy conversion. Ultimately, errors grow rapidly and further induce the overestimation of the KE strength.