Fig. 1

Model setup and wind forcing schemes. (a) Spatial distributions of initial sea surface density (shading) and surface velocity (black arrows). Arrows denote the direction of wind stress; the white dashed box indicates the nested grid region. (b) Initial vertical temperature profiles at the eddy center (red line) and the background field (blue line). (c) Time series of wind stress magnitude across different experiments. Subgrid nesting is activated after 480 h."

Tab. 1

Configuration of each sensitivity experiment"

Fig. 2

Time series of submesoscale kinetic energy variation of eddy fronts in different wind forcing experiments. (a) Time series of the prescribed wind stress forcing; (b) frontal-averaged vertical mixing coefficient (Kv) across different wind frequency experiments; (c) frontal-averaged submesoscale eddy kinetic energy (SEKE) evolution."

Fig. 3

Kinetic energy budget diagnostics and time series of selected dynamic parameters in frontal regions in different wind forcing experiments. The figure shows the time series of VBF (a), kinetic energy flux (Π) (b), normalized convergence (c) and normalized frontogenetic rate (d) averaged over the frontal region at 10 m depth from different experiments. The frontal region is selected according to the frontal intensity, with a threshold of 1×10-7s-2. The yellow shaded area corresponds to the phase of weakened low-frequency wind."

Fig. 4

Submesoscale frontogenesis in each experiment. (a—f) Spatial distributions of frontogenetic rate (Tb/f) and horizontal convergence (-δ/f) at 10 m depth at the 780th hour of the simulation; (g—i) scatter plots of frontogenetic rate versus horizontal convergence in the frontal region for each experiment, where R denotes the correlation coefficient"

Fig. 5

Spatial distribution of frontal intensity and kinetic energy budget diagnostics for each experiment a—c. Frontal intensity $ (|\nabla \mathrm{b}|)$ at 10 m depth at the 780th hour of the simulation, with gray contours representing 1×10-7s-2; d—f. Spatial distributions of vertical buoyancy flux (VBF) for each experiment; g—i. Spatial distributions of kinetic energy flux (Π) for each experiment. Each row corresponds to the stwind (steady wind experiment), windf10 (low-frequency wind experiment) and windf1 (high-frequency wind experiment), respectively."

Fig. 6

Temporal trends of frontal submesoscale kinetic energy and its budget diagnostics for each experiment. The figure shows the time series of 10-day moving averaged SEKE (a), VBF (b) and kinetic energy flux (Π) (c) averaged over the frontal region at 10 m depth from different experiments. The frontal region is selected according to the frontal intensity, with a threshold of 1×10-7s-2"

Fig. 7

Temporal variation trends of the horizontal and vertical terms of kinetic energy flux (Π) for each experiment. The figure shows the temporal variation trends of 10-day moving averaged frontogenetic rate (a), frontal intensity (b), horizontal component of Π (c) and vertical component of Π (d) averaged over the frontal region at 10 m depth from different experiments. The frontal region is selected according to the frontal intensity, with a threshold of 1×10-7s-2"