Oceanic mesoscale eddies interact strongly with the atmosphere, inducing heat flux that acts to dissipate the eddy potential energy. So far it remains unknown how well this oceanic mesoscale eddy‐atmosphere (OMEA) interaction is represented in the current generation of general circulation models. Here we evaluate the intensity of OMEA interaction in numerical models widely used by the community. It is found that the intensity of OMEA interaction differs significantly among models in its overall magnitude and spatial distribution. In eddy‐rich regions such as Kuroshio Extension and Antarctic Circumpolar Current, the intermodel difference can reach 40%. Surface wind strength and marine atmospheric boundary layer adjustment to mesoscale heat flux anomaly are two important factors accounting for the intermodel difference. Models with stronger surface wind tend to have higher OMEA interaction. Moreover, neglecting the marine atmospheric boundary layer adjustment, ocean‐alone model simulations overestimate OMEA interaction especially at middle and high latitudes by 20%–50%.

Oceanic eddies play a crucial role in transporting heat from the subsurface to surface ocean. However, dynamics responsible for the vertical eddy heat transport (QT) have not been systematically understood, especially in the mixed layer of western boundary current extensions characterized by the coincidence of strong eddy activities and air-sea interactions. In this paper, the winter (December–March) QT in the Kuroshio extension is simulated using a 1-km regional ocean model. An omega equation based on the geostrophic momentum approximation and generalized to include the viscous and diabatic effects is derived and used to decompose the contribution of QT from different dynamics. The simulated QT exhibits a pronounced positive peak around the center of the mixed layer (~60 m). The value of QT there exhibits multi-timescale variations with irregularly occurring extreme events superimposed on a slowly varying seasonal cycle. The proposed omega equation shows good skills in reproducing QT, capturing its spatial and temporal variations. Geostrophic deformation and vertical mixing of momentum are found to be the two major processes generating QT in the mixed layer with the former and the latter accounting for its seasonal variation and extreme events, respectively. The mixed layer instability and the net effect of frontogenesis/frontolysis contribute comparably to the geostrophic deformation induced QT. The contribution of QT from vertical mixing of momentum can be understood on the basis of turbulent thermal wind balance.

Encountering of energetic ocean eddies and atmosphere storms makes the winter Kuroshio extension a hotspot for air-sea interactions. This second part investigates the regulation of vertical eddy heat transport QT in the winter Kuroshio extension mixed layer by different types of air-sea interactions, including the atmosphere synoptic forcing, eddy thermal feedback resulting from eddy-induced surface heat flux anomalies, and eddy current feedback from eddy current’s imprint on wind stress.Atmosphere synoptic forcing modulates intra-seasonal variation of QT by boosting its component contributed by the turbulent thermal wind balance during strong cooling events associated with intense winds. In addition, the magnitude of QT is influenced by the direction of synoptic wind stress primarily via , with the latter exhibiting enhancement both in the downfront- and upfront-wind forcing. Enhanced by the downfront-wind forcing is attributed to increased turbulent vertical viscosity and front intensity caused by the destabilizing wind-driven Ekman buoyancy flux, whereas interaction of uniform wind stress with smaller turbulent vertical viscosity at the front center than periphery (a so-called internal Ekman pumping) accounts for the increased in the upfront-wind forcing. The eddy thermal feedback reduces QT significantly through weakening the fronts. In contrast, the eddy current feedback exerts negligible influences on QT, although it weakens eddy kinetic energy (EKE) evidently. This is due to the much reduced effect of eddy current feedback in damping the fronts compared to EKE and also due to the compensation from Ekman pumping induced by the eddy current feedback.