The Nordic seas are commonly described as a single basin to investigate their dynamics and sensitivity to environmental changes when using a theoretical framework. Here, we introduce a conceptual model for a two-basin marginal sea that better represents the Nordic seas geometry. In our conceptual model, the marginal sea is characterized by both a cyclonic boundary current and a front current as a result of different hydrographic properties east and west of the midocean ridge. The theory is compared to idealized model simulations and shows good agreement over a wide range of parameter settings, indicating that the physics in the two-basin marginal sea is well captured by the conceptual model. The balances between the atmospheric buoyancy forcing and the lateral eddy heat fluxes from the boundary current and the front current differ between the Lofoten and the Greenland Basins, since the Lofoten Basin is more strongly eddy dominated. Results show that this asymmetric sensitivity leads to opposing responses depending on the strength of the atmospheric buoyancy forcing. Additionally, the front current plays an essential role for the heat and volume budget of the two basins, by providing an additional pathway for heat toward the interior of both basins via lateral eddy heat fluxes. The variability of the temperature difference between east and west influences the strength of the different flow branches through the marginal sea and provides a dynamical explanation for the observed correlation between the front current and the slope current of the Norwegian Atlantic Current in the Nordic seas.

We investigate the characteristics of the sinking of dense waters in the North Atlantic Ocean that constitute the downwelling limb of the Atlantic Meridional Overturning Circulation (AMOC) as simulated by two global ocean models: an eddy-permitting model at 1/4° resolution and its coarser 1° counterpart. In line with simple geostrophic considerations, it is shown that the sinking predominantly occurs in a narrow region close to the continental boundary in both model simulations. That is, the regions where convection is deepest do not coincide with regions where most dense waters sink. The amount of near-boundary sinking that occurs varies regionally. For the 1/4° resolution model, these variations are in quantitative agreement with a relation based on geostrophy and a thermodynamic balance between buoyancy loss and alongshore advection of density, which links the amount of sinking to changes in density along the edge of the North Atlantic Ocean. In the 1° model, the amount and location of sinking appears not to be governed by this simple relation, possibly due to the large impact of overflows and nonnegligible cross-shore density advection. If this poor representation of the processes governing the sinking of dense waters in the North Atlantic Ocean is a generic feature of such low-resolution models, the response of the AMOC to changes in climate simulated by this type of models needs to be evaluated with care.