Item

Abstract

Global Earth System Models generally have problems with the representation of currents and hydrography in the Arctic Ocean, especially with the placement and properties of the Atlantic Water layer. One severe limitation of these models is that they don't resolve mesoscale eddy transport which control the exchanges between shelf and deep ocean in the Arctic. Eddy transport is parameterized, but using theory largely based on lessons from mid-latitude oceans. A particularly severe limitation is their assumption of a flat ocean floor, even though observations and high-resolution modelling show that eddy transport is sensitive to the potential vorticity gradients associated with the sloping bottom topography. Analyzing idealized eddy-resolving simulations from the ocean component of the Norwegian Earth System Model (NorESM), we confirm that sloping bottom indeed reduces diagnosed buoyancy diffusivity. We also show that the reduction can be skillfully captured by mixing length parameterization when scaling diffusivity with the Eady growth rate and using the topographic Rhines scale as a representation of the eddy length scale. The effect is further enhanced when buoyancy diffusivity is scaled by a mean-flow dependent efficiency factor. Applying such parameterization at coarse resolution reduces eddy diffusivity over bottom slopes and thus strengthens the geostrophic mean flow. In realistic simulations following the OMIP-II protocol, the Arctic Ocean boundary currents are enhanced, winter sea ice cover is reduced, and the cold temperature bias in the Atlantic layer is reduced. Overall, introducing a bottom slope aware length scale to the mesoscale eddy parameterization improves the model performance.