Pathways of Nordic Overflows from climate model scale and eddy resolving simulations

The overflows of cold, heavy waters from the Nordic Seas across the Greenland-Iceland-Scotland Ridges are simulated using the Hybrid Coordinate Model in a North Atlantic configuration. Results at three different horizontal model resolutions are compared to each other, to recent hydrographic sections and moored observations. Simulations in the finest grid employed, 1/12° resolution, show realistic overflow pathways, reasonable overflow and Deep Western Boundary Current mean velocities and transports, and overall reasonable North Atlantic three-dimensional temperature and salinity fields, namely the Atlantic Meridional Overturning Circulation (AMOC). In contrast, simulations at coarser grids of 1/3° and 1° resolution exhibit a range of significant problems owing to unresolved, dynamically vital features in the seafloor topography. This lack of resolution, for example of the Faroe Bank Channel, leads to unrealistic overflow pathways between Iceland and Scotland in the 1/3° and 1° cases. Accordingly, overflow mass transports are also unrealistic in this area. In the Denmark Strait Overflow the underlying topographical scales are larger, and pathways are reasonable even at coarse resolution. However, overflow speeds are too small in the 1/3° and 1° cases. Underestimated velocities in the 1° simulations are compensated by an overestimated sill cross-section, whereas it is too small in 1/3°. As such, the 1/3° and 1° simulations show both large under- and overestimations of volume transport at several locations. No significant improvement in modeled overflows takes place when the grid spacing is decreased from 1° to 1/3°. An experiment conducted with hand-tuned topography shows improved volume transports near the regions of modification, but somewhat increased errors in other parts of the deep circulation, indicating the complex response of the system to perturbations in bathymetry. These results demonstrate the importance of an accurate representation of the domain geometry, in particular the channels of the complex Iceland-Scotland ridge system, in order to reproduce the pathways of the deep AMOC.