SEARCH

SEARCH BY CITATION

Numerical experiments are conducted with a coarse-resolution global ice-ocean model in order to determine to what degree the sea ice-ocean exchanges of heat, salt/freshwater, and momentum control the general circulation of the world ocean on long timescales. These experiments reveal that the formation of North Atlantic Deep Water (NADW) in the model results from the strong heat losses that occur at the oceanic surface in the high-latitude North Atlantic. The large-scale ice-ocean interactions have nearly no influence on this process. In particular, neglecting the freshwater flux associated with the southward ice transport at Fram Strait does not impact seriously on the salinity of the Greenland and Norwegian Seas. At equilibrium the absence of this freshwater flux is balanced by an enhanced oceanic freshwater transport from the Arctic. Furthermore, it appears that the model NADW formation does not critically depend on the media (ice or ocean) transporting the freshwater. Besides, both the salt/freshwater and heat exchanges between sea ice and ocean are crucial in the Southern Ocean for the deep water production, properties, and export. The large amount of brine released during ice formation on the model Antarctic continental shelf leads to very high salinities there. The resulting dense shelf waters are then transported toward great depths after some mixing with ambient waters, finally forming the Antarctic Bottom Water body. On the other hand, the net ice melting associated with ice convergence in some areas, such as the southwestern Pacific, stabilizes the water column and forbids deep mixing in these regions. Furthermore, the contact with the ice imposes that the polar surface waters must be maintained very close to their freezing point temperature. Our results suggest that this process takes an important part in increasing the density of the Antarctic Bottom Water. We also show that the modifications of the stress at the ocean surface induced by the internal ice stress have only a regional effect.