Abstract
The sensitivity of the circulation and surface properties of the North Atlantic to different forms of surface forcing is investigated using a Bryan and Cox type model. Forcing taking the form of heat and freshwater fluxes calculated using bulk aerodynamic formulae (with an effective salt flux representing freshwater forcing) failed to simulate the low salinity region in the Labrador Sea and north of the Gulf Stream. By contrast, the use of this form of surface forcing, but with direct freshwater forcing and transport instead of the more commonly used salt flux, reproduces the low salinity in these regions and gives a circulation in good agreement with that obtained using the robust diagnostic mode. Since coupled climate models currently use an artificial salt flux form for the freshwater fluxes between ocean and atmosphere, the poor performance of this surface forcing reported here has significant implications for climate prediction.
