Thermohaline Structure of an Eddy-Resolving North Atlantic Model: The Influence of Boundary Conditions

Abstract

A T?S volumetric census, with a resolution of 0.2°C and 0.1 psu, for years 20-25 of the World Ocean Circulation Experiment Community Modeling Effort eddy-resolving simulation of the equatorial and North Atlantic Ocean, reveals how the thermohaline character of the model has changed from the initial conditions, which were taken from the Levitus climatology. Any changes in the thermohaline structure, other than stirring, mixing, or geostrophic adjustment of smoothed climatology, must be due to the boundary conditions, which are imposed at the surface and at four sponge layers (northern boundary, southern boundary, Labrador Sea and Mediterranean Sea), where water temperature and salinity are nudged toward climatological conditions. Several unrealistic thermohaline features appear in the solution, which can be traced to these surface and lateral sponge boundary conditions. 1) Water masses from the Arctic Ocean are overrepresented in the model. The volume transport across the northern sponge is twice the value estimated from observations. The heat flux is approximately correct, while the salt flux is large by a factor of 4.2) Water masses from the South Atlantic are underrepresented. The transport of water across the southern sponge is about two-thirds of the observed value, but the salt flux is comparable with estimates. However, the heat flux is only 10% of measured values due to a missing equatorward motion of warm surface waters. 3) Water masses from the Labrador Sea and Baffin Bay are overrepresented. The volume flux is twice that observed, while the heat flux from the sponge is realistic. The salt flux is about 20% of the observed value. 4) Finally, Mediterranean Water is underrepresented. Even though the volume transport across the sponge is eight times the observed value, the net salt flux is small by a factor of 400, leading to an insufficient production of salt. All of these difficulties with the model T-S structure are traced to three general problems. First, the flow at the outer edge of the sponges is strongly barotropic in spite of the fact that the temperature and salinity fields are from climatology. Part of the problem with the sponges may be the smoothed nature of the climatology, which has the effect of reducing density gradients, thereby reducing geostrophic shears In all case, except the southern sponge, the volume transport across the sponge is two to eight times larger than the value expected from other analyses or observations. Since the vertical structure of the Bow is set by the climatology, the only way to create this additional transport is through barotropic flow. The reason for the additional transport is not entirely clear, but it may be due to the excessive vertical velocities that are demanded by the conversion process in the sponges. These vertical motions create bound vortices in the sponge layers that drive recirculation in the vicinity of the sponges, increasing the transport without changing the heat or salt flux. The second problem is due to geometric effects within the sponges. One such problem is that Iceland blocks the exchange along the northern sponge. Another problem is that the ocean bathymetry is specified in the sponge layer. For example, the inner Mediterranean sponge is so shallow (around 100 m) that there is very little area in which to modify the water. Similar conditions occur in the Labrador sponge where the water is also 100 m deep. The third general problem is the use of relaxation to climatology to represent surface freshwater Fluxes, which leads to unrealistic surface forcing it the currents are displaced from climatological locations. The combination of a displaced Gulf Stream and the relaxation of surface salinity to climatology produces mode waters that are unrealistically cool and fresh.