Mean Circulation

Large-scale cyclonic circulation is necessary preconditioning for convection (Marshall and Schott, 1997). A cyclonic mean current is seen at all depths in the model. The mean sea surface height shows an intensified geostrophic rim current off the coast of Greenland and a slightly less intense Labrador Current. The influence of warm and salty water of the Irminger Current at depths of 200-700 meters extends well into the Labrador Sea convecting region. The depth-integrated mean model streamfunction shows that the rim current transports roughly 30 Sverdrups (see Fig. 1). 15-20 Sverdrups recirculates in the Labrador and Irminger Seas with 10-15 Sverdrups of throughflow. Calculations based on Sverdrup balance of the wind-driven ocean gyre indicate that 30-40 Sverdrups should be returned southward in the subpolar western boundary current (R. Pickart, personal communication). Previous findings from surface drifters indicate sustained rim current velocities as large as 25 $cm/s$ (Lab Sea Group, 1998), while maximum model mean currents are also between 20-30 $cm/s$. Recently-published results from PALACE float data (Lavender et al., 2000) show a 15 $cm$ geostrophic height difference across the rim current at 700 meters depth; the model predicts nearly 20 $cm$ and has the same path (see Fig. 2). The model's deep western boundary current can be seen at depths between 1500-3500 meters, in contrast with the relatively quiescent abyssal circulation elsewhere. Even at 3000 meters, mean deep western boundary current velocities near Labrador approach 10 $cm/s$. Deep currents transporting Denmark Strait overflow water are much weaker. The North Atlantic Current flows through the southeast corner of the model domain, but does not penetrate as far north as the true current. The model North Atlantic Current is 2-3 degrees of latitude too far south at the surface. Although the North Atlantic Current begins as a sharply defined density front, it gradually becomes a diffuse, broad current in this model run. The prescribed boundary conditions are interpolated from a global model run which simply does resolve the sharp features of the Gulf Stream extension. One possible solution is the use of a higher resolution velocity field at the boundaries, but data from floats becomes quite sparse around the domain edges. However, it can be seen that the bulk circulation moves in a believable fashion.

Smaller scale features of the mean circulation sometimes trap and expose surface waters to repeated extreme atmospheric forcing and drive convection. Seasonal means of the model temperature and salinity show eddy-like features propagating around the rim current. One large anomaly travelled from the West Greenland Current around the basin to the Labrador Current in 3 months. This corresponds to an advective speed of roughly 10 $cm/s$ and is consistent with the expected rim current velocity. Rhines and Lazier (1995) report that pulse-like variations in the hydrography of the rim current can be traced upstream to the Irminger Sea. They also report that large incursions of newly-formed LSW can interrupt the rim current's structure after the convective season. The salinity anomaly reported here is present before convection has occurred and therefore is not an incursion of newly-formed LSW. The magnitude of the anomaly decays rapidly over a 4 month timespan. It is unclear if this represents the realistic advection of anomalous properties around the rim current or transient features of a model out of equilibrium. Stationary eddy-like features have been observed in PALACE float data along the rim current also (Lavender et al., 2000). The mean circulation deduced from PALACE floats deployed between 1994-1999 show a succession of cyclonic eddies along the rim at 700 meters depth. The Eulerian description of the PALACE float circulation contains unknown mapping errors, so it is unclear how consistent the model streamfunction should be. The mean model streamfunction at 700 meters does show the cyclonic recirculations in the rim current directly east of Labrador and directly east of Cape Farewell, Greenland. The cyclonic recirculation on the southwest side of Labrador Sea basin may be especially important in trapping surface water to extreme forcing. Because the model spins up from zero velocity during the 1-year run, eddies may not immediately form. However, the mean circulation of the model does not depend greatly on the averaging interval. The mean circulation averaged over the second half of the year shows the same, slightly weak succession of stationary eddies in the model rim current. The model presented here does not predict any interior anticyclonic movement. However, the rim current does have the right flow strength and recirculating tendencies.