Mid-Depth Ventilation Along the Western Boundary of the Sub-Polar Gyre

A late-winter hydrographic survey northeast of the Grand Banks in 1991 revealed the presence of sub-mesoscale eddies of newly-formed water near 500-1000 m depth. Such eddies have not been observed before, and are believed to represent a major source of the pronounced anthropogenic tracer signal observed at this stratum throughout much of the western North Atlantic basin (e.g., Weiss et al., 1985). The density of the lenses is too light to be classical Labrador Sea water (LSW) which is formed during dee p convection in the central Labrador Sea. To investigate the source of these mid-depth eddies and their downstream evolution, a mixed-layer model was applied to historical hydrographic data from the Labrador Sea region, and a primitive equation numerical model of the western boundary circulation was implemented. To determine the winter-time water mass products of the Labrador Sea, a grid of CTD stations occupied in February 1978 (Figure 1) was used as input to a mixed-layer model. Clarke and Gascard (1983) have previously shown that classical LSW can be formed w ithin the western Labrador Sea gyre (Figure 1). The model removes buoyancy specified by the climatological average total heat flux observed at Ocean Weather Station BRAVO (Figure 1), modulated spatially according to the ECMWF heat flux pattern in the re gion. Three general water classes are produced by winter cooling: one within the western Labrador Sea gyre, the second in the interior south of the gyre, and the third along the density front of the Labrador Current on the upper continental Slope (Figur e 1).

Figure 1. Historical CTD stations used in the mixed-layer analysis (crosses). The contours map the potential density averaged over the top 100 m (27.0, 27.2, 27.4, 27.6, 27.65, 27.70 kg/m3) revealing the main branch of the Labrador Current (along the continental slope) and western Labrador Sea gyre (in the interior). Also shown is weather station BRAVO and the location where the mid-depth was observed in 1991.

Figure 2. Temporal evolution of the mixed layer in the three regions discussed in the text.

The western Gyre water mass (region, 1, Figure 2) has the characteristics of classical LSW, i.e., mixed layers on the order of 1500-2000 m in the proper density range. Note however that the water so formed will tend to recirculate, making it difficul t to quickly ventilate the southern regions. The water formed south of the gyre in the interior (region 2, Figure 2) is somewhat lighter than this, though the mixed layers are substantially shallower (order 500 m). Thus while this water has more direct access to regions outside the Labrador Sea, it occupies only the upper-most portion of the water column. The water mass formed within the Labrador Current front (region 3, Figure 2) has the proper T-S and density to be the source of the mid-depth eddies observed further south. The mixed layers are on the order of 1000 m, and reach equilibrium about a month sooner than the classical LSW. This implies that the formation of the water will be less sensitive to the severity of a given winter and thus be fo rmed on a more regular basis. This is consistent with historical observations further south along the western boundary (Pickart, 1992).

Figure 3. Primitive equation model results. (a) model topography and domain, (b) horizontal velocity field (only every other point is plotted), (c) thickness of layer 2. Note splitting of baroclinic Labrador Current into two branches at Flemish Pas s, with the offshore branch characterized by elongated, low potential vorticity eddies consistent with observations.

To investigate the fate of this newly ventilated water within the Labrador Current, as well as the origin of the eddy observed downstream (Figure 1), a primitive equation model of the regional circulation was implemented. The model is based on the is opycnal coordinate model of Bleck et al. (1992), with inflow/outflow conditions forced through relaxation terms in the momentum and continuity equations, as described by Spall (1994). The model domain covers the region around Flemish Cap (Figure 3a) wit h horizontal resolution of 4 km and four isopycnal layers in the vertical (potential densities of 27.5, 27.7, 27.75, 27.82 kg/m3). Note that layer 2 corresponds to the winter-time product of region 3 discussed above. The mid-depth western boundary circu lation in this region consists of two major currents. The main branch of the Labrador Current flows equatorward along the 1000 m isobath and has a baroclinic transport of approximately 4.e+6 m3/s. Lazier and Wright (1993) also report a southward barotro pic current offshore of this (over the 2500 m isobath) with maximum velocities of 20 cm/s, which they term the deep Labrador Current. It is believed that this represents the return flow of the wind-driven sverdrup transport of the sub-polar gyre. The ed dy observed seaward of Flemish Cap was embedded in such a barotropic flow centered over the 2200 m isobath (Figure 4). The layer thicknesses and transports forced from the north in the regional model are based on the available observations of these two b oundary currents.

Figure 4. Mid-depth eddy sampled in late-winter 1991, located seaward of Flemish Cap. Large-dashed lines map the potential temperature at 500 m (3.0-3.15 C); arrows measure the average velocity over the top 1000 m.

The model has been run for a period of 200 days. The Labrador Current becomes baroclinically unstable and develops large amplitude meanders over the first 2-3 weeks. The meanders often pinch off anticyclonic low potential vorticity eddies just north of Flemish Pass near 48N, 48W (Figure 3a) whose maximum signature is within layer 2, the density class of the observed eddies. These eddies are then entrained into the barotropic deep Labrador Current and advected to the outside of Flemish Cap, approxim ately following the topography. The horizontal velocity field in layer 2 is shown in Figure 3b on day 170. The main branch of the Labrador Current is evident flowing along the western boundary, and the deep Labrador Current follows the topography around the outside of Flemish Cap. The thickness of Layer 2 (Figure 3c) indicates the baroclinic structure, and clearly shows a splitting of the low potential vorticity water of the Labrador Current with one branch flowing through Flemish Pass and the other fl owing to the outside of Flemish Cap (the small Region of decreased layer thickness near 47N, 45W is the top of Flemish Cap). This splitting is caused by the entrainment and advection of the large amplitude meanders and eddies of the Labrador Current into the deep barotropic flow just north of Flemish Pass. The eddies then become elongated due to the horizontal shear in the barotropic flow (Figure 3c).

The location, scale, strength, and water properties of these eddies are consistent with the observed lens sampled during the winter 1991 cruise (Figure 4). The model implies that the eddies erode very quickly and that the downstream signature is quit e patchy. This is also consistent with the observations, which reveal a strong downstream decay of the lenses. Based on these model results, it is estimated that approximately 1/3 of the intermediate depth transport in the Labrador Current gets entraine d into the deep Labrador Current and is advected around the outside of Flemish Cap. It is possible that this splitting will impact the basin-scale ventilation of this density class because the waters diverted to the outside of Flemish Cap may (1) mix mor e strongly due to the energetic barotropic flow and small spatial scales of the eddies and (2) get partially entrained into the North Atlantic Current and remain in the subpolar gyre rather than being transported to lower latitudes. Clearly additional obs ervations are needed to sort out the complex spreading and mixing of recently ventilated waters in this portion of the sub-polar gyre.

Robert S. Pickart
Michael A. Spall
Woods Hole Oceanographic Institution
Woods Hole, Massachusetts 02543

REFERENCES

Bleck, R., C. Rooth, D. Hu, L. Smith. 1992. Salinity-driven thermohaline transients in a wind-and thermohaline-forced isopycnic coordinate model of the North Atlantic. J. Phys. Oceanogr., 22: 1486-1505.

Clark, R.A. and J. Gascard, 1983. The formation of Labrador Sea Water. Part I: Large-scale processes. J. Phys. OCeanogr. 1: 1764-1778.

Pickart, R.S. 1992. Water mass Components of the North Atlantic Deep Western Boundary Current. Deep-Sea Res. 9: 1553-1572.

Spall, M.A. 1994. Wave-induced abyssal recirculations. J. Mar. Res., 52:1051-1080.

Weiss, R.F., J.L. Bullisert, R.H. Gammon, and M.J. Warner. 1985. Atmospheric Chlorofluoromethanes in the deep equatorial Atlantic. Nature 314: 608-610.

Return to ACCP notes.