lower densities by water inflowing from the
boundaries.
While the focus of this chapter has been on the
Labrador Sea, numerous aspects of the convective
processes discussed here apply as well to the other
two locations of open-ocean convection in the
North Atlantic: the Greenland Sea and the Mediterranean Sea. For example, all three regions are subject to strong wintertime winds blowing off the
adjacent continental boundaries. These seasonal
winds bring cold, dry air over the relatively warm
ocean (the air–sea temperature contrast is especially
large in the Mediterranean Sea), leading to the
high buoyancy loss that drives the overturning. A
second common factor among the regions is the
‘layering’ of in-situ water masses within the water
column. Each sea is characterized by a relatively
cold and fresh upper layer, warmer and saltier
water beneath this, and a larger body of weakly
stratified water occupying the middle of the water
column (i.e. the remnant of the convective product
from earlier years). The relative volumes and properties of these layers can modulate the occurrence
of convection. For example, a large abundance of
water remaining from an earlier period of convection predisposes the water column to further overturning. On the other hand, enhanced flux of the
highly buoyant surface water to the convection
region can help inhibit the overturning. This effect
appeared to be important during the late 1960s
when the Great Salinity Anomaly invaded the
Labrador Sea (Lazier, 1980).
A third important similarity between the three
areas of convection is the regional circulation; each
of them is characterized by a cyclonic flow regime.
This is undoubtedly a crucial aspect of the preconditioning for convection. The isopycnal doming in
the centre of each of the ‘gyres’ allows the more
weakly stratified subsurface waters to reside closer
to the sea surface and hence be more readily subject to the atmospheric buoyancy forcing. Furthermore, in each sea, part of the cyclonic flow regime
involves boundary currents (e.g. in the Mediterranean Sea it is the westward flowing Northern
Mediterranean Current). This has several ramifications. It enhances the ability for newly formed
water masses to exit the region, and influences the
restratification process since the boundary currents
contain more highly stratified water. Additionally,
there is increasing evidence that convection may at
times occur directly within the boundary currents
(Mauritzen, 1996a; Schott et al., 1996; Pickart
et al., 2000a), since after all this is where the heat
loss is greatest.
But as there are common threads to the overturning in the Labrador, Greenland, and Mediterranean Seas, there are also significant regional
differences. The Greenland Sea is unique in that
ice plays a significant role in the preconditioning
phase. As discussed in Marshall and Schott (1999),
the deepest overturning occurs in late-winter just
after the ice-free ‘Nord Bukta’ region opens up. In
the Mediterranean the winds are more localized in
space than in the other two regions (Candela,
Chapter 5.7). This clearly impacts the convection,
as the region of deepest mixed layers generally lies
in the path of the Mistral winds. Furthermore,
these winds are not cold enough to cause convection during daylight hours, so the Mediterranean
has a daily cycle of overturning that is not present
in the other two seas. Also, the impact of the basinscale NAO wind pattern impacts the Labrador and
Greenland Seas to a much greater extent than it
does the Mediterranean. Finally, the respective
sizes of the convection zones and convected water
masses differ greatly. In the Mediterranean, the
convecting patch is of order 50 km wide; in
the Greenland Sea it is of order 100 km wide; and
in the Labrador Sea the zone of convection
approaches 500 km width, and includes the
boundary currents.
Convection at each of these three Atlantic sites
contributes to the global meridional overturning
circulation, although the quantitative measures are
not yet known. In the Greenland Sea particularly,
the deep convection into the cyclonic gyre seems
rather isolated from the processes that produce the
dense overflows (Mauritzen, 1996a). Mediterranean
water and Labrador Sea water both make an obvious contribution to the Upper North Atlantic Deep
Water, respectively, as very high- and low-salinity
end points. Distant identification of Labrador Sea
Water follows its low salinity, potential vorticity,
and low nutrient concentration, high dissolved
oxygen, tritium and CFCs. The rate of efflux of
Labrador Sea Water into the open Atlantic is
known to vary greatly, with the intensity of wintertime forcing. Johns et al. (1997a) observe velocity
and CFC maxima associated with both deep and
shallow NADW near Abaco, some 5000 km south
of the Labrador Sea. A striking increase in CFCs
(even after adjusting for time-dependence of the
5.5 Deep Convection
399
Lazier, Pickart and Rhines
boundaries.
While the focus of this chapter has been on the
Labrador Sea, numerous aspects of the convective
processes discussed here apply as well to the other
two locations of open-ocean convection in the
North Atlantic: the Greenland Sea and the Mediterranean Sea. For example, all three regions are subject to strong wintertime winds blowing off the
adjacent continental boundaries. These seasonal
winds bring cold, dry air over the relatively warm
ocean (the air–sea temperature contrast is especially
large in the Mediterranean Sea), leading to the
high buoyancy loss that drives the overturning. A
second common factor among the regions is the
‘layering’ of in-situ water masses within the water
column. Each sea is characterized by a relatively
cold and fresh upper layer, warmer and saltier
water beneath this, and a larger body of weakly
stratified water occupying the middle of the water
column (i.e. the remnant of the convective product
from earlier years). The relative volumes and properties of these layers can modulate the occurrence
of convection. For example, a large abundance of
water remaining from an earlier period of convection predisposes the water column to further overturning. On the other hand, enhanced flux of the
highly buoyant surface water to the convection
region can help inhibit the overturning. This effect
appeared to be important during the late 1960s
when the Great Salinity Anomaly invaded the
Labrador Sea (Lazier, 1980).
A third important similarity between the three
areas of convection is the regional circulation; each
of them is characterized by a cyclonic flow regime.
This is undoubtedly a crucial aspect of the preconditioning for convection. The isopycnal doming in
the centre of each of the ‘gyres’ allows the more
weakly stratified subsurface waters to reside closer
to the sea surface and hence be more readily subject to the atmospheric buoyancy forcing. Furthermore, in each sea, part of the cyclonic flow regime
involves boundary currents (e.g. in the Mediterranean Sea it is the westward flowing Northern
Mediterranean Current). This has several ramifications. It enhances the ability for newly formed
water masses to exit the region, and influences the
restratification process since the boundary currents
contain more highly stratified water. Additionally,
there is increasing evidence that convection may at
times occur directly within the boundary currents
(Mauritzen, 1996a; Schott et al., 1996; Pickart
et al., 2000a), since after all this is where the heat
loss is greatest.
But as there are common threads to the overturning in the Labrador, Greenland, and Mediterranean Seas, there are also significant regional
differences. The Greenland Sea is unique in that
ice plays a significant role in the preconditioning
phase. As discussed in Marshall and Schott (1999),
the deepest overturning occurs in late-winter just
after the ice-free ‘Nord Bukta’ region opens up. In
the Mediterranean the winds are more localized in
space than in the other two regions (Candela,
Chapter 5.7). This clearly impacts the convection,
as the region of deepest mixed layers generally lies
in the path of the Mistral winds. Furthermore,
these winds are not cold enough to cause convection during daylight hours, so the Mediterranean
has a daily cycle of overturning that is not present
in the other two seas. Also, the impact of the basinscale NAO wind pattern impacts the Labrador and
Greenland Seas to a much greater extent than it
does the Mediterranean. Finally, the respective
sizes of the convection zones and convected water
masses differ greatly. In the Mediterranean, the
convecting patch is of order 50 km wide; in
the Greenland Sea it is of order 100 km wide; and
in the Labrador Sea the zone of convection
approaches 500 km width, and includes the
boundary currents.
Convection at each of these three Atlantic sites
contributes to the global meridional overturning
circulation, although the quantitative measures are
not yet known. In the Greenland Sea particularly,
the deep convection into the cyclonic gyre seems
rather isolated from the processes that produce the
dense overflows (Mauritzen, 1996a). Mediterranean
water and Labrador Sea water both make an obvious contribution to the Upper North Atlantic Deep
Water, respectively, as very high- and low-salinity
end points. Distant identification of Labrador Sea
Water follows its low salinity, potential vorticity,
and low nutrient concentration, high dissolved
oxygen, tritium and CFCs. The rate of efflux of
Labrador Sea Water into the open Atlantic is
known to vary greatly, with the intensity of wintertime forcing. Johns et al. (1997a) observe velocity
and CFC maxima associated with both deep and
shallow NADW near Abaco, some 5000 km south
of the Labrador Sea. A striking increase in CFCs
(even after adjusting for time-dependence of the
5.5 Deep Convection
399
Lazier, Pickart and Rhines