becomes restratified more quickly. Theoretical considerations suggest that convection adjacent to a
sloping boundary should result in a greater net
vertical sinking of water than overturning in the
interior (Hallberg and Rhines, 1996; Spall and
Pickart, 2000).
Subsequent to formation, the new LSW spreads
to other regions of the ocean at intermediate depths.
Our knowledge of the speed of this flow and its
influence increased during the World Ocean Circulation Experiment (WOCE) due to the widespread
high quality observations of temperature, salinity
and CFCs across the North Atlantic. One study by
Sy et al. (1997) suggested that the newly ventilated
water formed in the Labrador Sea during the severe
winters of the early 1990s moved into the Irminger
Sea within months, to the Iceland Basin within a
year, and to Rockall Trough on the eastern boundary in approximately 5 years, yielding a transAtlantic speed of approximately 2 cm s
91
. This
speed is about three to four times greater than the
previous estimate, and one conclusion is that the
intermediate flows are much faster than previously
thought. An alternative explanation, however, is
that convection occurred over a far greater extent of
the Labrador Basin during this high NAO (North
Atlantic Oscillation) period. This would allow for
more direct (and much faster) contact of the new
water with both the North Atlantic Current, and
Deep Western Boundary Current (including convection directly into the latter as noted above). In this
scenario, the ‘initial condition’ for escape from
the Labrador Sea is greatly enhanced. Regarding the
surprisingly short time lag to the Irminger Basin
(6 months according to Sy et al., 1997), yet another
scenario is possible: that deep convection occurred
within the Irminger Basin itself. This possibility is
raised by Pickart et al. (2000b).
The far-field influence of LSW in the subtropical gyre was examined by Curry et al. (1998) in a
comparison of six decades of data from the
Labrador Sea and from the ocean near Bermuda.
A correlation between these two regions suggests
the products of deep convection in the Labrador
Sea impact the waters off Bermuda after about 6
years. These studies along with others such as
those by Cunningham and Haine (1995a,b) confirm the LSW flow patterns mapped by Talley and
McCartney (1982) and sketched in Figure 5.5.1.
For the reader wishing a more thorough review
of deep convection, Marshall and Schott (1999)
provide an excellent summary of studies of observations, laboratory experiments, theory and modelling. Also the Labrador Sea is the focus of two
recent descriptions. The first, by the Lab Sea
Group (1998), concentrates on the observations
obtained in conjunction with the cruise of the
Knorr to the Labrador Sea in January and February 1997. This includes high-resolution numerical
simulations and data from meteorological measurements, CTD surveys and neutrally buoyant
floats. The second by Lilly et al. (1999) concentrates on the observations obtained from a mooring placed in the central part of the sea between
1994 and 1995, and by drifting, profiling
P-ALACE (Profiling-Autonomous Lagrangian Circulation Explorer) floats.
In the next section some features of plumes, the
convective mixing agents, are reviewed. This is followed in Sections 5.5.3 and 5.5.4 by discussions of
two phenomena associated with the convection.
The first is the increase in the amplitude of temperature and salinity variability during convection
observed in moored records. The second is the
restratification of the density field following the
cessation of convection.
5.5.2 Plumes – the mixing agent
Convection begins to increase the depth of the
mixed layer in the Labrador Sea near the end of
September when the surface net buoyancy flux
from the surface turns from positive to negative.
Deepening continues until about the end of
March, when the buoyancy flux again becomes
positive. The effect of this deepening on the density structure is illustrated, for part of the water
column, by the year-long time series of 1.5 at five
depths between 260 m and 2000 m plotted in
Figure 5.5.5 (see Plate 5.5.5, p. 428). The deepening mixed layer is indicated to have reached the
instrument at 260 m by the rapid increase in 1.5 in
the first week in February. This is followed a week
or so later by a similar increase at 510 m as the
mixed layer reaches that instrument. The fact that
the 1.5 at these two depths continues to increase
until the end of the first week in March suggests
the mixed layer continues to increase in depth and
density during this interval. After the first week in
March, 1.5 at 260 and 510 m remain relatively
constant for about 3 weeks. This likely indicates
convection is continuing but that the buoyancy
5.5 Deep Convection
391
Lazier, Pickart and Rhines
sloping boundary should result in a greater net
vertical sinking of water than overturning in the
interior (Hallberg and Rhines, 1996; Spall and
Pickart, 2000).
Subsequent to formation, the new LSW spreads
to other regions of the ocean at intermediate depths.
Our knowledge of the speed of this flow and its
influence increased during the World Ocean Circulation Experiment (WOCE) due to the widespread
high quality observations of temperature, salinity
and CFCs across the North Atlantic. One study by
Sy et al. (1997) suggested that the newly ventilated
water formed in the Labrador Sea during the severe
winters of the early 1990s moved into the Irminger
Sea within months, to the Iceland Basin within a
year, and to Rockall Trough on the eastern boundary in approximately 5 years, yielding a transAtlantic speed of approximately 2 cm s
91
. This
speed is about three to four times greater than the
previous estimate, and one conclusion is that the
intermediate flows are much faster than previously
thought. An alternative explanation, however, is
that convection occurred over a far greater extent of
the Labrador Basin during this high NAO (North
Atlantic Oscillation) period. This would allow for
more direct (and much faster) contact of the new
water with both the North Atlantic Current, and
Deep Western Boundary Current (including convection directly into the latter as noted above). In this
scenario, the ‘initial condition’ for escape from
the Labrador Sea is greatly enhanced. Regarding the
surprisingly short time lag to the Irminger Basin
(6 months according to Sy et al., 1997), yet another
scenario is possible: that deep convection occurred
within the Irminger Basin itself. This possibility is
raised by Pickart et al. (2000b).
The far-field influence of LSW in the subtropical gyre was examined by Curry et al. (1998) in a
comparison of six decades of data from the
Labrador Sea and from the ocean near Bermuda.
A correlation between these two regions suggests
the products of deep convection in the Labrador
Sea impact the waters off Bermuda after about 6
years. These studies along with others such as
those by Cunningham and Haine (1995a,b) confirm the LSW flow patterns mapped by Talley and
McCartney (1982) and sketched in Figure 5.5.1.
For the reader wishing a more thorough review
of deep convection, Marshall and Schott (1999)
provide an excellent summary of studies of observations, laboratory experiments, theory and modelling. Also the Labrador Sea is the focus of two
recent descriptions. The first, by the Lab Sea
Group (1998), concentrates on the observations
obtained in conjunction with the cruise of the
Knorr to the Labrador Sea in January and February 1997. This includes high-resolution numerical
simulations and data from meteorological measurements, CTD surveys and neutrally buoyant
floats. The second by Lilly et al. (1999) concentrates on the observations obtained from a mooring placed in the central part of the sea between
1994 and 1995, and by drifting, profiling
P-ALACE (Profiling-Autonomous Lagrangian Circulation Explorer) floats.
In the next section some features of plumes, the
convective mixing agents, are reviewed. This is followed in Sections 5.5.3 and 5.5.4 by discussions of
two phenomena associated with the convection.
The first is the increase in the amplitude of temperature and salinity variability during convection
observed in moored records. The second is the
restratification of the density field following the
cessation of convection.
5.5.2 Plumes – the mixing agent
Convection begins to increase the depth of the
mixed layer in the Labrador Sea near the end of
September when the surface net buoyancy flux
from the surface turns from positive to negative.
Deepening continues until about the end of
March, when the buoyancy flux again becomes
positive. The effect of this deepening on the density structure is illustrated, for part of the water
column, by the year-long time series of 1.5 at five
depths between 260 m and 2000 m plotted in
Figure 5.5.5 (see Plate 5.5.5, p. 428). The deepening mixed layer is indicated to have reached the
instrument at 260 m by the rapid increase in 1.5 in
the first week in February. This is followed a week
or so later by a similar increase at 510 m as the
mixed layer reaches that instrument. The fact that
the 1.5 at these two depths continues to increase
until the end of the first week in March suggests
the mixed layer continues to increase in depth and
density during this interval. After the first week in
March, 1.5 at 260 and 510 m remain relatively
constant for about 3 weeks. This likely indicates
convection is continuing but that the buoyancy
5.5 Deep Convection
391
Lazier, Pickart and Rhines