source) was seen in 1995–96, subsequent to the
intensely cold winters of 1988–94, and before this
a path change was observed by their current meter
array. The 8- to 10-year transit time is slower than
the core-current speeds would suggest, yet faster
than earlier estimates. For the era ending 1977,
Jenkins and Rhines (1980) observed dilution of the
tritium maxima of the deep western boundary currents by factors of order 10, from the subpolar gyre
to the Blake–Bahama Outer Ridge; this indicated
dilution and delay (by recirculation) mechanisms
en route, which were later explored in model studies by Pickart (1988).
Model studies (Hallberg and Rhines, 1996) suggest that when convection is initiated or increased
at the high-latitude source, a pressure wave propagates down the western boundary, as a topographic Rossby wave, well before the arrival of
tracer-tainted, identifiable water mass. Such model
studies (also that of Winton, 1997, and Spall and
Pickart, 2000) point out that sloping continental
rise topography acts as a waveguide, and rather
gently leads dense water masses equatorward from
high latitude, as they gently sink. Thus, ‘sinking’ is
minimal in the near-field of the convection, but
occurs downstream. Production of kinetic energy
of the overturning circulation by potential energy
created by buoyancy forcing requires that dense
water sink and less dense water rise, but the sites
of sinking and rising are, at least in model studies,
often distant from the convection. In a diapycnal/
epipycnal coordinate system, however, convection
is more locally associated with time-averaged
diapycnal transport (‘water mass conversion’).
Such analyses are beginning to be carried out with
models and are an insightful way to approach the
link between convection, sinking and global
meridional overturning.
Thus we are still seeking to quantify production
rates of the constituent water masses of the global
meridional overturning. Outward transport of
Labrador Sea Water is even difficult to define,
because of extensive recirculation within the
sub-polar gyre and entrainment once the water
mass has left the subpolar gyre. Estimates have
ranged from less than 1 Sv to more than 10 Sv. By
using the relaxation rate of the central Labrador
Sea Water after cessation of convection (either
annually or, through periods of negative NAO
index, interannually), we estimate that renewal of
that mass occurs at a rate that decreases with depth
from the surface and residence times ranging from
roughly 2 years above, to greater than 5 years at
2000 m (Lilly et al., 1999). Much remains unknown
about the detailed geography of deep convection
and circulation. In the Labrador Sea, both interior
and boundary currents are known to participate in
the deep convection (Spall and Pickart, 2000, estimate 1–2 Sv of boundary current production). The
boundary current, however, is shielded from deep
convection by low-salinity shelf waters at some
sites. Where the circulation crosses from Greenland
to Labrador, the boundary currents broaden and
slow, and are generally exposed to some of the
most intense air–sea heat flux in the Sea; here and
over the wide continental slope near Labrador,
convection may be particularly deep (Cuny et al.,
2000).
Direct velocity and transport measurements are
needed to augment water mass observations.
Unfortunately the Lagrangian movement of water
masses is difficult to observe, even with modern
‘quasi-Lagrangian’ floats and drifters. Lavender
et al. (2000c) describe the Labrador/Irminger Sea
circulation as observed with P-ALACE floats.
They concentrate on the Eulerian ensemble mean
velocity field, and remark that ‘no floats travelled
southward to the subtropical gyre in the deep
western boundary current, the putative main pathway of dense water in the meridional overturning
circulation.’ If the boundary current is concentrated to a narrow width, for example at the
Flemish Cap, then these profiling floats may have
difficulty staying within it; tracer observations
assure us that the transport does take place.
SECTION 5 FORMATION AND TRANSPORT OF WATER MASSES
400
intensely cold winters of 1988–94, and before this
a path change was observed by their current meter
array. The 8- to 10-year transit time is slower than
the core-current speeds would suggest, yet faster
than earlier estimates. For the era ending 1977,
Jenkins and Rhines (1980) observed dilution of the
tritium maxima of the deep western boundary currents by factors of order 10, from the subpolar gyre
to the Blake–Bahama Outer Ridge; this indicated
dilution and delay (by recirculation) mechanisms
en route, which were later explored in model studies by Pickart (1988).
Model studies (Hallberg and Rhines, 1996) suggest that when convection is initiated or increased
at the high-latitude source, a pressure wave propagates down the western boundary, as a topographic Rossby wave, well before the arrival of
tracer-tainted, identifiable water mass. Such model
studies (also that of Winton, 1997, and Spall and
Pickart, 2000) point out that sloping continental
rise topography acts as a waveguide, and rather
gently leads dense water masses equatorward from
high latitude, as they gently sink. Thus, ‘sinking’ is
minimal in the near-field of the convection, but
occurs downstream. Production of kinetic energy
of the overturning circulation by potential energy
created by buoyancy forcing requires that dense
water sink and less dense water rise, but the sites
of sinking and rising are, at least in model studies,
often distant from the convection. In a diapycnal/
epipycnal coordinate system, however, convection
is more locally associated with time-averaged
diapycnal transport (‘water mass conversion’).
Such analyses are beginning to be carried out with
models and are an insightful way to approach the
link between convection, sinking and global
meridional overturning.
Thus we are still seeking to quantify production
rates of the constituent water masses of the global
meridional overturning. Outward transport of
Labrador Sea Water is even difficult to define,
because of extensive recirculation within the
sub-polar gyre and entrainment once the water
mass has left the subpolar gyre. Estimates have
ranged from less than 1 Sv to more than 10 Sv. By
using the relaxation rate of the central Labrador
Sea Water after cessation of convection (either
annually or, through periods of negative NAO
index, interannually), we estimate that renewal of
that mass occurs at a rate that decreases with depth
from the surface and residence times ranging from
roughly 2 years above, to greater than 5 years at
2000 m (Lilly et al., 1999). Much remains unknown
about the detailed geography of deep convection
and circulation. In the Labrador Sea, both interior
and boundary currents are known to participate in
the deep convection (Spall and Pickart, 2000, estimate 1–2 Sv of boundary current production). The
boundary current, however, is shielded from deep
convection by low-salinity shelf waters at some
sites. Where the circulation crosses from Greenland
to Labrador, the boundary currents broaden and
slow, and are generally exposed to some of the
most intense air–sea heat flux in the Sea; here and
over the wide continental slope near Labrador,
convection may be particularly deep (Cuny et al.,
2000).
Direct velocity and transport measurements are
needed to augment water mass observations.
Unfortunately the Lagrangian movement of water
masses is difficult to observe, even with modern
‘quasi-Lagrangian’ floats and drifters. Lavender
et al. (2000c) describe the Labrador/Irminger Sea
circulation as observed with P-ALACE floats.
They concentrate on the Eulerian ensemble mean
velocity field, and remark that ‘no floats travelled
southward to the subtropical gyre in the deep
western boundary current, the putative main pathway of dense water in the meridional overturning
circulation.’ If the boundary current is concentrated to a narrow width, for example at the
Flemish Cap, then these profiling floats may have
difficulty staying within it; tracer observations
assure us that the transport does take place.
SECTION 5 FORMATION AND TRANSPORT OF WATER MASSES
400