with the sea. These droplets are sufficiently small
that they dissolve before they have settled an appreciable distance. The apparatus is designed to
allow the accurate delivery of SF 6 at rates of up to
3 kg h
À1 onto a given ‘target’ density surface at any
depth greater than 200 m in the ocean, when towed
behind a ship on a conducting cable.
In use, the injector was controlled by a computer
in the laboratory of the ship. The output of the CTD
was used to calculate in real time the density of the
water at the package, and compare it to the ‘target’
density. The computer issued commands to the
automated winch to haul in wire if the density was
higher than the target, or pay out if it was significantly lower. During the North Atlantic Tracer Release Experiment (NATRE) this system was able to
deliver tracer with an overall RMS accuracy of 72 m
from the target surface. With such an injection system, it is practical to initiate experiments using several hundred kilograms of tracer, sufficient to enable
investigations at the ocean-basin scale.
NATRE: Overview of Results
The tracer results from NATRE have been reported
in detail. Major findings were that the diapycnal
diffusivity was 0.12 cm
2 s
À1 for the first 6 months,
and then 0.17 cm
2 s
À1 for the subsequent 24 months.
The mean vertical profile for each survey was nearly
Gaussian, and as a set they illustrate an approximate
solution of the diffusion equation in one dimension
(Figure 1).
The result that the diapycnal diffusivity in the
pycnocline is of order 0.1 cm
2 s
À1 confirms estimates
based on internal wave dynamics and on measurements of turbulent dissipation rates. Some analyses of
the penetration of transient tracers into the deep
pycnocline also have implied diffusivities on the order
of 0.1 cm
2 s
À1
. Values of diffusivity of 1 cm
2 s
À1 were
inferred by Munk’s classic ‘abyssal recipes’ analysis,
but this was for depths between 1000 and 4000 m
and included boundary processes as well as interior
processes. It is now clear that 1 cm
2 s
À1 is an overestimate for the interior pycnocline.
The lateral dispersion of the tracer revealed surprisingly efficient mechanisms of stirring at scales
from 0.1 to 30 km. The lateral diffusivity setting the
width of tracer streaks at 6 months was found to be
about 2 m
2 s
À1
. The mechanism is not well understood, but may be due to shear dispersion by vortices
generated during the adjustment to diapycnal mixing
events. The experiment did confirm the predictions
of C. Garrett, that a tracer patch remains streaky
only for a year or so, after which time the exponential growth of the area actually tainted by the
tracer streaks catches up with a power-law growth of
the overall area occupied by the tracer patch.
It is important in a tracer study of mixing in the
ocean to measure the hydrodynamic forcing, and
also to measure hydrodynamic parameters that are
believed to be useful for estimating diffusivities, so
that existing theories can be tested. Several groups
were involved in profiling fine structure and microstructure during NATRE. Dissipation of turbulent
kinetic energy and temperature variance measured
by profiling instruments gave estimates of diapycnal
diffusivity which agreed closely with the tracer results. Measurements of the fine structure have helped
reveal the roles of shear and double diffusive gradients in driving the mixing.
Dependency of Diapycnal Mixing and Buoyancy
Frequency from Tracer Release Experiments
Figure 2 shows diapycnal diffusivities as a function of
buoyancy period for the deep ocean tracer release
experiments so far published. Except for the recent
0
0.005 0.010 0.015 0.020 0.025 0.030
_ 150
_ 100
_ 50
0
50
100
150
C (Normalized)
Height above target density surface (m)
30
12
6
0
5 mo
Figure 1 Mean vertical profiles from NATRE at 0, 5, 6, 12 and
30 months after the initial survey. The SF 6 concentration has
been averaged on isopycnal surfaces, approximately, and plotted
versus height above the target isopycnal surface using the mean
relation between depth and density for the 12-month survey.
The profiles are normalized to have equal areas. The initial profile
(y .) is allowed to run off the graph so that the others are clear.
176 TRACER RELEASE EXPERIMENTS
that they dissolve before they have settled an appreciable distance. The apparatus is designed to
allow the accurate delivery of SF 6 at rates of up to
3 kg h
À1 onto a given ‘target’ density surface at any
depth greater than 200 m in the ocean, when towed
behind a ship on a conducting cable.
In use, the injector was controlled by a computer
in the laboratory of the ship. The output of the CTD
was used to calculate in real time the density of the
water at the package, and compare it to the ‘target’
density. The computer issued commands to the
automated winch to haul in wire if the density was
higher than the target, or pay out if it was significantly lower. During the North Atlantic Tracer Release Experiment (NATRE) this system was able to
deliver tracer with an overall RMS accuracy of 72 m
from the target surface. With such an injection system, it is practical to initiate experiments using several hundred kilograms of tracer, sufficient to enable
investigations at the ocean-basin scale.
NATRE: Overview of Results
The tracer results from NATRE have been reported
in detail. Major findings were that the diapycnal
diffusivity was 0.12 cm
2 s
À1 for the first 6 months,
and then 0.17 cm
2 s
À1 for the subsequent 24 months.
The mean vertical profile for each survey was nearly
Gaussian, and as a set they illustrate an approximate
solution of the diffusion equation in one dimension
(Figure 1).
The result that the diapycnal diffusivity in the
pycnocline is of order 0.1 cm
2 s
À1 confirms estimates
based on internal wave dynamics and on measurements of turbulent dissipation rates. Some analyses of
the penetration of transient tracers into the deep
pycnocline also have implied diffusivities on the order
of 0.1 cm
2 s
À1
. Values of diffusivity of 1 cm
2 s
À1 were
inferred by Munk’s classic ‘abyssal recipes’ analysis,
but this was for depths between 1000 and 4000 m
and included boundary processes as well as interior
processes. It is now clear that 1 cm
2 s
À1 is an overestimate for the interior pycnocline.
The lateral dispersion of the tracer revealed surprisingly efficient mechanisms of stirring at scales
from 0.1 to 30 km. The lateral diffusivity setting the
width of tracer streaks at 6 months was found to be
about 2 m
2 s
À1
. The mechanism is not well understood, but may be due to shear dispersion by vortices
generated during the adjustment to diapycnal mixing
events. The experiment did confirm the predictions
of C. Garrett, that a tracer patch remains streaky
only for a year or so, after which time the exponential growth of the area actually tainted by the
tracer streaks catches up with a power-law growth of
the overall area occupied by the tracer patch.
It is important in a tracer study of mixing in the
ocean to measure the hydrodynamic forcing, and
also to measure hydrodynamic parameters that are
believed to be useful for estimating diffusivities, so
that existing theories can be tested. Several groups
were involved in profiling fine structure and microstructure during NATRE. Dissipation of turbulent
kinetic energy and temperature variance measured
by profiling instruments gave estimates of diapycnal
diffusivity which agreed closely with the tracer results. Measurements of the fine structure have helped
reveal the roles of shear and double diffusive gradients in driving the mixing.
Dependency of Diapycnal Mixing and Buoyancy
Frequency from Tracer Release Experiments
Figure 2 shows diapycnal diffusivities as a function of
buoyancy period for the deep ocean tracer release
experiments so far published. Except for the recent
0
0.005 0.010 0.015 0.020 0.025 0.030
_ 150
_ 100
_ 50
0
50
100
150
C (Normalized)
Height above target density surface (m)
30
12
6
0
5 mo
Figure 1 Mean vertical profiles from NATRE at 0, 5, 6, 12 and
30 months after the initial survey. The SF 6 concentration has
been averaged on isopycnal surfaces, approximately, and plotted
versus height above the target isopycnal surface using the mean
relation between depth and density for the 12-month survey.
The profiles are normalized to have equal areas. The initial profile
(y .) is allowed to run off the graph so that the others are clear.
176 TRACER RELEASE EXPERIMENTS