They also result in relatively simple equations for
the mean ocean circulation. Thus if we consider the
vertically integrated effect, we obtain Sverdrup’s
(1947) relation between the curl of the wind stress
and the ocean meridional transport. The addition
of a western boundary current then leads to the
Stommel (1948, 1965), Munk (1950) and Munk
and Carrier (1950) models of the gyre circulation,
the Godfrey (1989) ‘Island Rule’ and the Stommel
and Arons (1960a,b) theory for the abyssal circulation of the ocean.
The next level of understanding comes from
applying the same ideas to the conservation of
potential vorticity on density surfaces. The results
emphasize the role of ventilation in determining the
vertical structure of the subtropical gyres (Price,
Chapter 5.3). They also highlight the importance
of potential vorticity mixing within the ocean.
These theories often work well for the nearsurface flows in the subtropics, but they tend to be
less useful at high latitudes and at depth because
there the effect of topography is important. Topography can both steer currents and produce isolated
Taylor columns. Elsewhere divergences and convergences associated with the bottom Ekman layer
can also have an effect.
In trying to understand the mean circulation we
also need to consider the effect of the mesoscale
eddy field, internal waves and tides. At one stage it
was thought that the mesoscale eddies were so
widespread that they might have an important
effect on the dynamics (Warren, 1981b). However,
this now appears to be true only for the regions
where eddy kinetic energy is highest (Stevens and
Ivchenko, 1997) or where non-linear interactions
with topography generate an along-slope current
(Holloway, 1992). Elsewhere the eddies are very
effective at mixing properties on density surfaces,
but otherwise their dynamical effect is small.
For a long time the background internal wave
field was thought to be important for vertical mixing in the deep ocean (Munk, 1966). This is the
field usually associated with the Garrett and Munk
(1979) spectrum that has a relatively uniform
amplitude throughout the ocean. However, both
theory and direct measurements show that the
cross-isopycnal mixing it produces is an order of
magnitude too small to explain the observed
properties of the ocean (Toole and McDougall,
Chapter 5.2). Instead there is now evidence that
topography (Polzin et al., 1997; see also Hogg,
Chapter 4.5) and tides (Egbert, 1997; Munk and
Wunsch, 1998) can produce large amounts of vertical mixing in localized regions of ocean. This is a
new field of study and the implications have still
to be fully investigated.
4.2.2 Observational evidence
As stated above, the interior flow of the ocean is
difficult to measure directly with current meters
because of the long integration time needed to
overcome the masking effect of the mesoscale eddy
field and other motions. Because of this, our
knowledge about the ocean circulation has come
primarily from hydrographic sections and tracer
studies. In recent years, and especially during the
World Ocean Circulation Experiment (WOCE),
this has been complemented by high-quality data
from deep floats (Autonomous Lagrangian Circulation Explorer (ALACE) floats) and surface
drifters. Very long current meter data records are
also becoming available.
4.2.2.1 Classical hydrography
Some of the best early descriptions of the largescale Atlantic circulation resulted from the Meteor
expedition of the 1920s. Two main approaches
were used to analyse the data, one based on the
distribution of tracers (Wüst, 1935) and the other
on the thermal wind equation (Defant, 1941). The
dynamic height method (see also Fofonoff, 1962)
works well in the top few hundred metres of
the ocean, where currents are strong. As a result
Defant’s (1941, 1961) analysis of the surface
dynamic height field of the Atlantic still compares
well with modern results. Defant also found that
down to about 900 m the amplitude of the field
slowly declined but its overall shape changed little.
Below 1000 m the method tends to fail. This is
because the method only defines current shears
and so it needs an absolute current to be defined at
some level. Water properties can be used to estimate a level of no motion (or zero reference current) but, given the low currents in the deep ocean,
it is difficult to do this to sufficient accuracy.
However, where currents are strong, or the level of
no motion estimates are good ones, the method
may still be successful. Thus, despite the limited
amount of data available, Defant’s (1941) analysis
of the flow at 2000 m shows a well-defined deep
western boundary current in the North and South
SECTION 4 THE GLOBAL FLOW FIELD
206
the mean ocean circulation. Thus if we consider the
vertically integrated effect, we obtain Sverdrup’s
(1947) relation between the curl of the wind stress
and the ocean meridional transport. The addition
of a western boundary current then leads to the
Stommel (1948, 1965), Munk (1950) and Munk
and Carrier (1950) models of the gyre circulation,
the Godfrey (1989) ‘Island Rule’ and the Stommel
and Arons (1960a,b) theory for the abyssal circulation of the ocean.
The next level of understanding comes from
applying the same ideas to the conservation of
potential vorticity on density surfaces. The results
emphasize the role of ventilation in determining the
vertical structure of the subtropical gyres (Price,
Chapter 5.3). They also highlight the importance
of potential vorticity mixing within the ocean.
These theories often work well for the nearsurface flows in the subtropics, but they tend to be
less useful at high latitudes and at depth because
there the effect of topography is important. Topography can both steer currents and produce isolated
Taylor columns. Elsewhere divergences and convergences associated with the bottom Ekman layer
can also have an effect.
In trying to understand the mean circulation we
also need to consider the effect of the mesoscale
eddy field, internal waves and tides. At one stage it
was thought that the mesoscale eddies were so
widespread that they might have an important
effect on the dynamics (Warren, 1981b). However,
this now appears to be true only for the regions
where eddy kinetic energy is highest (Stevens and
Ivchenko, 1997) or where non-linear interactions
with topography generate an along-slope current
(Holloway, 1992). Elsewhere the eddies are very
effective at mixing properties on density surfaces,
but otherwise their dynamical effect is small.
For a long time the background internal wave
field was thought to be important for vertical mixing in the deep ocean (Munk, 1966). This is the
field usually associated with the Garrett and Munk
(1979) spectrum that has a relatively uniform
amplitude throughout the ocean. However, both
theory and direct measurements show that the
cross-isopycnal mixing it produces is an order of
magnitude too small to explain the observed
properties of the ocean (Toole and McDougall,
Chapter 5.2). Instead there is now evidence that
topography (Polzin et al., 1997; see also Hogg,
Chapter 4.5) and tides (Egbert, 1997; Munk and
Wunsch, 1998) can produce large amounts of vertical mixing in localized regions of ocean. This is a
new field of study and the implications have still
to be fully investigated.
4.2.2 Observational evidence
As stated above, the interior flow of the ocean is
difficult to measure directly with current meters
because of the long integration time needed to
overcome the masking effect of the mesoscale eddy
field and other motions. Because of this, our
knowledge about the ocean circulation has come
primarily from hydrographic sections and tracer
studies. In recent years, and especially during the
World Ocean Circulation Experiment (WOCE),
this has been complemented by high-quality data
from deep floats (Autonomous Lagrangian Circulation Explorer (ALACE) floats) and surface
drifters. Very long current meter data records are
also becoming available.
4.2.2.1 Classical hydrography
Some of the best early descriptions of the largescale Atlantic circulation resulted from the Meteor
expedition of the 1920s. Two main approaches
were used to analyse the data, one based on the
distribution of tracers (Wüst, 1935) and the other
on the thermal wind equation (Defant, 1941). The
dynamic height method (see also Fofonoff, 1962)
works well in the top few hundred metres of
the ocean, where currents are strong. As a result
Defant’s (1941, 1961) analysis of the surface
dynamic height field of the Atlantic still compares
well with modern results. Defant also found that
down to about 900 m the amplitude of the field
slowly declined but its overall shape changed little.
Below 1000 m the method tends to fail. This is
because the method only defines current shears
and so it needs an absolute current to be defined at
some level. Water properties can be used to estimate a level of no motion (or zero reference current) but, given the low currents in the deep ocean,
it is difficult to do this to sufficient accuracy.
However, where currents are strong, or the level of
no motion estimates are good ones, the method
may still be successful. Thus, despite the limited
amount of data available, Defant’s (1941) analysis
of the flow at 2000 m shows a well-defined deep
western boundary current in the North and South
SECTION 4 THE GLOBAL FLOW FIELD
206