4.2.4 The abyssal circulation
At present, most theoretical studies of the deep circulation are based on the ideas of Stommel (1958)
and Stommel and Arons (1960a,b). Their work
incorporated a number of key ideas. The first was
that surface water downwells into the deep ocean in
a few localized regions near the poles. Second, they
assumed that upwelling occurs fairly uniformly
throughout the deep ocean as a result of the vertical
mixing due to breaking internal waves. This is an
important concept and is consistent with Munk’s
(1966) ideas on abyssal mixing, the apparent uniform nature of the internal wave field (Garrett
and Munk, 1979) and observations of the sporadic
breaking of internal waves (Woods, 1968).
Another key idea was the realization that uniform upwelling in the deep ocean acts like Ekman
suction at the sea surface and so produces similar
Sverdrup gyres. Finally, they realized that the flow
could be completed by a series of deep western
boundary currents with the three main oceans connected by the flows in the Antarctic Circumpolar
Current. The result, taken from Stommel (1958),
is shown in Fig. 4.5.1.
The pattern of deep western boundary currents
has been confirmed by measurements. There is evidence of similar potential vorticity conserving systems producing flows around the deep trenches in
the Pacific (Mitsuzawa and Holloway, 1998) and
there have been a number of publications extending the theory to more complicated basins and
topography (Pedlosky, 1996).
However although this view of the deep ocean
circulation is very appealing, observational evidence indicates that something is wrong. First, the
tracer observations referred to earlier indicate that
the flow is much more zonal than is expected from
theory. Second, Munk (1966) and Munk and
Wunsch (1998) found that a diapycnal diffusivity
of 10
94 m
2 s
91 was required in the deep ocean to
support the thermohaline circulation. However,
most observational estimates from microscale profiling (Gregg, 1989) and tracer release experiments
(Ledwell et al., 1998) give values near 10
95 m s
91
,
an order of magnitude lower.
There have been two key observations to fill the
gap. The first comes from satellite measurements of
the tides (Egbert, 1997). These show that the tides in
the deep ocean lose a significant amount of energy
where the tidal wave crosses major topographic
features such as the mid-ocean ridges. The energy
loss is almost certainly due to the generation of
internal tides. These are strongly non-linear and so
will eventually break, causing local vertical mixing.
This is confirmed by measurements of mixing
made in the South Atlantic as part of the Deep
Basin Experiment (Hogg, Chapter 4.5; Toole and
McDougall, Chapter 5.2). Polzin et al. (1997)
found values of 10
95 m
2 s
91 over much of the deep
Brazil Basin but above the rough Mid-Atlantic
Ridge the values were much larger, reaching
10
93 m
2 s
91 in the bottom-most 150 m (see
Fig. 5.2.5, Plate 5.2.5, p. 428). The values near
10
95 m
2 s
91 were consistent with mixing due to
the background isotropic internal wave field. In
contrast, the high values were associated with
upward-propagating internal waves generated
by the interaction of tide and other currents with
the rough bottom topography (J. Toole, personal
communication).
Such observations imply that the upwelling in
the deep ocean is not uniform and so the Stommel
and Arons’ hypothesis about the abyssal circulation needs revision. In fact, where we do have evidence for mean flows in the deep ocean away from
the western boundary currents, the flows appear
to be mainly zonal. Examples are the Pacific tracer
measurements discussed earlier and the float observations in the South Atlantic Deep Basin Experiment (see Hogg, Chapter 4.5). In the Pacific,
Talley and Johnson (1994) demonstrated that the
zonal flows are not associated with hydrothermally driven abyssal circulation. Here and in the
South Atlantic the zonal flows may be due to the
enhanced vertical mixing near topography and
the resulting upwelling.
If the upwelling is non-uniform, then on the basis
of the theory discussed earlier, one would expect
meridional or depth contour crossing flows in the
regions where the vertical velocity is divergent and
more zonal or contour-following flows where mixing is weak. Such behaviour is seen in numerical
model studies of rectangular basins with enhanced
mixing near the side wall boundaries (Marotzke,
1997; Samelson, 1998). It is also seen in a global
ocean model (Hasumi and Suginohara, 1999a) that
tries to represent realistically the variations in mixing due to the interaction of tides and topography.
Although this idea of inhomogeneous vertical
mixing in the deep ocean helps to explain the
4.2 The Interior Circulation of the Ocean
211
Webb and Suginohara
At present, most theoretical studies of the deep circulation are based on the ideas of Stommel (1958)
and Stommel and Arons (1960a,b). Their work
incorporated a number of key ideas. The first was
that surface water downwells into the deep ocean in
a few localized regions near the poles. Second, they
assumed that upwelling occurs fairly uniformly
throughout the deep ocean as a result of the vertical
mixing due to breaking internal waves. This is an
important concept and is consistent with Munk’s
(1966) ideas on abyssal mixing, the apparent uniform nature of the internal wave field (Garrett
and Munk, 1979) and observations of the sporadic
breaking of internal waves (Woods, 1968).
Another key idea was the realization that uniform upwelling in the deep ocean acts like Ekman
suction at the sea surface and so produces similar
Sverdrup gyres. Finally, they realized that the flow
could be completed by a series of deep western
boundary currents with the three main oceans connected by the flows in the Antarctic Circumpolar
Current. The result, taken from Stommel (1958),
is shown in Fig. 4.5.1.
The pattern of deep western boundary currents
has been confirmed by measurements. There is evidence of similar potential vorticity conserving systems producing flows around the deep trenches in
the Pacific (Mitsuzawa and Holloway, 1998) and
there have been a number of publications extending the theory to more complicated basins and
topography (Pedlosky, 1996).
However although this view of the deep ocean
circulation is very appealing, observational evidence indicates that something is wrong. First, the
tracer observations referred to earlier indicate that
the flow is much more zonal than is expected from
theory. Second, Munk (1966) and Munk and
Wunsch (1998) found that a diapycnal diffusivity
of 10
94 m
2 s
91 was required in the deep ocean to
support the thermohaline circulation. However,
most observational estimates from microscale profiling (Gregg, 1989) and tracer release experiments
(Ledwell et al., 1998) give values near 10
95 m s
91
,
an order of magnitude lower.
There have been two key observations to fill the
gap. The first comes from satellite measurements of
the tides (Egbert, 1997). These show that the tides in
the deep ocean lose a significant amount of energy
where the tidal wave crosses major topographic
features such as the mid-ocean ridges. The energy
loss is almost certainly due to the generation of
internal tides. These are strongly non-linear and so
will eventually break, causing local vertical mixing.
This is confirmed by measurements of mixing
made in the South Atlantic as part of the Deep
Basin Experiment (Hogg, Chapter 4.5; Toole and
McDougall, Chapter 5.2). Polzin et al. (1997)
found values of 10
95 m
2 s
91 over much of the deep
Brazil Basin but above the rough Mid-Atlantic
Ridge the values were much larger, reaching
10
93 m
2 s
91 in the bottom-most 150 m (see
Fig. 5.2.5, Plate 5.2.5, p. 428). The values near
10
95 m
2 s
91 were consistent with mixing due to
the background isotropic internal wave field. In
contrast, the high values were associated with
upward-propagating internal waves generated
by the interaction of tide and other currents with
the rough bottom topography (J. Toole, personal
communication).
Such observations imply that the upwelling in
the deep ocean is not uniform and so the Stommel
and Arons’ hypothesis about the abyssal circulation needs revision. In fact, where we do have evidence for mean flows in the deep ocean away from
the western boundary currents, the flows appear
to be mainly zonal. Examples are the Pacific tracer
measurements discussed earlier and the float observations in the South Atlantic Deep Basin Experiment (see Hogg, Chapter 4.5). In the Pacific,
Talley and Johnson (1994) demonstrated that the
zonal flows are not associated with hydrothermally driven abyssal circulation. Here and in the
South Atlantic the zonal flows may be due to the
enhanced vertical mixing near topography and
the resulting upwelling.
If the upwelling is non-uniform, then on the basis
of the theory discussed earlier, one would expect
meridional or depth contour crossing flows in the
regions where the vertical velocity is divergent and
more zonal or contour-following flows where mixing is weak. Such behaviour is seen in numerical
model studies of rectangular basins with enhanced
mixing near the side wall boundaries (Marotzke,
1997; Samelson, 1998). It is also seen in a global
ocean model (Hasumi and Suginohara, 1999a) that
tries to represent realistically the variations in mixing due to the interaction of tides and topography.
Although this idea of inhomogeneous vertical
mixing in the deep ocean helps to explain the
4.2 The Interior Circulation of the Ocean
211
Webb and Suginohara