enhanced mixing through much of the water column, the surmised bottom-generated internal
waves must also carry momentum upwards and
thus re-distribute the bottom stress up into the
water column. If so, an Ekman-type balance might
support net meridional flow above the height of
the bathymetric sills of the zonal flow in the
Southern Ocean.
The observed bottom-intensified profiles of K
over rough topography have an interesting consequence for the diapycnal velocity. Neglecting
(demonstratively small) consequences of the nonlinear equation of state and curvature effects, the
steady-state density equation in a coordinate system aligned with the isopycnals is w* z :(K z ) z ϳ
(⌫ ) z (St Laurent et al., 2000). In the Brazil
Basin above the Mid-Atlantic Ridge flank, is
observed to increase with depth below about
2500 m, implying that w*:0. Moreover, w* is
increasingly negative with depth, indicating water
column stretching. Vorticity considerations in turn
imply poleward flow at these levels; possibly consistent with the southward displacement experienced by the centre of mass of the tracer injected
at this site (Ledwell et al., 2000; St Laurent et al.,
2000). But mass conservation for the bottom
waters, in light of the flow entering the Brazil
Basin through the Vema Channel, requires net
upwelling across deep isopycnals (a positive average w*). Mass balance may be achieved by strong
‘upward’-directed flow across isopycnals within
the many fracture zones of the ridge driven by
mixing immediately above the (insulating) bottom.
Before departing the deep ocean, it is important
to call attention to other potentially significant
mixing processes: those associated with enhanced
abyssal currents. Under some conditions, Deep
and Bottom waters may be locally accelerated,
the resulting strong shears supporting enhanced
mixing. For example, significant turbulent dissipation was observed near the bottom at the
Blake Bahama Outer Rise where the deep western
boundary current is enhanced by local recirculation and topographic effects (Stahr and Sanford,
1999). Flow acceleration is also observed in narrow ocean passages, particularly those that include
a vertical sill that can result in downstream flow
acceleration by gravity. Such flows may internally
produce low Richardson number and develop
shear instability. Internal lee waves and hydraulic
jumps might also play a mixing role within such
currents. An example here is provided by the
flow through the Romanche Fracture Zone in
the equatorial Atlantic (Polzin et al., 1996;
Ferron et al., 1998). Mixing in the Romanche Fracture Zone warms the coldest waters passing over
this Mid-Atlantic Ridge sill from about 0.9°C to
nearly 1.5°C. Thus turbulence acting at centimetre
scale appears as important as the depths of the sills
in the Romance Fracture Zone in defining the bottom water properties of the Eastern Atlantic
basins.
5.2.6.3 Abyssal summary
While deep-ocean dissipation estimates are too few
to attempt a quantitative global estimate of abyssal
mixing intensity and diapycnal transport, recent
results motivate speculation. The Brazil Basin study
re-emphasizes that not all baroclinic tides and
internal lee waves link directly to enhanced turbulent dissipation and mixing. The connection is most
direct for waves at high vertical wavenumber
(9ϳ0.01 cycles m
91
). Although energy at these
fine scales can come from larger motions via
wave–wave interactions, such small-vertical-scale
waves are directly generated by flow over bottom
irregularities with 1–5 km horizontal scale. Existence of this near-direct energy source suggests
that it is small-scale bottom roughness, as opposed
to broad slopes or step-change-like bathymetric
structures, that is key to mixing in the abyss.
Unfortunately, bathymetric information at these
small spatial scales is today limited to localized
regions surveyed with specialized acoustic sounders.
If, however, the bottom roughness is self-similar
(areas of 1 km scale roughness correspond to areas
that are rough at 10 km scales), the global bathymetric map inferred by Smith and Sandwell (1994)
provides some guidance. The map suggests that
perhaps half of the seafloor can be characterized
as rough, which implies that significant diapycnal
transport is widespread at depth. But these rough
areas are distributed in a complex fashion around
the globe; the field of diapycnal velocity that drives the abyssal circulation is thus likely to be
equally complicated. Although their theory does
not require uniform upwelling, the Stommel and
Arons (1960b) schematics, which for clarity were
based on this simplification, are impressed on our
subconscious. It is time to revisit the venerable
Stommel–Arons abyssal flow patterns with an eye
towards developing new paradigms. In regions of
5.2 Mixing and Stirring in the Ocean Interior
353
Toole and McDougall
waves must also carry momentum upwards and
thus re-distribute the bottom stress up into the
water column. If so, an Ekman-type balance might
support net meridional flow above the height of
the bathymetric sills of the zonal flow in the
Southern Ocean.
The observed bottom-intensified profiles of K
over rough topography have an interesting consequence for the diapycnal velocity. Neglecting
(demonstratively small) consequences of the nonlinear equation of state and curvature effects, the
steady-state density equation in a coordinate system aligned with the isopycnals is w* z :(K z ) z ϳ
(⌫ ) z (St Laurent et al., 2000). In the Brazil
Basin above the Mid-Atlantic Ridge flank, is
observed to increase with depth below about
2500 m, implying that w*:0. Moreover, w* is
increasingly negative with depth, indicating water
column stretching. Vorticity considerations in turn
imply poleward flow at these levels; possibly consistent with the southward displacement experienced by the centre of mass of the tracer injected
at this site (Ledwell et al., 2000; St Laurent et al.,
2000). But mass conservation for the bottom
waters, in light of the flow entering the Brazil
Basin through the Vema Channel, requires net
upwelling across deep isopycnals (a positive average w*). Mass balance may be achieved by strong
‘upward’-directed flow across isopycnals within
the many fracture zones of the ridge driven by
mixing immediately above the (insulating) bottom.
Before departing the deep ocean, it is important
to call attention to other potentially significant
mixing processes: those associated with enhanced
abyssal currents. Under some conditions, Deep
and Bottom waters may be locally accelerated,
the resulting strong shears supporting enhanced
mixing. For example, significant turbulent dissipation was observed near the bottom at the
Blake Bahama Outer Rise where the deep western
boundary current is enhanced by local recirculation and topographic effects (Stahr and Sanford,
1999). Flow acceleration is also observed in narrow ocean passages, particularly those that include
a vertical sill that can result in downstream flow
acceleration by gravity. Such flows may internally
produce low Richardson number and develop
shear instability. Internal lee waves and hydraulic
jumps might also play a mixing role within such
currents. An example here is provided by the
flow through the Romanche Fracture Zone in
the equatorial Atlantic (Polzin et al., 1996;
Ferron et al., 1998). Mixing in the Romanche Fracture Zone warms the coldest waters passing over
this Mid-Atlantic Ridge sill from about 0.9°C to
nearly 1.5°C. Thus turbulence acting at centimetre
scale appears as important as the depths of the sills
in the Romance Fracture Zone in defining the bottom water properties of the Eastern Atlantic
basins.
5.2.6.3 Abyssal summary
While deep-ocean dissipation estimates are too few
to attempt a quantitative global estimate of abyssal
mixing intensity and diapycnal transport, recent
results motivate speculation. The Brazil Basin study
re-emphasizes that not all baroclinic tides and
internal lee waves link directly to enhanced turbulent dissipation and mixing. The connection is most
direct for waves at high vertical wavenumber
(9ϳ0.01 cycles m
91
). Although energy at these
fine scales can come from larger motions via
wave–wave interactions, such small-vertical-scale
waves are directly generated by flow over bottom
irregularities with 1–5 km horizontal scale. Existence of this near-direct energy source suggests
that it is small-scale bottom roughness, as opposed
to broad slopes or step-change-like bathymetric
structures, that is key to mixing in the abyss.
Unfortunately, bathymetric information at these
small spatial scales is today limited to localized
regions surveyed with specialized acoustic sounders.
If, however, the bottom roughness is self-similar
(areas of 1 km scale roughness correspond to areas
that are rough at 10 km scales), the global bathymetric map inferred by Smith and Sandwell (1994)
provides some guidance. The map suggests that
perhaps half of the seafloor can be characterized
as rough, which implies that significant diapycnal
transport is widespread at depth. But these rough
areas are distributed in a complex fashion around
the globe; the field of diapycnal velocity that drives the abyssal circulation is thus likely to be
equally complicated. Although their theory does
not require uniform upwelling, the Stommel and
Arons (1960b) schematics, which for clarity were
based on this simplification, are impressed on our
subconscious. It is time to revisit the venerable
Stommel–Arons abyssal flow patterns with an eye
towards developing new paradigms. In regions of
5.2 Mixing and Stirring in the Ocean Interior
353
Toole and McDougall