How is the bulk of the ocean mixed? The process of turbulence in stratified fluids thought to be responsible for ocean mixing is a highly random and intermittent process that is itself poorly understood. But it must be somehow represented in larger-scale models of ocean circulation. Because these models cannot possibly resolve turbulence scales (millimeters to centimeters), mixing is represented on much larger scales (at best 10 km horizontally, 100 m vertically) as subgrid scale (SGS) processes [ Garrett, 1994]. The methods of simulating SGS processes and the magnitude of diffusion is critical to the models' conclusions [ Gargett, 1984, 1989; Bryan, 1987]. If a model provides a reasonable representation of the observed circulation, it is usually considered that SGS processes have been well represented. But this agreement may be coincidental. If the models are to be used for prediction, the physics of the SGS processes must be understood well enough for us to know that they are well-represented. This is crucial to understanding global circulation, climate change, pollutant dispersal and primary productivity in the world's oceans. Certainly in order to predict the behavior of an ocean different from today's, we need to understand the physics.
From global considerations, Munk [1966] made an estimate of
the diffusivity, K, in the deep (1--4 km) ocean of 10
m
s
that has provided a standard for turbulence estimates of
K. We thought for many years that the mechanism for this diffusivity
was turbulent mixing across isopycnals.
Turbulence measurements below the surface layer, even in the upper
1 km, have been sparse relative to the great expanse of the ocean and
its many scales of variability. However, they have consistently
yielded estimates of K ten times smaller than Munk's [ Moum
and Osborn, 1986; Gregg, 1987]. Results from the North
Atlantic Tracer Release Experiment (NATRE) support these lower
estimates [ Ledwell et al., 1993, 1994]. A dye tracer released at
a depth of 300 m spread across density surfaces at a rate equivalent to
a K of 
m
for the
period May--October, 1992 and a K of 
m
for October 1992--May 1993. Estimates of K
from microscale and finescale measurements made during the dye
spreading yielded remarkably similar values [ Oakey et al., 1994;
Schmitt et al., 1994; Duda and Jacobs, 1994; Sherman
and Davis, 1994]. Mixing in the top few hundreds of meters (maybe even
the top 1 km or so) may well take place by stirring along sloping
density surfaces rather than by turbulent mixing across them [
Garrett, 1993]. At depths shallower than about 1 km, isopycnals outcrop
at the sea surface in the subtropics, providing a source for mixed fluid
in the interior.
As important as the NATRE result is in providing an integral determination of the rate at which turbulent mixing diffuses the thermocline (at least, at that time and place), it does not reveal the physics of the process. Each estimate of K results from a straight line fit to two data points. The path taken between the end points is unknown. Ultimately, it is the path between the end points that we need to understand if we are ever to properly represent SGS processes.
Below 1 km there have been even fewer measurements. Only three
reports exist of ocean microstructure measurements below 1.5 km depth.
Two short stretches of data from 2000 m depth at 28
N,
155
W and 3500 m depth at 19
S, 150
W indicated
that K was 
m
or less [ Gregg, 1977].
In contrast, Moum and Osborn [1986] reported an increase of K
with depth to 2200 m in the vicinity of the Kuroshio Extension
approaching Munk's value of 
m
in the abyss.
Toole et al. [1994] report new estimates of K from experiments in
the NE Pacific and the NE Atlantic. They found a depth-independent K
of 
m
in the ocean interior where the
internal-wave level was normal. However, they found enhanced mixing
in regions of elevated internal-wave energy, particularly
near steeply-sloping boundaries. There, K exceeded 
m
, increasing towards the bottom. These results suggest
that we look for deep mixing near strong, deep western boundary currents
or near sloping boundaries.