As has often been remarked in the context of global climate, a mere 2.5 m depth of water has the same heat capacity as the whole depth of the atmosphere. The absorption, storage and release of this heat has a crucial long-term effect on climate. In his review of air-sea interactions, Donelan [1990] concluded that the widest gap in our knowledge of air-sea interactions lies in the aqueous boundary layer and its relation to surface waves. How well do we understand the nature of turbulence near the surface? We frequently look for guidance to the atmospheric boundary layer. But most studies of the atmospheric boundary layer have been made over solid boundaries, while the ocean lies beneath a free surface. So it is not, a priori, clear that the two should be similar.
But there are distinct similarities. In the convecting
oceanic boundary layer we find a superadiabatic surface layer [ Anis
and Moum, 1992] overlying a well-mixed layer, like the atmospheric
boundary layer during convection. Dissipation measurements in the
well-mixed layer do scale with the surface buoyancy flux, as they
do in the convecting atmospheric boundary layer [ Shay and Gregg,
1986; Imberger, 1985]. Within the surface layer,
dissipation measurements sometimes show agreement with atmospheric
scaling laws, in which the flow over the solid boundary results in
a quasi-constant-stress layer, with corresponding logarithmic
velocity profile and 
, where z is
the depth [ Dillon et al., 1981; Imberger, 1985;
Soloviev et al., 1988, 1989; Lombardo and Gregg,
1989].
However, there is an increasing body of observations near the
surface of both lakes and oceans that suggest that mixing is much
more energetic than predicted by constant-stress scaling.
Kitaigorodskii et al. [1983] measured values of 
beneath wind waves in Lake Ontario that were two orders of magnitude
greater than 
. Moreover, in their measurements the
depth dependence of 
was inconsistent with pure
shear-produced turbulence, and turbulence velocities depended strongly
on wave energy. Intense near-surface dissipation rates in a
strongly convecting mixed layer observed by Gregg [1987] were
also several orders of magnitude greater than 
. These
large dissipation rates extended to depths of at least 30 m. It
was suggested that the large values of 
could be due
to convective plumes, surface breaking waves, Langmuir circulations or
some combination of all three. Gargett [1989] presented
vertical profiles of 
near the surface which showed a
depth dependence like a constant-stress layer on calm days
but closer to 
on windy days. In submarine measurements,
Osborn et al. [1992] also found values of 
much greater
than 
near the sea surface, accompanied
by acoustically-detected bubble clouds presumed to be produced by
breaking surface waves. Agrawal et al. [1992] obtained time series
at the sea surface and found 
to be highly intermittent,
with background values near constant-stress layer predictions but with
a substantial fraction of values exceeding 
by at least
an order of magnitude, so that mean values at a particular depth were
much greater than 
. Anis and Moum [1995]
found examples of both cases in open ocean conditions and suggested
two possible mechanisms for the generation of turbulence by surface waves
in the presence of swell. In the first, turbulence generated by surface
wave breaking is carried downward by the swell, while diffusing
and decaying, causing a net downward transport of turbulence kinetic
energy, which then dissipates. In the second, in a slightly rotational
wave field, energy drawn from the wave field to the mean flow, via
wave stresses, may in turn be drawn from the mean flow by
turbulence production, and then dissipated.
So our current picture is that some elements of the oceanic surface boundary layer are similar to those in the atmospheric layer above it, but that waves may force turbulence in the ocean that is much more energetic than in the atmosphere.