Mc. Gregg et al., STATISTICS OF SHEAR AND TURBULENT DISSIPATION PROFILES IN RANDOM INTERNAL WAVE-FIELDS, Journal of physical oceanography, 23(8), 1993, pp. 1777-1799
Because breaking internal waves produces most of the turbulence in the
thermocline, the statistics of epsilon, the rate of turbulent dissipa
tion, cannot be understood apart from the statistics of internal wave
shear. The statistics of epsilon and shear are compared for two sets o
f profiles from the northeast Pacific. One set, PATCHEX, has internal
wave shear close to the Garrett and Munk model, but the other set, PAT
CHEX north. has average 10-m shear squared, [S-10(2)] about four times
larger than the model. The 10-m shear components, S(x) and S(y), were
measured between 1 and 9 MPa and referenced to a common stratificatio
n by WKB scaling. The scaled components, S(x) and S(y), are found to b
e independent and normally distributed with zero means, as assumed by
Garrett and Munk. This readily leads to analytic forms for the probabi
lity densities of S-10(2) and S-10(4). The observed probability densit
ies of S-10(2) and S10(4) are close to the predicted forms, and both a
re strongly skewed. Moreover, sigma(lnS10(2)) and sigma(lnS10(4)) are
constants, independent of the standard deviations of S(x) and S(y). Th
e probability density of the inverse Richardson number, Ri10(-1) = S-1
0(2)/[N2], is a scaled version of the probability density of S-10(2).
The PATCHEX distribution cuts off near Ri10(-1) = 4, as found by Eriks
en, but the PATCHEX north distribution extends to higher values, as pr
edicted analytically. Consequently, a cutoff at Ri10(-1) = 4 is not a
universal constraint. Over depths where [N2] is nearly uniform, the pr
obability density of 0.5-m epsilon can be approximated, to varying deg
rees of accuracy, as the sum of a noise variate with an empirically de
termined distribution and a lognormally distributed variate whose para
meters can be estimated using a minimum chi-square fitting procedure.
The 0.5-m epsilon, however, are far from being uncorrelated, a circums
tance not considered by Baker and Gibson in their analysis of microstr
ucture statistics. Obtaining approximately uncorrelated samples requir
es averaging over 10 m for PATCHEX and 15 m for PATCHEX north. These l
engths correspond approximately to reciprocals of the wavenumbers at w
hich the respective shear spectra roll off. After correcting the uncor
related epsilon samples for noise and then scaling to remove the depen
dence on stratification, the scaled dissipation rates, epsilon = epsil
on(N-0(2)/[N2]), are lognormally distributed. (Without noise correctio
n and [N2] scaling the data are not lognormal; e.g., noise correction
and scaling with [N1] and [N3/2] do not produce lognormality.) It is h
ypothesized that the approximate lognormality of bulk ensembles of eps
ilon results from generation of turbulence in proportion to S-10(4). L
ognormality is well established for isotropic homogeneous turbulence (
Gurvich and Yaglom), and Yamazaki and Lueck show that it also occurs w
ithin individual turbulent patches. Bulk ensembles from the thermoclin
e, however, include samples from many sections lacking turbulence as w
ell as from a wide range of uncorrelated turbulent events at different
evolutionary stages. Consequently, the bulk data do not meet the crit
eria used to demonstrate lognormality under more restricted conditions
. If the authors are correct, the high-amplitude portion of [N2]-scale
d bulk ensembles is lognormal or nearly so owing to generation of the
turbulence by a highly skewed shear moment. As another consequence, si
gma(lnS10(4)) = 2.57 should be an upper bound for 10 m sigma(lnepsilon
) when the turbulence is produced by the breaking of random internal w
aves. Because many parts of the profile lack turbulence, sensor noise
limits the epsilon distribution to smaller spreads than those of S-10(
4). In practice we observe sigma(lnepsilon) almost-equal-to 1.2 when S
10(2) equals GM76, and sigma(lnepsilon) almost-equal-to 1.5 when S-10(
2) is about three times GM76. For the larger spread, 95% confidence li
mits require n almost-equal-to 60 for accuracies of +/- 100%, n almost
-equal-to 140 for +/-50%, and n almost-equal-to 2000 for +/-10%. Owing
to instrumental uncertainties in epsilon estimates, the authors sugge
st accepting less restrictive confidence limits at one site and sampli
ng at multiple sites to estimate average dissipation rates in the ther
mocline.