Φ = 101.3. This increase with Φ also mirrors the
do is estimate the orientation of roughness ele-
directional sensitivity we reported in AC95,
ments by tracking the history of surface winds that
have been inferred from analyzed geostrophic
though the predicted maximum is about 7%
winds. According to AC95, if the surface wind
higher than our measured maximum for CDN10,
blows at 8 m/s or higher with constant direction, it
will build sastrugi parallel to it and, thus, stream-
The upshot of this modeling is that the small,
line the surface. If such conditions persist for 12
10-cm-high roughness elements associated with
hours, CDN10 will fall from its initial value to about
drifting snow--rather than pressure ridges or
1.5 103 and will remain here as long as the wind
other mesoscale features--determine the local
remains aligned with the sastrugi.
drag coefficient over snow-covered sea ice.
If subsequent surface winds fall below 6 m/s,
Though some have stated this conclusion explic-
Figure 4 shows how CDN10 will vary as the wind
itly (e.g., Joffre 1982) or implicitly (e.g., Banke et
turns with respect to the axis of the sastrugi.
al. 1976, 1980), my model is the first to provide a
The difficult part of the parameterization will
theoretical foundation for it.
be treating directionally variable winds of 68 m/s,
These model predictions and the observations
or higher. These are strong enough to begin re-
in AC95 suggest how we should attempt to para-
working the surface--eroding the snowdrifts
meterize CDN10 in terms of what we can measure.
Because CDN10 depends crucially on how the
present--but, because of their variability, will not
mean wind is oriented with respect to the sas-
necessarily build new drifts and thereby stream-
trugi, the key is predicting sastrugi orientation.
line the surface in another direction. Neither the
Our observations showed that this orientation
data in AC95 nor the model developed here pro-
depends on the history of the surface wind. By
vides a clear suggestion on how to evaluate CDN10
for such winds.
inferring surface winds from analyzed fields
In summary, except for the complication just
of the geostrophic wind, we should be able to de-
mentioned, it seems possible to now estimate
duce this orientation. The event to look for is a
CDN10 over snow-covered sea ice by tracking the
directionally constant surface wind of at least
history of the surface wind computed from ana-
8 m/s that persists for 12 hours. At the end of
such an event, CDN10 will be roughly 1.5 103.
lyzed pressure fields. I have described the algo-
rithm in words; coding and testing it is the next
Figure 4 then shows how to estimate CDN10 when
the wind direction subsequently changes.
I have adapted R92's model for the form drag
Allen, J.R.L. (1965) Scour marks in snow. Journal of
on a field of interacting roughness elements and
Sedimentary Petrology, 35: 331338.
applied it to rudimentary sastrugi-like snow-
Andreas, E.L and K.J. Claffey (1995) Airice drag
drifts. The key step was treating the complex ge-
coefficients in the western Weddell Sea: 1. Values
ometry of the sastrugi. In the absence of laborato-
deduced from profile measurements. Journal of
ry values for the form drag on the various faces of
Geophysical Research, 100: 48214831.
the sastrugi, I assumed that the form drag reflects
Andreas, E.L, M.A. Lange, S.F. Ackley and P.
additive contributions from the various faces of
Wadhams (1993) Roughness of Weddell Sea ice
the individual sastrugi (see eq 13).
and estimates of the airice drag coefficient. Jour-
The resulting model does quite well in repro-
nal of Geophysical Research, 98: 12,43912,452.
ducing the values of CDN10 that we observed in
Arie, M. and H. Rouse (1956) Experiments on two-
AC95. For sastrugi of height 10 cm and with an
dimensional flow over a normal wall. Journal of
areal coverage of 15%, the model predicts that
Fluid Mechanics, 1: 129141.
CDN10 = 1.43 103 when Φ is less than 12. In
Arya, S.P.S. (1973) Contribution of form drag on
AC95, the nominal head-on value of CDN10 that
pressure ridges to the air stress on Arctic ice. Jour-
we observed was 1.5 103. In the model, as Φ
nal of Geophysical Research, 78: 70927099.
gets larger than 12, CDN10 increases rapidly. In
Arya, S.P.S. (1975) A drag partition theory for de-
this example, CDN10 increases by 17% as Φ turns
termining the large-scale roughness parameter
from 12 to 20. As Φ continues increasing, CDN10
and wind stress on the Arctic pack ice. Journal of
eventually reaches a maximum of 2.73 103 at
Geophysical Research, 80: 34473454.