free model parameters, R92's model can account
drag can, at times, be an important means for
for the behavior of CDN10 at ISW that we docu-
transferring momentum to a surface. For exam-
ple, Lettau (1969) offered a heuristic formula in-
mented in AC95. In particular, the three main ob-
volving obstacle geometry and silhouette area to
servations that I hope to explain are as follows:
1. CDN10 is near 1.5 103 when the wind is
parameterize the roughness lengths that Kutz-
bach (1961) measured during his famous bushel
aligned with a well-developed field of drifted
basket experiment on Lake Mendota. Arya (1973,
snow.
2. CDN10 is near 2.5 103 when the mean wind
1975) assumed that pressure ridges were the
direction makes a large angle (say 50140) with
roughness elements responsible for most of the
form drag over Arctic sea ice and developed a
the dominant axis of the drifts.
model to partition the total stress into skin fric-
3. CDN10 can increase by roughly 20% if, after
tion and form drag on the basis of ridging inten-
being aligned with a developed (or developing)
sity. Banke et al. (1976, 1980) investigated the
field of drifted snow, the wind turns by as little as
20.
pressure distribution around Arctic pressure
ridges 13 m high in an attempt to determine the
form drag coefficient for these. But their ultimate
RAUPACH'S MODEL ADAPTED TO
parameterization for CDN10 (also Banke and
DRIFTED SNOW
Smith 1973) is more in line with our findings;
they showed that
Geometry
Because during our deployment on ISW the
103 CDN10 = 1.10 + 0.072 ξ
(1)
wind was directionally quite variable, the sur-
face never developed the deep, elaborate fields
where ξ (in centimeters) is the root-mean-square
of sastrugi that Mellor (1965) and Jackson and
surface roughness for roughness elements with
Carroll (1978) observed on the Antarctic conti-
wavelengths shorter than 13 m. That is, the
nent. Rather, on ISW, the drifts were sparsely dis-
smaller scale roughness dominates the momen-
tributed and rudimentary. When the wind
tum exchange. Jackson and Carroll (1978)
reached 68 m/s, linear sastrugi-like drifts
reached the same conclusion. They showed that
would begin forming behind any protruding
the angle between the mean wind and the domi-
obstacles. The drifts would grow as the wind
nant axis of the sastrugi at the South Pole deter-
persisted; but when the wind direction changed,
mined the roughness length; they then devel-
these drifts would erode in winds above the 6- to
oped a model based on Lettau's to explain this
8-m/s threshold. Thus, the drifts that deter-
effect.
mined the surface roughness on ISW were gener-
More recently, Raupach (1992; henceforth,
ally small. The scour marks that Allen (1965) de-
R92; also Raupach et al. 1993) developed a more
scribed and named scour-remnant ridges may
sophisticated model, based on dimensional anal-
well be precursors to these rudimentary sastrugi.
ysis and two physically reasonable hypotheses,
Figure 2 shows a typical leveling profile of the
to partition the total surface stress into contribu-
snow and sea ice surfaces in the vicinity of the
tions from form drag and from the stress on the
profiling mast described in AC95. We made sev-
underlying surface. That model is the basis of the
en such leveling surveys during our deployment
analysis here. I adapt Raupach's model to evalu-
on ISW. The profiles show that snow tends to col-
ate the stress partitioning for a field of sastrugi-
lect around obstacles in the underlying ice. The
like snowdrifts. That model predicts the drag co-
root-mean-square (rms) roughness of the snow
efficient evaluated at the top of the roughness el-
surface in this particular profile is 11 cm; its ice-
ements; I show how this quantity is related to
surface roughness is 9 cm. In the seven surface
CDN10, which, in turn, is monotonically related to
profiles that we collected, the rms snow-surface
the roughness length z0 through
roughness ranged from 9 to 26 cm; the ice-sur-
face roughness ranged from 9 to 28 cm. A value
CDN10 = k2 {ln[(10 d)/z0]}2 .
(2)
in this range is a reasonable choice for the nomi-
nal height of the snow-surface roughness ele-
Here, k (= 0.4) is the von Krmn constant, and z0
ments.
and the displacement height d must be in meters.
Figure 3 shows the idealized geometry of the
My objective is to see whether, with reason-
roughness elements that I will treat with R92's
able choices for snowdrift geometry and other
model. These are rudimentary sastrugi with
2
Previous Page
