30 ≤ Φ ≤ 160, the differences were only 23%.

range, 1.361.51 103.

Consequently, the choice of *C*S10 does not have a

Figure 4b investigates the sensitivity of the

big effect on the results, as long as the value is a

model to γ when *h *= 10 cm. For this value of *h*,

reasonable one. Secondly, my handling of the dis-

γ = 0.10 yields *C*DN10 values for head-on flow of

placement height in converting from *C*S10 to *C*Sh is

1.36 103. Although we did measure four *C*DN10

appropriate, given this weak sensitivity to *C*S10.

values in this range, the more likely lower limit is

Figure 5 shows some examples of how the

about 1.5 103. Also, as Φ increases, the predict-

stress is partitioned between form drag and skin

ed *C*DN10 values reach a maximum that is 6% be-

low our observed maximum. When γ is 0.20 and

figure, the (γ = 0.10, *h *= 10 cm) and the (γ = 0.20,

0.25, the predicted head-on values for *C*DN10 are

roughly correct, 1.501.55 103. But for these γ

and maximum values of τR/τ that I obtained for

values, the maximum predicted values are, re-

the range of model parameters considered. My

spectively, 17 and 26% above our observed maxi-

computations suggest that, even for a very rough

mum. Notice, γ = 0.25 means that the sastrugi

surface, γ = 0.20 and *h *= 20 cm, skin friction pro-

cover half the area that they would if they were

duces more than half of the total stress when the

packed as tightly as possible. This value is, there-

flow is approximately head-on. In other words,

fore, probably an unrealistically large sastrugi

the streamlined shape of the sastrugi is quite ef-

coverage for ISW.

fective at reducing form drag. When the flow is at

For the γ = 0.15 and *h *= 10 cm case--the one

right angles to the sastrugi, however, form drag

that fits the data best--I also ran the model with

dominates the total stress because the wind has

so much more surface to push against. This dom-

est value compatible with the surveys I cited ear-

inance of form drag for such wind orientations

lier. Resulting predictions for *C*DN10 were 46%

explains why the drifts erode so quickly (AC95).

lower for 0 ≤ Φ ≤ 20 and for 160 ≤ Φ ≤ 180 than

Equation 46 yields the displacement height;

for calculations based on *C*S10 = 1.10 103. For

Figure 6 shows several computations of *d*/*h*. For

1.0

0.8

0.6

(γ, h) = (0.20, 20)

τR

(0.15, 10)

τ

(0.10, 10)

0.4

0.2

0

0

20

40

60

80

100

120

140

160

180

Wind Direction, Φ ()

0.5

(γ, h) = (0.20, 20)

0.4

(0.20, 10)

(0.15, 20)

(0.15, 10)

0.3

(0.15, 5)

d

h

0.2

0.1

0

0

20

40

60

80

100

120

140

160

180

Wind Direction, Φ ()

9