Fig 1. Sand entering leeside eddy
Wind blowing over a dune can create an eddy current ('recirculation cell') in the lee of the dune (see Lee-side eddy currents). Sand blown over the brink accumulates in deposits on the upper slipface, forming a low mound or cornice, until eventually the sand avalanches, and the process repeats.
Fig 2. Leeside eddy in gale-force wind
The video clip to the right shows sand blowing over the brink of a dune approximately 3m high, carried by wind estimated to be moving at about 50 to 70 km/h. Some sand moves upward after crossing the brink, presumably lifted by a leeside eddy, as illustrated in figure 1. (Click on the 'play' icon to begin the video.)
The frame of this video clip includes most of the dune height. Avalanches typically started roughly a meter below the brink, implying that most grains were depositing above that -- though some grains were falling out of suspension even meters beyond the toe of the slipface (the rain of grains striking the plastic bag protecting the camera is audible in the video clip; note also the rustling of the plastic bag, reflecting the sharp but low-amplitude gustiness of air in the wind shadow -- much different from the steady roar just a meter or two higher).
The black material on the slipface in the video is easily-moved light organic debris (mostly small seed husks). It rests largely undisturbed on the surface despite the gale-force wind crossing the brink.
After a windy episode, often there are cornice deposits remaining on the upper slopes of slipfaces, sometimes cut by an avalache. Figure 3 shows the profile of one such deposit on a 7m-high dune, determined by measuring the depth of the trough relative to the undisturbed shoulders of an avalanche trough that cut through it (the floor of the trough is presumably at a constant angle, the angle of repose):
Fig 3. Depth of a cornice deposit, downslope of brink
The deposit measured above (fig 3) presumably formed as the wind was losing strength and thus probably differs from the shape of deposits made during strong wind. Even so, it extends much further downslope than the deposits in the photo below (fig 4) that were formed by wind barely able to move sand. The length of the slipface from brink to toe of the left-most avalanche is ~115cm.
Fig 4. Cornice deposits formed by low-speed wind
The shape of a cornice deposit is likely influenced by:
A cornice deposit would presumably spread further downslope in the following conditions:
A slipface can form only if a cornice can obtain a slope at an angle greater than the angle of yield (avalanching). This can only occur if the dune has enough length on the leeward side for the failure point of the deposit to be on the slope. The shortest cornice would probably occur with minimum wind able to move sand, low-density easily-moved sand, angular sand, and steady wind.
Fig 5. Secondary brink-eddy deposit?
Perhaps in strong wind the stream of sand blowing over the brink is divided by the upstream 'nose' of the eddy current into two sand streams -- one (the majority) that follows gravity and the slope downward, under the eddy current and depositing near the brink in 'dead air', and another sand stream that goes over the nose of the eddy, is kept aloft by the rising eddy current, and is carried perhaps quite a distance downstream by the combined eddy and main currents before falling into the leeward part of the recirculation cell, perhaps creating a second (smaller) cornice mound (or extending the first). This effect would presumably be greater with stronger wind. Evidence for this speculation includes the significant sand accumulation in the interdune area, attributed to grainfall, observed during the 2005 Hanksville storm [link here to the interdune photo and the video with the audio track of sand grains hitting the camera]. Evidence against are the numerous reports in the literature that little sand is deposited further than a meter or two from the brink (in what strength of wind, size of dune, type of sand?).
Development of a compound cornice deposit
Grainflow layers of up to 17cm thick are reported at Zion National Park, Utah (Navajo sandstone). Were such thick layers deposited in one avalanche event from the collapse of one cornice deposit (a drawn-out continuous flow until 17cm accumulated), or were they deposited in several cornice-avalanche cycles but with no visible distinction between cycles (eg., due to the nature of that particular sand and circumstances)? The preserved strata sets are often many meters thick and the dune must have been still higher, so the former (one cycle) implies an extraordinarily large cornice deposit -- is that possible?
A mechanism to create a large upper slope deposit might be this: The angle of repose of sand increases when the flow is thin (in the neighbourhood of 10 grain diameters, eg., 5mm), so it's possible that repeated avalanches could taper out and stop flowing before reaching the lower slopes, building up a wedge. The probability this wedge avalanching would increase with its slope, and eventually this 'super cornice' would fail, feeding grainflow with an extraordinary amount of sand via a receding scarp.
Another possible mechanism might be a secondary deposit (as illustrated in figure 4) acting as a barrier, creating a super-cornice.