Image by NASA/JPL/MSSS
Those very curious transverse barchan dunes on so many canyon floors, as fresh as daisies, would be the result of the wind re-arranging the residue from the last flow that dropped out at the end of the event, without making it to the canyon mouth.
Photo after photo shows what looks to me unmistakably like dry slumping and slippage of poorly consolidated material at the heads of canyons and their tributaries, and on the flanks of both as well.
If the higher areas catch the sun sooner than the lower, which seems to be the way of the worlds, the dry ice would sublimate in the high areas, destabilising material there, while the lower remained locked in solid dry ice. Hence the frozen material lower down the channel would remain resistive to that entrained, and when eroded would retain sharp and steep sides, as if it were soft rock. That seems to be what the photos show.
As noted above, the Martian atmosphere is about 95 percent carbon dioxide. It didn't grow there, but presumably degassed from the interior. What so much carbon is doing down there is a mystery, but that is beyond me.
There is an insignificant little volcano, Ceraunius Tholus, merely the size of the island of Hawaii, that may show the degassing is not entirely over. It has an odd elliptical depression right at its base, directly at the end of a very distinct and narrow erosion gully running from the crater.
That has been interpreted as a near bullseye hit by a meteorite if aimed at the volcano, and a perfect hit if aimed at the end of the channel. There is a small delta of sand or silt or dust, right where the channel enters the depression.
My guess is that the depression was formed by material that flowed down the channel from the crater. Whatever flowed down, if it did, was heavier than the dust at the foot of the volcano, because it flowed under that dust, probably quite rapidly, or at least slopped large quantities of it as a slurry over the edges, where it remains in irregular, spilled sheets.
The solid material in the flow seems to have dumped out first, to form the interior delta; then the gas vanished. Water has a specific gravity of 1, by definition. Carbon dioxide as dry ice has a specific gravity of about 1.5, so in theory the gas can be heavier than water, I think, if very cold.
Subterranean carbon dioxide gas reaching the surface of Mars in midwinter through a volcanic vent may get very chilly, stay dense, and flow downhill like a liquid. Maybe that is what lifted the dust, slopped it away, and then vanished into thin air, as the saying goes.
Maybe some of the smaller craters of Mars are also de-gassing slump structures. When basalt cools, shrinks and cracks, or mud dries, we at times get regular polygons on the surface.
When unconsolidated aeolian sediment shrinks by material loss resulting from sub-surface sublimation, which does not seem to happen here on planet Earth, maybe on Mars we could get neat round craters, at similar spacings.
One argument for the existence of large amounts of water is that the summer temperature at the poles, -68 degrees C, is too high for dry ice to remain solid, though water ice would. So what does not melt every year is assumed to be water ice.
The short answer may be that surface temperature is not automatically sub-surface temperature. Simple calorimetry may show that the incoming solar radiation each summer is insufficient to sublimate all the dry ice in the icecap, given that what is lost each summer is replaced each winter, more or less.