Image by NASA/JPL/MSSS
Also, less force is required to start a light particle moving than a heavy one. The dust storms, so extensive and so slow to settle, seem to bear this out. So I contend that, where not tectonically controlled, the canyons and channels were and still are being carved by slumping and by related moving masses of abrasive material, more like dry powdery snow in avalanche mode than by similar material entrained by liquid water.
Between them, the Hubble Space Telescope Wide Field/Planetary Camera (HST/WFPC) and the Mars Global Surveyor Orbiter Camera (MGS/MOC) tracked a "dust storm" in the Valles Marineris that "extended as far as the chaotic eastern end of the system",(Release MOC2-1A), but which did not leave the canyon.
I would like to respectfully suggest this may have been dust raised by a collapse of the canyon walls, proving it is alive and well and getting bigger.
These "storms' should be common, as the largest pile of unstable and entrainable surface material on the planet is that forming the steep, unstable canyon edges. I would be delighted to see barchan dunes along the valley bottom in that dust storm area.
Some of the physics may in some senses be more like that of undersea turbidity currents than river flows, in so much as, with those the lubricant and the overlying environment are of the same phase. In our oceans both are liquid water, on Mars both are and perhaps always were the local air.
When dry ice sublimates on Mars, if the atmosphere is still, or a blanket of atmospheric dust is present from a dust storm, it may in some circumstances hang close to the surface as a denser layer of gas than the rest of the atmosphere, being far colder.
If it does and starts to flow downhill, like a katabatic airflow here, it could entrain dust, with the mixture becoming denser still and remaining cohesive, as with undersea turbidity currents.
Such flows may be triggered by temperature changes, by dust storms or occasionally by meteorite impacts. They would account for the braided and meandering channels and all, very nicely.
In the great tension chasms, all that is required to start material moving is the collapse of steep walls under their own weight. Once moving, the heat of friction would rapidly round the contained dry ice, providing millions of roller bearings for the flow to coast on.
Martian air is about 95 percent carbon dioxide. Phase diagrams indicate it will skip the liquid phase on Mars as on Earth, and freeze from the gaseous state, or sublimate. The effects of this on the surface processes of a faraway lithosphere are difficult to conceptualise, but one thing that should happen is that the gas in the pore spaces between the dust grains should almost all freeze solid.
This is far different from a minute amount of interstitial water vapour freezing on dry slopes here. Carbon dioxide freezes and sublimates on the surface of Mars at the poles, so it almost certainly does the same to some depth below the surface, at lower latitudes.
I don't know enough of carbon dioxide's expansion and contraction at low temperatures to have any idea of its expansion coefficients as a solid, but they are probably quite sufficient to destabilise steep slopes, so that when the temperature rises sufficiently to cause sublimation, away it all goes.
Maybe the expansion of both the dry ice and the dust particles as the temperature rises from the lowest point reached, is sufficient to push the dust grains apart so that they lose their point-contacts with other dust grains. When the dry ice evaporates, you would then have a very unstable dust pile, all ready to go travelling down any handy slope.