Despite the fact that, apart from greenhouse effects, Mars has had temperatures that never exceeded about minus 50 degrees C over its lifetime, it also has had some quite unexpected fluid systems. One of the longest river systems starts in several places at approximately 60 degrees south in the highlands and drains into Argyre, thence to the Holden and Ladon Valles, then stops and apparently dropped massive amounts of ice in the Margaritifer Valles, which are at considerably lower altidude and just north of the equator. Why does a river start at one of the coldest places on Mars, and freeze out at one of the warmest? There is evidence of ice having been in the fluid, which means the fluid must have been water. (Water is extremely unusual in that the solid, ice, floats in the liquid.) These fluid systems flowed, although not necessarily continuously, for a period of about 300 million years, then stopped entirely, although there are other regions where fluid flows probably occurred later. To the northeast of Hellas (the deepest impact crater on Mars) the Dao and Harmakhis Valles change from prominent and sharp channels to diminished and muted flows at –5.8 k altitude that resemble terrestrial marine channels beyond river mouths.
So, how did the water melt? For the Dao and Harmakhis, the Hadriaca Patera (volcano) was active at the time, so some volcanic heat was probably available, but that would not apply to the systems starting in the southern highlands.
After a prolonged period in which nothing much happened, there were catastrophic flows that continued for up to 2000 km forming channels up to 200 km wide, which would require flows of approximately 100,000,000 cubic meters/sec. For most of those flows, there is no obvious source of heat. Only ice could provide the volume, but how could so much ice melt with no significant heat source, be held without re-freezing, then be released suddenly and explosively? There is no sign of significant volcanic activity, although minor activity would not be seen. Where would the water come from? Many of the catastrophic flows start from the Margaritifer Chaos, so the source of the water could reasonably be the earlier river flows, but why did it ice?
If we start with the source of the water, that would presumably be volcanism, and the evidence is there was plenty of volcanic activity about four billion years ago. Water and gases would be thrown into the atmosphere, and the water would ice/snow out predominantly in the coldest regions. That gets water to the southern highlands, and to the highlands east of Hellas. There may also be geologic deposits of water. From the previous posts, the gases would contain methane and ammonia. The methane would provide some sort of greenhouse effect, but ammonia on contact with ice at minus 80 degrees C or above, dissolves in the ice and makes an ammonia/water solution. This, I propose, was the fluid. As the fluid goes north, winds and warmer temperatures would drive off some of the ammonia so oddly enough, as the fluid gets warmer, ice starts to come out. Ammonia in the air will go and melt more snow. (This is not all that happens, but it should happen.) Eventually, the ammonia has gone, and the water sinks into the ground and freezes out into a massive buried ice sheet.
If so, we can now see where the catastrophic flows come from. We have the ice deposit where required. We now require at least fumaroles to be generated underneath the ice. The Margaritifer Chaos is within plausible distance of major volcanism, and of tectonic activity (near the mouth of the Valles Marineris system). Now, let us suppose the gases emerge. Methane immediately clathrates with the ice (enters the ice structure and sits there), because of the pressure. The ammonia dissolves ice and forms a small puddle below. This keeps going over time, but as it does, the amount of water increases and the amount of ice decreases. Eventually, there comes a point where there is insufficient ice to hold the methane, and pressure builds up until the whole system ruptures and the mass of fluid pours out. With the pressure gone, the remaining ice clathrates start breaking up explosively. Erosion is caused not only by the fluid, but by exploding ice. The temperature never gets near the freezing point of water.
The point then is, is there any evidence for this? The answer is, so far, no. However, if this mechanism is correct, there is more to the story. The methane will be oxidised in the atmosphere to carbon dioxide by solar radiation and water. Ammonia and carbon dioxide will combine and form ammonium carbonate, then urea. So if this is true, we expect to find buried where there had been water, deposits of urea, or whatever it converted to over three billion years. (Very slow chemical reactions are essentially unknown – chemists do not have the patience to do experiments over millions of years, let alone billions!) There is one further possibility. Certain metal ions complex with ammonia to form ammines, which dissolve in water or ammonia fluid. These would sink underground, and if the metal ions were there, so might be the remains of the ammines now. So we have to go to Mars and dig.
I am hoping to attach two images taken by the NASA satellites. One shows the Dao and Harmakhis Valles. The other shows a small section of one. As you can see, the greatest erosion happens near the sources, or in isolated sections, and these can have extremely steep sides. In the second, the flow is from top right to bottom left, and you an see it has to link through a very narrow channel. Additionally, you should be able to see what looks like signs of slight run-off erosion similar to that you might expect from snow melting on the close-up. My view is that these great cavities were simply huge ice masses that eventually melted and flowed away.