Along crater and canyon wall

Liquid water cannot exist on the surface of Mars due to low atmospheric pressure, except at the lowest elevations for short periods.[55][56] The two polar ice caps appear to be made largely of water.[57][58] The volume of water ice in the south polar ice cap, if melted, would be sufficient to cover the entire planetary surface to a depth of 11 meters.[59] A permafrost mantle stretches from the pole to latitudes of about 60°.[57]
Large quantities of water ice are thought to be trapped within the thick cryosphere of Mars. Radar data from Mars Express and the Mars Reconnaissance Orbiter show large quantities of water ice both at the poles (July 2005)[20][60] and at mid-latitudes (November 2008).[21] The Phoenix lander directly sampled water ice in shallow Martian soil on July 31, 2008.[23]
Landforms visible on Mars strongly suggest that liquid water has at least at times existed on the planet's surface. Huge linear swathes of scoured ground, known as outflow channels, cut across the surface in around 25 places. These are thought to record erosion which occurred during the catastrophic release of water from subsurface aquifers, though some of these structures have also been hypothesised to result from the action of glaciers or lava.[61][62] One of the larger examples, Ma'adim Vallis is 700 km long and much bigger than the Grand Canyon with a width of 20 km and a depth of 2 km in some places. It is thought to have been carved by flowing water early in Mars's history.[63] The youngest of these channels are thought to have formed as recently as only a few million years ago.[64] Elsewhere, particularly on the oldest areas of the Martian surface, finer-scale, dendritic networks of valleys are spread across significant proportions of the landscape. Features of these valleys and their distribution very strongly imply that they were carved by runoff resulting from rain or snow fall in early Mars history. Subsurface water flow and groundwater sapping may play important subsidiary roles in some networks, but precipitation was probably the root cause of the incision in almost all cases.[65]
Along crater and canyon walls, there are also thousands of features that appear similar to terrestrial gullies. The gullies tend to be in the highlands of the southern hemisphere and to face the Equator; all are poleward of 30° latitude. A number of authors have suggested that their formation process demands the involvement of liquid water, probably from melting ice,[66][67] although others have argued for formation mechanisms involving carbon dioxide frost or the movement of dry dust.[68][69] No partially degraded gullies have formed by weathering and no superimposed impact craters have been observed, indicating that these are very young features, possibly even active today