The Viking Orbiter photographed much of the surface of Mars and revealed many areas covered by channels that could not have been produced by lava (right image). These dendritic channels become more complex on downward slopes just as water channels appear on Earth. Fragments of martian rock recovered from Antarctic meteorites show considerable water locked up in the crystals indicating that there was water on Mars at one time. The best dating that can be done from counts of impact craters on these 'stream beds' suggests running water about 3-4 billion years ago in the so-called 'Noachen period'. There may be considerable sub-surface water based on the 'mud flows' seen in some impact craters. Also, the permanent part of the southern ice cap is believed to be water ice.
Here is the text from aNASA report by Dr. Vivien Gornitz summarizing the evidence.
Water on Mars By Vivien Gornitz
On July 4, 1997, Mars Pathfinder touched down on a rocky, rusty, wind-blown surface, reminiscent of terrestrial deserts. A trained eye, however, can spot signs that water was once abundant on Mars. Many of the rocks are rounded. (above left image) Some could be conglomerates, i.e., pebbles embedded in silt, sand, or clay, deposited by streams or floods. In places, soils appear to be crusts or hardpans -- cemented by clays or precipitation of silica, iron oxides, or sulfates. All of these features are evidence for former water activity. Today, however, Mars is bone-dry. If all the water in the atmosphere precipitated on the surface, it would form a layer only 10 thousandths of a millimeter thick, on average. Liquid water is not presently stable at the martian surface, due to the thin CO2 atmosphere (around 6-7 mbars) and low temperatures (-60°C, on average). Yet, a small water ice cap remains at the north pole, after the more volatile carbon dioxide frost sublimes into the atmosphere each summer. Near the north pole, the summertime water vapor content of the atmosphere increases to 80 thousandths of a millimeter of precipitatable water.
Pictures of the surface taken by orbiting satellites point to many features showing that Mars once had a much wetter, more clement climate. Twenty-one years ago, two Viking Orbiter spacecraft circled Mars, mapping its surface. The satellite imagery revealed the presence of two types of channels that were probably scoured by running water: networks of branched river-like channels and huge outflow channels, created by catastrophic flooding. More recently, pictures of a meandering martian channel, Nanedi Vallis, taken by the Mars Orbiter Camera (MOC), aboard the Mars Global Surveyor spacecraft, on Jan. 8, 1998, (above right) show that flowing water cut down canyons over prolonged periods of time. The canyon is about 2.5 km wide. The enlarged image shows a tight oxbow meander (bend) and flat terraces along the canyon walls. In order to develop these features, repeated episodes of running water are needed.
Teardrop-shaped "islands" in outflow channels recorded by the Viking Orbiter indicate streamlining by catastrophic floods. The Pathfinder landing site lies at the mouth of one such large outflow channel -- Ares Vallis. Sediments carried down by the floods could still hold moisture at depth and contain traces of possible fossil martian life-forms. Other indirect evidence for a once wetter Mars abounds. Many martian craters are surrounded by ejecta that look like mudflows. Near the equator, only the larger craters (over 4 km diameter) show such "muddy" ejecta, whereas near the poles, even craters as small as one kilometer display them. This suggests the existence of a subsurface permafrost layer (frozen water and soil) which is deeper near the equator, where it is warmer (and hence, only the larger craters can penetrate to this layer), but becomes shallower near the colder poles. Certain features on Mars, surrounding the Argyre and Hellas basins, resemble terrestrial landforms that were carved by glaciers. Some erosional and depositional features in the northern lowland plains may have been produced by large transient lakes. The rusty red color on the martian surface and in the dusty sky yields yet another clue. The Pathfinder lander study of the magnetic properties of airborne martian dust indicates the presence of several percent maghemite, a red magnetic iron oxide (-Fe2O3). Maghemite is common in tropical and subtropical soils on Earth, but also occurs in soils of temperate regions and in weathered basalts from Greenland. The nearly-ubiquitous occurrence of ferric iron on the martian surface, as inferred from Earth-based spectral observations, implies extensive surface weathering involving water under oxidizing conditions. Clay minerals are another product of surface weathering. Spectral absorption bands of clay minerals have been detected by the Mariner 9 IRIS instrument.
Where did all the martian water go? Mars may have lost to space a substantial amount of its atmosphere, including carbon dioxide, nitrogen, and water vapor. The loss of carbon dioxide -- a potent greenhouse gas -- would have caused Mars to become much colder and drier. While some of the water may have been irretrievably lost to space, most of it may still reside in various reservoirs on Mars. Among these are a permafrost layer up to 1 km thick, clay minerals and oxidation products, and layered deposits at the poles. Taken together, these reservoirs could hold the equivalent of a layer of water around 0.5 km thick, if spread uniformly over the martian surface.
The presence of water fulfills an essential condition for the evolution of life on Mars. The possibility that Mars may once have harbored primitive life-forms was dramatically illustrated by the discovery of fossil-like features in a martian meteorite recovered in Antarctica. Water is also a key resource for any future manned mission to the Red Planet.