One of the most exciting revelations from the mapping of Mars in the 1970s was the discovery of ancient, dry riverbeds. We know these channels were carved by water because they have the same structure as networks of terrestrial rivers and they are associated with minerals that only form in the presence of liquid. Some channels also have islands with characteristic streamlined shapes, like teardrops. This only occurs when water flows around a raised piece of land, for example a crater rim. These dry riverbeds seem to be common in older (more heavily cratered) regions, but not in the youngest regions. This indicates water flowed one or more times across the surface during some ancient era, carving the river channels before disappearing. This ancient water may have been frozen over with ice, but virtually every planetary scientist agrees that liquid water did flow at one time on ancient Mars.
Teardrop shaped islands, as seen by Viking. Click here for original source URL
In recent years, mission after mission has found new evidence for this water. In the 1990s, the Mars Global Surveyor (MGS) took images with a resolution of 2 meters per pixel, which is good enough to see features as small as a boulder. At southern latitudes of the planet, there are hundreds of gullies with signs of liquid seepage and runoff. They look very much like places on Earth where sub-surface water from a spring creates a small landslide. This carries rocks and dirt downhill and deposits an "apron" of debris at the bottom. Since these gullies are superimposed on older features, and since almost none of them show any craters, they seem to be geologically young. They might be only a few million years old, or even younger. In fact, scientists can't rule out the existence of liquid water under the present-day Martian surface. The orbiters have seen transient surface features consistent with water erupting from sub-surface aquifers through an escarpment. The water would have run downhill, carved out a gully, and quickly boiled away into the thin Martian atmosphere. Even more recently, modern missions, such as Mars Reconnaissance Orbiter, have observed black streaks appearing on Mars surface. These changing apparitions are thought to perhaps be caused by briny water melting and soaking surface soils. (This is not a unique explanation.)
While surface liquid no longer exists on Mars, water ice has been definitively found by the Mars Phoenix Lander. This little lander dug beneath the top-most layer of dirt on the Mars Northern Polar Cap, and revealed a pocket of water ice/snow, that quickly sublimated away. The orbiters and landers from the past decade have given us an inventory of the frozen reserviors of warter on Mars. Instruments on Mars Odyssey have shown that ice is widespread and abundant on the modern surface. At mid latitudes ice has concentrations of 10-20%, approaching 100% at the poles. If all the ice in the surface soil were melted and distributed evenly it would form a water layer 14 centimeters deep over the whole planet. That would barely get your ankles wet, but Mars is not a bone dry planet! More recently, and more controversially, researchers have claimed that Mars had shallow seas or oceans covering more than a third of its surface, and lasting for hundreds of millions of years. There has even been evidence of water and debris runoff within the last million years in Istok Cater, within the time span of the human species. The "Mars ocean" hypothesis is attracting a lot of attention and many people are working to see if it "holds water."
The source of the past water — the liquid that formed the dry river beds now observed — isn't fully understood. Current temperatures are too low for liquid water to exist on the Martian surface, but it could exist in frozen form beneath the ground, like permafrost. Some of the channels emerge from closed, collapsed regions. This suggests that the water came from local melting of permafrost, perhaps caused by geothermal activity. The heat from a meteorite impact might also melt ice beneath the surface, and some craters actually appear to have had lakes in them at one time. Prominent theories include volcanic activity and related melt, impacts and related melt, and combinations thereof.
Some of the riverbeds appear to have gathered their water from widely dispersed tributaries, suggesting rainfall occurred on ancient Mars. A few researchers have mapped areas they believe contain glacial features, and even shoreline deposits of long-vanished seas. In 1998, Mars Global Surveyor showed that the topography dropping off from the Martian uplands onto the smooth lowland plains is remarkably similar to profiles dropping from the Earth's continents onto smooth sea floor sediments. Rainfall or oceans would require a climate vastly different from the current cold, dry Martian climate. This radical change in Mars from a once wet world to today's cold desert demonstrates how radically planets can change. Were we to terraform Mars (via imaginary technologies) into a wet world again, various factors would cause it to again revert back to its current state.