Mike Caplinger, Malin Space Science Systems February 1995
Under current conditions, liquid water on the surface of Mars would either freeze or evaporate almost immediately. The atmosphere, too, is almost without water. Thus, one would not expect to find features that look like those carved by rivers and floods on Earth. But, surprisingly, these can be found almost everywhere on the planet. How and when were they formed?
These features are divided into two types: outflow channels and valley networks. On the following map, the outflow channels are colored red, and the valley networks are yellow.
Compare the locations of these features with our discussion of the age of surfaces on Mars derived from the cratering record. Note that the outflow channels occur mainly in the young surfaces in the northern lowlands (from the Amazonian period of martian history), while the valley networks occur throughout the older heavily-cratered terrains of the Noachian and Hesperian periods. This suggests that the outflow channels were formed after the valley networks.
(As an aside, it must be noted that "young" is a relative term. Since the main techniques we have to date surfaces on Mars are based on the cratering record, it is difficult to resolve ages after the period of heavy bombardment that created most of the craters, and this period ended perhaps as long as 3.8 billion years ago. So even though the outflow channels are younger than the valley networks, they are probably not young in any absolute sense.)
As one can see from the map, most of the outflow channels are in northern lowlands north of Valles Marineris, just west of the Chryse region, landing site of Viking Lander 1.
The outflow channels are large, often more than 100 km wide and as much as 2000 km long. They emanate from cracked or jumbled terrain (termed chaotic terrain) and have distinct flow features such as eroded craters with teardrop-shaped tails, scour marks, and "islands".
As an example, consider Tiu Vallis.
Tiu Vallis appears to have started from an area of collapsed terrain (a region known as Hydaspis Chaos), moved northward through a fairly narrow channel, and then spread out and eroded a large area to the north and west. A more detailed view of the source shows the chaotic terrain, the initial channels, and various eroded features:
Source of Tiu Vallis
Erosional features in the outflow channels often form around obstacles such as craters, such as these at the mouth of Ares Vallis:
Eroded Craters in Ares Vallis
As generally interpreted, the streamlined islands were protected from the fluid flow's erosion by the craters. (An alternative hypothesis is that the tails are depositional features where material was laid down in the lee of the crater during the flow.)
Not all of the outflow channels start in chaotic terrain; some, such as Mangala Vallis, appear to start at deep cracks known as grabens:
Source of Mangala Vallis
What formed the outflow channels? The most commonly-accepted view is that they were formed by catastrophic floods of water released from large groundwater reservoirs. The water would have flowed across the terrain, simultaneously freezing and evaporating; some speculate that the chunks of ice that would have rapidly formed enhanced the erosive power of the flood. The flow might have frozen over at the surface, continuing to flow underneath, much as a frozen river might.
There are some objections to the catastrophic flood explanation. There are no obvious deposits at the ends of the channels; all the material that was eroded away by the flood would have presumably been left there, but it is not seen in the orbital photos. In addition, the volumes of the source areas don't seem to be large enough to account for all the water that would have been required to erode the affected areas, based on models of the efficiency of erosion by water.
Catastrophic floods have occurred on Earth. For example, the Channeled Scabland of Washington State was formed by the breakout of water from the Pleistocene Lake Missoula, and this area resembles the martian outflow channels in many respects. However, it is much easier to understand how standing water could accumulate in a lake and then break out of the lake's boundaries than it is to see how large amounts of groundwater could suddenly be released. Since lakes on Mars are impossible under current atmospheric conditions, the groundwater hypothesis has the advantage of being possible under current conditions, without requiring a period of denser atmosphere and a wetter climate in early martian history.
As with many other aspects of Martian geology, we really won't be sure exactly how the outflow channels formed until we can do fieldwork on the planet. It may be possible to get a better idea using higher-resolution imaging data taken from orbit, and this is one of the things we hope to examine using the Mars Observer Camera. The images taken at the surface by Mars Pathfinder, which will land in the outflow region of Ares and Tiu Valles, will also shed light on this issue.
As can be seen in the initial map, most of the outflow channels are isolated to a fairly small area of the planet, in the younger terrains on Mars. In contrast, the valley networks are present over almost half the planet, mostly in the ancient heavily-cratered southern highlands.
The valleys can be loosely divided into two subtypes: long, winding valleys with few tributaries, and smaller valley networks, often with complex, multiply-branched patterns of tributaries. A good example of the first is Nirgal Vallis, south of the eastern part of Valles Marineris:
An example of a small network is this one found at latitude 42 south, 93 west in the Thaumasia region:
Small valley network
Superficially, the valley networks resemble river-cut valleys on Earth, and initial speculation focused on this explanation for them. Despite the fact that there is no running water or rain on Mars at the present time, earlier in martian history such conditions might have prevailed. However, on further examination, there are significant differences between the martian valleys and river valleys on Earth. First and most important, a terrestrial river valley contains a river, or at least a dry river bed, and no such features have been seen on Mars at the resolution limit of our current images. (It is important to note that a valley is not a channel -- the fluid never filled the valley up to its rim, but was carried only in the channel that cut the valley over time.) In addition, even the densest tributary networks on Mars are much sparser than their terrestrial counterparts. These facts argue against a purely running-water origin for the martian valleys.
An alternate explanation involves sapping processes, the weathering and erosion of terrain by emerging groundwater. When the underlying soil is weakened by groundwater flow, the overlying surface collapses. Similar processes have acted on Earth in, for example, the Navajo Sandstone of the Colorado Plateau. This explanation works well for the long winding valleys such as Nirgal Vallis. For the more complex small valley networks, a mixture of the two mechanisms may be required, in which the valleys were initially formed by runoff of water, and then enhanced by sapping.
Today, based on our observations from orbit, Mars appears to be very dry. There is little water in the atmosphere and only a small amount of water ice in evidence on the surface. Yet the planet is covered with features that are best explained by the movement of water, either in catastrophic floods or the slow movement of groundwater. Whether that water was present early in the history of Mars and was lost to space over eons, or is still present in great underground deposits of ice and groundwater, is a question whose answer must be left for the future exploration of Mars.
Mike Caplinger (firstname.lastname@example.org)