Evidence for Precipitation on Mars from Dendritic Valleys in the Valles Marineris Area

See allHide authors and affiliations

Science  02 Jul 2004:
Vol. 305, Issue 5680, pp. 78-81
DOI: 10.1126/science.1097549


Dendritic valleys on the plateau and canyons of the Valles Marineris region were identified from Thermal Emission Imaging System (THEMIS) images taken by Mars Odyssey. The geomorphic characteristics of these valleys, especially their high degree of branching, favor formation by atmospheric precipitation. The presence of inner channels and the maturity of the branched networks indicate sustained fluid flows over geologically long periods of time. These fluvial landforms occur within the Late Hesperian units (about 2.9 to 3.4 billion years old), when Mars was thought to have been cold. Our results suggest a period of warmer conditions conducive to hydrological activity.

The formation of valley networks on Mars has been the subject of considerable scientific debate (14). Valleys were attributed to fluvial erosion implying a warm and wet climate on early Mars (5), possibly conducive to biological activity (6). Several recent observations argue in favor of such conditions on early Mars (7, 8), but valley networks could also have formed by water-lubricated debris flows (4), hydrothermal activity (9, 10), or groundwater sapping (1113) due to geothermal activity (14), and these hypotheses do not require conditions warmer than the current cold climate.

Here, we describe terrestrial-like dendritic valleys identified using THEMIS images (15) of the Mars Odyssey mission. These valleys are located in the Valles Marineris region (Fig. 1) on the plateau west of Echus Chasma (Fig. 2) and on the inner plateau west of Melas Chasma (Fig. 3). These landforms occur within Late Hesperian units (16), about 2.9 to 3.4 billion years old (17); they are thus unexpectedly younger than the Noachian (16) period, which is considered to be the potential primitive warm period (4).

Fig. 1.

MOLA shaded relief map of West Valles Marineris and the neighboring canyons. White squares indicate valley networks observed over Hesperian geologic units on THEMIS images of Figs. 2 and 3.

Fig. 2.

Dense valley networks of West Echus Chasma plateau (0°N, 81°W). (A) Night IR THEMIS images (I02119003 and I0857003). At the lower left of the image, a dark area is due to the low temperature of loose material (sand or dust), whereas the right part of the image is rocky with a higher temperature. The plateau is located at the transition between regions of low and high thermal inertia, which makes the identification of valleys easier from differential mantling. Valleys at top left are dissected by deep canyon tributary of Echus Chasma. (B) Geomorphic map of West Echus Chasma. The observed networks can be separated into nine drainage basins from A to I. (C) Close-up over G, H, and I drainage basins on day IR THEMIS image (I06419047). Contrasts in brightness correspond to differences in albedo and/or topography. The arrow displays the junction of the main valley over the plateau with the head of the tributary canyon. (D) THEMIS day IR (I01950002). The central arrow indicates a linear hill covered by gullies that organize downstream in branching valleys on the illuminated flank. The top arrow indicates a zone of erosion of the upper layers with apparent preservation of an inverted channel (29).

Fig. 3.

Dense valley networks on West Melas Chasma inner plateau (77.5°W, 10°S). (A) Day IR THEMIS images (I06631017 and I06227001). (B) Valley network map drawn from a combination of IR and visible-light THEMIS images. Valleys are located on a perched surface around 2 km above the floor of Melas Chasma. They are separated by a divide in two drainage basins (West and East). (C) THEMIS visible-light image (V3249001) shows meandering valleys over East Melas drainage basin. (D) Close-up over the central valley with inner channels.

The plateau west of Echus Chasma (0°N, 81°W) is covered by densely branched valleys that are frequently sinuous and extend over tens of kilometers (Fig. 2). Infrared (IR) images taken at night by THEMIS (Fig. 2A) show mainly intrinsic thermal properties of the ground (15). Valleys buried under loose material are outlined by variations of the thermal signal and resemble terrestrial deserts where sand covers the floor of dry valleys (fig. S1). On drainage basin G, the main valley is larger upstream than downstream close to the mouth, implying a thicker mantling at this location. With the exception of this place, most valleys have widths increasing from their sources to their mouth, as seen in terrestrial valleys. An IR image taken during the day by THEMIS shows that these dendritic valleys have their heads scattered at random points on the plateau (Fig. 2C). Hills are also gullied by small valleys with heads at the crestline of the hill (Fig. 2D). These characteristics are similar to terrestrial features of surface runoff due to atmospheric precipitation.

These characteristics are inconsistent with subsurface seepage induced by hydrothermal activity because water would not seep at the crest of hills. Moreover, no valleys with theater-shaped heads are observed, as would be the case if sapping had occurred (11, 12). Sapping has been invoked to explain the development of tributary canyons in the Valles Marineris region (18), such as the narrow tributaries of Echus Chasma, 1 to 3 km deep, that dissect the plateau borders. The main valleys of basins E, G, and I connect to the heads of tributaries, implying that these valleys were active during the formation of tributaries, as observed on Earth (fig. S1B). This contemporaneous activity suggests that the backward recession of tributary canyons by sapping was connected to the hydrological processes existing over the plateau.

Two drainage basins with dendritic valleys are also observed on THEMIS images (Fig. 3) of the inner plateau of west Melas Chasma (77.5°W, 10°S). Several valley heads located at the foot of the southern wall slopes would suggest subsurface seepage. However, most valley heads are located on the opposite perched interior plateau and along the divide separating both drainage basins, thus suggesting another source of water than seepage from canyon walls (19). THEMIS visible-light images also show the presence of meandering valleys (Fig. 3C) and inner channels (Fig. 3D) on the floor of some of these valleys, indicating surface conditions with stable liquid water and sustained fluvial activity (20, 21).

The drainage densities (i.e., the total length of valleys divided by the area of each basin) vary from 0.6 to 1.0 km–1 for the nine basins of Echus Plateau measured using IR images at 100 m/pixel (table S1). The drainage densities in Melas Chasma are 1.1 km–1 and 1.5 km–1, as measured using visible-light images at 18 m/pixel. By comparison, drainage densities of terrestrial valley networks are usually from 2 to 100 km–1. However, such high densities are obtained with maximum-resolution mapping, whereas the same terrestrial networks mapped at the scale of Viking image mosaics (22) have densities of only 0.1 to 0.2 km–1. Thus, densities measured at THEMIS resolution, a scale slightly better than Viking mosaics, are equivalent to terrestrial fluvial valleys mapped at the same scale. Additionally, morphometric parameters such as valley order, bifurcation ratio, and valley length ratio (23, 24) give values similar to terrestrial river networks for the drainages of both regions (table S1 and fig. S2).

Precise depths of valleys are not measurable at the resolution of the Mars Observer Laser Altimeter (MOLA), but they can be estimated roughly from THEMIS images to a few tens of meters for widths of several hundreds of meters. These valleys, like rivers, are always oriented in the direction of the local slope (fig. S3). Moreover, the variations in the geometry of valleys are consistent with the variations of the slope, with dendritic valleys on nearly flat areas and subparallel valleys on slopes >1.3° (fig. S3). This relationship is observed in experimental streams and terrestrial rivers formed by surface runoff due to precipitation by rainfall or snowmelt subsequent to snow deposition (25).

Snowmelt under a cold climate was proposed to explain some valley networks (26, 27). However, alluvial valleys with inner channels (Fig. 3D) require stable liquid water, favoring subaerial flows under a warmer climate. Moreover, snow accumulation would create glaciers (27), but subglacial valleys are discontinuous with large undissected areas and abrupt inception and termination of valleys [e.g., (28)], unlike such dendritic valleys. The maturity of the valley networks is also inconsistent with short episodes of glacial outburst with large and braided channels. For example, terrestrial networks become mature (i.e., with fully branching patterns) only after several tens of thousands of years of activity (24). The formation of dendritic valleys thus likely involves a relatively warm climate with liquid water stable at the surface.

If atmospheric processes are invoked, similar valleys should be observed elsewhere than in the Echus and Melas areas. Nonetheless, dust deposition could completely fill in valleys that are only a few tens of meters deep. Other valleys may be hidden beneath this mantle. No valleys exist west of the Echus plateau valleys except for some relics present in inverted topography (29) (Fig. 2D), demonstrating the role of wind erosion in the removal of the uppermost deposits. Differences in the strength of eroded rocks may play a role in the development or preservation of fluvial landforms. Mesoscale variations of the climate could also affect their distribution.

Echus Chasma plateau valleys formed over a Late Hesperian volcanic unit (30), and they are buried at the north by Early Amazonian lava flows, thus restricting their formation to that interval of time. The connections of the main valleys and the heads of tributary canyons (Fig. 2C) also imply an age contemporaneous with or slightly younger than the Echus canyon, which formed during the Late Hesperian. Melas Chasma valleys are younger than the development of Valles Marineris, as they developed on inner plateaus dated to the Late Hesperian epoch (18). Thus, they may have formed at the same period as Echus Chasma plateau valleys. The transition from a possible warm early Mars to a colder climate is usually dated to the Late Noachian–Early Hesperian boundary (4), about 3.6 billion years ago (17). The apparent age of the valleys suggests that significant fluvial activity occurred until the Late Hesperian.

Surface runoff in the Late Hesperian epoch could correspond either to a progressive transitional climate after the warmer Noachian epoch (7) or to episodic warmer periods, such as those that could be related to the increase in atmospheric water vapor due to the outburst of outflow channels (31). Stable liquid water at the surface in the Hesperian epoch was previously suggested to explain potential lake deposits inside Candor Chasma (18) or in highland craters [e.g., (32)], and other Hesperian valley networks were noticed on highlands (7, 10) and volcanoes (7, 9). Such conditions could also solve the paradox of the ocean in the northern lowlands (33), which is dated to the Late Hesperian. Finally, longer term hydrological activity until the Late Hesperian would lead to interesting exobiological consequences, because life—if it ever existed on Mars—would have benefited from a longer period of warmer conditions.

Supporting Online Material

Figs. S1 to S3

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article