Martian hillside gullies and icelandic analogs

Martian hillside gullies and icelandic analogs

Available online at www.sciencedirect.com R Icarus 162 (2003) 259 –277 www.elsevier.com/locate/icarus Martian hillside gullies and Icelandic analog...

2MB Sizes 16 Downloads 82 Views

Available online at www.sciencedirect.com R

Icarus 162 (2003) 259 –277

www.elsevier.com/locate/icarus

Martian hillside gullies and Icelandic analogs William K. Hartmann,a,* Thorsteinn Thorsteinsson,b and Freysteinn Sigurdssonc b

a Planetary Science Institute, 620 North Sixth Avenue, Tucson, AZ 85705, USA Science Institute, Department of Geophysics, University of Iceland, Dunhaga 3, IS-107, Reykjavik, Iceland c Energy Resources Division, National Energy Authority, Grensasvegur 9, IS-108, Reykjavik, Iceland

Received 5 March 2002; revised 7 November 2002

Abstract We report observations of Icelandic hillside gully systems that are near duplicates of gullies observed on high-latitude martian hillsides. The best Icelandic analogs involve basaltic talus slopes at the angle of repose, with gully formation by debris flows initiated by ground water saturation, and/or by drainage of water from upslope cliffs. We report not only the existence of Mars analog gullies, but also an erosional sequence of morphologic forms, found both on Mars and in Iceland. The observations support hypotheses calling for creation of martian gullies by aqueous processes. Issues remain whether the water in each case comes only from surficial sources, such as melting of ground ice or snow, or from underground sources such as aquifers that gain surface access in hillsides. Iceland has many examples of the former, but the latter mechanism is not ruled out. Our observations are consistent with the martian debris flow mechanism of F. Costard et al. (2001c, Science 295, 110 –113), except that classic debris flows begin at midslope more frequently than on Mars. From morphologic observations, we suggest that some martian hillside gully systems not only involve significant evolution by extended erosive activity, but gully formation may occur in episodes, and the time interval since the last episode is considerably less than the time interval needed to erase the gully through normal martian obliteration processes. © 2003 Elsevier Science (USA). All rights reserved. Keywords: Mars, gullies; Mars, erosion; Mars, aqueous processes; Mars, ground ice

I. Background Malin and Edgett (2000) announced the discovery of martian hillside gullies that they interpreted as possible evidence of recent water release on Mars. These gullies tend to include three common features: an alcove in the cliff wall at the uphill source, a hundreds-meter-scale channel often narrowing downhill, and a debris fan at the base. Channels are common poleward of 30° latitude, but more so in the southern highlands, and occur primarily on pole-facing, shaded (colder) slopes. Malin and Edgett noted that gullylike features on Earth can form by headward sapping, debris flow, and other mass movements, and argued that the martian features probably involve water. They proposed underground water moving along bedrock layers, with a buildup of ice plugs at the cold surface layers forming contacts with * Corresponding author. Fax: ⫹1-520-622-8060. E-mail address: [email protected] (W.K. Hartmann).

cliff faces. Release of the plugs could cause water outbursts. From lack of craters and from overlap of debris fans on dunes, they suggested that many gullies are geologically very young, possibly ⬍1 My in age (Malin and Edgett, 2000, fn. 31). Hartmann et al. (2000) and Hartmann (2001) pointed out gullies on basaltic hillsides in Iceland that appear to be exact duplicates of the martian examples, in terms of scale and morphology (Fig. 1). Similarities include dimensions, origin in alcove-like features in blocky layers tens of meters down from the cliff top, some channels narrowing downhill, and debris fans at the bottom of some but not all gullies. The similarities strengthened the case for water origins of gullies on Mars. Hartmann emphasized that geologically young martian volcanism, recently confirmed both from martian meteorite samples (Nyquist et al., 2001) and from crater counts on martian lava flows (Hartmann and Neukum, 2001), implies modern sources of sporadic subsurface heating that could melt ground ice and produce aquifers. Very

0019-1035/03/$ – see front matter © 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0019-1035(02)00065-9

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

261

Fig. 1 (continued)

modest fluctuations in subsurface geothermal temperature by only a few tens of degrees from the subsurface norm could produce stable, liquid brines (cf. also discussion of subsurface regolith as a rechargeable ice repository by Clifford (1993) and by Berman and Hartmann (2002). Thus, current evidence confirms that subsurface aquifers could be common or sporadic throughout martian geologic time. This strengthens the case for the underground water flow in aquifers proposed by Clifford (1993), as well as by Malin and Edgett and others. If such aquifers reach the surface on cliff faces, briny composition would lead to freezing points as low as ⫺20 to ⫺50°C, allowing geothermally heated water the opportunity to saturate soils and/or create muddy slurries that would run downhill to create channels and debris flows. This scenario agrees with the finding that various martian meteorites have been exposed to brines (Bridges and Grady, 2000; Bridges et al., 2001; Sawyer et al., 2000). Water reaching a Sun-facing cliff could evaporate rapidly, and never accumulate adequate liquid to create a gully, but an aquifer reaching a colder pole-facing cliff could initially freeze and form a plug, allowing hydrostatic pressure to

build, causing a blow-out with large volume and rate of water release, along the lines suggested by Malin and Edgett (2000). Adding to the reported examples of Iceland gullies, Costard et al. (2001a, b) and Lee et al. (2001) at the Lunar and Planetary Science Conference XXXII pointed out additional hillside gullies in Greenland and northern Canada, respectively, which also matched those of Mars. Both groups suggested additional mechanisms of formation. Costard et al. (2001b, c) described the Greenland gullies as debris flows features. They state that the ground is permanently frozen and no water sources were observed. They concluded that these landforms result not from subsurface water but from surface or near-surface processes (upper 10 m), when soils begin to be heavily saturated with water, due to snow melting and ground ice melting at the surface. They propose the same mechanism for Mars, arguing that the gullies appear primarily poleward of 30° latitude in areas where they suggest ground ice is less than 10 m deep (based on estimates by Fanale et al., 1986), and that the gullies result only from melting of this near-surface material. Costard et al. (2001c) invoke obliquity cycles and calculate that pole-

Fig. 1. “Classic” gullies of the style described on Mars by Malin and Edgett (2000), beginning at the interface of talus and coarsely layered, blocky outcrops. (a) Example, on the wall of Nirgal Vallis at ⫺30°S, 39°W, shown in the Malin/Edgett discovery paper. MGS M03-02290. (b) Aerial view of south side of Reynivallaha´ ls. One prominent gully (left center) originates from a drainage on the plain above the blocky basaltic outcrops, unlike Mars, but other gullies are more Mars-like, originating at the base of the outcrops where no obvious drainage network is visible. (c) Ground view, south face of Esja Plateau, showing Mars-like features such as wider alcove at upper end, and narrowing downslope. A range of preservation states is more apparent than on Mars.

262

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

Fig. 2. Example of Icelandic gully on north side of Langahlı´d, less Mars-like than Fig. 1b. This has formed on a more convex slope, and starts with small furrows partway down the hill, widening downhill. Such forms may be primarily influenced by runoff from melting snow cover.

ward-facing slopes at summer solstice, during high obliquity, above 30° latitude, are the only places that have a daily mean surface temperature above freezing (high obliquity leads to more total insolation of these sites). There is also 50 times as much precipitable water at these times, allowing H2O ice accumulation in the soil. Summer disappearance of CO2 ice cover could lead to sudden (few-day) warming of liquid water release, launching debris flows. Lee et al. (2001) also describe orientation and solar insolation effects on snow cover in the Canadian examples, including the possible development of gullies by melted snow or ice runoff on initially uneven slopes. A radically different view was proposed by Hoffman (2000), who posited that the dominant fluid on Mars has been not liquid water, but liquid CO2, and proposed that gullies and other “fluvial” features of Mars may be formed by high-pressure CO2-powered density flows, where channels are cut by high-density, high-pressure slurries maintained in the liquid state by the weight of overlying material,

as in nue´ es ardentes and similar density flows on Earth. Hoffman’s work is valuable in pointing out that liquid CO2 could play a role at depths as low as a few hundred meters, where ambient gas pressures could rise above the liquid CO2 limiting condition of 5 bars, if adequate pressure seals are maintained between depth and surface. However, three problems exist with the liquid CO2 hypothesis: (1) To maintain liquid CO2 at depth, no open access to the atmosphere can be allowed between the deep reservoir and the surface, in order to maintain the required 5-bar pressure. (Hoffman has appealed to a seal of H2O ice to do this.) Otherwise the CO2 in pore spaces and cavities remains gaseous. We note that caverns on Earth typically do not maintain such seals and have ordinary near-surface ambient air pressure, not pressures corresponding to the weight of overlying rock. (2) It has not been demonstrated that liquid CO2 at 5-bar pressure released on martian hillsides into 0.007-bar ambient pressure would run downhill in tidy debris flows instead of decompressing explosively, more like a geyser eruption. (3)

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

263

Fig. 3. Classic debris flow on Mjo´ afell, Iceland. (a) Distant view, showing winding downhill path, origin on the talus slope, somewhat downhill from the talus/outcrop interface, and levees on each side. (b) Closer wide-angle view, emphasizing the small deltaic deposit at the foot of the slope.

A necessary but insufficient test of H2O vs CO2 hypotheses is to examine whether similar gully-like features due to H2O fluvial activities exist or are excluded on Earth. We argue that they do exist. Our goal in this work has been to carry out a preliminary survey of Icelandic features, both on the ground and from the air, and to gather available data that may shed more light on both martian and Icelandic gully origins. Our initial studies indicate that little formal work had been done on these features, and even our present work gives only an overview, carried out during several field trips within 100 km of Reykjavik. We will point out several issues that need further investigation.

II. Field observations We observed Icelandic hill slopes in August, 2001, at two broadly different geological settings: (1) hillsides of hyaloclastic materials, formed in subglacial basaltic eruptions that create brecciated, glassy, and easily erodible materials, including palagonites; and (2) talus slopes of basaltic

debris on cliffs surmounted by layered Tertiary basalts. Both categories of hillside were observed to have a range of gully-like features, some resembling the martian gullies more than others. We especially sought examples of “classic” martian gully morphologies, as described by Malin and Edgett (2000). As shown in Fig. 1, we found many similar features. We also looked for conditions that produced gullies with other morphologies, less like the martian cases, to understand the range of conditions that produce Mars-like gullies. In the area of our study, around the Esja plateaus, annual precipitation averages were 1500 to 3000 mm, depending on elevation, with minimum in summer and maximum in autumn and early winter (Einarsson, 1988). In the lowlands at the base of the plateau slopes, 5 to 10% of this falls as snow, but we estimate 50% or more falls as snow on the upper surfaces of the plateau. The average annual temperature in the area is ⬃⫹5°C; the January average is close to 0°C and the July average close to ⫹12°C. These data are near sea-level values. The lapse rate is about 0.6°C/100 m. There are many discussions of debris flows in the geomorphological literature, but most illustrated examples deal with broader, avalanche-style wet soil movements and do

264

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

give way and move over deeper frozen ground. He notes that these conditions are favored in spring when heavy rains may soak freshly thawed soils. In principle, a range of different water sources, such as upslope runoff, snow melt, ground ice melting, or aquifer-fed springs could produce water-saturated soils both on Mars and in Iceland, leading to similar styles of debris flows. Below, we discuss various observed examples of Icelandic gully-like forms. Example 1

Fig. 4. Double debris flow on west end of Esja Plateau, with unusually clear display of diagnostic features. The flow starts not at the talus/outcrop interface, but in a distinct patch of dislodged talus soil (dark) somewhat below the outcrops. The patch of waterlogged soil became dislodged, then split into two debris flows, which are clearly outlined by levees consisting of dark rocks.

not resemble the Malin–Edgett martian gullies (Allen, 1997; Innes, 1983). Examples dealing with arctic debris flow morphology (Akerman, 1984; Beylich, 2000; Harris and Gustafson, 1993; Jahn, 1976; Rapp, 1992) may be more relevant but have little discussion of specifically Mars-like examples. Intriguingly, the developmental phenomenology of debris flows on Icelandic slopes may be better known to local farmers than to the scientific community. Future work may profit by exploring these sources of information. This may be more the case for Icelandic Mars analog than for others that have been reported in remote areas in Greenland (Costard et al., 2001c) and Canada (Lee et al., 2001). In general, we find a range of origins for hillside gullies, depending on types of hillside materials, slopes (nearness to angle of repose), and range of water/soil ratios. The latter parameter creates a range of phenomena from ordinary clear stream gullying (high water/soil ratio and narrow incised gullies unlike Mars) to classic debris flow gullies (reportedly around 20 –30% water content in a muddy slurry, producing shallow Mars-like furrows). Interpreting the range of origins under initial conditions is made more complicated by an apparent range of erosional evolutionary states, discussed in Part III. Einarsson (1994) remarks that in Icelandic debris flows, water soaking increases the weight and reduces the cohesion of surface soil layers, which then

Many Icelandic gullies, especially in the hyaloclastic setting, have the opposite morphology from the martian case, starting in a narrow furrow partway down the slope and growing wider and deeper downhill. Examples of these were found in hyaloclastic hillsides northeast of Reykjavik, with various degrees of weak and altered basaltic materials (Fig. 2). Some of these features form on convex ridge-top parts of slopes, as well as in natural concave valleys whose shapes would focus runoff. These types of features appear likely to form under snow pack during springtime melting, so that the eroding water is produced locally on the slope, starting with a trickle upslope and with more water available downslope. This type of feature seems uncommon on Mars, and the external snow-melt origin is unlikely there. Example 2 A more Mars-like class of gully was observed in detail on the northwest face of Mjo´ afell, a hyaloclastic mountain north of the well-known Thingvellir graben and ancient parliament site. Two striking parallel gullies start in oval, cirque-like disturbances, roughly 10 m across, partway up the slope (not associated with bedding exposed further upslope). As shown in Fig. 3, the narrow gullies run in nearly linear channels downslope, ending in very small debris fans at the base of the slope, which is defined by a sharp contact with a very smooth and level dry lake-bed. The betterpreserved channel is about 4 m wide, with a striking, raised rim on each side, composed of cobble-sized rocks involved in a mud-like matrix. The small debris fans at the bottom have the same composition and texture. A visually obvious discrepancy, in which the debris fan had less volume than the gully, led to the conclusion that much of the material had been deposited along the two margins of the gully, on the way downslope. The levee-like ridges were characteristically 1 m high and 2 m wide. We made the following rough estimates of volumes, based on visual measurements and pacing distances: ● Gully volume ⬃ 4 m wide ⫻ 1 m deep ⫻ 100 m long ⫽ 400 m3 ● Debris-fan volume ⬃ 25 m wide ⫻ 0.5 m thick ⫻ 12 m long ⫽ 150 m3 ● Levee volume ⬃ 2 m wide ⫻ 60 m long ⫻ 1 m high ⫻ 2 levees ⫽ 240 m3.

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

265

Fig. 5. Martian-like gullies on the west face of Esja Plateau. Note tendency of many gullies to begin at positions without obvious uphill drainage systems, and Mars-like debris fan systems at the distal end. Cliff face is 500 m high. Note that many gullies begin close to stratigraphic layers outcropping in the talus, supporting the possible delivery of water into the talus through aquifers in these layers.

We interpret this feature as a classic debris flow, where water-saturated material on the slope breaks loose and carries a slurry of rocks downhill. Larger rocks and mud finding themselves on the edge of the flow may come to rest, so that banks or levees of rock and mud are left standing on

the sides. One of us (F.S.) also notes seeing similar flows utilizing older debris flow channels, as springtime runoff resaturates hillside slopes. Multiple activation of the above channel was supported by apparent overlapping structure of tongues of mud and rock of different stratigraphic age at the

266

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

Fig. 6. Schematic cartoon showing four stages of evolutionary development. Stage 1 shows a talus slope with no gullies, but minor albedo features created by downslope movements. This is typical of low-latitude martian slopes. Stage 2 shows gully formation, usually starting near the talus/outcrop interface and ending either partway downhill or in debris fans. Stage 3 shows further development of the upper end of the gullies, eroding the talus and carving it into triangular wedges pointing uphill, with facets of the original talus still exposed. Stage 4 shows late development characterized by deeper uphill alcoves, removal of the original talus, and/or coalescing debris fans.

foot of the gully. Thus, a gully formed by an initial debris flow event could be transformed by rain or snow melt into a more typical hillside stream channel with intermittent flow. Fig. 4 shows another debris flow, where the movement obviously started at mid-slope, and in this case divided into separate debris flows bounded by prominent levees. A patch of water-saturated soil, about 10 to 30 m across, became detached and started moving downslope, with the two debris flows petering out part way down the slope. Example 3 About 15 km northeast of Reykjavik lie tall cliffs (height few hundred meters), the west and south faces of the Esja Plateau, capped by Tertiary basalt flows. These cliffs are faced with long talus slopes of rocks and fine basaltic debris near the angle of repose, and, as seen in Fig. 5, contain gullies strikingly reminiscent of the martian examples. These slopes show a wide variety of Mars-like features produced by downslope motion, including vague “albedo” streaks defined by brightness and/or texture, and massive, broad rock slides caused by cliff collapse (Einarsson, 1994), in addition to fresh gullies produced by debris flow, often lined with parallel raised levies of boulders. The headward ends of the gullies are typically aligned near outcrops of blocky strata. Also visible are faint traces of older debrisflow gully systems, where later downslope movements had softened the outlines and partially hidden the rock alignments. The latter could be seen best in slanting light. As shown in Fig. 5, some of the gullies start along certain stratigraphic layers, without obvious upslope drainage networks to feed them. These offer some suggestion of local,

midslope, hillside origin for the water coming out of the cliff-forming layers, but we have so far been unable to confirm spring-fed aquifer origins for Iceland gullies. Example 4 Some gully-like Icelandic channels originate from what appear to be ordinary fluvial drainage systems on the upper basaltic cliffs, ranging from narrow vertical notches to broader, V-shaped drainages. In some such cases it appeared that simple narrow drainage valleys, without raised rims, had been cut in the talus slope by runoff streams of relatively pure water flowing from above. These did not resemble martian gully systems, having greater depth/diameter ratio. Example 5 On October 2, 2002, one of us (T.T.) observed the Esja gullied hillsides during a rainstorm and confirmed water running in the gullies. In some cases, where loose talus extended virtually to the top of the scarp, water could be seen coming from the plateau surface above, pouring over the rim and down the slope within the types of gullies we have described. Presumably the runoff from the plateau was focused onto certain areas of the scarp by the drainage pattern on the upper surface. In at least one place the talus appeared to have been wetted or saturated, and water was observed coming out of the talus itself, part way down the slope, and running downward in a gully. While these observations do not solve the issue of water delivery on Mars, they establish the general contention that these gullies are formed by water flow, and that various routings of water

Fig. 7. (a) Stage 1 martian hillside at ⫺9°S, 45°W, typical of low latitudes. Slopes at such low latitudes apparently do not develop gullies at the present time. MGS 09-04082. (b) Aerial view of similar stage 1 Icelandic hillside, showing talus slopes below bedded Tertiary basalts, west end of south face of Akrafjall.

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

269

Fig. 8 (continued)

into the talus, either from external water or water within the talus, may produce the conditions that result in Mars-like gully formation on the talus surface. General conclusion from observations of individual gullies The martian gullies are very similar to at least some of the water-based debris flow systems in Iceland. The best match to the martian systems appears to come from steep slopes of either basaltic or hyaloclastic debris in which water has gained access to the surface soils on the slopes. If the slopes are not near the angle of repose, or if the water source is too spread out, as in general snow melt on more uneven slopes, the gullies begin at random points with narrow furrows and widen downhill, or can form more dendritic systems, thus lacking the common martian form. More like the martian cases are traditional debris flows on talus near the angle of repose, where a restricted, upslope focus of soil failure initiates the downslope movement. Martian and some Icelandic gullies can have points of origin at the base of blocky cliffs, but without well-developed upslope drainage systems on the cliff. A central remaining

problem is the delivery mechanism for the water. In any of these cases, the point of gully origin appears associated with sources of water (as in drainage from the cliffs, a spring from an aquifer, or more general saturation of the talus by water), and the gully origin appears to involve water/mud systems organizing into a slurry in a channel system, sometimes atop frozen ground, often narrowing and/or running out of material and stopping its motion partway down the slope. The martian systems may also narrow downhill due to more dramatic evaporation or freezing of some of the water during the motion than would be true of that in Iceland. We thus reach the overall conclusion that the Icelandic evidence favors an origin of martian gullies involving slopes where a source of water has wetted or saturated soils and triggered debris-flow-like features. While we favor many aspects of the Costard et al. (2001c) mechanism, a remaining problem with their equating of martian gullies to classic terrestrial debris flow channels lies in the position of the starting point on the talus slope. As seen in Fig. 4, and less clearly in Fig. 3a, Icelandic debris flows can start at mid-slope positions, at any point where water-saturated soils break loose. We believe this is primarily due to infiltration of water from the rain and snow melt, producing saturated soils on the slopes. Both figures

Fig. 8. Stage 2 gully systems. (a) “Classic” martian gully systems at ⫺41°S, 159°W, from MGS M17-00877. Note origin below a specific stratum, but with lesser development of gullies and drainage systems upslope. These gullies narrow to invisibility downslope, before creating debris fans. (b) Icelandic analogs on west face of mjo´ fell (near location of debris flows in Fig. 3). Note origin along a specific outcropping stratum, and tapering of some systems (especially right center) from alcove-like upper end to nearly invisible lower end. Levees of debris are more visible in this slanting light than in most martian systems. (c) Ground view of similar Icelandic gullies at mjo´ fell, near location of (b).

270

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

Fig. 9. (a) Stage 3 martian gullies show advanced erosion, dissecting the talus into triangular facets. MGS M15-01616, ⫺41°S, 163°W. (b) Aerial view of similar Icelandic examples at the western edge of mjo´ fell, showing triangular remnants of the original talus face. (c) Ground view of similar Icelandic formations on south face of Esja Plateau.

show distinct oval patches well downslope from the rocky cliffs, whereas classic martian gullies (and many other Icelandic gullies—Figs. 1a, 8a) start at the top of the talus slope, i.e., at the base of the rocky outcrops. This difference may favor water delivery on Mars through aquifers in the blocky layers, in at least some cases, rather than uniform saturation of the talus slopes.

III. Evolution of martian and Icelandic talus slope systems Malin and Edgett (2000) noted the relatively pristine appearance of most martian gullies. While muted gullies can sometimes be seen among the more sharply defined ones, they are in the minority. To put it another way, we do not

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

commonly see a sequence of fading gullies, from fresh to muted to vestigial, with equal numbers in each class. On the other hand, we do see various forms of deeper, much more highly developed gullies, which have carved into the talus slopes. These observations suggest clues about relative timescales of martian gully creation and removal process. The situation seems consistent with a case in which gully creation and deepening proceeds mainly in global or widespread episodes, when water is available. Assuming this is the case, consider the relation between the time since the last episode of gully formation, TG, and the timescale needed for gradual gully removal by typical processes of erosion, deposition, and downslope motion, TR. To the extent that observable gullies look fresh, we can infer that TG ⬍⬍ TR. Furthermore, in this model, the lack of a distinct older generation of vestigial narrow gullies would suggest that TR ⬍ the time since the gully-forming episode prior to the last one, so that older, unenlarged gullies do not survive. This model is consistent with the process proposed by Costard et al. (2001c), in which TG would be the time since the last high-obliquity cycle, in which gully formation was widespread in the high latitudes of the planets. Costard et al. suggest this may be about 6 Myr, based on studies by several other workers. In this scenario, the last gullies formed about 6 Myr ago, and the removal processes have been too slow to have degraded them very significantly, but the gullies formed during the preceding high-obliquity cycle either have widened into larger systems (see below) or have been degraded to low visibility. Another model, or perhaps a supplemental model, is that a short timescale, episodic removal process, such as large landslides, could also be operating in addition to the long process such as dust infill. To explain that most channels look fresh, we could then hypothesize that gullies last until this brief process (such as a large landslide) occurs, which forms a new talus surface and removes old gullies entirely, all at once. In contrast to the above models, it is hard to defend a model in which martian gullies are continually forming and continually eroding, on comparable timescales, because we would expect to see a more uniform array of fresh, muted, and vestigial gullies on Mars, as is more the case in Iceland. Thus we infer that the Icelandic processes appear to be more continuous than the martian processes, based on these observations. In cases where gullies do not reflect simple, single waterrelease episodes, both the Icelandic and the martian hillsides show evidence of a more complex evolutionary sequence in the erosion of talus slopes by gully systems, apparently due to continued water release and ongoing gully-forming activity (perhaps the cumulative result of many formative episodes). This finding suggests that many martian systems are not single-event minimalistic releases of water, but rather have continued over extended periods, sufficient to remove substantial amounts of material. Here, we arrange

271

some of our images in terms of four stages of development of steep hillsides, and propose processes that may be involved. An overall schematic view of the evolutionary stages is shown in Fig. 6 in cartoon form. Stage 1. Pregullying In this stage, erosion of upper layers exposes bedding in cliff faces and produces a smooth talus slope of debris lying near the angle of repose. Downslope movements of dry materials produces striations visible as brightness and textural variations (Fig. 7). This stage is commonly found on Mars among low-latitude valley walls and crater walls, and among moderate-to high-latitude walls that presumably have not seen recent release of water. Stage 2. Initial simple gully formation Release of water (from aquifers or ground ice melt on Mars and from runoff, in situ snow melt, or aquifers in Iceland) saturates soils with water, at least locally, and leads to downslope debris flow of a slurry, creating one or more gullies on a hillside (Fig. 8). The slurry may pile up a debris flow at the base of the hill, or the gully may narrow and diminish to nothing, due to loss of water by evaporation, percolation into underlying soil, or freezing. As pointed out by Malin and Edgett (2000), some of the debris fans lie atop dunes at the foot of the hills, suggesting geologically very recent formation. At an advanced level of stage 2, the debris fans may coalesce at the base of the steeper talus slope, forming a continuous apron at the base. Stage 3. Dissection of talus slope into wedge-shaped blocks Further water production and activity in the various gullies can widen the headward ends of gullies and carve the upper part of the talus slope into upslope-pointing triangular, wedge-shaped blocks (Fig. 9). On the remaining lower slopes, smooth segments of the original talus slope surface can still be seen. The parallels between martian and Icelandic examples are striking, and imply considerable erosive evolution on Mars, as opposed to one-time brief releases of small amounts of water. Stage 4. Cliff face erosion and final stage of talus slope destruction Martian gully erosion appears capable of cutting well back into cliff faces, if the process continues long enough (Fig. 10a). Similar enlarged alcoves in cliff faces are observed Iceland (Fig. 10b), and in both cases the removed volume of material forms a large debris fan engulfing much of the original talus slope. After such erosion and deposition, it may be difficult to distinguish the talus remnants from the accumulating, coalescing debris aprons at the bot-

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

Fig. 10 (continued)

tom. The original talus slope may be consumed by erosion. During this process, the cliff face and once-smooth talus slope may be fluvially dissected into “badlands” above a hummocky mass of coalesced debris fans at the bottom. Fig. 10c shows a martian example.

IV. Possible glacial connections Fig. 11 shows an intriguing feature at ⫺38° latitude. It is located on the sloping inner wall of a crater and has the strong appearance of a glacial tongue or rock glacier. This feature at first appears unrelated to our topic, but there are two possible connections. First, if aquifers produce water saturated soils on cliff faces, it is likely that under certain topographic conditions, water could pour out into a perched basin and freeze, making a perched lens of ice or ice-rich soil, instead of a gully. This lens, if it breaches the initial basin-forming barrier, could then begin to break out and flow down the slope as a glacial structure. Turtle et al. (2001) have shown that the timescales for such flow, for either ice or ice/soil, are very short, in the range of thousands of years. Plausibly, as the structure spreads into a flat tongue, as in Fig. 11, ice is continually brought within a few meters of the porous layers of surface debris, so that the ice

273

sublimes and the ice content of the formation decreases, with the ice-rich central part of the flow collapsing due to sublimation loss of the ice — unlike the situation on Earth. This could explain the depressed central part of the tongue, surrounded by pronounced periglacial ridges of debris, and softer-looking mounds of debris, perhaps material pushed downhill ahead of the glacier or remnants of dunes piled against the original front of the flow, on the downhill end. The second connection between Fig. 11 and gullies can be seen by examining other martian gully images. Many of them have remnant crescentric, ridge-like structures at the foot of the gullied slope, remarkably reminiscent of the distal end of the tongue in Fig. 11. Fig. 12 shows an especially impressive example, and the reader who has seen those can pick out vestigial examples at the foot of the cliff in Fig. 10c. These forms may be explicable from the above discussion. If gully-forming water release also produces occasional perched ice masses and/or glacial formations that last only, say, 10,000 years, and if sublimation causes the ice-containing mass to collapse, the form may self-destruct and the main glacial morphology be buried or carried away by aeolian activity on timescales of thousands of years. The distal crescentric ridges of coarser debris, on the flats at the foot of the slopes, are much more common and must have longer lifetimes. While the crescentric distal ridge in Fig. 12 seems clearly associated with the glacier-like tongue, the hypothetical great difference in lifetime would explain why the latter are rare and the former common. In any case, the features in Figs. 11 and 12 are reminiscent of rock glaciers, which are also common in Iceland, and we suggest further searches for these features in the MGS images, studies of their systematics vis-a`-vis martian gullied slopes, and searches for terrestrial analogs.

V. Conclusions Icelandic hillsides offer an array of gully-like forms and talus slopes, some of which appear nearly identical to martian examples, in terms of both scale and morphology. It appears that the Icelandic examples grade into forms not matched on Mars, and these probably involve water sources not available on geologically recent Mars, such as rain runoff and seasonal melt from H2O snow packs. The most Mars-like examples are narrow, shallow, well-defined gullies or furrows created on steep talus slopes, near the angle of repose, by debris flow and/or drainage from upslope cliffs. Some Icelandic debris flows deposit part of their mass in lateral walls, typically composed of rocks embedded in mud, and end partway down the slope, as observed on Mars.

Fig. 10. (a) Stage 4 martian gully, with heavily eroded alcove in the cliff face, and large debris apron covering lower part of hill. MGS M17-00207 at ⫺39°S, 164°W. (b) Icelandic example on south face of Reynivallaha´ ls showing highly eroded alcove and with triangular debris fan covering the hillside just below it. (c) Late stage 4 martian gullies, showing almost complete destruction of the original smooth talus slope. MGS M16-00658, ⫺38°S, 170°W.

274

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

Fig. 11. Possible glacial-like feature at ⫺38°S, 247°W. Note depressed central zone and sharply defined lateral ridges, forming crescentric ridge at the base with softer crescentic ridge beyond it. The latter features are found at the foot of many martian cliff-face gullies, as illustrated in Fig. 12. MGS 18-00897.

Other Icelandic debris flows have enough mass to reach the bottom of the slope, depositing a debris fan of material, as also observed on Mars. Vestigial, softened traces of older debris flows can often be seen on Icelandic slopes, sometimes outlined by parallel rows of rocks that initially were part of the lateral walls. Few examples of vestigial gullies are seen on Mars, suggesting that on Mars, unlike Iceland,

gully-forming episodes are sporadic, and the time since the last one is considerably less than the time required to obliterate the gullies by typical martian processes of erosion, deposition, or gradual downslope creep. When the Icelandic gully systems have been active for a long time, with large-scale sources of water, the whole upper part of the talus slope can be eroded away, creating triangular wedges of remnant slope, pointing uphill, divided by widened gully systems. Upslope from the wedge-shaped talus remnants, apparent headward erosion can create rugged eroded landforms. Both of these evolved morphologies, the wedge-shaped talus slope remnants and the headwarderoded gully complexes, can also be seen on Mars, implying that some martian systems are more complex than single, one-time debris flows. These observations support, but do not prove, the hypothesis that martian gully systems are created by waterbased, geologically recent, fluvial erosion on Mars. Many features of the martian examples are consistent with the hypothesis of Costard et al. (2001c) that they mark debris flows initiated by water released from melting ground ice during high-obliquity episodes. The main problem we see with that hypothesis is that the talus slope should become saturated fairly uniformly and debris flows should often commence in discrete patches partway down the slope. As Costard et al. have developed that model in considerable detail, we offer here a second working hypothesis. In this view, martian subsurface water is created from sporadic partial melting of ground ice masses at some distance from the cliff faces and delivered to the cliff faces by aquifers. Ground ice is known to exist near the martian surface in the same latitudes where the gullies are observed (Kuzmin, 1980; Squyres et al., 1992). Youthful martian volcanic and igneous activity, known from Mars meteorites and crater counts (Hartmann and Berman, 2000; Nyquist et al., 2001), proves that modern heat sources, causing modest subsurface temperature variations, are available as magma moves at depth. Mild heating by only tens of degrees would melt ice and release water on the underside of existing ground ice masses, without creating full-fledged volcanic systems or surface expressions (Hartmann, 2001). Masses of ground ice are likely to inhabit high-porosity megaregolith materials (Clifford, 1993), so that melted ice would find itself mobile within an aquifer layer. If lower porosity layers, or aquicludes, underlie the ground ice, the water could move laterally and gain access to cliffs and crater walls in these latitudes. If such water arrives on the surface at modest flow rates on sun-facing slopes, it would be likely to sublime; but if it arrives on poleward-facing slopes, it would form ice plugs in the very cold surface soils on shaded, pole-facing slopes (Malin and Edgett, 2000). This would allow hydrostatic pressure to build, eventually blowing out the plug and causing a period of significant, rapid water release. Alternatively, water could be released into the thin, upper parts of a talus slope from the exposed or thinly covered blocky strata, saturate the soil, and freeze, only to melt later as per

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

275

Fig. 12. Example of sharply defined crescentric ridges, with softer crescentric ridges beyond them, from MGS M18-00303 at ⫺39°S, 166°W. This duplicates distal features associated with the possible glacier in Fig. 11. The dramatic similarity may suggest that gullies can be associated with short-lived glacial phenomena, as discussed in the text. Vestigial examples of such features are also seen at the foot of gullied slopes in Figs. 1 and 10a.

Costard et al. (2001c). Such water release could erode channels directly but also may moisten upslope soils in such a way as to trigger debris flows of muddy slurries, as

discussed here. This hypothesis explains why martian gullies tend to start at the top of the talus. A continuing flow of warm water exiting a cliff in an aquifer at the top of a talus

276

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

Fig. 13. Apparent aquifer-fed sources of water on Icelandic cliffs (Bru´ ara´ rsko¨ rd, SW Iceland). The waterfalls in the right half of the image originate at the interface between blocky, light-colored strata (possible aquifer) and underlying dark strata (possible aquiclude). This testifies to the lateral transport of water in basaltic layers, and release in cliff faces. (Photograph courtesy of Hja´ lmar R. Ba´ rdarson.)

slope could saturate the thin soil cover there, whereas water exiting from an aquifer deeply buried near the base of the talus slope would have to work its way to the surface through a longer path length of cold martian talus, resulting in freezing. Icelandic gullies, on the other hand, originate in soil saturated by snow melt or rain, and debris flows may be triggered at any point on the slope. Such water release events may thus also create frozen masses of ice/soil mixtures that may explain some features possibly caused by downslope glacial movement. Sublimation of water, as well as sinking of water into the soil, would explain why many martian gullies grow narrower downhill and some disappear before reaching the base of the slope and forming debris fans. Support for the martian aquifer hypothesis of producing martian channels would come from clear evidence of similar processes in Iceland. A number of examples of water production from the middle of cliff faces, by aquifers outcropping on cliffs, are known in Iceland (Fig. 13), but we have so far been unable to confirm Mars-like gully formation specifically from Icelandic aquifer sources of water. We propose that further studies of Icelandic aquifers and further examination of water sources at the headward end of

Icelandic gullies, at the base of cliffs, and in the upper talus slopes could shed further light on martian gully formation.

Acknowledgments We thank an anonymous reviewer and Franc¸ ois Costard for helpful reviews and appreciate the latter’s suggestion of additional references. We also thank Icelandic/martian channel researcher Devon Burr for accompanying us on two of the field trips and offering insights into the issues discussed here. Thanks to PSI Research Assistant Daniel C. Berman for help with processing many of the MGS images used in connection with our search for martian analogs. This work was supported by NASA Grant NAG5-8342 from the NASA Mars Data Analysis Program, and by JPL Grant 961161 from the Mars Global Surveyor program. MGS images are courtesy of Malin Space Science Systems, Jet Propulsion Lab, and NASA. We especially acknowledge the work of Ken Edgett, Mike Malin, and the staff at Malin Space Science Systems in handling and processing the enormous flow of MGS/MOC camera images under tight budget conditions. Th. Thorsteinsson thanks the Icelandic Research Council (RANNIS) for support. This is PSI Contribution 360.

W.K. Hartmann et al. / Icarus 162 (2003) 259 –277

References Akerman, J., 1984. Notes on talus morphology and processes in Spitzbergen. Geograf. Annal. 66, 267–284. Allen, P.A., 1997. Earth Surface Processes. Blackwell Science, Oxford. Berman, D.C., Hartmann, W.K., 2002. Recent fluvial, volcanic, and tectonic activity on the Cerberus Plains of Mars. Icarus 159, 1–17. Beylich, A., 2000. Geomorphology, sediment budget and relief development in Austdalur, Austfirdir, East Iceland. Alp. Ant. Alp. Res. 32 (3– 4), 466 – 477. Bridges, J.C., Grady, M.M., 2000. Evaporite mineral assemblages in the nakhlite (martian) meteorites. Earth Plan et. Sci. Lett. 176, 267– 279. Bridges, J.C., Catling, D.C., Saxton, J.M., Swindle, T.D., Lyon, I.C., Grady, M.M., 2001. Alteration assemblages in martian meteorites: implications for near-surface processes, in: Kallenbach, R., Geiss, J., Hartmann, W.K. (Eds.), Chronology and Evolution of Mars, Kluwer Academic, Netherlands, pp. 365–392. Clifford, S.M., 1993. A model for the hydrologic and climatic behavior of water on Mars. J. Geophys. Res. 98, 10973–11016. Costard, F., Forget, F., Mangold, N., Mercier, D., Peulvast, J.P., 2001a. Debris flows on Mars: analogy with terrestrial periglacial environment and climatic implications. Lunar Planet. Sci. XXXII, 1534 (abstract). [CD-ROM]. Costard, F., Forget, F., Mangold, N., Mercier, D., Peulvast, J.P., 2001b. Debris flows on Mars: comparison with terrestrial analogs, in: Conference on the Geophysical Detection of Subsurface Water on Mars, August 6 –10, 2001, Houston, Abstract 7017. Costard, F., Forget, F., Mangold, N., Peulvast, J.P., 2001c. Formation of recent martian debris flows by melting on near-surface ground ice at high obliquity. Science 295, 110 –113. Einarsson, M.A., 1988. Precipitation in southwestern Iceland. Jo¨ kull 38, 61–70. Einarsson, T., 1994. Geology of Iceland. Ma´ l og menning. Reykjavik, Iceland. Fanale, F.P., Salvail, J.R., Zent, A.P., Postawko, S.E., 1986. Global distribution and migration of subsurface ice on Mars. Icarus 67, 1–18. Harris, S.A., Gustafson, C.A., 1993. Debris flows characteristics in an area of continuous permafrost, St. Elias Range, Yukon territory. Zeit. Geomorphol. 37, 41–56. Hartmann, W.K., 2001. Martian seeps and their relation to youthful geothermal activity, in: Kallenbach, R., Geiss, J., Hartmann, W.K. (Eds.),

277

Chronology and Evolution of Mars, Kluwer Academic, Netherlands, pp. 405– 410. Hartmann, W.K., Berman, D.C., 2000. Elysium Planitia lava flows: crater count chronology and geological implications. J. Geophys. Res. 105, 15011–15026. Hartmann, W.K., Neukum, G., 2001. Cratering chronology and evolution of Mars, in: Kallenbach, R., Geiss, J., Hartmann, W.K. (Eds.), Chronology and Evolution of Mars, Kluwer Academic, Netherlands, pp. 165–194. Hartmann, W.K., Grier, J.A., Berman, D.C., Esquerdo, G.A., 2000. The case for youthful geological activity on Mars. Bull. Am. Astron. Soc. 32, 5802. Hoffman, N., 2000. White Mars: a new model for Mars’ surface and atmosphere based on CO2. Icarus 146, 326 –342. Innes, J.L., 1983. Debris flows. Prog. Phys. Geogr. 7, 469 –501. Jahn, A., 1976. Contemporaneous geomorphological processes in Longyeardalen Vest-Spitsbergen. Biuletyn Periglacjalny 26, 253–268. Kuzmin, R.O., 1980. Determination of frozen soil depth on Mars from the morphology of fresh craters. Akad. Nauk SSSR Dokl. 252, 1445–1448. Lee, P., Cockell, C.S., Marinova, M.M., McKay, C.P., Rice Jr., J.W., 2001. Snow and ice melt flow features on Devon Island, Nunavut, Arctic Canada as possible analogs for recent slope flow features on Mars. Lunar Planet. Sci. XXXI, 1809 (abstract). [CD-ROM]. Malin, M., Edgett, K., 2000. Evidence for recent groundwater seepage and surface runoff on Mars. Science 288, 2330 –2335. Nyquist, L.E., Bogard, D.D., Shih, C.-Y., Greshake, A., Sto¨ ffler, D., Eugster, O., 2001. Ages and geologic histories of martian meteorites, in: Kallenbach, R., Geiss, J., Hartmann, W.K. (Eds.), Chronology and Evolution of Mars, Kluwer Academic, Netherlands, pp. 105–164. Rapp, A., 1992. Frequency and importance of major debris flows in Arctic and other mountains. Bull. Assoc. Geogr. Fr. 3, 249 –252. Sawyer, D.J., McGehee, M.D., Canepa, J., Moore, C.B., 2000. Water soluble ions in the Nakhla martian meteorite. Meteor. Planet. Sci. 35, 743–748. Squyres, S.W., Clifford, S., Kuzmin, R., Zimbelman, J., Costard, F., 1992. Ice in the martian regolith, in: Kieffer, H., Jakosky, B.M., Snyder, C., Matthews, M.S. (Eds.), Mars, Univ. of Arizona Press, Tucson, pp. 523–554. Turtle, E., Pathare, A.V., Hartmann, W.K., Esquerdo, G., 2001. Investigating creep of ground ice as a cause of crater relaxation in martian high-latitude softened terrain. Lunar Planet. Sci. XXXI, 2044 (Abstract). [CD-ROM].