Geologically recent gully–polygon relationships on Mars: Insights from theAntarctic Dry Valleys on the roles of permafrost, microclimates, and water sourcesfor surface flow
J.S. Levy a,∗, J.W. Head a, D.R. Marchant b, J.L. Dickson a, G.A. Morgan a
a Department of Geological Sciences, Brown University, Box 1846, Providence, RI 02912, USAb Department of Earth Science, Boston University, 675 Commonwealth Ave., Boston, MA 02215, USA
a r t i c l e i n f o a b s t r a c t
Article history:Received 29 May 2008Revised 17 October 2008Accepted 22 December 2008Available online 21 January 2009
We describe the morphology and spatial relationships between composite-wedge polygons and Mars-like gullies (consisting of alcoves, channels, and fans) in the hyper-arid Antarctic Dry Valleys (ADV),as a basis for understanding possible origins for martian gullies that also occur in association withpolygonally patterned ground. Gullies in the ADV arise in part from the melting of atmospherically-derived, wind-blown snow trapped in polygon troughs. Snowmelt that yields surface flow can occurduring peak southern hemisphere summer daytime insolation conditions. Ice-cemented permafrostprovides an impermeable substrate over which meltwater flows, but does not significantly contributeto meltwater generation. Relationships between contraction crack polygons and sedimentary fans at thedistal ends of gullies show deposition of fan material in polygon troughs, and dissection of fans byexpanding polygon troughs. These observations suggest the continuous presence of meters-thick ice-cemented permafrost beneath ADV gullies. We document strong morphological similarities betweengullies and polygons on Mars and those observed in the ADV Inland Mixed microclimate zone. Onthe basis of this morphological comparison, we propose an analogous, top–down melting model for theinitiation and evolution of martian gullies that occur on polygonally-patterned, mantled surfaces.
Gullies on Mars are a class of geologically young features, ini-tially interpreted to have formed by surficial flow of releasedgroundwater (Malin and Edgett, 2000, 2001; Mellon and Phillips,2001), and which may still be active (Malin et al., 2006). Mar-tian gullies are geomorphic features composed of a recessed al-cove, one or more sinuous channels, and a depositional fan orapron (Malin and Edgett, 2000). Alternative hypotheses for thesource of gully-carving fluids include obliquity-driven melting ofnear-surface ground ice (Costard et al., 2002), melting of dust-richsnow deposits (Christensen, 2003), and melting of atmosphericallyemplaced frost and/or snow (Hecht, 2002; Dickson et al., 2007a;Head et al., 2007; Dickson and Head, 2008; Williams et al.,2008). Complementing gully formation models, recent GCM re-sults (Forget et al., 2007) predict the deposition and potential formelt of up to 25 mm/yr of water ice at martian northern midlat-itudes (∼30–50◦ N) during obliquity conditions modeled to haveoccurred within the past 10 My (and potentially within the past<1 My; Laskar et al., 2004). Other workers have proposed that
gullies can form by dry avalanche processes alone (Treiman, 2003;Pelletier et al., 2008).
Concurrent with advances in understanding of gully processeson Mars, modeling and observational studies have documented thedistribution and origin of various types of martian thermal con-traction crack polygons (Mellon, 1997; Mangold, 2005; Levy et al.,2008a). Despite the observation of polygonally patterned groundin gullied terrains on Mars and Earth (Malin and Edgett, 2000,2001; Bridges and Lackner, 2006), and an increasing awarenessof the importance of polygonally patterned permafrost in the de-velopment of terrestrial polar fluvial systems (Fortier et al., 2007;Levy et al., 2007a; Levy et al., 2008b), there has been little analy-sis of the interactions between thermal contraction crack polygonsand gullies on Mars.
In this contribution, we explore interactions between gulliesand polygons in the Mars-like Antarctic Dry Valleys (Marchant andHead, 2007), and then assess similarities and differences with fea-tures observed on Mars. We first summarize recent research onthe spatial distribution, formation, and modification of gullies andpolygons in selected regions of the Antarctic Dry Valleys (ADV). Inthe next section we show how gully development on polygonallypatterned ADV surfaces affects gully morphology and enhanceswater-flow processes. Further, we show how the morphology of
Fig. 1. Perspective view of a portion of the South Fork study area in upper Wright Valley, Antarctica. Black arrows indicate channels on the southern wall of the valley andwhite arrows indicate large alcoves present in the dolerite bedrock, approximately 1000 m above the valley floor. The dark, tongue-shaped lobe of dolerite boulders at centeris approximately 300 m wide. Inset. Boxed region showing a small concavity present in the colluvium slope. White arrow indicates the center of the depression. Channelsenter into and emanate from the concavity.
polygons is altered by proximity to developing gullies. We describesuch reciprocal modification relationships as “gully–polygon sys-tems.”
We then analyze HiRISE images that document the interplaybetween polygonally patterned ground, ice-cemented permafrost,and gullies on Mars. If strong morphological similarities exist be-tween gullies and polygons observed on Mars and those docu-mented in the ADV, then this evidence would suggest that, toa first order, some martian gullies formed and were modifiedby processes analogous to those occurring in ADV gully–polygonsystems. Such morphological comparisons can help constrain thephysical and hydrological properties of gully flow. We addressconcerns over equifinality (similar morphologies produced by dif-ferent processes) by focusing our analysis on morphological rela-tionships that illustrate specific spatial and stratigraphic relation-ships.
2. The Antarctic Dry Valleys (ADV)
The Antarctic Dry Valleys are a suitable laboratory for under-standing the geomorphological effects of water moving throughtemperature-dependent phase transitions (freezing, melting, sub-limation, evaporation). On the basis of summertime air tempera-ture, relative humidity, soil temperature, and soil moisture con-ditions, the ADV region is divided into three microclimate zones.The three zones include a coastal thaw zone, an inland mixedzone, and a stable upland zone (Marchant and Denton, 1996;Marchant and Head, 2007). In the inland mixed zone, melting,evaporation, and sublimation occur, whereas in the stable up-land zone, sublimation is the dominant phase transition (Ragotzkieand Likens, 1964; Marchant et al., 2002; Kowalewski et al., 2006;Marchant and Head, 2007). The stable upland zone is interpretedto be closely analogous to Mars under current, average climateconditions, whereas the inland mixed zone may be a good analogfor more clement martian conditions produced by orbitally-drivenclimate change (Marchant and Head, 2007) or for short durationpeak temperature and insolation conditions. Landforms that areproduced in equilibrium with microclimate conditions in each zone
are termed equilibrium landforms (Marchant and Head, 2007). Gul-lies and polygons are the two dominant equilibrium landforms oninland-mixed zone valley walls.
2.1. Gully–polygon systems in the ADV
In the inland mixed zone of the ADV, gullies are character-ized by a recessed alcove, sinuous channels with seasonally moisthyporheic zones (McKnight et al., 1999; Gooseff et al., 2002;Levy et al., 2008b), and one or more distal fans (Figs. 1 and 2). Thehyporheic zone is the area marginal to and beneath a stream thatexchanges water with the stream channel. Within and adjacent tomost gullies, dry, ice-free sediment overlies sediment that is ce-mented by pore ice. The lower depth of this pore ice is unknown,but its surface, called the “ice-cement table,” is fairly uniform andoccurs on average at about 15–20 cm depth (Bockheim et al., 2007;Levy et al., 2007b). Typically, the ice-cement table deepens withincreasing distance from isolated snow banks and gully chan-nels.
In ADV areas with extensive pore ice, the ground commonlyshows well-developed thermal contraction crack polygons (Bergand Black, 1966). All gullies save one observed in the Wright Valleystudy site are present on polygonally-patterned slopes (Levy et al.,2008b; Morgan et al., 2008). Across the ADV, active and recentlyactive gullies are typically present in association with contraction-crack polygons; relict gullies in the coldest and driest portion ofthe ADV that have been inactive for up to 10 My (Lewis et al.,2007) typically lack polygons characteristic of the Wright Valleysite. The most common polygons present in the South Fork area arecomposite-wedge polygons (Levy et al., 2008b). Composite-wedgepolygons are those in which alternating layers of sand and ice fillthermal contraction cracks (Berg and Black, 1966). Importantly, ar-eas in the Dry Valleys that lack pore ice within the upper ∼1 mof soils tend to lack all varieties of thermal contraction crack poly-gons (Marchant and Head, 2007).
2.1.1. ADV gully water sourcesSoil-temperature measurements indicate that melting along the
ice-cement table in the inland mixed zone is uncommon, and
Geologically recent gully–polygon systems on Mars and Earth 115
Fig. 2. Summary of key observations from gully–polygon systems in the South Fork of upper Wright Valley, Antarctica. (a) Polygon troughs accumulate wind-blown snowbanksthat contribute meltwater to gully flow. (b) Trough excavated through dry colluvium across a downslope-oriented polygon trough. The ice-cement table is depressed alongpolygon troughs, channelizing flow over the ice-cement table. Snow-derived meltwater moves down-slope (from left to right) along the top of impermeable ice-cement table.(c) Stratigraphic relationships between an Antarctic gully and surrounding polygons. Black arrows indicate embayment of surrounding polygons by gully fan material. The fanis dissected by underlying polygons. A white arrow indicates a polygon trough that has been annexed by the channel. The fan is ∼100 m wide.
is not a significant source for surface meltwater (Marchant andHead, 2007; Morgan et al., 2007b). Rather, surface water arisesfrom insolation-driven snowmelt (Head et al., 2007; Morgan etal., 2007a; Levy et al., 2007b). Snowbanks accumulate in poly-gon troughs and in gully channels through seasonal capture andpreservation of windblown snow (Dickson et al., 2007b; Head etal., 2007; Levy et al., 2007b; Morgan et al., 2007a). A compara-ble volume of snow to that stored in gully channels can be storedin polygon troughs adjacent to gully channels, and in the polygontroughs present in gully alcoves (Levy et al., 2008b). Large snow-banks in gully channels and polygon troughs endure for weeksdespite high rates of summertime sublimation (see Kowalewskiet al., 2006). Southern hemisphere peak-summer daytime insola-tion causes melting of snowbanks that produces ephemeral waterflow capable of eroding and redistributing sediments (Dickson etal., 2007b; Head et al., 2007; Levy et al., 2007b; Morgan et al.,2007a).
Walking surveys at the field site were conducted over ∼3 km ofpolygonally-patterned valley wall, rising from the valley floor to el-evations of ∼800 m, in order to ascertain the presence or absenceof deep groundwater sources for gullies. Inspection of five gully–polygon systems in the field resulted in no observations of over-land flow associated with springs emanating from faulted bedrockexposed at the surface or of high-pressure scouring of sediment orbedrock due to catastrophic release from confined aquifers.
In the South Fork study area of upper Wright Valley, gully–polygon systems are overwhelmingly present on north-facing(equator-facing) valley slopes (Morgan et al., 2007b). This distri-bution reflects enhanced peak-summer warming and subsequentmelting of wind-blown snow and perennial snowbanks on warm,equator-facing slopes; shadowed, pole-facing slopes accumulateless abundant snowbanks and produce minimal volumes of sum-mer meltwater (Morgan et al., 2008).
These observations of gully water sources in the ADV establishtwo important processes to gully hydrology in permafrost environ-ments. First, networks of polygon troughs can accumulate melt-water derived from broadly distributed melt source (e.g., polygon-trough snowbanks) into a concentrated, channelized flow. Withoutthe presence of polygons, snowbanks would have fewer valley wallaccumulation sites, and any snowbank meltwater would simplypercolate downslope. Second, all the water present in the Antarctic
gully systems analyzed is derived from sources lying above the im-permeable permafrost ice-cement table—there is no deep aquifercomponent to these cold desert gullies.
2.1.2. ADV gully–polygon morphological relationships: Alcoves,channels, and fans
Large alcoves in the Antarctic Dry Valleys commonly form indolerite bedrock cliffs (Fig. 1, inset; Head et al., 2007; Morgan etal., 2007b). These alcoves have little to no polygonal patterning dueto thin to non-existent sediment cover (Levy et al., 2007b). Largealcoves present in the ADV span ∼100–400 m in width, and can beup to ∼400 m long, with aspect ratios of close to 1. Below largebedrock alcoves, small concavities are present on valley walls inthe ADV (Fig. 1; Dickson et al., 2007b; Head et al., 2007; Levy etal., 2007b; Morgan et al., 2007a). Concavities may form by erosionof colluvium by braided channels or as typical nivation hollows.Concavities exhibit a dessicated, near-surface sediment layer overice-cemented debris (Fig. 1). The concavity located in the studyarea is ∼290 m long, ∼150 m wide, and has an aspect ratio of∼1.6. Channels cut the concavity from upslope and emanate fromit. Bank erosion of braided channels is intense within and aroundthe concavity. Composite-wedge polygons occur within concavitiesas well as in adjacent colluvium, and are commonly ∼16 m indiameter, spanning 12–24 m in diameter, with a standard deviationof ∼2.8 m (Fig. 1; Levy et al., 2008b).
Surface-generated meltwater follows local topography and maybe captured in polygon troughs. Where flow is concentrated inpolygon troughs, erosion of trough/wedge sediments is enhanced(Fig. 2c), a process termed “trough annexation” (e.g., Levy et al.,2008b). Annexed polygon troughs are generally wider and deeperthan unaffected troughs, and typically have rounded trough in-tersections (in contrast to angular intersections between pristinepolygon troughs; Levy et al., 2008b). Annexed polygon troughscommonly have sandy floors, composed of layers of bedded andcross-bedded sediment (Levy et al., 2008b). Water flow directly ob-served within annexed troughs is restricted to the ground surfaceand to the shallow subsurface (Fig. 2b); the impermeable bound-ary at the top of the ice-cement table (∼10–20 cm depth) pre-vents deeper meltwater infiltration. Meltwater that drains to theice-cement table may flow downslope along its surface for tensof meters before emerging as overland flow (Levy et al., 2008b).
116 J.S. Levy et al. / Icarus 201 (2009) 113–126
Over some reaches of the gullies, meltwater derived from snow-bank melting was observed to flow several meters over the surface,before infiltrating into the colluvium, only to emerge from thesediment and resume overland flow several meters downstream(usually at small breaks in slope; Levy et al., 2007b).
At the base of ADV valley walls, fan deposits overprint andembay active composite-wedge polygons. Some fan material is de-posited in topographic lows (troughs) between high polygon cen-ters, suggesting that polygon troughs provided a topographic bar-rier to fan emplacement (Fig. 2). Other polygons are completelycovered by fan material. These stratigraphic relationships (Fig. 2c)indicate that some fan material was deposited over existing poly-gons. Continued polygon development results from elevation of theice-cement table through the fan, and enables contraction cracksto propagate upward, dissecting overlying fan material. This pro-cess is analogous to the formation of syngenetic contraction-crackpolygons first described by MacKay (1990) and also by Levy et al.(2008b).
In summary, these observations are interpreted to indicate thatWright Valley gully development is strongly influenced by thepresence of polygonally patterned ground. The presence of polyg-onally patterned ground does not directly cause the formation ofgullies, however, polygons enhance the accumulation of snow thatfeeds gully flow, concentrate and direct the flow of gully melt-water, and modify depositional fans. Stratigraphic relationships be-tween polygons and gully fans indicate that the Wright Valley gul-lies studied formed on surfaces continuously underlain by meters-thick, ice-cemented, impermeable permafrost, effectively removingthe possibility of groundwater contributions to gully flow.
3. Distribution of gully–polygon systems on Mars
Prompted by the developmental relationship between ADV gul-lies and polygons, we undertook a comprehensive survey of MRO-HiRISE images in order to assess morphological and stratigraphicrelationships between martian gullies and polygons. A surveyof HiRISE primary science phase images of the martian surface(McEwen et al., 2007), spanning orbits PSP_001330 to PSP_007207,and ranging between 30–80◦ north and south latitude (a regionknown for concentrations of gullies, e.g., Milliken et al., 2003;Balme et al., 2006; Dickson et al., 2007a) forms the basis for thisanalysis. After selecting images based on latitude, images spanningorbits 001330–003824 were analyzed sequentially by orbit number(∼530 images). Next, a subset of images from orbits 003825–007207 were selected on the basis of geographical location withinthe 30–80◦ latitude bands in order to increase the density of ana-lyzed images in locations sparsely sampled in early orbits. A senseof the magnitude of the dataset, and the degree of morphologi-cal detail present in each image is achieved by considering that,a typical ∼1 GB HiRISE image contains approximately 200 timesmore information than a typical 5 MB Mars Orbiter Camera image,resulting from increased spatial resolution within a comparablesurface footprint. Of the 722 images studied, 168 contain gullies,and 93 contain gullies present on clearly polygonally patternedsurfaces (Figs. 3 and 4; Supplementary data). Features present inHiRISE images were classified as gullies if they were composed ofat least two of the three gully structural elements defined by Malinand Edgett (2000), namely: a recessed alcove, a sinuous channel,and a distal fan or apron.
Geographically, gullies are predominantly observed in HiRISEimages in the southern hemisphere (125 southern hemisphere oc-currences compared to 43 northern hemisphere occurrences), asare gully–polygon systems (71 southern hemisphere occurrencescompared to 22 northern hemisphere occurrences). We partiallycorrect for targeting bias by dividing the number of occurrencesof gullies and gully–polygon systems by the number of survey
images in the latitude bands in which the features are present.Gullies are present in 48% of survey images between 30–55◦ S,and in 21% of survey images between 30–55◦ N; gully–polygonsystems are present in 26% of survey images between 30–55◦ S,and in 9% of survey images between 35–55◦ N. These distributions(Fig. 4) are consistent with rougher topography in the southernhighlands (e.g., Neumann et al., 2003) providing steep-sloped sur-faces important for gully formation (Kreslavsky and Head, 2002;Dickson et al., 2007a; Kreslavsky, 2008).
The distribution of gullies and polygons between 30◦–80◦ lat-itude strongly correlates with the distribution of dissected andcontinuous latitude-dependent mantle terrain, a meters thick dustand ice-rich deposit interpreted to have been emplaced during re-cent ice ages caused by spin-axis obliquity excursions (Mustard etal., 2001; Head et al., 2003; Fig. 4). Polygonally patterned groundis present on both the continuous and dissected mantles, whilegullies are concentrated in dissected mantle terrains. These dis-tributions suggest that polygons form over a wide range of zonalclimate conditions, with and without gullies. In contrast, withinthe limited latitude range of gullies, the abundance of images con-taining gullies interacting with polygons suggests that gullies maypreferentially form on polygonally patterned surfaces. Using in-sight from the ADV, gullies and polygons may be landforms whichcan be used to interpret the range of cold-desert geomorphologicalprocesses that have modified latitude dependent mantles on Mars.Where morphological similarities exist between spatially associ-ated gullies and polygons on Mars and ADV gully–polygon systems,we suggest that these features may have formed by analogous pro-cesses.
4. Morphological relationships between gully–polygon systemson Mars
4.1. Alcoves and polygons
The largest gully alcoves observed in this survey, character-ized by lengths of ∼1000 m and widths in excess of ∼500 m,are localized in a latitude band between ∼40–50◦ S and areless common elsewhere. These alcoves (Fig. 5) are triangular inshape and have aspect ratios of <1 to ∼3, comparable to al-coves described in previous surveys (e.g., Malin and Edgett, 2000;Dickson et al., 2007a). These large alcoves uniformly lack poly-gons, particularly on steep slopes where mantle material has beeneroded, exposing bedrock (Fig. 5).
Alcoves with polygonally patterned surfaces are commonlyelongate and have a rectangular shape (Fig. 6), comparable tothe “lengthened alcoves” of Malin and Edgett (2000). Elongate al-coves mapped in this study average ∼820 m in length (n = 29,minimum = 250 m, maximum = 2200 m) and have high length-to-width aspect ratios (mean = 6, minimum = 4, maximum = 12).Elongate alcoves form within a surficial mantling unit (Fig. 6),and do not generally expose underlying bedrock or crater-fracturedmaterial and boulders (e.g., Fig. 5). One or more channels are com-monly present within these elongate alcoves (Fig. 6), and emanateout from the alcoves.
Thermal contraction crack polygons commonly form in the ice-rich sediments of the martian latitude-dependent mantle (Mustardet al., 2001; Milliken et al., 2003; Mangold, 2005) or pasted-on terrain within or adjacent to gully alcoves (Costard et al.,2002; Christensen, 2003; see Fig. 4: most HiRISE images in man-tled terrains polewards of 40◦ feature polygons). Pasted-on ter-rain is a relatively smooth-surfaced unit typical of martian mid-latitudes, that is commonly superposed on pole-facing surfacesand is thought to have formed by atmospheric deposition ofice and/or dust (Malin and Edgett, 2001; Mustard et al., 2001;Christensen, 2003). Boulders present on some pasted-on terrain
Geologically recent gully–polygon systems on Mars and Earth 117
Fig. 3. Distribution of polygonally patterned ground, gullies, and gully–polygon systems mapped using HiRISE images. Small black dots indicate HiRISE images which donot contain gullies or polygons. (Top) HiRISE images containing polygonally patterned ground (triangles). (Middle) HiRISE images featuring gullies (circles). (Bottom) HiRISEimages with gully–polygon systems (circles with black and white fill). Gully–polygon systems tend to occur in the region between regions with gullies and regions whichhave polygonally patterned ground.
outcrops have been interpreted to indicate a rock-glacier origin forpasted-on terrain (e.g., McEwen et al., 2007); however, the wast-ing of fractured crater-rim materials located upslope from pasted-on terrain may also account for the presence of boulders atoppasted-on surfaces. Polygons present in pasted-on terrain and onmantle surfaces are commonly flat-topped, with elevated interi-ors and depressed troughs. This morphology is consistent withsand-wedge polygon or sublimation-polygon structures that formpreferentially in fine-grained and ice-rich substrates (Lachenbruch,1962; Washburn, 1973; Maloof et al., 2002; Marchant et al., 2002;Marchant and Head, 2007). Analysis of 136 alcove polygons onMars, in 8 HiRISE images, indicates a mean martian alcove polygondiameter of ∼11 m, spanning ∼5–21 m, with a standard deviationof 3.4 m.
Some martian alcove polygons are outlined by bright depositsthat are present preferentially in polygon troughs (Fig. 7). “Bright”
indicates pixel DN values in processed HiRISE images that areseveral times higher than proximal pixels sampled from polygoncenters or gully channels. These deposits may be water-ice de-posited seasonally as frost (for images taken during winter periods;Mangold, 2005), salt deposits (Burt and Knauth, 2007), dusty lagdeposits (Williams et al., 2008), or some other form of high-albedo,particulate deposit, such as snow, that accumulates preferentiallyin shielded topographic lows (Head et al., 2008). In some im-ages, bright material is distributed broadly over surfaces containinggully–polygon systems (Figs. 7a–7b); in others bright material ispresent in polygon troughs within gully alcoves (e.g., Fig. 7c). Dustcover is not pronounced in the analyzed images (e.g., dust ripplesare uncommon and boulders are clearly visible), and no salts havebeen spectroscopically detected in the examined HiRISE images(e.g., Osterloo et al., 2008). Rather, these deposits are seasonallypresent, and are commonly blue-toned in HiRISE color data: ob-
118 J.S. Levy et al. / Icarus 201 (2009) 113–126
Fig. 4. Histogram of gully, polygon, and gully–polygon system distribution by latitude. The number of feature-containing images in each latitude band has been normalizedto the total number of HiRISE images in the latitude band in order to remove bias in spatial coverage (thus, a normalized value of 0.5 indicates that half the HiRISEimages surveyed in a latitude band contain the feature plotted). Gullies and gully–polygon systems are found primarily in latitudes where dissected mantle terrain is present(Mustard et al., 2001; Head et al., 2003), while polygonally patterned ground spans the continuous and dissected mantles.
Fig. 5. Large, triangular alcoves. Layered outcrops interpreted to be exposed crater wall bedrock surfaces are visible within the alcoves (arrows). (a) Portion ofPSP_002368_1275, located at 52◦ S, 247◦ E, on a crater wall. Ls = 174.0◦: southern winter. (b) Portion of PSP_002054_1325, located at 47◦ S, 177◦ E, on a crater wall.Ls = 160.7◦: southern winter. (c) Portion of PSP_001882_1410, located at 39◦ S, 194◦ E, on a crater wall. Ls = 153.7◦: southern winter.
servations consistent with the seasonal deposition of water ice(Gulick et al., 2008). On the basis of these observations, we in-terpret bright, trough-filling material present in or around alcoves,that has been imaged during winter or early spring time periods,to be atmospherically-emplaced frost or, possibly, particulate ice.
4.2. Channels and polygons
Channel-like features were observed in analyzed HiRISE imagesthat are (1) continuous and sub-linear; (2) present in widened,curved, and down-slope-oriented polygon troughs; and (3) present
Geologically recent gully–polygon systems on Mars and Earth 119
Fig. 6. Elongate alcoves with thermal contraction crack polygons. Elongate alcoves commonly have a length-to-width aspect ratio of 6 or greater. (a) Portion ofPSP_001846_1415, located at 38◦ N, 97◦ E, on a crater wall. (b) Portion of PSP_001882_1410 located at 39◦ S, 194◦ E, on a crater wall. Ls = 153.7◦: southern winter. (c)Portion of PSP_001357_2200, located at 40◦ N, 105◦ E. Illumination is from the left in all images.
nearby to typical gully channels (Fig. 8). These features are presentindividually and in braided groups on polygon-surfaced slopes, andcan be distinguished from typical polygon troughs by variationsin surface texture, relief, and continuity (Fig. 9). These channelsare commonly ∼200–500 m in length. Deposits in these linearfeatures are light-toned in HiRISE red-filter images (using DN-comparison of unstretched images), and are distinguished fromblue-toned bright deposits present in polygon troughs in alcovesby a difference in texture (trough-channel deposits are slightly rip-pled with possible small boulders present), a lower albedo, anddifferent color. Given their morphological similarity to terrestrialgully channels that have formed through the annexation of pre-existing polygon troughs (Levy et al., 2008b), we interpret thesemartian features to be remnants of polygon troughs annexed byincipient gully channels. We interpret in-trough deposits to be flu-vially deposited sediments, similar to fan-forming sediments, de-posited during periods of channelized gully flow.
4.3. Fans and polygons
Several gully fans observed in HiRISE images embay polygo-nally patterned ground (Fig. 10), consistent with observations offans overprinting polygons in MOC image analyses (e.g., Malinand Edgett, 2000; Heldmann et al., 2007). Some fans terminateabruptly at relatively deep polygon troughs. Other fans appear towrap around elevated polygon centers and elevated outcrops ofpatterned ground (Fig. 10a). These relationships suggest that poly-gon topography provided a barrier to fan emplacement.
Small gully fans on polygonally patterned slopes range in sur-face area from 1 × 104–2 × 105 m2. These fans have little topo-graphic relief and appear to be thin, surficial deposits (Fig. 10).
Polygon troughs are visible through small fan surfaces. Thesepolygon troughs are typically continuous with, and are exten-sions of, surrounding trough networks (Fig. 10). These observationssuggest that polygons have remained active through fan deposi-tion, and have winnowed fan sediments into underlying polygonwedges.
In contrast to small-scale fans, several large fans were observedin HiRISE images that show different morphological and strati-graphic relationships with polygons (Fig. 10c). Large fans havesurface areas spanning 2.8 × 105–1.1 × 106 m2 (one to two or-ders of magnitude larger than small fans described above). Largefans generally have significant topographic relief, and rise convexlyup from inter-fan slopes. Typically, these large fans lack modi-fication by surface polygons, though some are characterized byan array of fine-scale fractures that superficially resemble smallpolygon troughs. These fine fractures are not continuous withtroughs from adjacent polygon networks (Fig. 10c). These obser-vations suggest that large fans have buried surrounding polygonnetworks.
Light-toned material can be seen displaced from the fan insome HiRISE images, suggesting partial redistribution of the finefraction of fan-forming sediments by subsequent aeolian processes(Figs. 10a and 10b). The continued presence of bright fan ma-terial in polygon centers, rather than preferential redistributionof fan material into polygon troughs, suggests that much of thepolygon-interior fan material has been preserved in place, despitewinnowing of fan sediments into polygon troughs and aeolian ero-sion of fines.
Lastly, the majority of polygons present on gully fans, andon surfaces topographically lower than the fans, do not displaymorphologies characteristic of seasonal saturation of sediments
120 J.S. Levy et al. / Icarus 201 (2009) 113–126
Fig. 7. Bright material present in polygon troughs (white arrow) in proximity to alcoves and channels (black arrows). Images were taken during winter/spring in all cases, sug-gesting bright material is frost, ice, or snow. Downslope is towards image bottom in all images. (a) Portion of PSP_003920_1095, located at 70◦ S, 2◦ E. Ls = 246.8◦: southernspring. Illumination from left. (b) Portion of PSP_3511_1115, located at 69◦ S, 1◦ E. Ls = 226.7◦: southern spring. Illumination from above. (c) Portion of PSP_002165_1270,located at 53◦ S, 28◦ E. Ls = 165.3◦: southern winter. Illumination from left.
by gully flow (e.g., Lyons et al., 2005; Levy et al., 2008b), suchas concentration of boulders at the surface (heaving), sorting ofsediments through cryoturbation (which might be detectable aschanges in surface brightness or texture), or the formation of ice-wedge polygons with upturned shoulders. These observations sug-gest that water involved in the transport and deposition of fansediments rapidly froze-on to the ice-cement table within the fanand/or sublimated until local equilibrium conditions were met forwater stability.
In summary, we interpret these overprinting and cross-cuttingrelationships to indicate the following formational sequence. Small-scale gully fans formed by deposition of sediments over previouslyexisting polygonally-patterned ground (consistent with MOC obser-vations, e.g., Malin and Edgett, 2000; Heldmann et al., 2007), andcrack expansion continued throughout and after fan deposition,dissecting gully fans from beneath. This implies the continuouspresence of ice-cemented permafrost beneath gully fans duringtheir development and aggradation of permafrost concurrent withthe growth of gully fans. Larger fans formed from the emplace-ment of sediment at a rate that resulted in the burial of previouslyextant polygon networks, resulting in polygon development limitedto fine-scale networks that are discontinuous with the surroundingpolygonal network.
4.4. Slope orientation
Although preliminary reports conflicted on the presence or ab-sence of orientation preferences for gullies at the hemisphere scale(e.g., Malin and Edgett, 2000; Edgett et al., 2003), binning of gullyorientations by latitude by Heldmann and Mellon (2004) discov-ered a latitude-dependence for the orientation of gullies in the
southern hemisphere. Gullies between 30–44◦ S predominantlyface polewards and gullies between 45–60◦ S generally face equa-torwards (Heldmann and Mellon, 2004). This observation was ver-ified in the Newton Crater region by Berman et al. (2005). Dicksonet al. (2007a) further confirm these observations, finding that∼86% of gullies in the 30–45◦ S latitude band occur on pole-facingslopes, and noting that the few gullies mapped on equator-facingslopes are confined to above ∼40◦ S. One interpretation of theorientation data is that gullies form on protected slopes wheresnow/ice, if available, would tend to accumulate (Hecht, 2002;Dickson et al., 2007a; Head and Marchant, 2008; Head et al., 2008),and where protected ice reservoirs could be rapidly exposed topeak insolation, leading to melting. Hecht (2002) demonstratedthat peak insolation sufficient to cause melting can be achievedon either pole- or equator-facing slopes on Mars, depending onlatitude and slope inclination.
In some HiRISE images gullies on polygonally-patterned sur-faces can be observed on both pole-facing and equator-facingslopes (Fig. 11). “Pole-facing” and “equator-facing” are qualitativemeasurements of orientation, indicating that gully–polygon sys-tems were present on slopes oriented within ∼30◦ of north orsouth. These occur most commonly on interior crater walls. OnlyHiRISE images in which gully–polygon systems are present onnear-diametrically opposite slopes (> ∼150◦ angular separation)were included in orientation analyses.
The morphology of gullies and polygons on Mars differs withslope orientation (Fig. 11). Gullies on pole-facing slopes generallyhave sharply-defined channels and fans, and polygons on pole-facing slopes are crisply delineated. On equator-facing slopes (im-aged at the same resolution) gullies and polygons have subdued
Geologically recent gully–polygon systems on Mars and Earth 121
Fig. 8. Features interpreted to be polygon troughs annexed by gully channels. Annexed troughs are continuous and sub-linear, are present on polygonally patterned surfacesin widened and curved polygon troughs, and are preferentially filled with bright deposits. Upslope is toward image top in all panels, and annexed troughs are orienteddownslope. Arrows indicate some of the annexed polygon troughs. (a) Portion of PSP_001846_2390, located at 59◦ N, 82◦ E, on a crater wall. (b) Portion of PSP_001846_2390,adjacent to part a, located at 59◦ N, 82◦ E, on a crater wall. (c) Portion of PSP_001508_2400, located at 60◦ N, 302◦ E, on a crater wall. (d) Portion of PSP_001938_2265,located at 46◦ N, 92◦ E, in Utopia Planitia. Gully features are present in a large, scalloped depression.
Fig. 9. Braided annexed polygon troughs from PSP_001548_2380, located at 58◦ N, 292◦ E. (a) Small annexed troughs present within crater-slope-oriented polygons on theupper slope of a crater rim. Illumination is from the left. Boulders of various sizes are visible. (b) Larger braided channels adjacent to small channels shown in part a (box).
and softened morphologies (Fig. 11). Gully channels on pole-facingslopes tend to be narrower and are bounded by steeper walls thanthose on equator-facing slopes. In addition, the surface texture offans on equator-facing slopes is commonly indistinguishable fromthat of inter-fan surfaces. Lastly, polygons on equator-facing slopes
generally lack sharp trough boundaries, and commonly grade fromnetworks of raised mounds surrounded by low troughs, to irreg-ular, linear albedo patterns interpreted to be incomplete and de-graded crack networks. We interpret the softening of gully–polygonsystem morphologies on equator-facing slopes to indicate removal
122 J.S. Levy et al. / Icarus 201 (2009) 113–126
Fig. 10. Stratigraphic relationships between fans and polygons. (a–b) Small, thin fans, with little topographic relief, are cut by underlying thermal contraction crack polygons.Dissecting polygons are continuous with polygons present on inter-fan surfaces. (c) A large, convex-up fan is subtly patterned with a polygon network discontinuous with theinter-fan network (inset showing white boxed region, arrows highlight two of many polygon troughs). (a) Portion of PSP_001846_2390, located at 59◦ N, 82◦ E. (b) Portionof PSP_001548_2380, located at 58◦ N, 292◦ E. (c) Portion of PSP_002368_1275, located at 52◦ S, 247◦ E.
of near-surface, permafrost-cementing ice by sublimation subse-quent to gully–polygon system formation.
5. Discussion
Analysis of gully–polygon systems on Earth suggests the follow-ing stratigraphic and temporal relationships within gully–polygonsystems: (1) polygons pre-date alcove formation; (2) polygontroughs have been annexed by some gully channels, indicatingoverland flow and channel development on polygonally patternedsurfaces; (3) many fans formed on a polygonally-patterned sur-face; and (4) polygon development has continued during fanaggradation. Identical stratigraphic and temporal relationships areobserved between gullies and polygons on Mars. These resultsfrom HiRISE are consistent with stratigraphic interpretations madeusing MOC image data (e.g., Malin and Edgett, 2000; Costardet al., 2002; Christensen, 2003; Heldmann and Mellon, 2004;Berman et al., 2005; Balme et al., 2006; Dickson et al., 2007a;Head et al., 2008), but provide an unprecedented view of the de-tailed morphological relationships between martian gullies andpolygons. Over half of the gullies imaged in this survey interactwith underlying polygonally patterned ground, showing evidenceof polygon-influenced water ice accumulation, polygonal pattern-ing of elongate alcoves, annexation of polygon troughs by gullychannels, and dissection of fans by underlying polygons. These re-ciprocal changes in gully and polygon morphology suggest a linkeddevelopmental history for martian gullies and polygons, analogousto that observed in terrestrial gully–polygon systems. Accordingly,we interpret these spatially-linked landforms observed in HiRISEimages to be gully–polygon systems. Thus, understanding the ef-fects of polygonally patterned permafrost on gully development on
Earth may be important for understanding the hydrological andmicroclimatological processes involved in gully formation on Mars.
On the basis of our observations in the Antarctic Dry Val-leys and on Mars, we propose the following model for the initi-ation and evolution of martian gully–polygon systems describedin this survey (Fig. 12). On Mars, a climate-related, latitude-dependent, ice-rich mantling unit composed of atmospheric dust,ice, and ice-cemented regolith is deposited regionally above 30◦latitude (Mustard et al., 2001; Hecht, 2002; Christensen, 2003;Head et al., 2003; Milliken et al., 2003). This mantling unit ispreferentially preserved in sheltered environments on steep slopesand at low elevations (<3 km above the datum; Hecht, 2002;Dickson et al., 2007a), and likely has an ice-free sublimation lagat its surface (Mustard et al., 2001; Hecht, 2002; Head et al., 2003;Williams et al., 2008). This mantling unit may be analogous to theunconsolidated debris layer overlying ice-cemented colluvium inthe ADV. However, the mantle differs in that its primary mode ofemplacement is atmospheric deposition, rather than typical collu-vial transport.
Next, thermal cycling generates polygons (Mellon, 1997) that,owing to relatively dry climatic conditions, may be analogousto sand-wedge and/or composite-wedge polygons (Mellon, 1997;Mangold et al., 2004; Mangold, 2005; Levy et al., 2008b; Marchantand Head, 2007). Polygons disturb mantling sediments at troughlocations. Although calculations of thermal wave propagation onMars have demonstrated the potential for wet active layers dur-ing geologically recent time (Kreslavsky et al., 2007), the lack ofextensively water-related polygon structures in these units (suchas markedly raised polygon shoulders), coupled with a lack ofsolifluction features, suggests that near-surface warming did notresult in significant melting of buried ice-cemented permafrost.
Geologically recent gully–polygon systems on Mars and Earth 123
Fig. 11. Orientation dependence of morphology in gully–polygon systems. Image pairs (a and b, c and d) are portions of the same HiRISE image, with identical resolution,illumination, and signal-to-noise conditions. (a) Sharply-defined gully–polygon systems on a pole-facing slope. Portion of PSP_001846_2390, located at 59◦ N, 82◦ E, on acrater wall. Illumination is from the right. (b) Softened gully features with little to no polygonal patterning present on the crater wall opposite the gullies shown in part (a).Illumination is from the left. (c) Sharply-defined gully–polygon systems on a pole-facing slope. Elongate alcoves between polygon-covered topographic spurs are present.From PSP_001357_2200, located at 40◦ N, 105◦ E. Illumination is from the right. (d) Smoothed gullies with sparse polygonal patterning located on crater rim opposite gulliesshown in part (c). Topographic spurs are present, but lack mantling material. Illumination is from the left.
Fig. 12. Schematic diagram of a polygon-influenced model for gully initiation and evolution. (a) Initial topography is generated; in this case, the inside wall of a crater. Thepole-facing slope is illustrated. (b) Thermal contraction crack polygons form in ice-cemented regolith and sediments on the crater wall. (c) Accumulation of atmosphericallydeposited frost or wind-blown snow occurs in sheltered polygon troughs. Localized melting of this ice is channelized by polygon troughs, creating small-scale annexedpolygon trough channels that produce few to no distal fans. Small fans are readily cut by continuing thermal contraction cracking. (d) Ice deposition and localized meltingcontinues in sheltered polygon troughs, resulting in the growth of annexed channels and the braiding of nearby channels into anastomosing groups. Growing distal fans arecut by expansion of underlying thermal contraction cracks. (e) Continued snow and ice deposition and localized melting in braided annexed channels erodes inter-channelwalls, creating an elongate alcove with one or more internal channels. The distal fan is still thin enough to permit dissection by polygon growth. (f) Continued erosion ofcrater-mantling sediments exposes crater wall surfaces and results in rapid alcove erosion by water/ice-assisted mass-wasting. The rapid deposition of poorly sorted andlarge-grain-size sediments overwhelms underlying thermal contraction cracks, burying the original polygon network.
124 J.S. Levy et al. / Icarus 201 (2009) 113–126
Subsequently, atmospherically-derived water is introduced tothe gully–polygon system. Topographically-depressed polygontroughs are shaded environments that could cold-trap atmosphericwater frost (Hecht, 2002) and/or act as topographic obstacles,concentrating wind-blown particulate ice (Head et al., 2008)—expanding the extent and thickness of seasonal frost accumulationsimaged in this survey. Under appropriate obliquity- and slope-dependent peak insolation conditions (Hecht, 2002; Kuzmin, 2005;Levy et al., 2007a; Morgan et al., 2008) this ice, concentrated inpolygon troughs, could melt to produce short-lived, ephemeral, liq-uid water. Peak insolation conditions occur when the solar angle isnormal to the surface slope. Gully–polygon systems commonly oc-cur on steep, ∼30◦ slopes (Dickson et al., 2007a), making themstrong candidates for recent surface melting of atmosphericallyemplaced volatiles (Hecht, 2002). Crater-retention age dating of re-cent gully deposits similar to those included in this study indicatesgully activity within the past 1–2 Ma, although flows may have oc-curred as recently as 300 ka (Riess et al., 2004; Schon et al., 2009).
As in the Antarctic Dry Valleys, martian polygon troughs appearto concentrate and direct the transport of metastable surface- andnear-surface, meltwater, creating annexed polygon troughs (Levyet al., 2008b). Short-lived water transport in annexed troughs(Figs. 8 and 9) could easily transport unconsolidated polygontrough and wedge sediments, contributing to sediment depositionin small terminal fans. Over time, annexed trough channels wouldconverge towards trunk channels (Fig. 9b, compare left and right),forming braided annexed channels. Braided channels would pro-duce larger fans than those formed from isolated annexed troughsby collecting sediment from several separate channels, and by in-creasing the volume of water available to move sediment. Thesefan deposits would still be relatively thin, and would be easily cutor dissected by the continued growth of underlying polygon cracks(Fig. 10b).
Continued erosion within annexed troughs would provide anincreasingly large sheltered environment for accumulation andsubsequent melting of ice and windblown snow (Figs. 6 and 9b).Eventually, braided channel walls would be widened, creating anelongated alcove with one or more incised channels (Fig. 6). Ma-terial eroded from elevated alcoves would be deposited in a distalfan, which could remain thin enough to permit continuing dissec-tion by thermal contraction crack expansion. Elongate alcoves areanalogous to gully-related concavities or nivation hollows in theADV (Figs. 1 (inset) and 2b). The presence of polygons in elon-gate alcoves, but not in widened alcoves (below) suggests thatsome ice-rich latitude-dependent mantle material remains intactin these alcoves.
This process would continue for as long as climate conditionsremained capable of generating liquid water or brines that re-mained metastable long enough to flow (e.g., Mellon and Phillips,2001; Hecht, 2002; Costard et al., 2002; Christensen, 2003;Kreslavsky and Head, 2007; Burt et al., 2008). The longevity ofbriny fluids on Mars is strongly dependent on solute depres-sion of the freezing temperature (potentially supporting liquidflow at temperatures as low as −20 to −50 ◦C, depending onsalt chemistry and concentration; Burt and Knauth, 2007) andreduction in evaporation rate to support persistent fluvial activ-ity (potentially as slow as 0.04 mm/h at −25 ◦C; Ingersoll, 1970;Sears and Chittenden, 2005). Eventually, alcove mantle materialwould be fully eroded, exposing original scarp/crater-wall surfaces(Fig. 5). Exposed crater and mantle material in these large, steepalcoves might fail in response to gravitational sliding, as well as inresponse to surficial fluvial erosion, producing large fans beneathleveed channels (Morgan et al., 2007b). Extensive fan depositionmay bury polygons, cutting off underlying thermal contractioncracks from the fan surface exposed to seasonal thermal cycling,and leading to the generation of a network of new polygons (fine
fractures present on some large fan surfaces that are discontinu-ous with the polygon network surrounding the fan). Other largefans lack any polygonal patterning, suggesting that fan emplace-ment may have been rapid enough to prevent the formation ofsyngenetic polygons (MacKay, 1990).
As climate conditions became colder and drier (Forget et al.,2007), fluvial and erosive processes would decrease and eventuallycease in the gullies. In the absence of gully flow and infiltration re-freshing buried ice-cemented permafrost, enhanced sublimation onequator-facing slopes would desiccate shallow permafrost, reduc-ing sediment cohesion, and ultimately resulting in subdued gullyand polygon textures in response to aeolian erosion. In degradedgully–polygon systems (e.g., Fig. 11) polygons are lost from viewbefore gullies (owing to the larger size of gullies) suggesting thatpolygons may have interacted with gullies even more commonlythan is currently observed.
In summary, this model provides a mechanism for the devel-opment of martian gullies that occur in association with polygo-nally patterned ground. Morphological similarities between mar-tian gully–polygon systems and the closest morphological and cli-matological analog on Earth (gully–polygon systems in the ADV),suggest that the martian examples may have formed and devel-oped on slopes underlain by ice-cemented permafrost. In bothcases, a top–down source for gully-carving water is implied, asgeomorphological evidence suggests limited melting of the under-lying ice-cemented substrate.
6. Conclusions
Observations of gullies and polygons from Antarctic field workand analysis of HiRISE image data suggests the following strati-graphic and temporal relationships between gullies and polygons:(1) polygons pre-date alcove excavation in some gullies; (2) poly-gon troughs form traps for ice and windblown snow that can be-come sources of meltwater for gullies; (3) polygon troughs havebeen annexed and eroded by some channels, indicating that chan-nel formation occurred on a polygonally patterned surface; (4) fanembayment and dissection relationships indicate that some fansformed on polygonally patterned surfaces; and (5) polygon devel-opment continued during fan emplacement. Using morphologicallysimilar gully–polygon systems in the ADV as a guide, the strati-graphic relationships between gullies and polygons observed inthe HiRISE images suggest that the martian gullies analyzed inthis study developed on slopes underlain by polygonally-patterned,ice-cemented permafrost. Interactions between martian gullies andpolygons are analogous to those documented in ADV gully–polygonsystems.
No evidence was seen for significant melting of underlying ice-cemented permafrost on Earth or Mars. Additionally, no evidenceof subsurface groundwater release (e.g., Malin and Edgett, 2000;Heldmann and Mellon, 2004; Heldmann et al., 2007) from beneaththe ice table was observed at HiRISE resolution. No paired aqua-cludes or intensive substrate layering abutting gully channels or al-coves was observed in gully–polygon system sites, which includedcrater rims, crater walls, and isolated central peaks, nor was scourassociated with high-pressure water release observed. Rather, thelocations of gullies on Mars are strongly associated with the pres-ence of a mantling unit that is commonly polygonally-patterned.These lines of evidence suggest an atmospherically emplaced, top–down source for fluids involved in martian gully evolution onpolygonally-patterned surfaces, comparable to hydrological pro-cesses observed in the Antarctic Dry Valleys. On Earth and Mars,the presence of polygons is not shown to be directly causal of mar-tian gully formation, but to be diagnostic of top–down gully watersources, and to amplify the key processes of gully formation: accu-mulation of water ice and the channelized transport of melt water.
Geologically recent gully–polygon systems on Mars and Earth 125
These observations provide new challenges to the modelingcommunity to incorporate detailed treatment of landscape mi-crorelief and substrate composition into water cycling models forthe martian surface. Conditions permitting localized accumulationand peak-insolation melting of surface ice are broadly consis-tent with peak climate conditions modeled to have prevailed atgully–polygon sites during the last ∼1–10 My (Forget et al., 2007;Schon et al., 2009). Additional analysis of HiRISE images, coupledwith ongoing modeling of late Amazonian climate conditions, willenhance our understanding of gully–polygon system morphologyas an indicator of past climate processes on Mars.
Acknowledgments
This work was made possible with support of JSL by theRhode Island Space Grant Consortium, by NSF Grant ANT-0338291to D.R.M. and J.W.H., NASA MDAP Grants NNG04GJ99G andNNG05GQ46G to J.W.H., NASA MFRP Grant NNX06AE32G to D.R.M.and J.W.H., and NASA Applied Information Systems Research GrantNNG05GA61G to J.W.H. Thanks are extended to Caleb Fassett andJames Dickson for HiRISE image processing and to James Dick-son, Douglas Kowalewski, Gareth Morgan, David Shean, and KateSwanger for field support. Also, thanks to the helicopter pilots,technicians, and ground crew of PHI, Inc., as well as to the staff ofRaytheon Polar Services Company, and the personnel of McMurdoStation.
Supplementary data
Supplementary data for this article may be found on ScienceDi-rect, in the online version.
Please visit DOI: 10.1016/j.jcarus.2008.12.043.
References
Balme, M., Mangold, N., Baratoux, D., Costard, F., Gosselin, M., Masson, P., Pinet,P., Neukum, G., 2006. Orientation and distribution of recent gullies in thesouthern hemisphere of Mars: Observations from High Resolution Stereo Cam-era/Mars Express (HRSC/MEX) and Mars Orbiter Camera/Mars Global Surveyor(MOC/MGS) data. J. Geophys. Res. 111, doi:10.1029/2005JE002607.
Berg, T.E., Black, R.F., 1966. Preliminary measurements of growth of non-sorted poly-gons, Victoria Land, Antarctic. In: Tedrow, J.C.F. (Ed.), Antarctic Soils and Soil-Forming Processes. In: Antarctic Research Series, vol. 8. American GeophysicalUnion, Washington, DC. 177 pp.
Berman, D.C., Hartmann, W.K., Crown, D.A., Baker, V.R., 2005. The role of arcuateridges and gullies in the degradation of craters in the Newton Basin region ofMars. Icarus 178, 465–486.
Bockheim, J.G., Campbell, I.B., McLeod, M., 2007. Permafrost distribution and active-layer depths in the McMurdo Dry Valleys, Antarctica. Permafrost Periglac. Pro-cess. 18, 217–227.
Bridges, N.T., Lackner, C.N., 2006. Northern hemisphere martian gullies and man-tled terrain: Implications for near-surface water migration in Mars’ recent past.J. Geophys. Res. 111, doi:10.1029/2006/JE002702.
Burt, D.M., Knauth, L.P., 2007. Impacts, salts, and ice on Mars: How brine flows inyoung gullies and elsewhere could be related to impact cratering. Lunar Planet.Sci. 38. Abstract 2054.
Burt, D.M., Knauth, L.P., Wohletz, K.H., 2008. Martian gullies and salty sidewalks. In:Workshop on Martian Gullies, League City, TX, February 4–5. Abstract 1301.
Christensen, P.R., 2003. Formation of recent martian gullies through melting of ex-tensive water-rich snow deposits. Nature 422, 45–48.
Costard, F., Forget, F., Mangold, N., Peulvast, J.-P., 2002. Formation of recent mar-tian debris flows by melting of near-surface ground ice at high obliquity. Sci-ence 295, 110–113.
Dickson, J.L., Head, J.W., 2008. Global synthesis of Mars gully observations: Evidencefor top–down formation from morphology, distribution, topography, and analogsfrom the Antarctic Dry Valleys. In: Workshop on Martian Gullies: Theories andTests, Houston, TX, February 4–5. Abstract 8010.
Dickson, J.L., Head, J.W., Kreslavsky, M.A., 2007a. Martian gullies in the southernmidlatitudes of Mars: Evidence for climate-controlled formation of young fluvialfeatures based upon local and global topography. Icarus 188, 315–323.
Dickson, J.L., Head, J.W., Marchant, D.R., Morgan, G.A., Levy, J.S., 2007b. Recent gullyactivity on Mars: Clues from late-stage water flow in gully systems and channelsin the Antarctic Dry Valleys. Lunar Planet. Sci. 38. Abstract 1678.
Edgett, K.S., Malin, M.C., Williams, R.M.E., Davis, S.D., 2003. Polar- and middle-latitude martian gullies: A view from MGS MOC after two Mars years in themapping orbit. Lunar Planet. Sci. 34. Abstract 1038.
Forget, F., Montmessin, F., Levrard, B., Haberle, R.M., Head, J.W., Madeleine, J.-B.,2007. Glaciers, polar caps, and ice mantling: The effect of obliquity on martianclimate. In: Seventh International Conference on Mars, Pasadena, CA. Abstract3028.
Fortier, D., Allard, M., Shur, Y., 2007. Observation of rapid drainage system de-velopment by thermal erosion of ice wedges on Byot Island, Canadian ArcticArchipelago. Permafrost Periglac. Process. 18, 229–243.
Gooseff, M.N., McKnight, D.M., Lyons, W.B., Blum, A.E., 2002. Weathering re-actions and hyporheic exchange controls on stream water chemistry in aglacial meltwater stream in the McMurdo Dry Valleys. Water Resour. Res. 38,doi:10.1029/2001WR000834.
Gulick, V.C., the HiRISE Science Team, 2008. A closer look at valley, channel, andgully formation on Mars with HiRISE. Lunar Planet. Sci. 39. Abstract 2411.
Head, J.W., Marchant, D.R., 2008. Formation of gullies on Mars: Link to recent cli-mate history implicates surface water flow origin. In: Workshop on MartianGullies: Theories and Tests, Houston, TX, February 4–5. Abstract 8009.
Head, J.W., Marchant, D.R., Dickson, J.L., Levy, J.S., Morgan, G.A., 2007. Mars gullyanalogs in the Antarctic Dry Valleys: Geological setting and processes. LunarPlanet. Sci. 38. Abstract 1617.
Head, J.W., Marchant, D.R., Kreslavksy, M.A., 2008. Formation of gullies on Mars:Link to recent climate history and insolation microenvironments implicate sur-face water flow origin. Proc. Natl. Acad. Sci. USA, submitted for publication.
Hecht, M.H., 2002. Metastability of water on Mars. Icarus 156, 373–386.Heldmann, J.L., Mellon, M.T., 2004. Observations of martian gullies and constraints
on potential formation mechanisms. Icarus 168, 285–404.Heldmann, J.L., Carlsson, E., Johansson, H., Mellon, M.T., Toon, O.B., 2007. Obser-
vations of martian gullies and constraints on potential formation mechanisms.II. The northern hemisphere. Icarus 188, 324–344.
mertime sublimation rates for buried glacier ice in Beacon Valley, Antarctica.Antarct. Sci. 18, 421–428.
Kreslavsky, M.A., 2008. Slope steepness of channels and aprons: Implications fororigin of martian gullies. In: Workshop on Martian Gullies: Theories and Tests,Houston, TX, February 4–5. Abstract 8034.
Kreslavsky, M.A., Head, J.W., 2002. Mars: Nature and evolution of young,latitude-dependent water-ice-rich mantle. Geophys. Res. Lett. 29, doi:10.1029/2002GL015392.
Kreslavsky, M.A., Head, J.W., 2007. Slope streaks on Mars: An assessment of “wet”scenarios and the role of concentrated brines. In: Seventh International Confer-ence on Mars, Pasadena, CA, July 9–13. Abstract 3203.
Kreslavsky, M.A., Head, J.W., Marchant, D.R., 2007. Periods of active permafrostlayer formation during the geological history of Mars: Implications for circum-polar and mid-latitude surface processes. Planet. Space Sci. 56, doi:10.1016/j.pss.2006.02.010.
Kuzmin, R.O., 2005. Ground ice in the martian regolith. In: Tokano, T. (Ed.), Wateron Mars and Life. Springer-Verlag, Berlin, pp. 155–189.
Lachenbruch, A.H., 1962. Mechanics of thermal contraction cracks and ice-wedgepolygons in permafrost. In: Geological Society of America Special Papers, vol.70, pp. 1–69.
Laskar, J., Correia, A.C.M., Gastineau, M., Lervard, B., Robutel, P., 2004. Long termevolution and chaotic diffusion of the insolation quantities of Mars. Icarus 170,343–364.
Levy, J.S., Head, J.W., Marchant, D.R., Morgan, G.A., Dickson, J.L., 2007a. Gully–polygon interactions and stratigraphy on Earth and Mars: Sand-wedge polygonsas part of cold-desert, near-surface fluvial systems. In: Seventh InternationalConference on Mars, Pasadena, CA, July 9–13. Abstract 3059.
Levy, J.S., Head, J.W., Marchant, D.R., Morgan, G.A., Dickson, J.L., 2007b. Gully surfaceand shallow subsurface structure in the south fork of Wright Valley, AntarcticDry Valleys: Implications for gully activity on Mars. Lunar Planet. Sci. 38. Ab-stract 1728.
Levy, J.S., Head, J.W., Marchant, D.R., Kowalewski, D.E., 2008a. Identification ofsublimation-type thermal contraction crack polygons at the proposed NASAPhoenix landing site: Implications for substrate properties and climate-drivenmorphological evolution. Geophys. Res. Lett. 35, doi:10.1029/2007GL032813.
Levy, J.S., Head, J.W., Marchant, D.R., 2008b. The role of thermal contractioncrack polygons in cold desert fluvial systems. Antarct. Sci. 20, doi:10.1017/S0954102008001375.
Lewis, A.R., Marchant, D.R., Ashworth, A.C., Hemming, S.R., Machlus, M.L., 2007. Ma-jor middle Miocene global climate change: Evidence from East Antarctica andthe Transantarctic Mountains. Geol. Soc. Am. Bull. 119, 1449–1461.
Lyons, W.B., Welch, K.A., Carey, A.E., Wall, D.H., Virginia, R.A., Fountain, A.G., Do-ran, P.T., Csathó, B.M., Tremper, C.M., 2005. Groundwater seeps in Taylor ValleyAntarctica: An example of a subsurface melt event. Ann. Glaciol. 40, 200–206.
Mackay, J.R., 1990. Some observations on the growth and deformation of epigenetic,syngenetic and antisyngenetic ice wedges. Permafrost Periglac. Proces. 1, 15–29.
Malin, M.C., Edgett, K.S., 2000. Evidence for recent groundwater seepage and surfacerunoff on Mars. Science 288, 2330–2335.
Malin, M.C., Edgett, K.S., 2001. Mars Global Surveyor Mars Orbiter Camera: Inter-planetary cruise through primary mission. J. Geophys. Res. 106, 23429–423540.
Maloof, A.C., Kellogg, J.B., Anders, A.M., 2002. Neoproterozoic sand wedges: Crackformation in frozen soils under diurnal forcing during a snowball Earth. EarthPlanet. Sci. Lett. 204, 1–15.
Mangold, N., 2005. High latitude patterned grounds on Mars: Classification, distri-bution and climatic control. Icarus 174, 336–359.
Mangold, N., Maurice, S., Feldman, W.C., Costard, F., Forget, F., 2004. Spatial rela-tionships between patterned ground and ground ice detected by the neutronspectrometer on Mars. J. Geophys. Res. 109, doi:10.1029/2004JE002235.
Marchant, D.R., Denton, G.H., 1996. Miocene and Pliocene paleoclimate on the DryValleys region, southern Victoria Land: A geomorphological approach. Mar. Mi-cropaleontol. 27, 253–271.
Marchant, D.R., Lewis, A.R., Phillips, W.M., Moore, E.J., Souchez, R.A., Denton, G.H.,Sugden, D.E., Potter Jr., N., Landis, G.P., 2002. Formation of patterned groundand sublimation till over Miocene glacier ice in Beacon Valley, southern VictoriaLand, Antarctica. Geol. Soc. Am. Bull. 114, 718–730.
McEwen, A.S., and 32 colleagues, 2007. A closer look at water-related geologic ac-tivity on Mars. Science 317, 1706–1709.
McKnight, D.M., Niyogi, D.K., Alger, A.S., Bomblies, A., Conovitz, P.A., Tate, C.M., 1999.Dry valley streams in Antarctica: Ecosystems waiting for water. Bioscience 49,985–995.
Mellon, M.T., 1997. Small-scale polygonal features on Mars: Seasonal thermal con-traction cracks in permafrost. J. Geophys. Res. 102, 25617–625628.
Mellon, M.T., Phillips, R.J., 2001. Recent gullies on Mars and the source of liquidwater. J. Geophys. Res. 106, 23165–23180.
Milliken, R.E., Mustard, J.F., Goldsby, D.L., 2003. Viscous flow features on the surfaceof Mars: Observations from high-resolution Mars Orbiter Camera (MOC) images.J. Geophys. Res. 108, doi:10.1029/2002JE002005.
Morgan, G.A., Head, J.W., Marchant, D.R., Dickson, J.L., Levy, J.S., 2007a. Gully forma-tion on Mars: Testing the snowpack hypothesis from analysis of analogs in theAntarctic Dry Valleys. Lunar Planet. Sci. 38. Abstract 1656.
Morgan, G.A., Head, J.W., Marchant, D.R., Dickson, J.L., Levy, J.S., 2007b. Gully forma-tion on Mars: Testing the snowpack hypothesis from analysis of analogs in theAntarctic Dry Valleys. In: Seventh International Conference on Mars, Pasadena,CA. Abstract 1353.
Morgan, G.A., Head, J.W., Marchant, D.R., Dickson, J.L., Levy, J.S., 2008. Gully for-mation and evolution in the Antarctic Dry Valleys: Implications for Mars. In:Workshop on Martian Gullies, Houston, TX. Abstract 1301.
Mustard, J.F., Cooper, D.C., Rifkin, M.K., 2001. Evidence for recent climate change onMars from the identification of youthful near-surface ground ice. Nature 412,411–414.
Ragotzkie, R.A., Likens, G.E., 1964. The heat balance of two Antarctic lakes. Limnol.Oceanograph. 9, 412–425.
Riess, D., van Gasselt, S., Neukum, G., Jaumann, R., 2004. Absolute dune ages andimplications for the time of formation of gullies in Nirgal Valles, Mars. J. Geo-phys. Res. 109, doi:10.1029/2004JE002251.
Sears, D.W.G., Chittenden, J.D., 2005. On laboratory simulation and the temperaturedependence of the evaporation rate of brine on Mars. Geophys. Res. Lett. 32,doi:10.1029/2005GL024154.
Schon, S.C., Head, J.W., Fassett, C.I., 2009. Unique chronostratigraphic maker in de-positional fan stratigraphy on Mars: Evidence for ∼1.25 Ma old gully activityand surficial meltwater origin. Geology 37, 207–210.
Treiman, A.H., 2003. The geologic setting for martian gullies: Implications for theirorigins. J. Geophys. Res. 108, doi:10.1029/2002JE001900.
Washburn, A.L., 1973. Periglacial Processes and Environments. St. Martin’s Press,New York, pp. 90–99.
