Icarus 178 (2005) 56–73
www.elsevier.com/locate/icarus
Young (late Amazonian), near-surface, ground ice features near the
equator, Athabasca Valles, Mars
Devon M. Burr a,b,∗ , Richard J. Soare c,d , Jean-Michel Wan Bun Tseung d , Joshua P. Emery e
a USGS Astrogeology Program, 2255 N. Gemini Dr., Flagstaff, AZ 86001, USA
b The SETI Institute, 515 N. Whisman Dr., Mountain View, CA 94043, USA
c Department of Geography, Concordia University, 1455 de Maisonneuve Blvd. W., Montreal, Quebec H3G 1M8, Canada
d Department of Geography, Burnside Hall, McGill University, 805 Sherbrooke St. W., Montreal, Quebec H3A 2K6, Canada
e NASA Ames Research Center/The SETI Institute, Mail Stop 245-6, Moffett Field, CA 94035, USA
Received 9 July 2004; revised 31 March 2005
Available online 13 June 2005
Abstract
A suite of four feature types in a ∼20 km2 area near 10◦ N, 204◦ W in Athabasca Valles is interpreted to have resulted from near-surface
ground ice. These features include mounds, conical forms with rimmed summit depressions, flatter irregularly-shaped forms with raised rims,
and polygonal terrain. Based on morphology, size, and analogy to terrestrial ground ice forms, these Athabascan features are interpreted as
pingos, collapsing pingos, pingo scars, and thermal contraction polygons, respectively. Thermal Infrared Mapping Spectrometer (THEMIS)
data and geological features in the area are consistent with a sedimentary substrate underlying these features. These observations lead us
to favor a ground ice interpretation, although we do not rule out volcanic and especially glaciofluvial hypotheses. The hypothesized ground
ice that formed the mounds and rimmed features may have been emplaced via the deposition of saturated sediment during flooding; an
alternative scenario invokes magmatically cycled groundwater. The ground ice implicit in the hypothesized thermal contraction polygons
may have derived either from this flooding/ground water, or from atmospheric water vapor. The lack of obvious flood modification of the
mounds and rimmed features indicates that they formed after the most recent flood inundated the area. Analogy with terrestrial pingos
suggests that ground ice may be still extant within the positive relief mounds. As the water that flooded down Athabasca Valles emerged via
a volcanotectonic fissure from a deep aquifer, any extant pingo ice may contain evidence of a deep subsurface biosphere.
 2005 Elsevier Inc. All rights reserved.
1. Introduction
Geological features which imply the present or previous
existence of ground ice are key to understanding some important geologic processes on Mars. They are markers for
paleo- and present climate (e.g., Baker, 2001; Head et al.,
2003b). They aid in delineating the near-surface distribution of water (e.g., Carr and Schaber, 1977; Squyres and
Carr, 1986). They inform our understanding of volatile inventory and cycling, potential biotic or prebiotic activity, and
possible future human exploration (Carr, 1996). Features at* Corresponding author. The SETI Institute, 515 N. Whisman Dr. Mountain View, CA 94043, USA. Fax: +1 (650) 962 8120.
E-mail address: [email protected] (D.M. Burr).
0019-1035/$ – see front matter  2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.icarus.2005.04.012
tributed to present or previous near-surface ground ice have
been inferred for the mid- (e.g., Lucchitta, 1981; Costard
and Kargel, 1995; Seibert and Kargel, 2001; Mustard et al.,
2001) and high-latitudes (e.g., Kreslavsky and Head, 2002;
Soare et al., 2005). Ground ice at these latitudes is consistent
with atmospheric and climatic modeling (e.g., Clifford and
Hillel, 1983; Fanale et al., 1986; Clifford, 1993; Mischna
et al., 2003; Mellon et al., 2004). Recent data indicate the
presence of significant amounts of near-surface ground ice
poleward of ±60◦ and a relative dearth of it in the equatorial region (Boynton et al., 2002; Feldman et al., 2002). In
agreement with these models of the present general ground
ice distribution, young, near surface, equatorial ground ice
features have not previously been documented with highresolution data. Equatorial features interpreted as due to
Young ground ice features, Athabasca Valles, Mars
57
(a)
Fig. 1. (a) Context MOLA image of the Athabasca Valles region, with
50 m contours. The black asterisk in the center-right of the image indicates the center location of (b). (b) MOC image R01-00745, centered near
9.67◦ N 203.95◦ W. Boxes show approximate locations of subsequent figures as labeled. Streamlined form, dune field, and fissure are also labeled.
Black dotted and then dashed line indicates the continuation of the fissure
across the dune field and then the channel floor as inferred from the faint
break in density of the mounds and rimmed forms on the channel floor (see
also Fig. 12). Illumination is from the left. (Comparison with MOC image
E10-01384 shows that the two data drop-outs across the middle of the image cover an insignificant area.) Formation of the dune field during diluvial
(flood) sedimentation is discussed in Burr et al. (2004).
ground ice (Cabrol et al., 2000; Levoy et al., 2004) are old,
buried under a thick over-burden, and/or interpreted on the
basis of lower resolution Viking imagery alone.
On the basis of Mars Global Surveyor Mars Orbiter Camera (MOC) and Mars Odyssey Thermal Emission Imaging
System (THEMIS) data, we interpret a suite of features near
the equator as resulting from near surface, young ground ice
and associated processes. This suite of features is centered
near 9.67◦ N 203.95◦ W within Athabasca Valles (Fig. 1a),
a late Amazonian outflow channel system that originates
from the Cerberus Fossae (Tanaka, 1986; Burr et al., 2002b;
Plescia, 2003; Lanagan, 2004), young volcanotectonic fissures from which lava also recently emerged (Plescia, 1990;
Burr et al., 2002a; Berman and Hartmann, 2002; Head et
al., 2003a). The hypothesized ground ice features to be
discussed may have resulted from deposition of saturated
sediments during flooding and subsequent exposure of this
wet sediment to freezing conditions. An alternative scenario involves the magmatic emplacement and then freezing of ground water. For one of the four feature types, an
atmospheric source of water is also possible. In any case,
we do not see evidence of modification of the mounds and
rimmed forms by subsequent Athabascan floodwaters. Thus,
they apparently post-date the most recent (late Amazonian)
flooding.
(b)
Fig. 1. (Continued.)
The Athabascan features are morphologically similar to
terrestrial landforms resulting from ground ice and associated processes. For each Athabascan feature type, we char-
58
D.M. Burr et al. / Icarus 178 (2005) 56–73
acterize the features, describe potential terrestrial ground ice
analogues, and discuss possible alternative hypotheses from
the literature. This outflow channel is in a volcanically active
region, and these alternative hypotheses generally involve
volcanic processes (Lanagan et al., 2001; Jaeger et al., 2003,
2005; Lanagan, 2004). A glaciofluvial interpretation of these
features as ice-block kettle holes has also been advanced
(Gaidos and Marion, 2003). Morphologically, and on the basis of THEMIS data and other local surficial features, we
find volcanic origins less satisfactory than ground ice. However, the sedimentary substrate we interpret to underlie these
features is equally consistent with both the previous kettle
hole hypothesis (Gaidos and Marion, 2003) and the ground
ice hypothesis presented here. The morphological data presented below were gathered from measurements on overlapping MOC images E10-01384 (res. 3.10 m/pixel), R0100745 (res. 1.54 m/pixel), R12-03203 (res. 1.50 m/pixel
cross-track, 0.50 m/pixel downtrack), and R13-03976 (res.
3.10 m/pixel).
Athabasca Valles and Marte Vallis (Lanagan et al., 2001;
Jaeger et al., 2003), but we limit our present discussion to
the forms grouped within this ∼20 km2 area.
The effective basal diameter for these mounds and
rimmed forms ranges from a low of ∼15 m for the smallest circular mounds up to a high of ∼128 m for the largest
irregularly shaped rimmed forms. Vertical dimensions were
derived from both two-dimensional photoclinometry (Kirk
et al., 2003) and shadow measurements. The photoclinometry, which assumes a uniform surface albedo, was
performed with a Lambertian photometric function, using
two different methods to account for haze: the assumption of no haze, and use of the minimum pixel data number (DN) value for the entire MOC image to represent
2. Athabascan features and comparison with terrestrial
analogues
2.1. Mounds and rimmed features
2.1.1. Description
Perhaps the most intriguing features at this location in
Athabasca Valles are numerous interspersed mounds and
rimmed forms (Fig. 1b). Included in this grouping are all
the mesoscale (100-m-scale) positive relief features superposed on the channel floor. These features range from positive relief circular mounds, to shorter conical forms with
rims surrounding summit depressions, to even flatter irregularly shaped forms with only slightly raised rims (Fig. 2;
see also Fig. 3). Approximately 250 of the mounds and
rimmed features are found within several square kilometers. They are concentrated in the area just to the north
of a streamlined form, with a density of up to ∼70/km2 .
Similar features are found in various locations throughout
Fig. 2. Examples of the features as discussed in the text. The black arrow
points to a positive relief mound; the gray arrow points to a conical form
with a summit depression; the white arrow points to a flatter, more irregularly shaped form.
Fig. 3. Three features chosen to represent the suggested evolution of form from taller circular mounds to flatter more irregular features. We suggest these three
stages loosely correlate to pingos (a), collapsing pingos (b), and pingo scars (c).
Young ground ice features, Athabasca Valles, Mars
59
is used in both variables. However, the plot also supports the
idea of a continuum of form by showing no discrete groupings. These data are not intended as a representation of an
evolutionary sequence in time (a chronosequence), only as
indicating a continuous variation in form. On the basis of
their continuum of form, we sought a single explanation for
all of these forms.
2.1.2. Terrestrial analogs
Like the Athabascan features, terrestrial pingos display a
continuous range of morphologies within a spatially limited
area (e.g., Mackay, 1979; Holmes et al., 1968). Pingos are
perennial ice-cored mounds that result from injection (or ‘intrusion’) and freezing of groundwater under pressure. Two
types are commonly recognized.
Fig. 4. Graphs of circularity vs area and circularity vs perimeter for
Athabascan forms. Circularity is defined as 4π × area/(perimeter2 ), with a
perfect circle having a value of 1. The Pearson product-moment correlation
coefficient for a group of randomly sized, nearly circular features would be
nearly zero.
the haze. These two methods give minimum and maximum elevation values, respectively. (See Kirk et al., 2003;
Beyer et al., 2003, for more details.) These maximum photoclinometric values, minimum photoclinometric values, and
shadow measurement were roughly similar, giving a value
of ∼24 m for the largest mounds, decreasing to ∼1.5 m for
the rims of the irregular shapes. These rims were often lower
on the upslope than the downslope side, even in a small number of cases grading into the channel floor.
These mounds and rimmed features are grouped together
because they appear in plan view to lie on a continuum. That
is, they display a variety of plan view form without clear or
discrete distinctions between the various forms. The continuum appears to follow a trend of flattening, and of becoming more irregular with size (Fig. 4). To assess this visual
impression, we measured the circularity of the rims of the
conical features with summit depressions and of the flatter
more irregularly shaped features. Circularity is here defined
as 4π × area/(perimeter2 ) (http://rsb.info.nih.gov/ij/plugins/
circularity.html). A perfect circle has a circularity value of 1,
and more irregular shapes have decreasing circularity values.
Plotting the features’ circularity against either their area or
perimeter shows a negative slope (n = 202; Fig. 4), supporting the impression of increasing irregularity with size. Some
degree of correlation is built into these measurements (for
circularity vs perimeter, in particular), because the same data
Hydrostatic pingos Hydrostatic (‘closed system’) pingos
are most abundant in Arctic America and Siberia. Specifically, the Tuktoyaktuk Peninsula of the Northwest Territories, Canada, hosts ∼1350 pingos in Pleistocene silts and
sands, about one-quarter of the world’s total and the largest
known concentration (Mackay, 1998). Yakutia, Siberia, also
hosts an exceptional number of pingos, located on the region’s glaciofluvial terraces (Soloviev, 1973). For hydrostatic pingos, the required groundwater pressure comes from local permafrost aggradation in saturated sediments (Soloviev,
1973; Mackay, 1979; French, 1996). They are found in flat
terrains in areas of continuous permafrost at depth, which
constrains the expelled groundwater to dome upward.
Hydrostatic pingo form is influenced in the vertical dimension largely by temporal evolution and in plan view
largely by initial conditions. Hydrostatic pingo evolution has
been documented through decades of monitoring hydrostatic pingo growth in the Tuktoyaktuk Peninsula (summarized
by Mackay, 1998). Ninety-eight percent of Tuktoyaktuk hydrostatic pingos form in residual ponds in the basins of
drained thaw lakes, which, when suddenly exposed to cold
atmospheric conditions, begin to freeze around the edges.
A residual pond or central unfrozen zone (‘talik’) retards
freezing of the underlying saturated sediments and minimizes permafrost development. Freezing progresses inward
from the lake edges toward the residual pond or talik, expelling pore water ahead of the freezing front. This produces
hydrostatic pressure, which eventually uplifts the overlying
sediments in the area of thinnest permafrost. As the pressurized water freezes, it results in a pingo (Fig. 5a). A hydrostatic pingo takes on and will maintain the approximate
shape of the former residual pond in which it forms, which
may range from circular to elongate or curvilinear (see, e.g.,
Mackay, 1963, 1987, 1998, discussion and pictures of pingos 4 and 12). It will increase in height, however, as the ice
core continues to grow with continued freezing of underlying saturated sediments and resultant pore water expulsion
(Fig. 5b). This growth may continue to the point of tensional
cracking of the insulating overlying sediments, resulting in
sublimation or melting of the ice core (Figs. 5c and 6a).
60
D.M. Burr et al. / Icarus 178 (2005) 56–73
(a)
(b)
(c)
(d)
Fig. 5. (a) Incipient hydrostatic pingo in the town of Tuktoyaktuk. Person at left is 1.9 m tall. (b) Hydrostatic pingo, Tuktoyaktuk Peninsula. Person on top
is 1.9 m tall. (c) Pingo showing signs of initial collapse, as indicated by summit crater. Probable dilation cracks (outlined by snow) suggest this collapse
was initiated through exposure and melting of the ice core. Pingo is roughly 20 m high. (d) Collapsed pingos (in foreground, and in back left), Tuktoyaktuk
Peninsula, photographed from atop a neighboring pingo. Person on back rim in image center is 1.9 m tall.
Young ground ice features, Athabasca Valles, Mars
Slumping of the overlying sediments or erosion along the
cracks may also contribute to exposing the ice core to thaw.
In cases of a subpingo groundwater lens, rupture of the overburden around the base of the pingo may result in spring
discharge from the lens. As a result of these effects, the pingo
begins to collapse, developing a summit depression (Fig. 5c).
Hydrostatic pingos may pulsate if groundwater recharges the
subpingo water lens (Mackay, 1979).
Eventually, pingos collapse entirely (Fig. 5d). The results
are closed depressions with rims up to a few meters in height.
Found in many now temperate climes (see numerous references in (Flemal, 1976; De Gans, 1988)), these features are
referred to as pingo scars. This temporal evolution does not
originate simultaneously nor proceed at the same rate for all
pingos within the same climatic or sedimentological zones.
As a result, pingos in various states of collapse are closely
interspersed on the Tuktoyaktuk Peninsula (Mackay, 1998)
and in Yakutia (Soloviev, 1973). Mackay (1998), in Fig. 4
provides an excellent diagram of hydrostatic pingo evolution.
Hydraulic pingos For hydraulic (‘open system’) pingos,
the required groundwater pressure is artesian. Thus, hydraulic pingos are generally found in valleys or at the base
of slopes, which can provide the necessary head. Alluviumfilled valleys of Alaska contain a few hundred hydraulic
pingos (Holmes et al., 1968), and glaciofluvial outwash
plains in Greenland also host large concentrations (Müller,
1959). Greenland pingos have been documented as forming
in coarse sediment, including cobbles, boulders and in one
case sandstone bedrock (Müller, 1959). Hydraulic pingos
are found in areas of discontinuous permafrost, where openings in the permafrost (beneath ponds, river flood plains, or
equator-facing slopes) permit the groundwater under artesian pressure to dome up the overlying substrate (Holmes et
al., 1968; French, 1996). Fig. 9 of Holmes et al. (1968) provide a cut-away diagram of hydraulic pingos in Alaska.
Hydraulic pingo evolution has not been documented as
carefully as that of hydrostatic pingos, but generally appears
to mirror that of hydrostatic pingos. Although the controls
on their plan view morphology are not as well known, their
vertical relief is also a result of growth over time. After initiation of collapse, hydraulic pingos may undergo two or three
episodes of rejuvenation (Holmes et al., 1968) before collapsing entirely. The majority of pingos scars discussed in
the literature (Flemal, 1976; De Gans, 1988, and references
therein) were attributed to hydraulic pingos.
As a result of plan view controls and temporal evolution, pingos within the same area and same climatic regime
display a variety of morphologies, ranging from fresh to collapsed in the vertical dimension, and from circular to irregular in plan view (Mackay, 1979; Soloviev, 1973). Unfortunately, our inability to confidently distinguish a statistically
representative number of pingos in air photos precludes us
from presenting a plan view circularity distribution for terrestrial pingos such as is presented in Fig. 4 for the Athabas-
61
can rimmed forms. Description of pingos in the literature
suggests that a similar inverse and continuous circularity distribution would be reasonable.
2.1.3. Comparison between Athabascan features and
terrestrial pingo-related forms
In addition to this continuous range of morphologies,
pingo-related forms on Earth show other characteristics similar to the Athabascan features. Plan view and vertical dimensions of the Athabascan features are both within the
values of terrestrial pingo forms. In plan view, terrestrial pingos (Müller, 1959; Holmes et al., 1968; Mackay, 1998) and
pingo scars (Flemal, 1976) generally range from ∼10 m up
to a few hundreds of meters in basal diameter. This encompasses the range of the Athabascan features. Heights of terrestrial and these hypothesized Athabascan pingos are also
similar. This similarity is expected, based on the physics of
pingo growth. In the vertical dimension, the driving force
for pingo growth is the pore water pressure in the subpingo water lens. This is equivalent to the normal force
at the bottom of the pingo ice core, and thus is proportional to gravity (Mackay, 1987). The force resisting pingo
growth is the overburden weight, and thus is also proportional to gravity. At the point of maximum pingo growth,
the resisting force equals the driving force, such that the
effect of gravity is nullified. So for all other conditions being similar, martian pingo heights should be similar to terrestrial pingo heights. The Tuktoyaktuk hydrostatic pingos
range up to 49 m in height for Ibyuk, the world’s second
tallest known hydrostatic pingo (Fig. 6a) (Mackay, 1986);
the Alaskan hydraulic pingos range up to ∼30 m in height
(Holmes et al., 1968); and the Greenland hydraulic pingos
have grown up to ∼50 m in height (Müller, 1959). The
heights of the Athabascan mounds are within these ranges.
Terrestrial pingo scars are classified as having rims equal to
or greater than ∼1.5 m in height (De Gans, 1988). This compares closely with the heights for the flat irregularly shaped
rims in Athabasca.
The Athabascan rimmed forms and terrestrial pingorelated forms also have similar morphologies. As mentioned
above, terrestrial pingos may collapse consequent to dilational cracking of the overburden (Fig. 6a). This can produce
an undulating or broken rim collapse morphology, similar to
that observed for a few of the Athabascan cones (Fig. 6b).
Slumping of the overburden or rupture of a subpingo water lens can lead to a collapse morphology having a more
fully intact rim without discernable cracking (Fig. 7a). This
smoother inverted morphology is common for the Athabascan rimmed features (Fig. 7b).
The Athabascan features exhibit repeated examples of
overlapping, contiguous, or conjoining relationships (Figs. 2
and 8a). This is similar to many hydraulic terrestrial pingo
scars, which occur in groups of “mutually interfering structures” (Fig. 8b) (Watson (1971), see also De Gans (1988)
and references therein). This is attributed to multiple generation and non-circularity of the original pingos (Flemal, 1976;
62
D.M. Burr et al. / Icarus 178 (2005) 56–73
(a)
(b)
(a)
(b)
Fig. 6. (a) Aerial photo of Ibyuk Pingo, Tuktoyaktuk Peninsula, showing
dilational cracking and a summit depression. Illumination from the right.
(b) Conical features in Athabasca Valles, with interpreted cracking and collapse, similar to (a). Illumination from the left.
Fig. 7. (a) Oblique aerial photograph of collapsed pingo, Tuktoyaktuk
Peninsula. Few-meter-tall pine trees to left of form suggest scale. (b) Closed
rimmed feature in Athabasca Valles, with interpreted collapse similar to the
collapsed feature in (a).
De Gans, 1988). The maximum density of the Athabascan features is within the maximum density range of 50–
100 forms/km2 for terrestrial pingo scars (Flemal, 1976).
Close to half the Athabascan rimmed features are surrounded by lower elevation annuli or ‘moats,’ which average
∼20 m in maximum width (Fig. 9). These exterior moats often appear to have their own raised rims along some part of
their exterior edges, which are higher than either the annuli
to their inside or the channel floor to their outside. However,
this seems not always to be the case, and the determination
of a raised exterior edge of a moat, being made at the limit
of image resolution, is often ambiguous (Fig. 9). Terrestrial
pingo scars are often surrounded by moats (Flemal, 1976;
De Gans, 1988), although the process that forms these exterior moats is not well known. A collapsed pingo in the
Tuktoyaktuk Peninsula has a moat surrounding a central
plug located interior to adjoining ridges, and the formation
of this feature is attributed to up-and-down oscillations or
pulsations of the central plug (Mackay, 1998). This ∼1 m
deep and few-meters-wide terrestrial moat is several times
narrower than the Athabascan annuli. If the features are analogous, such a difference in size might be caused by larger
pulsations on Mars, perhaps due to more extreme seasonal
temperature changes. Alternatively, for hydrostatic pingos,
moats might be formed from the portion of the thaw lake bottom that was not uplifted. Figs. 61 and 63 of Mackay (1998)
show the typical arrangement of pingos located within thaw
lake basins. However, the distance from the pingos to the
terrestrial thaw lake edges is several times greater than the
width of the Athabascan moats and the lake edges do not exhibit uplifted edges that are seen around some of the martian
moats.
About one-third of the flatter more irregularly shaped features have smaller, non-concentric, non-contiguous, interior
rings or cones with summit depressions (Fig. 10). The existence of such features has been noted in terrestrial collapsed
pingos, but their exact genesis seems ambiguous. They seem
to occur in a variety of situations. In the hydrostatic pingos of the Tuktoyaktuk Peninsula, Mackay (1998) notes a
common association with frost mounds. These ice cored
hills are smaller than pingos and seasonal, in contrast to the
perennial nature of pingos. The two form through a similar
process—pressurization of ground water during freezing of
the ground (Pollard and van Everdingen, 1992)—and frost
Young ground ice features, Athabasca Valles, Mars
(a)
63
(b)
Fig. 8. (a) Image of two of the flatter more irregular Athabascan features. This contiguous and conjoining structure is similar to that of many terrestrial pingo
scars. (b) Drawing of Welsh hydraulic pingo scars from Watson, 1971, Fig. 2. (Copyright John Wiley & Sons Limited, reproduced with permission.)
Fig. 9. Closed ring feature (in the center) with a lower elevation annulus or
moat surrounding it. Illumination is from the left. Shadowing indicates that
this moat has a raised rim on its straight eastern side, but this raised rim is
discontinuous, missing or ambiguous elsewhere around the form.
Fig. 10. A closed ring feature enclosing three smaller conical features with
summit depressions. These interior cones range at their summit from ∼8 to
∼16 m in maximum diameter.
mounds frequently precede pingo genesis or grow on pingo
flanks (Mackay, 1998). A similar spatial relationship applies
to some of the Athabascan features, for which a smaller ring
or cone is located on the rim of a larger more irregular form
(Fig. 11), but the majority of smaller cones are more centrally located within the larger features (Fig. 10). Fig. 5.47
of Washburn (1980) provides a photograph of a Canadian
Fig. 11. A closed ring feature with a smaller conical or ring feature at its
northwestern edge.
(hydrostatic?) pingo with an interior cone with a summit
depression, although without explanation. (This photograph
is reproduced in Lucchitta, 1981, Fig. 9e.) Watson (1971)
includes an image of a hydraulic pingo in Greenland and
pingo scars in Wales, both with interior mounds or cones;
this is attributed to the tendency for new pingos to be “repeatedly born from the ruins of earlier ones” (Müller, 1959),
although the cause for the tendency is not detailed. Fig. 4.20
of Davis (2001) shows several cones with summit depressions attributed to mud-charged springs interior to a collapsed hydraulic pingo in Alaska. Thus, evidence exists for
the formation of smaller rings or cones within collapsed pingos. However, the exact mechanism for their formation—
hydrostatic or hydraulic, contemporary with or after pingo
formation—is currently unclear to us.
In summary, the Athabascan features show similar vertical and horizontal dimensions, and similar densities to terrestrial pingos, collapsing pingos, and pingo scars. They
show a similar range and type of morphology, and occur in
complexes of interconnected structures, similar to terrestrial
pingo-related forms. They have exterior annuli or moats,
which may have resulted from pulsation or might be the
64
D.M. Burr et al. / Icarus 178 (2005) 56–73
floors of the drained lakes in which they formed. Analogs for
the non-concentric interior rings or cones in the Athabascan
features have been documented within terrestrial pingos, but
their exact mechanism of formation remains enigmatic.
2.1.4. Alternative hypotheses
The Cerberus plains have been recently volcanically
resurfaced (e.g., Tanaka, 1986; Plescia, 1990; Keszthelyi et
al., 2000, 2004a, 2004b; Lanagan, 2004), and volcanic hypotheses have been proposed for these and similar mesoscale
rimmed forms across the Cerberus plains. Rimmed features with summit depressions in the Cerberus region have
been classified as rootless cones (Lanagan et al., 2001;
Greeley and Fagents, 2001). Terrestrial rootless cones may
lack summit craters (Greeley and Fagents, 2001) or the summit craters may be below the limit of resolution, so simple
mound forms could also potentially be classified as rootless
cones. However, rootless cones are not identified as having
external moats. Furthermore, the martian rootless cone identification was made in part on inferred geologic context. i.e.,
the hypothesized rootless cones sit on surfaces classified as
lava. As explained below, we interpret the substrate here to
be sediment, not lava.
The irregular rimmed features have also been suggested
to be analogous to basaltic ring structures (Jaeger et al.,
2005). These structures, found in the diluvially eroded
Columbia River basalts of the Channeled Scabland, are
described as “circular structures defined by arcuate, concentric ridges and scarps that surround hills, mesas, or
crater-like depressions” (Hodges, 1978). Their formation is
attributed to the interaction of lava with surficial ground
water, followed by diluvial erosion (Hodges, 1978; Jaeger
et al., 2005). One population of the Channeled Scabland
structures shows concentric interior rings containing autointrusive dikes (McKee and Stradling, 1970; Hodges, 1978;
Jaeger et al., 2005). However, as discussed above, ∼30% of
the Athabascan features have non-concentric interior rings
or cones (e.g., Fig. 10). Furthermore, the Channeled Scabland structures are not simple mounds, which account for
∼10% of the forms in Athabasca, although the terrestrial
mesas and hills surrounded by concentric rings might appear
as simple mounds in low resolution narrow-angle MOC images. The Channeled Scabland basaltic ring structures rarely
have prominent moats, although one group does appear to
have exterior moats partially filled with aeolian sediments.
We also favor a sedimentary material for the substrate in
which these forms are found based on our interpretation of
relative thermal inertia and surficial geology. A qualitative
sense of thermal inertia can be obtained by comparing day
and night surface thermal response as seen in 100 m/pixel
THEMIS IR images (Christensen et al., 2003). A surface
that is relatively warm in daytime and cold at night has relatively low thermal inertia. This is the case for the substrate on
which the bulk of the mounds and rimmed forms are located
(Fig. 12). Only the forms within the dune field (Fig. 13) lie in
higher thermal inertia material. Thermal inertia can be used
to distinguish bedrock, larger-grained, or cohesive sediments
from finer-grained non-cohesive sediments (e.g., Kieffer et
al., 1977; Christensen et al., 2003). The relatively low thermal inertia of the substrate for these forms points toward a
surface dominated by fine-grained sedimentary materials.
Comparison of the thermal inertia with the surficial geology further indicates a sedimentary substrate for the mounds
and rimmed forms. In visible wavelength images, a fissure is
apparent on the streamlined form to the south of the dunes
(Fig. 1b). This fissure disappears beneath the dune field, but
beyond the dunes, a faint break in the pattern of the mounds
and rimmed forms hints at its continuation northwestward
(Figs. 1 and 12a). In both day and night IR images, this filled
fissure blends with the substrate for the mounds and rimmed
forms (Fig. 12). This blending suggests that the material
in the fissure and the substrate of the mounds and rimmed
forms is the same low thermal inertia material. Still further
to the northwest, the fissure fill continues as a relatively low
thermal inertia trace (Figs. 12b–12d).
We interpret these observations is as indicating that the
fissure is filled with a sedimentary material, and that the
most likely material is flood sediments. Deposits in terrestrial flood channels (e.g., Carling et al., 2002) show that
this location beside the streamlined form would be a reasonable place for flood sediment deposition, which is commonly
associated with in-channel flow obstacles or found along
the margins of the channel itself. Under martian gravity,
floodwaters would be expected to carry substantial amounts
of sediment if such sediment were available for transport
(Komar, 1980). The flood-formed dunes (Burr et al., 2004)
demonstrate that sediment was available for transport and
that sediment was deposited in this area.
Other interpretations are possible. Lava may fill the fissure (e.g., Keszthelyi et al., 2004a) and, as this region of
Mars is dusty (e.g., Mellon et al., 2000), be covered with dust
resulting in the observed low thermal inertia trace. However, we do not find this explanation for this particular area
as reasonable as the flood sediments hypothesis because of
the context and the morphology of the lower inertia area.
Dust is broadly distributed on Mars, generally by large or
global dust storms (Greeley and Iverson, 1985); the smallest
dust storm detected by MOC was a few 100 km2 (Cantor et
al., 2001). However, the low thermal inertia substrate of the
mounds and rimmed forms is only ∼20 km2 in area, and we
would not expect dust storms to produce thermal variations
on this small scale. Even if dust deposits were particularly
localized, perhaps due to aeolian redistribution (Greeley
et al., 2000), we would expect these residual deposits to
have diffuse boundaries. However, the boundary between the
lower and higher thermal inertia units is fairly distinct, occurring within 1–2 THEMIS pixels (∼100–200 m); this is
particularly apparent where the fissure’s western stretch contrasts with the surrounding channel floor (Fig. 12). Finally,
no source is apparent for such a localized lava flow. Even if
filled, a source for the lava might be expected to be visible—
just as this filled fissure is—on THEMIS IR imagery. If the
Young ground ice features, Athabasca Valles, Mars
65
(a)
(b)
(c)
(d)
Fig. 12. (a) Visible image mosaic, (b) stretched day IR, (c) stretched night IR, and (d) sketch of the fissure and streamed form adjacent to the rimmed forms.
Visible image mosaic includes MOC images E10-01384, R11-00065, and R13-03976, overlain on THEMIS day IR I08483034. Each figure image (a–d) is
∼7.3 km; see scale bar on (a).
fissure itself were the source of the lava, the lava would have
to have flowed ∼2 km up the channel. We do not see an obvious source for a lava flow in this location.
Thus, we find that a fissure-filling material of dustcovered lava is not as consistent with observations or context
as flood-sediments. The fissure-filling material and the substrate of the mounds and rimmed forms blend together in
day and night IR imagery. This IR imagery is noisy, but
the simplest explanation for this blending is that the fissurefilling material and the substrate are comprised of the same
material. This interpretation is supported by the observation that both the fissure-filling material and the mounds and
rimmed forms substrate have the same polygonal morphology (Fig. 12) (see Section 2.2). The area of the mounds and
rimmed features is thus interpreted as being comprised of
the same material and that material is interpreted to be most
likely flood sediments.
A sedimentary substrate for the mesoscale rimmed forms
is also indicated by their presence in the sedimentary floodformed dunes northwest of the streamlined form (Fig. 1).
The appearance of the dunes and the rimmed forms suggests
that the forms post-date and are formed within the dunes.
The linear NW–SE strike of the dunes is interrupted by the
arcuate rims of the forms (Fig. 13). Also, the dune material
does not appear streamlined behind the forms, which would
be expected if the forms preceded the dunes. In a few locations, the form rims appear partially infilled or covered
by bright material; but, as this is a dusty region of Mars
66
D.M. Burr et al. / Icarus 178 (2005) 56–73
2.2. Polygonal terrain
Fig. 13. Rimmed form within the sedimentary dunes. The black lines highlight the dunes’ linear NW–SE strike; the circular forms are left in plain
view.
Fig. 14. High-centered polygonal terrain, which is interspersed with the
rimmed features.
(Mellon et al., 2000), we attribute this bright infill to aeolian dust. In some cases, the mesoscale rimmed forms on
the dunes appear slightly different—possibly more muted
or jumbled—than those on the channel floor. However, as
noted above, pingos can form in a variety of sedimentary
substrates, from fine lacustrine silts to coarse glaciofluvial
sands and cobbles, and their substrate would reasonably be
expected to influence their appearance. We hypothesize that
the slightly different appearance between the rimmed forms
on the dunes and those on the channel floor results from their
formation in slightly different sedimentary substrates.
We conclude that the rimmed forms post-date the dunes,
and are located formed within the dunes’ sediments. A sedimentary substrate is consistent with a ground ice, and not
a volcanic, origin for the forms. A sedimentary substrate is
equally consistent with a surficial ice (i.e., ice block) depositional origin for the forms, as proposed by Gaidos and
Marion (2003).
2.2.1. Description
Patterned or polygonal terrain is visible on the Athabasca
channel floor among the mounds and rimmed features
(Fig. 14). This terrain is characterized as having a bumpy appearance in which slightly raised “bumps” appear outlined
by slightly lower polygons. Photoclinometry shows the average maximum heights of the raised polygon centers to be
within 0.5 to ∼5.0 m above the average elevation of the surrounding troughs. (The lower value results from assuming
no haze; the higher value results from subtracting the minimum pixel DN value as haze.) The polygons have an average
maximum diameter of ∼25 m (standard deviation = ∼10 m;
n = 212) which appears generally parallel to the channel
slope. However, this preferred orientation may be in part an
effect of the illumination angle, which is approximately perpendicular to the channel slope.
2.2.2. Terrestrial analogs
Terrestrial high-centered thermal contraction polygons
are of the same size range and have the same high-centered
morphology as these Athabascan forms. Thermal contraction polygons result from repeated, rapid lowering of the
subzero temperature of ice-rich ground; this produces tensile stresses in the ground, which are relieved by cracking.
As these thermal contraction cracks coalesce, they form thermal contraction polygons (Lachenbruch, 1962).
In humid cold climates, these polygons are initially lowcentered (French, 1996; Davis, 2001). The thermal contraction cracks fill seasonally with water, e.g., snow melt,
leading to ice wedge development (Lachenbruch, 1962).
The edges of these cracks become raised, due to seasonal thermal expansion, which pushes up the soil on either side of the wedge (Davis, 2001). The coalescence
of these thermal contraction cracks with raised edges results in thermal contraction polygons with low centers.
High-centered polygons can form indirectly from degradation of low-centered polygons due to an increase in
seasonal thaw depth, which causes melting and preferential settlement at ice-wedge edges. Commonly attributed
causes for this melting or thaw settlement on Earth are
a warming climate, a change in the overlying vegetation,
or an improvement in drainage conditions (French, 1996;
Davis, 2001). These high-centered degradation polygons are
up to a few tens of meters in diameter (Washburn, 1980;
French, 1996; Davis, 2001), and up to a few meters in relative relief (French, 1996; Davis, 2001).
In dry cold climates, high centered polygons have been
hypothesized to have formed directly (Marchant et al.,
2002). In the Dry Valleys of Antarctica, seasonal thermal
contraction has produced polygonal cracks in near-surface
Miocene glacial ice. Coeval sublimation of the underlying
Miocene glacial ice has resulted in an overlying lag of glacial
till. Passive sorting of this till into the thermal contraction
cracks has filled them with a coarse, porous gravel and cob-
Young ground ice features, Athabasca Valles, Mars
67
Fig. 15. Three ice wedges along the Tuktoyaktuk coast, in very close proximity to thermal contraction polygons. Three people at left provide scale.
Fig. 16. Aerial view of high-centered polygons, Tuktoyaktuk Peninsula.
ble debris. This allows an increased rate of ice sublimation
from the cracks. The unsorted, finer-grained sediment over
polygon centers is of much lower permeability. This process
results in lower sublimation rates, which leaves the centers high-standing. The results are high-centered sublimation
polygons with maximum diameters ranging from 9 to 35 m,
and bounding trough depths of 1.2 to 3.0 m (Marchant et al.,
2002).
2.2.3. Comparison
Based on their plan view size, relief, and morphology, the
high-centered Athabascan polygons could be either degradation or ‘sublimation’ polygons. The collapse of the lowcentered polygons into high-centered degradation polygons
would accord with the inferred transition of pingos into collapsing pingos and pingo scars. On Earth, thermal contraction ice-wedges (e.g., Fig. 15) and high-centered degradation
polygons (e.g., Fig. 16) are commonly found in close spatial
association with pingos in saturated sands (Mackay, 1973,
1979; Soloviev, 1973). The precipitation implicit in terrestrial degradation ice wedge polygons could have come in the
form of either snow (cf. Christensen, 2003) or as rain precip-
itated by volatiles from the Cerberus Plains lavas (cf. Plescia,
1993).
The Antarctic ‘sublimation’ polygons documented by
Marchant et al. (2002) form in the absence of precipitation,
which would be more consistent with the pervasive climate
of Mars. However, the sublimation polygons have formed
on massive glacial ice, which would not provide the groundwater mobility necessary for pingo formation. Elsewhere in
Antarctica, sediment-filled thermal contraction cracks contain windblown sand instead of glacial till (Péwé, 1959).
This aeolian infill is attractive as an analogue for this dusty
area of Mars. Relict (Pleistocene) sand wedges are also
found along the Tuktoyaktuk coastline near pingo-related
features, where they extend 10 m deep into massive ice
(Murton and French, 1993, 1994). However, we have not
found documentation in the literature of high-centered polygons surrounded by sand- (instead of by ice-)wedges.
In summary, none of these terrestrial analogues seems the
complete analogue for the high-centered polygonal terrain in
Athabasca. If local precipitation occurred, degradation polygons from ice-wedges would have been possible. Any ice in
this equatorial region is likely short-lived, so that secondary
modification, such as sedimentary crack infill to create sandwedges, may apply. Alternatively, the formation of the sublimation polygons formed from sediment, if they could form
in ice-rich frozen sediments instead of pure glacial ice, might
also be possible. The thermal contraction cracking implicit
in periglacial polygonal terrain is seasonal on Earth, but may
be due to longer-term orbital variation on Mars (cf. Laskar
et al., 2002).
2.2.4. Alternative hypotheses
Lava flows often show polygonal cracking (columnar
jointing) resulting from contraction due to cooling. These
joint-bounded basaltic columns can range in diameter up to
a few hundred meters (DeGraff and Aydin, 1987), encompassing the size of the Athabascan polygons. However, when
68
D.M. Burr et al. / Icarus 178 (2005) 56–73
exposed by erosion, they do not show a high-centered morphology.
Primary lava surfaces on Earth can display polygonal terrain in which the individual polygons are several meters in
diameter (i.e., smaller than the Athabascan polygons on average) but do have centers that are a few tens of centimeters
higher than the edges (Carr and Greeley, 1980). Polygonal
terrain in the Cerberus region has been previously interpreted as a primary lava surface (Keszthelyi et al., 2004b),
and this classification also appears reasonable for this local
area, based on the morphology and measured vertical dimensions of these polygons. This volcanic classification of the
polygons appears inconsistent with the volcanic basaltic ring
structure classification of the rimmed forms, as basaltic ring
structures require catastrophic flood erosion of the lava to be
exposed (Jaeger et al., 2003, 2005). However, this volcanic
classification of the polygons is consistent with a rootless
cone classification, as the rootless cones sit on pristine lava
(Lanagan et al., 2001; Greeley and Fagents, 2001).
As presented in Section 2.2.4, we also interpret the
Athabascan polygonal substrate to be more likely sedimentary than lava bedrock, which would be inconsistent with a
rootless cone origin of the mounds and rimmed forms. Desiccation polygons, which form as a result of drying of wet
sediments, would be consistent with a sedimentary substrate.
However, desiccation polygons are noted for their extremely
flat, not high-centered, morphology (Neal et al., 1968).
Thus, we find that the size, morphology, and inferred
composition of the Athabascan polygonal terrain taken together are less consistent with volcanic morphologies or desiccation polygons than with thermal contraction polygons.
3. Source of the ground ice
3.1. Mounds and rimmed features
Other young (late Amazonian) ground ice features on
Mars are located in the mid- to polar-latitudes, where
they are interpreted to have derived from ground ice emplaced from the atmosphere during periods of high obliquity (Mustard et al., 2001; Kreslavsky and Head, 2002;
Head et al., 2003b). Gamma Ray Spectrometer (GRS) data
indicate that extensive near-surface ground ice is currently
present at these higher latitudes (Boynton et al., 2002;
Feldman et al., 2002). Modelers of this data interpret this
higher-latitude near-surface ground ice to have resulted from
equilibrium vapor diffusion into the surface from the atmosphere (Mellon et al., 2004), or atmospheric deposition of
surface ice and burial at recently higher obliquity (Mischna
et al., 2003). The GRS data show that the area around
Athabasca Valles is relatively hydrogen-poor (see Boynton
et al., 2002, Fig. 6). However, the hypothesized ground ice
features would only make up a small portion of the ∼300 km
pixel of the GRS composition maps. To be discernable in
GRS data, the ground ice features would have to cover a
significantly larger area. Thus, the GRS data are not inconsistent with localized near-surface ice. The GRS data do
argue against a broad ground ice layer, as would be emplaced from the atmosphere.
Pingo ice cores are comprised of predominately massive
ice, either intrusive ice, which is formed by downward freezing at the top of a subpingo water lens (Mackay, 1998, p.
277), or segregated ice, which is formed in frost susceptible soils from cryosuction (Mackay, 1973, 1998). Terrestrial
(McKay et al., 1998; Davis, 2001) and martian (Mellon and
Jakosky, 1995; Mellon et al., 2004) studies show that atmospherically derived water produces cavity or pore ice, not
massive ice. An atmospheric water source is unlikely to have
formed the massive ground ice implicated in the hypothesis that the mounds and rimmed features are pingo-related
forms. Thus, an atmospheric water source for these forms is
discounted.
The Athabascan features could have resulted from very
local emplacement and freezing of flood sediments, resulting in hydrostatic-type pingo genesis. The large number and
density of hydrostatic pingos on the Tuktoyaktuk Peninsula
and in Yakutia indicate that conditions at these locations
are most conducive to hydrostatic pingo genesis. The pingos in both of these environments are formed in saturated
sediments, either on catastrophically drained lake bottoms
(Mackay, 1979) or on glaciofluvial flood terraces (Soloviev,
1973). In Tuktoyaktuk, the sediments are coarse silts to fine
sands usually capped by lacustrine silts (Mackay, 1979). The
sands allow pore water expulsion ahead of a downward and
inward freezing front (Mackay, 1979), whereas the finer,
capping silts have a lower permeability. Both Tuktoyaktuk
drained lakes and Yakutian flood terraces are conducive to
the emplacement, sudden exposure, and freezing of saturated
sandy-silty material.
The pingo-related features in Athabasca Valles are found
in a setting that may likewise have experienced emplacement, sudden exposure, and rapid freezing of saturated
sand and silt and overlying finer-grained sediments. Martian floods could have been sediment-rich, containing up to
60–70% by weight sediment (Komar, 1980). Modeling suggests that the Athabascan dune flood sediments would have
been at least gravels (>2 mm) (Burr et al., 2004). However, finer material could also have been deposited during
the waning stages of the flood or during floodwater ponding
(Burr, 2005) if such sediment had been available for transport. Clay, which is a weathering product of basaltic lavas,
is a fine-grained material likely to have been available for
flood transport in this volcanically active region. Thus, sediment deposition from flood flow at this location in Athabasca
Valles could reasonably have produced a sedimentology and
even a relative sedimentary stratigraphy similar to that of
the terrestrial pingo-rich sites. (As noted above, terrestrial
hydraulic pingos can also develop in much coarser material, even in bedrock (Müller, 1959).) Cessation of flooding
would have resulted in sudden exposure of the sediments to
freezing atmospheric temperatures. Residual ponds or areas
Young ground ice features, Athabasca Valles, Mars
of saturated sediment, by minimizing permafrost development, would have provided sites for pingo growth, closely
analogous to those of terrestrial hydrostatic pingos.
As a second possibility, the Athabascan features could
have formed as hydraulic-type pingos, in which magmatic
cycling of groundwater could have provided the necessary
hydraulic head. This could have been artesian head from
Elysium Mons (cf. Gulick, 1998). Alternatively, the head
could have been provided by hydrothermal convection of
ground water (cf. Travis et al., 2003) in association with cryovolcanism (cf. Gaidos, 2001). Locally enhanced heat flow
is suggested by the nearby fissure (Figs. 1 and 12) and would
have thinned the otherwise continuous permafrost into local
discontinuity. By analogy with terrestrial hydraulic pingos,
this discontinuous permafrost would have provided a site for
pressurized ground water to well up, forming a pingo.
Either a hydrostatic-type or a hydraulic-type pingo genesis requires some period of mobile liquid water at or near
Mars’ surface. Terrestrial pingos take on the order of several tens of years to >1000 years to form (Mackay, 1998;
Holmes et al., 1968), during which near-surface groundwater is flowing towards and freezing onto the ice core. Modeling shows that the Elysium plains including Athabasca
Valles meet the minimum requirements for the existence of
liquid water under Mars’ present climate (Haberle et al.,
2001). The likely briny nature of martian groundwater (Burt
and Knauth, 2003), along with the freezing point depression that results from solute rejection during pingo formation
(Mackay, 1985), makes this especially tenable.
69
could be floodwater or magmatically cycled groundwater,
as with the mounds and rimmed features, or it could be atmospheric water, possibly including snow, during periods of
high obliquity.
4. Age of the ground ice
Multiple floods may have flowed down Athabasca Valles
(Burr et al., 2002b; Keszthelyi et al., 2004a). If the mounds
and rimmed features had pre-dated the most recent Athabascan flood, they should have been at least modified by the
flood. The apparent absence of streamlining of the features
or other evidence for erosion or deposition indicates that
these features developed after the last floodwater to cover
this area.
The most recent geomorphologically effective flood has
been dated at 2–8 Ma (Burr et al., 2002b). Age-dating from
small, sparse craters is uncertain, and Athabasca Valles may
be one to two orders of magnitude older than dated (McEwen
et al., 2005). If this were the case, it would still remain
the catastrophic outflow channel with the youngest model
age, and these ground ice features would still date from the
late Amazonian epoch. It is possible that this most recent
flood may not have covered the particular area of the channel
showing the hypothesized ground ice features. In this case,
the features would be older, dating to the last flood that inundated the area. The preservation of fine textures and small
details of these features in this dusty location argues for their
recent formation.
3.2. Polygonal terrain
As discussed above, the mounds and rimmed features
require mobile groundwater to form their massive ice
cores. However, this constraint does not apply to terrestrial polygonal terrain resulting from thermal contraction cracking. Polygons on Mars have been interpreted to
have resulted from surface water (e.g., Lucchitta, 1981;
Costard and Kargel, 1995; Seibert and Kargel, 2001), but
may also result from emplacement of atmospheric water.
Extreme temperature changes during periods of high obliquity could produce the necessary cooling to result in thermal
contraction cracks. Modeling (Mellon and Jakosky, 1995;
Mellon et al., 2004) indicates that atmospheric water could
have been deposited at Mars’ equator during such recent
(<5 Ma) periods of high obliquity. Snow (cf. Christensen,
2003) might have filled in the cracks, leading to ice-wedge
polygon formation as it does on Earth. This snow is not required, however; cryosuction of atmospheric water vapor
into the thermal contraction cracks might have created icewedge polygons. Alternatively, the cracks might have been
filled with sediment. These scenarios are consistent with
that discussed by Mellon and Jakosky (1995), pp. 11797–
11798, regarding the possible formation of polygonal terrain
at Mars’ equator. Thus the source of the ground ice implicit
in the formation of polygonal terrain is indeterminate. It
5. Summary and implications
We interpret a suite of four features in Athabasca Valles
as resulting from young ground ice. We suggest that the
mounds are analogous to terrestrial pingos, the conical features with summit depressions are analogous to pingos in
various stages of collapse, and the flatter more irregular features are analogous to pingo scars (Fig. 3). We interpret the
high-centered polygons among these hypothesized pingorelated forms as either degradation or sublimation thermal
contraction polygons. These interpretations are made on the
basis of (1) similar morphology and continuous range of
morphologies, (2) similar dimensions as measured in planview and with photoclinometry and shadow measurements,
and (3) evidence from the thermal IR data and geological
features for a sedimentary substrate. The close spatial relationship between pingo-related features and polygonal terrain here as in terrestrial periglacial environments mutually
reinforces these interpretations of the Athabascan features.
We do not categorically refute all other hypotheses, and in
fact find the surficial ground ice hypothesis of Gaidos and
Marion (2003) to be equally viable.
Ground ice features have been used to reconstruct paleoclimates in North American, Europe, Asia, and Antarctica on
70
D.M. Burr et al. / Icarus 178 (2005) 56–73
Earth (Flemal, 1976; De Gans, 1988; French, 1996, and references therein), and in the mid- to polar-latitudes on Mars
(e.g., Costard and Kargel, 1995; Seibert and Kargel, 2001;
Mustard et al., 2001; Head et al., 2003b; Soare et al.,
2005). On Mars, ground ice features have been ascribed
to atmospherically emplaced water. However, in Athabasca,
we argue that the massive ground ice for the mounds and
rimmed features was derived either from floodwater or from
magmatically cycled groundwater. Thus, we do not ascribe
the genesis of these features to climatic factors. Their secondary characteristics, such as the large exterior annuli or
the interior cones, may have some relation to climatic oscillations. The polygonal terrain, if correctly hypothesized
to have formed from thermal contraction cracks, likewise
implies some subzero temperature oscillation that may have
been climatic.
The rimmed forms, hypothesized to be collapsing or collapsed pingos, comprise 90% of the mounds and rimmed
forms in Athabasca Valles, suggesting a periglacial landscape in collapse. Degradation polygons, one possible interpretation of the high-centered Athabascan polygonal terrain,
likewise imply a periglacial landscape in collapse. On Earth,
such a transition over a large area is generally attributed to
a change in climate, although smaller areas of collapse may
result from thermal erosion after river avulsion or other local causes (French, 1996). In Athabasca Valles, where the
ground ice for at least the rimmed features may have derived
from a discrete (flood) event, the collapse may be the result
of the evaporation or sublimation of the groundwater or ice.
That mounds are extant suggests that, if the mounds are pingos, this collapse may be on-going.
In Athabasca, all of the flatter, irregular features are collapsed, whereas the taller, more circular features as a group
show less collapse. This distribution is consistent with current models of the physics of pingo growth for a group of
pingos that all originated at the same time. The application
of plate theory (Pollard and Johnson, 1973) to pingos indicates that subpingo water lenses should increase in height
with increasing pingo elongation (Mackay, 1987). Increasing water lens height leads to lens rupture and pingo collapse. For pingos initiated diachronously, this implies that
the more elongate pingos would collapse sooner relative to
initial growth, as is observed on the Tuktoyaktuk Peninsula
(Mackay, 1987). For a group of pingos initiated simultaneously, this would imply that the more irregularly shaped
pingos in the group would collapse first, leaving the more
circular pingos standing longer. Thus, at any one time, the
more irregular forms would have lower heights and the more
circular forms would have higher heights, as is observed in
Athabasca Valles.
About 10% (∼20) of the Athabascan features are entirely positive relief mounds, showing no evidence of collapse. These features are hypothesized above to be pingos.
Hydrostatic pingos appear to maintain their positive relief
form only with a massive ice core, the removal of which
causes collapse (Mackay, 1986, 1987, 1998). We infer that,
if the Athabascan mounds are hydrostatic-type pingos, they
contain massive ground ice. Regression of plotted measurements from Canadian, Siberian, and (hydraulic) Alaskan
pingos (see Mackay, 1987, Fig. 10) shows that the overburden thickness on top of pingos is linearly proportional to the
pingo size (radius), such that
To = 0.13r − 1.26,
where To is the overburden thickness and r is the pingo radius in meters. Applying this regression line to the Athabascan mounds (n = 22; average radius = ∼13 m; standard
deviation = 3.2 m) gives a maximum overburden thickness
of ∼1.6 m, with an average thickness of ∼0.5 m. Of course,
the minimum overburden thickness would equal 0 where the
regression line intercepts the abscissa or pingo radius, but the
regression line may not be applicable at these smallest sizes.
This regression line appears to describe well pingos greater
than ∼60 m in radius, but the data appear notably more scattered below this size (see Mackay, 1987, Fig. 10). Also, the
regression line is not drawn by Mackay (1987) as intercepting the abscissa, perhaps in recognition of this physical limit
to the overburden thickness. We also note the suggestion
that hydraulic pingos may replace their ice cores with finegrained sediments from groundwater and thereby retain their
positive relief mound form even without ice cores (Davis,
2001, p. 129). So if our hypothesis is correct, pingo ice may
lie very close to the surface of the mounds in Athabasca
Valles, but a more conclusive determination of this requires
further consideration, including determination of the type of
pingo and modeling of pingo formation under martian conditions.
As noted above, a sedimentary substrate would be consistent with the hypothesis that the rimmed forms are kettle holes (Gaidos and Marion, 2003). If this hypothesis
is correct, then the mounds could be ‘sediment mounds,’
features a few meters in diameter that are found interspersed with kettle holes on glacial outwash plains (Fay,
2002). They may be a result of melting of particularly
sediment-rich ice blocks to create positive-relief ‘till-fill’
kettle holes (cf. Maizels, 1992). If this hypothesis is correct
for the Athabascan features, there would not be ice in the
mounds.
The mounds and rimmed features have two sets of implications: for age-dating from small craters, and for astrobiology. The channel has been dated by crater counting areas
of the channel floor interpreted as flood carved (Burr et al.,
2002b). However, recent investigations of secondary cratering on Mars throw doubt on age-dating from small craters
and suggest that the age derived for Athabasca may be one
to two orders of magnitude too young (McEwen et al., 2005).
An independent age estimate of the channel would constrain
the age for the most recent flooding on Mars. The collapsed
pingo features could provide that independent estimate. Determining the length of time required for the sublimation of
pingo ice through a few meters’ overburden for this location on Mars would constrain the maximum age of the most
Young ground ice features, Athabasca Valles, Mars
recent flood. This independent age estimate would also constrain the amount of error associated with age dating from
small craters
If these hypothesized pingos in Athabasca are hydrostatic, the ice that forms them emerged as water from a very
recently active volcanotectonic fissure (Burr et al., 2002a;
Berman and Hartmann, 2002; Head et al., 2003a). This water
flow up the fissure was likely fed from a few-kilometersdeep aquifer (Burr et al., 2002b). Thus, this water may hold
evidence of a hypothesized deep subsurface biosphere (Fisk
and Giovannoni, 1999). The immediate burial of the water within the sediments would have protected this evidence
from UV radiation. Thus, the ground ice in Athabasca Valles
may be the most likely surface, equatorial, site yet documented to hold evidence of biotic or prebiotic activity on
Mars.
Acknowledgments
DMB acknowledges support from the NASA Planetary
Geology and Geophysics Program through the E.M. Shoemaker Fellowship at the U.S. Geological Survey during research and manuscript production and revision. The licenses
under which the included photos were collected were issued
to R.J.S. by Parks Canada (permit number 2004-02), the Inuvialuit Land Administration (right number ILA04TN022),
and the Aurora Research Institute (scientific research license
#13616N). We thank Michelle Crossfield (the Aurora Research Institute), Brian Johnston (Parks Canada) and Jim
Taggert at the ILA for their gracious assistance. We also
thank Windy Jaeger for pleasant and interesting discussions
on this subject, and Jeff Kargel, Baerbel Lucchitta, and Laszlo Keszthelyi for helpful internal reviews. Nathalie Cabrol
and an anonymous reviewer are thanked for their external
reviews.
References
Baker, V.R., 2001. Water and the martian landscape. Science 412, 228–236.
Berman, D.C., Hartmann, W.K., 2002. Recent fluvial, volcanic, and tectonic
activity on the Cerberus Plains of Mars. Icarus 159, 1–17.
Beyer, R.A., McEwen, A.S., Kirk, R.L., 2003. Meter-scale slopes of candidate MER landing sites from point photoclinometry. J. Geophys.
Res. 108 (E12), 8085, doi:10.1029/2003JE002120.
Boynton, W.V., 24 colleagues, 2002. Distribution of hydrogen in the near
surface of Mars: Evidence for subsurface ice deposits. Science 296, 81–
85.
Burr, D.M., 2005. Clustered streamlined forms in Athabasca Valles, Mars:
Evidence for sediment deposition during floodwater ponding. Geomorphology. In press.
Burr, D.M., McEwen, A.S., Sakimoto, S.E.H., 2002a. Recent aqueous
floods from the Cerberus Fossae, Mars. Geophys. Res. Lett. 29 (1),
doi:10.1029/2001GL013345.
Burr, D.M., Grier, J.A., McEwen, A.S., Keszthelyi, L.P., 2002b. Repeated aqueous flooding from the Cerberus Fossae: Evidence for
very recently extant, deep groundwater on Mars. Icarus 159, 53–73,
doi:10.1006/icar.2002.6921.
71
Burr, D.M., Carling, P.A., Beyer, R.A., Lancaster, N., 2004. Flood-formed
dunes in Athabasca Valles, Mars: Morphology, modeling, and implications. Icarus 171, 68–83.
Burt, D.M., Knauth, L.P., 2003. Electrically conducting, Ca-rich brines,
rather than water, expected in the martian subsurface. J. Geophys.
Res. 108 (E4), 8026, doi:10.1029/2002JE001862.
Cabrol, N.A., Grin, E.A., Pollard, W.H., 2000. Possible frost mounds in an
ancient martian lake bed. Icarus 145, 91–107.
Cantor, B.A., James, P.B., Caplinger, M., Wolff, M.J., 2001. Martian dust
storms: 1999 Mars Orbiter Camera observations. J. Geophys. Res. 106
(E10), 23653–23687.
Carling, P.A., Kirkbride, A.D., Parnachov, S., Borodavki, P.S., Berger,
G.W., 2002. Late Quaternary catastrophic flooding in the Altai Mountains of south-central Siberia: A synoptic overview and an introduction
to flood deposit sedimentology. Spec. Publs. Int. Ass. Sediment. 32, 17–
35.
Carr, M.H., 1996. Water on Mars. Oxford Univ. Press, New York. 229 p.
Carr, M.H., Schaber, G.G., 1977. Martian permafrost features. J. Geophys.
Res. 82 (28), 4039–4054.
Carr, M.H., Greeley, R., 1980. Volcanic Features of Hawai’i: A Basis for
Comparison with Mars, Publication SP-403. NASA, Washington, DC.
211 p.
Christensen, P.R., 2003. Formation of recent martian gullies through melting of extensive water-rich snow deposits. Nature 422, 45–48.
Christensen, P.R., 21 colleagues, 2003. Morphology and composition of the
surface of Mars: Mars Odyssey THEMIS results. Science 300, 2056–
2061.
Clifford, S.M., 1993. A model for the hydrologic and climatic behavior of
water on Mars. J. Geophys. Res. 98 (E6), 10973–11016.
Clifford, S.M., Hillel, D., 1983. The stability of ground ice in the equatorial
region of Mars. J. Geophys. Res. 88, 2456–2474.
Costard, F.M., Kargel, J.S., 1995. Outwash plains and thermokarst on Mars.
Icarus 114, 93–112.
Davis, N., 2001. Permafrost: A Guide to Frozen Ground in Transition. Univ.
of Alaska Press, Fairbanks, AK, 351 p.
De Gans, W., 1988. Pingo scars and their identification. In: Clark, M.J.
(Ed.), Advances in Periglacial Geomorphology. Wiley, New York,
pp. 299–322.
DeGraff, J.M., Aydin, A., 1987. Surface morphology of columnar joints and
its significance to mechanics and direction of joint growth. Geol. Soc.
Am. Bull. 99, 605–617.
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.
Fay, H., 2002. Formation of kettle holes following a glacial outburst flood
(jökulhlaup), Skeid̄arársandur, southern Iceland. In: The Extremes of
the Extremes: Extraordinary Floods (Proceedings of a Symposium Held
at Reykjavík, Iceland, July 2000), IAHS Publ. No. 271.
Feldman, W.C., 12 colleagues, 2002. Global distribution of neutrons from
Mars: Results from Mars Odyssey. Science 297, 75–78.
Fisk, M.R., Giovannoni, S.J., 1999. Sources of nutrients and energy for a
deep biosphere on Mars. J. Geophys. Res. 104 (E5), 11805–11815.
Flemal, R.C., 1976. Pingos and pingo scars: Their characteristics, distribution, and utility in reconstructing former permafrost environments.
Quat. Res. 6, 37–53.
French, H.M., 1996. The Periglacial Environment, second ed. Addison–
Wesley–Longman, Essex, 341 p.
Gaidos, E.J., 2001. Cryovolcanism and the recent flow of liquid water on
Mars. Icarus 153, 218–223.
Gaidos, E., Marion, G., 2003. Geological and geochemical legacy of
a cold, early Mars. J. Geophys. Res. 108 (E6), 5005, doi:10.1029/
2002JE002000.
Greeley, R., Iverson, J.D., 1985. Wind as a Geologic Process on Earth,
Mars, Venus and Titan. Cambridge Univ. Press, Oxford, Great Britain.
333 p.
Greeley, R., Fagents, S.A., 2001. Icelandic pseudocraters as analogs to
some volcanic cones on Mars. J. Geophys. Res., 20527–20546.
72
D.M. Burr et al. / Icarus 178 (2005) 56–73
Greeley, R., Kraft, M.D., Kuzmin, R.O., Bridges, N.T., 2000. Mars
Pathfinder landing side: Evidence for a change in wind regime from
lander and orbital data. J. Geophys. Res. 105 (E1), 1829–1840.
Gulick, V.C., 1998. Magmatic intrusion and a hydrothermal origin for fluvial valleys on Mars. J. Geophys. Res. 103 (E8), 19365–19387.
Haberle, R.M., McKay, C.P., Schaeffer, J., Cabrol, N.A., Grin, E.A., Zent,
A.P., Quinn, R., 2001. On the possibility of liquid water on present-day
Mars. J. Geophys. Res. 106 (E10), 23317–23326.
Head, J.W., Wilson, L., Mitchell, K.L., 2003a. Generation of recent massive water floods at Cerberus Fossae, Mars by dike emplacement,
cryospheric cracking, and confined aquifer groundwater release. Geophys. Res. Lett. 30 (11), doi:10.1029/2003GL017135.
Head, J.W., Mustard, J.F., Kreslavsky, M.A., Milliken, R.E., Marchant,
D.R., 2003b. Recent ice ages on Mars. Nature 426, 797–802.
Hodges, C.A., 1978. Basaltic ring structures of the Columbia Plateau. Geol.
Soc. Am. Bull. 89, 1281–1289.
Holmes, G.W., Hopkins, D.M., Foster, H.L., 1968. Pingos in central Alaska:
U.S. Geological Survey Bulletin 1241-H, 40 p.
Jaeger, W.L., Keszthelyi, L.P., Burr, D.M., McEwen, A.S., Baker, V.R.,
Miyamoto, H., Beyer, R.A., 2003. Ring dike structures in the Channeled
Scabland as analogs for circular features in Athabasca Valles, Mars. Lunar Planet. Sci. XXXIV. Abstract 2045.
Jaeger, W.L., Keszthelyi, L.P., Burr, D.M., Emery, J.P., Baker, V.R.,
McEwen, A.S., Miyamoto, H., 2005. Basaltic ring structures as an
analog for ring features in Athabasca Valles, Mars. Lunar Planet.
Sci. XXXVI. Abstract 1886.
Keszthelyi, L.P., McEwen, A.S., Thordarson, Th., 2000. Terrestrial analogs
and thermal models for martian flood lavas. J. Geophys. Res. 105,
15027–15049.
Keszthelyi, L., Burr, D.M., McEwen, A.S., 2004a. Geomorphologic/thermophysical mapping of the Athabasca region, Mars, using
THEMIS infrared imaging. Lunar Planet. Sci. XXXV. Abstract 1657.
Keszthelyi, L., Thordarson, T., McEwen, A., Haack, H., Guilbaud, M.-N.,
Self, S., Rossi, M.J., 2004b. Icelandic analogs to martian flood lavas.
Geochem. Geophys. Geosyst. 5, Q11014.
Kieffer, H.H., Martin, T.Z., Peterfreund, A.R., Jakosky, B.M., 1977. Thermal and albedo mapping of Mars during the Viking primary mission.
J. Geophys. Res. 82 (28), 4249–4291.
Kirk, R.L., Barrett, J.M., Soderblom, L.A., 2003. Photoclinometry
made simple. . .? Paper presented as ISPRS-ET Working Group IV/9
Workshop “Advances in Planetary Mapping 2003,” Int. Soc. for
Photogramm. and Remote Sens., Houston, TX (available at http:
//astrogeology.usgs.gov/Projects/ISPRS/MEETINGS/Houston2003/
abstracts/Kirk_irprs_mar03.pdf).
Komar, P.D., 1980. Modes of sediment transport in channelized water
flows with ramifications to the erosion of the martian outflow channels.
Icarus 42, 317–329.
Kreslavsky, M.A., Head III, J.W., 2002. Mars: Nature and evolution of
young latitude-dependent water-ice-rich mantle. Geophys. Res. Lett. 29,
doi:10.1029/2002GL015392.
Lachenbruch, A.H., 1962. Mechanics of thermal contraction cracks and icewedge polygons in permafrost. Geological Society of America, Special
Paper 70, 69 p.
Laskar, J., Levrand, B., Mustard, J.F., 2002. Orbital forcing of the martian
polar layered deposits. Nature 419, 375–377.
Lanagan, P.D., 2004. Geologic history of the Cerberus Plains, Mars. Ph.D.
thesis: University of Arizona, Tucson, 142 p.
Lanagan, P.D., McEwen, A.S., Keszthelyi, L.P., Thordarson, Th., 2001.
Rootless cones on Mars indicating the presence of shallow equatorial
ground ice in recent times. Geophys. Res. Lett. 28 (12), 2365–2367.
Levoy, C., Wood, S.E., Catling, D., Montgomery, D.R., Moore, J., Barnhart,
C., Ginder, E., Louie, M., 2004. Evidence for possible exposed water
ice deposits in martian low latitude chasms and chaos. Lunar Planet.
Sci. XXXV. Abstract 2016.
Lucchitta, B.K., 1981. Mars and Earth: Comparison of cold-climate features. Icarus 45, 264–303.
Mackay, J.R., 1963. Pingos in Canada. In: Proceedings of the International Conference on Permafrost, Lafayette, IN. National Academy of
Science-National Research Council, Washington, DC. Publ. no. 1287.
Mackay, J.R., 1973. The growth of pingos: Western Arctic coast, Canada.
Can. J. Earth Sci. 10, 979–1004.
Mackay, J.R., 1979. Pingos of the Tuktoyaktuk Peninsula area, Northwest
Territories. Géogr. Phys. Quat. 23 (1), 3–61.
Mackay, J.R., 1985. Pingo ice of the western Arctic coast, Canada. Can. J.
Earth Sci. 22, 1452–1464.
Mackay, J.R., 1986. Growth of Ibyuk Pingo, western arctic coast, Canada,
and some implications for environmental reconstructions. Quat. Res. 26,
68–80.
Mackay, J.R., 1987. Some mechanical aspects of pingo growth and failure,
western Arctic coast, Canada. Can. J. Earth Sci. 24, 1108–1119.
Mackay, J.R., 1998. Pingo growth and collapse, Tuktoyaktuk Peninsula
area, western arctic coast, Canada: A long-term field study. Géogr. Phys.
Quatern. 52 (3), 271–323.
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 ground and sublimation till over Miocene glacial ice
in Beacon Valley, southern Victoria Land, Antarctica. Geol. Soc. Am.
Bull. 114 (6), 718–730.
Maizels, J., 1992. Boulder ring structures produced during jökulhlaup flows.
Geogr. Ann. A 74 (1), 21–33.
McEwen, A.S., Preblich, B.S., Turtle, E.P., Artemieva, N.A., Golombek,
M.P., Hurst, M., Kirk, R.L., Burr, D.M., Christensen, P.R., 2005. The
Rayed crater Zunil and interpretations of small impact craters on Mars.
Icarus. In press.
McKay, C.P., Mellon, M.T., Friedmann, E.I., 1998. Soil temperatures and
stability of ice-cemented ground in the McMurdo Dry Valleys, Antarctica. Antarct. Sci. 10, 31–38.
McKee, B., Stradling, D., 1970. The sag flowout: A newly described volcanic structure. Geol. Soc. Am. Bull. 81, 2035–2044.
Mellon, M.T., Jakosky, B.M., 1995. The distribution and behavior of martian ground ice during past and present epochs. J. Geophys. Res. 100
(E6), 11781–11799.
Mellon, M.T., Jakosky, B.M., Kieffer, H.H., Christensen, P.R., 2000.
High-resolution thermal inertia mapping from the Mars Global
Surveyor thermal emission spectrometer. Icarus 148, 437–455,
doi:10.1006/icar.2000.6503.
Mellon, M.T., Feldman, W.C., Prettyman, T.H., 2004. The presence and
stability of ground ice in the southern hemisphere of Mars. Icarus 169,
324–340.
Mischna, M.A., Richardson, M.I., Wilson, R.J., McCleese, D.J., 2003. On
the orbital forcing of martian water and CO2 cycles: A general circulation model study with simplified volatile schemes. J. Geophys. Res. 108
(E6), 5062, doi:10.1029/2003JE002051.
Müller, F., 1959. Observations on pingos (in German). Meddeleser om
Grønland, 153, 127 p. Translated by D.A. Sinclair, National Research
Council of Canada, Ottawa, TT-1073, 1963, 117 p.
Murton, J.B., French, H.M., 1993. Sand wedges and permafrost history,
Crumbling Point, Pleistocene Mackenzie Delta, Canada. In: Proceedings of the Sixth International Conference on Permafrost, vol. 1. Science
Press, Beijing, pp. 482–487.
Murton, J.B., French, H.M., 1994. Cryostructures in permafrost, Tuktoyaktuk coastlands, western arctic Canada. Can. J. Earth Sci. 31, 737–747.
Mustard, J.F., Cooper, C.D., Rifkin, M.K., 2001. Evidence for recent climate change on Mars from the identification of youthful near-surface
ground ice. Nature 412, 411–414.
Neal, J.T., Langer, A.M., Kerr, P.F., 1968. Giant desiccation polygons of
Great Basin Playas. Geol. Soc. Am. Bull. 79, 69–90.
Péwé, T.L., 1959. Sand wedge polygons (tessellations) in the McMurdo
Sound region, Antarctica. Am. J. Sci. 257, 545–552.
Plescia, J.B., 1990. Recent flood lavas in the Elysium region of Mars.
Icarus 88, 465–490.
Plescia, J.B., 1993. An assessment of volatile release from recent volcanism
in Elysium, Mars. Icarus 104, 20–32.
Young ground ice features, Athabasca Valles, Mars
Plescia, J.B., 2003. Cerberus Fossae, Elysium, Mars: A source for lava and
water. Icarus 164, 79–95.
Pollard, D.D., Johnson, A.M., 1973. Mechanics of growth of some laccolithic intrusions in the Henry Mountains, Utah. II. Bending and failure
of overburden layers and sill formation. Tectonophysics 18, 331–354.
Pollard, W.H., van Everdingen, R.O., 1992. Formation of seasonal ice bodies. In: Dixon, J.C., Abrahams, A.D. (Eds.), Periglacial Geomorphology. Wiley, New York, pp. 281–304.
Seibert, N.M., Kargel, J.S., 2001. Small-scale martian polygonal terrain:
Implications for liquid surface water. Geophys. Res. Lett. 28 (5), 899–
902.
Soare, R., Burr, D.M., Wan Bun Tseung, J.-M., 2005. Possible pingos and a
periglacial landscape in northwest Utopia Planitia. Icarus 174, 373–382.
73
Soloviev, P.A., 1973. Thermokarst phenomena and landforms due to frost
heaving in Central Yakutia. Biuletyn Peryglacjalny 23, 135–155.
Squyres, S.W., Carr, M.H., 1986. Geomorphic evidence for the distribution
of ground ice on Mars. Science 231, 249–252.
Tanaka, K.L., 1986. The stratigraphy of Mars. J. Geophys. Res. 91, 139–
158.
Travis, B.J., Rosenberg, N.D., Cuzzi, J.N., 2003. On the widespread subsurface convection in bringing liquid water close to Mars’ surface. J. Geophys. Res. 108 (E4), 8040, doi:10.1029/2002JE001877.
Washburn, A.L., 1980. Geocryology. Wiley, New York. 406 p.
Watson, E., 1971. Remains of pingos in Wales and the Isle of Man. Geol.
J. 7 (2), 381–392.