Icarus 189 (2007) 118–135
www.elsevier.com/locate/icarus
Valley formation on martian volcanoes in the Hesperian: Evidence for
melting of summit snowpack, caldera lake formation, drainage and erosion
on Ceraunius Tholus
Caleb I. Fassett, James W. Head III ∗
Department of Geological Sciences, Brown University, Providence, RI 02912, USA
Received 21 August 2006; revised 20 December 2006
Available online 25 January 2007
Abstract
Ceraunius Tholus, a Hesperian-aged volcano in the Tharsis region, is characterized by small radial valleys on its flanks, and several larger
valleys originating near its summit caldera. All of these large valleys drain from near the lowest present portion of the caldera rim and down
the flanks of the volcano. The largest valley debauches into Rahe Crater (an oblique impact crater), forming a depositional fan. Recent study of
climate change on Mars suggests that many low-latitude regions (especially large volcanic edifices) were periodically the sites of snow accumulation, likely triggered by variations in spin orbital parameters. We apply a conductive heat flow model to Ceraunius Tholus that suggests that
following magmatic intrusion, sufficient heating would be available to cause basal melting of any accumulated summit snowpack and produce
sufficient meltwater to cause the radial valleys. The geometry of the volcano summit caldera suggests that meltwater would also accumulate in
a volumetrically significant caldera lake. Analysis of the morphology and volumes of the largest valley, as well as depositional features at its
base, suggest that fluvial erosion due to drainage of this summit caldera lake formed the large valleys, in a manner analogous to how valleys
were formed catastrophically from a lake in Aniakchak caldera in Alaska. Moreover, the event which carved the largest valley on Ceraunius
Tholus appears to have led to the formation of a temporary lake within Rahe Crater, at its base. The more abundant, small valleys on the flanks
are interpreted to form by radial drainage of melted ice or snow from the outside of the caldera rim. Comparison of Ceraunius Tholus with the
volcano-capping Icelandic ice sheet Myrdalsjokull provides insight into the detailed mechanisms of summit heating, ice-cap accumulation and
melting, and meltwater drainage. These observations further underline the importance of a combination of circumstances (i.e., climate change to
produce summit snowpack and an active period of magmatism to produce melting) to form the valley systems on some martian volcanoes and not
on others.
© 2007 Elsevier Inc. All rights reserved.
Keywords: Mars, surface; Geological processes
1. Introduction
Valley networks on Mars are integrated, branching valley
systems that occur primarily in the highlands and are interpreted as predominantly Noachian in age (e.g., Carr, 1995).
The characteristics of valley networks have been interpreted to
mean that they were formed by processes requiring atmospheric
temperatures and pressures sufficient to produce a “warmer
and wetter” climate (e.g., Baker, 2001; Craddock and Howard,
* Corresponding author. Fax: +1 401 863 3978.
E-mail address: [email protected] (J.W. Head III).
0019-1035/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.icarus.2006.12.021
2002). Younger, Hesperian-aged valley networks have been observed on some martian volcanoes, but not others, and these
younger features have also been interpreted to have formed
by surface water runoff and sapping (e.g., Gulick and Baker,
1990; Gulick et al., 1997; Gulick, 1998, 2001). If these valleys formed well after the earlier, Noachian-aged period of
postulated warmer, wetter climate, two questions are raised:
(1) does this later period of valley formation represent a return to warmer, wetter conditions, and (2) why did the valleys
form on some surfaces and volcanoes and not others? The presence of the valleys on these younger edifices could be explained
by one or more of the following scenarios: (1) ‘warm, wet’
conditions recurred subsequent to the Noachian, perhaps in an
Valley formation on Ceraunius Tholus, Mars
episodic manner (e.g., Gulick et al., 1997), (2) the younger valleys formed by a mechanism different than those formed earlier
(e.g., Baker, 2001; Gulick et al., 1997), and/or (3) the earlier,
Noachian networks may not require a “warmer, wetter” period
of climate history.
To address these important questions, we previously analyzed one of the examples of younger, Hesperian-aged, valley
systems interpreted to have formed by surface water runoff
and sapping (Gulick and Baker, 1990) on the martian volcano
Hecates Tholus (Fig. 1). We tested the hypothesis that these
radially arrayed valleys which extend from high on the volcano down the flanks to the base could have formed by melting
and drainage of summit snowpack (Fassett and Head, 2006;
see also Gulick et al., 1997; Gulick, 1998, 2001). This is motivated in part by recent analyses that indicate many low-to-mid
latitude areas have been the site of accumulation of abundant
snow and ice in the past history of Mars (e.g., Head et al.,
2005, 2006a, 2006b); the climatic forcing that causes deposition
of ice at low-to-mid latitudes is thought to be triggered by variations in spin orbital parameters (e.g., Laskar et al., 2004; Forget
et al., 2006). Furthermore, low-latitude mountain glaciers have
been observed on the flanks of the Tharsis Montes (e.g., Head
and Marchant, 2003; Shean et al., 2005) and climate models
have shown that periods of enhanced obliquity and the presence
of topographic prominences (the volcanoes) cause adiabatic upwelling and cooling of water-rich air, enhancing snow deposition (e.g., Forget et al., 2006). In the case of some of these other
Tharsis volcanoes, signatures of past glaciation are observed at
elevations above the maximum elevation of Hecates and Ceraunius Tholi (e.g., Shean et al., 2007). These findings support
the idea that volcanic edifices were likely sites for snow deposition during specific climatic periods related to spin orbital
excursions; indeed Hecates Tholus shows evidence of such deposits emplaced on its low flanks in the Amazonian (Hauber et
al., 2005).
We next assessed whether such snow and ice accumulation
could undergo melting under current environmental conditions
at Hecates Tholus. On the basis of the extremely cold, hyperarid conditions typical of the current Mars environment, topdown melting of snowpack is unlikely, and bottom-up melting
due to the existing average planetary thermal gradient is also
unlikely unless snowpack thicknesses exceed several kilometers (e.g., Carr and Head, 2003). On the other hand, Hecates
Tholus must have experienced periods of magmatic activity to
form the edifice, which could locally increase the thermal flux
and near-surface temperatures to cause basal melting of the
snowpack. We modeled a series of magma reservoir geometries within the volcano and found that intrusions would lead
to sufficient heat flux (consistent with terrestrial experience) to
cause basal melting of summit snowpack (Fassett and Head,
2006).
Using patterns of meltwater production and drainage associated with snowpack and glaciers in the Mars-like Antarctic Dry
Valleys, we further concluded that the distinctive radial channel patterns on the flanks of Hecates Tholus could result from
drainage of melting summit snowpack during periods when the
volcano was magmatically active. This is a plausible scenario
119
for explaining why valley networks are found only on some volcanoes since it involves a combination of two time-dependent
circumstances, (1) climatic conditions allowing deposition of
snow and ice on the volcano summit, and (2) sufficient magmatic activity to produce enhanced heat flow at the summit and
to trigger melting.
To test this hypothesis further, in this paper we examine Ceraunius Tholus (Figs. 1 and 2), another of the locations cited by
Gulick and Baker (1990) where valleys are found on a young
volcanic surface. We first compare the similarities and differences between Hecates and Ceraunius Tholi, and then assess
whether the observed features at Ceraunius could be explained
by the same mechanism as that proposed for Hecates Tholus
(Fassett and Head, 2006). We find that both volcanoes share
generally similar characteristics in terms of radial valleys and
their characteristics and distribution.
Differences include the much larger and asymmetric summit
caldera on Ceraunius, and the presence of several large valleys
that emanate from the low point in its caldera rim, and terminate in fan features at the base of the volcano. On the basis
of these observations, we conclude that like Hecates Tholus,
Ceraunius Tholus was the site of snow and ice accumulation,
and that during one or more of these periods basal melting
was triggered by local enhancement of the geothermal heat flux
(caused either by intrusive or extrusive volcanic activity and/or
hydrothermalism). This melting in the summit region led to the
formation of the small radial valleys, as we believe was also
the case at Hecates. Unlike Hecates, however, basal melting
filled the caldera and formed a substantial lake, which was overtopped and drained, forming the large canyons that emanate
from the lowest part of the caldera rim. These large drainages
from a caldera-filling lake are similar to the current situation at
Myrdalsjokull, an Icelandic subglacial volcano, which we explore as a potential analog.
2. Geological characteristics of Ceraunius Tholus
Ceraunius Tholus is a member of the Uranius group of
volcanoes located at the northeast edge of Tharsis (Fig. 1).
The edifice is smaller and more elongate than Hecates Tholus
(∼180 km diameter), with an elliptical plan-form of 130 km (E–
W) by 95 km (N–S), and rises ∼6.5 km above the surrounding
plains, with the summit reaching ∼8.7 km in elevation, compared to a height of ∼8 km and a summit elevation of ∼4.5 km
for Hecates. A broad, flat-floored caldera ∼25 km in diameter
tops the edifice, compared to ∼12 km diameter for the Hecates
caldera. The rim of the Ceraunius caldera is very asymmetrical
(Crumpler et al., 1996), with the eastern portion of the rim lying
more than 2 km below the western rim (Figs. 2 and 4). The edifice has no directly superposed impact craters greater than 7 km
in diameter, but the ejecta of one large, elongate crater formed
by oblique impact (Rahe Crater; ∼35 × 18 km) occurs along
its northern base (Figs. 2 and 3A). A new crater count gives a
surface age for Ceraunius of mid–late Hesperian [∼3.5 Gyr in
the Hartmann (2005) isochron system, Fig. 5]. This is broadly
consistent with the suggestion that it is late Hesperian in age
based on the interpretation that it formed coeval with Uranius
120
C.I. Fassett, J.W. Head III / Icarus 189 (2007) 118–135
Fig. 1. MOLA shaded relief map illustrating the location of Hecates Tholus and Ceraunius Tholus, with detailed MOLA shaded relief maps showing their regional
context.
Patera (Plescia, 2000). Lava plains with an early Amazonian
surface age (At4 ) flood the surrounding region and embay its
base (Scott and Tanaka, 1986).
As described by Crumpler et al. (1996), Ceraunius Tholus
(Figs. 2–4) is characterized by a simple caldera and a distinctive sector structure; the western margin of the circular ∼25 km
diameter caldera flares outward onto a portion of the flank that
is both lower and lower-sloped (∼5◦ ) than the rest of the radially channeled flanks, which have an average slope of 7◦ to 8◦
(Figs. 2 and 4). The caldera wall is steep-sloped (∼25◦ ; Fig. 2F)
and its maximum elevation (∼8700 m) occurs on its eastern rim
crest (Figs. 2C and 6). A small sector of an earlier caldera occurs on the northern rim (Crumpler et al., 1996). The highly
variable nature of the rim crest elevation (Fig. 6) shows some
specific trends: starting at the high point (∼8700 m) on the eastern rim, the rim crest descends smoothly to both the north and
south (to ∼7600 m), below which the rim crest changes significantly in character. The western part of the rim is characterized
by an irregular, terrace-like profile (with terrace segments at approximately 6800, 7150, 7200, and 7500 m) (Fig. 6). The lowest
of these (∼6800 m elevation) is 8 km wide and contains a large
sinuous valley that extends westward downslope to the base of
the edifice.
The floor of the caldera is very smooth, with slopes less than
∼0.5◦ at a ∼500-m baseline (Fig. 2F), but it shows some morphological diversity. Much of the floor is smooth-textured, but
large segments in the north central part are rough-textured at
the small scale, suggesting coalesced lava flows. Several pits
or secondary-crater clusters occur on the floor. Characteristics
similar to the caldera floor (including its low slope and smoothness) are also observed within the region immediately to the
west of the caldera and on individual terraced areas of the
caldera rim (Fig. 2F).
Crumpler et al. (1996) describe the history of Ceraunius
Tholus caldera as follows: (1) formation of an early, relatively
small diameter caldera(s); (2) sector-like collapse of the western summit to produce the observed amphitheater-shaped structure; (3) deposition of smooth materials on the western flank;
and (4) final collapse or subsidence of the central floor region.
The new data are broadly consistent with this sequence; on the
Valley formation on Ceraunius Tholus, Mars
121
Fig. 2. Ceraunius Tholus and its summit caldera: (A) HRSC orbits 1096, 1107, and 3144. (B) Geologic sketch map of the valleys on the flanks of Ceraunius Tholus,
associated depositional features, and Rahe Crater. Large and small valleys are both shown in blue and depositional features are in beige. Craters (generally circular)
are shown in gray. (C) MOLA altimetry map with contours. Bold lines are at 1 km intervals, thin lines are at 500 m intervals. (D) MOLA slope map with a 463-m
baseline. (E) Summit caldera region, portion of HRSC image 1107. (F) Enlargement of the MOLA slope map of the summit caldera region. Note the smoothness of
the caldera floor, deposits just outside the low western caldera rim, and on the terraces interpreted as old caldera sections on the north rim. North is at the top in all
figures, except where otherwise noted.
basis of the observations of the multiple levels of the western
rim, and the smooth, lava-flow-like deposits there, step 3 (deposition of smooth materials on the western flank) may have
taken place in several stages downward erosion and resurfacing
of the caldera rim (Figs. 2 and 6).
3. Thermal environment and snow melt
In a manner similar to our treatment of Hecates Tholus
(Fassett and Head, 2006), we have examined how the volcanic
environment of Ceraunius Tholus can alter the heat flux envi-
122
C.I. Fassett, J.W. Head III / Icarus 189 (2007) 118–135
Fig. 3. Perspective views of Ceraunius Tholus (vertical exaggeration of all perspective images is 2.5×). (A) Looking from the north over Rahe Crater to Ceraunius,
showing the asymmetry of the edifice and the summit caldera and the large channels flowing down toward and into the crater Rahe (Viking MDIM on MOLA).
(B) Looking toward Ceraunius from the south, small channels can be seen starting just below the summit and extending to the base to form small fans (HRSC on
MOLA). (C) Looking down toward Ceraunius from the NW; the distinctive western flank is in clear contrast to the rest of the edifice, and the distinctiveness of the
three larger valleys is apparent, one of which breaches Rahe Crater and forms a depositional fan. (D) View of Ceraunius from the NW at an altitude lower than (C),
showing the low point in the caldera rim toward the east, where the large valleys emerge and extend down the flank of the volcano.
ronment at the surface during or following periods of magmatic
activity. In particular, we have calculated how magmatic intrusions of plausible volume, depth, and temperature would affect
the surface environment by solving the conductive heat flow
equation. We utilize the forward-time, centered space finite difference method on a three-dimensional grid derived for Ceraunius, with a surface defined by MOLA (with horizontal grid
size of 1 km × 1 km and vertical resolution of 100 m). We have
followed the conventions of Fassett and Head (2006) for defining initial conditions, boundary conditions, and thermophysical
parameters. These are summarized in Table 1.
A first-order estimate of the plausible scale for an intrusion
can be derived from the caldera size (∼25 km diameter). This
is somewhat larger than the caldera on the summit of Hecates
Tholus, so if all else remains equal, we would expect a higher
heat flux in the intrusive cooling scenario for Ceraunius Tholus
than for an analogous scenario at Hecates Tholus. In Figs. 7
and 8, we see that this is indeed the case.
The peak heat flux that is achieved in the nominal case (for
parameters in Table 1) is ∼170 mW/m2 , though a range of 130–
220 mW/m2 is plausible and might occur with slight changes in
the choice of parameters. In particular, as discussed in Fassett
and Head (2006), the peak heat flux is strongly dependent on
the chosen intrusion depth (Fig. 8), and somewhat dependent on
the aspect ratio/volume of the intrusion. If we rely on the estimate of Carr and Head (2003) for the thickness of snow needed
to cause melting for a given heat flux, our nominal case would
require ∼100–200 m of snow thickness for melting to occur.
We can also follow the energy balance calculations of Fassett
and Head (2006), which suggest that sufficient energy is likely
transferred to the snowpack in this scenario to overcome the
latent heat of melting and cause significant meltwater production.
As can be observed in Figs. 7 and 8, melting and meltwater production may become most viable long after the peak
of a given volcanic episode for the deep sort of intrusion that
we model (peak surface heat flow takes more than ∼100 kyr
to be reached). In reality, volcanic activity and peak surface
heat flow may be more closely coupled. Melting becomes
significantly easier if much of the heat transport to the near
surface is advective—either because of hydrothermal or volcanic activity near to the surface (Harrison and Grimm, 2002;
Gulick, 1998). Even transient, relatively minor volcanic activity can be thermally important if it is close to or at the surface. Thus, advection is typically an important or even dominant mechanism for heat transport in terrestrial volcanic or hydrothermal areas (e.g., Manning and Ingebritsen, 1999), as we
discuss for the Icelandic analog below. Nonetheless, it is useful
to consider conduction because it is conservative—conductive
heat transfer will occur regardless of whether advective heat
transfer is ongoing, and any advective activity will only increase
the peak heat flux at the surface. In summary, in the circumstances we consider, more than enough heating is present to
allow snowmelt to occur. This supports the hypothesis that basal
melting of snowpack is a viable mechanism for forming the observed valleys.
Valley formation on Ceraunius Tholus, Mars
123
Fig. 4. MOLA topographic profiles across Ceraunius Tholus and the Crater Rahe derived from MOLA PEDR shot data (diamonds). A–A and B–B show the
asymmetry of the summit of Ceraunius Tholus. C–C and D–D are across the long and short axis of the Crater Rahe, respectively. E–E and F–F show two profiles
crossing from the flank of Ceraunius Tholus into the Crater Rahe, and contrast the valley floor and fan with a more “typical” section of volcano and crater. The
different density of points (e.g., compare A–A and B–B ) reflects variations in the density of MOLA shots at various locations on the edifice.
4. Description of the Ceraunius Tholus valleys
The valleys on Ceraunius Tholus were divided into four
groups by Gulick and Baker (1990): (i) relatively small, pristine
valleys with steep walls, (ii) small degraded valleys with eroded
walls, (iii) linear chains of connected pits, and (iv) the large
valleys or canyons (mainly on the north flank of Ceraunius).
Here we use new Mars Orbiter Camera (MOC), Mars Orbiter
Laser Altimeter (MOLA), Thermal Emission Imaging System
(THEMIS), and High-Resolution Stereo Camera (HRSC) data
to describe these valley types and their formation processes.
4.1. Small valleys
Small valleys incise much of the surface of Ceraunius Tholus (Fig. 2B), although they are less dense on the western flank,
124
C.I. Fassett, J.W. Head III / Icarus 189 (2007) 118–135
Fig. 5. Crater frequency diagram for Ceraunius Tholus,
√ with the number of
of 2 bins, with isochrons
craters per km2 plotted incrementally in powers
√
from Hartmann (2005). Error bars are from N . The results suggest a mid to
late Hesperian age for the edifice.
likely as a result of later volcanic resurfacing. Most of these
small valleys originate just below the summit rim (Figs. 3B
and 9) and extend down the flanks in an approximately subparallel manner, with few tributaries and small junction angles
(Figs. 3A, 3B, and 9). Typical widths are about 200–600 m
although the larger valleys in this class can be ∼1 km wide.
Widths of individual valleys appear to remain fairly constant.
The depth of valley incision appears to be somewhat more variable than the width for a given valley, and typical depths are on
the order of tens of meters. Thus, the majority of the small valleys are smaller than those measured by Williams and Phillips
(2001) for typical valley networks on Mars. This difference
could be a function of (1) differences in sampling [Williams
and Phillips considered only MOLA crossings of valleys in the
Carr (1995) database, which by virtue of its reliance on lower-
resolution Viking data is somewhat biased to large valleys],
and/or (2) a fundamental difference in the nature or duration of
the geological processes that resulted in smaller valleys on the
flanks of Ceraunius Tholus than in the highlands. Most valleys
extend to the base of the edifice and some valleys have small
fan-shaped deposits along the break in slope at the base of the
volcano (Figs. 2B and 9). Subsequent plains volcanism (Scott
and Tanaka, 1986) has obscured some portions of the base of
the edifice so valleys are hard to trace beyond the steep flanks.
On the basis of Viking data, Gulick and Baker (1990) suggest that some valleys on Ceraunius Tholus appear to be pristine
and others degraded, which they argue implies that some valleys were reactivated by late-stage activity that they attribute
to groundwater sapping. On the basis of the new data, especially the available THEMIS VIS and HRSC images, this
proposed sequence is less apparent. Many of the valleys that
look degraded in Viking images appear to do so because they
are smaller and harder to resolve than those mapped as pristine. Although there is certainly a range of valley degradation
states across the volcano, the distinction between pristine and
degraded valleys is not striking in the new data. We do not observe distinctive differences in morphology that would suggest
a transition from runoff to sapping. In general, however, the
description of the morphology and morphometry of the small
valleys on Ceraunius Tholus by Gulick and Baker (1990) remains apt in light of the new data. The radial symmetry of the
valleys (excepting the western flank), as well as the disappearance of valleys in the near-summit region, remains strong evidence that fluvial processes (most likely surface runoff) played
a significant role in valley formation, as Gulick and Baker proposed.
4.2. Pitted valleys
Gulick and Baker (1990) also drew attention to what appeared to be connected chains of pits on the western flank
Ceraunius Tholus, extending from near the summit caldera to
the base of the volcano. In the new THEMIS, MOC, and HRSC
images the chain of pits appears much more continuous than in
Fig. 6. Summit caldera topography. (A) Topographic profile showing rim elevations around the summit caldera of Ceraunius Tholus. Note the several terraces on
the western and northern rim and the location of the large valleys originating near the topographic lows. (B) Location map for the rim elevation topographic profile.
Valley formation on Ceraunius Tholus, Mars
125
Table 1
Properties of the nominal conductive heat flux model for Ceraunius Tholus
Thermal and physical conditions
Surface temperature
Intrusion temperature
Crustal thermal diffusivity
Crustal thermal conductivity
Hesperian heat flux
210 K
1100 K
10−6 m2 /s
2 W/mK
35 mW/m2
Intrusion geometry and location
Location
Shape
Depth
Extent
Centered beneath summit
Flattened ellipse, 10:1 width
to height aspect ratio
8 km
Caldera diameter (d = 25 km)
Fig. 8. Peak heat flow with time for Ceraunius Tholus intrusions of the same
volume at different depths. The 8.0 km case (also illustrated in Fig. 7) has peak
surface heat flux >150 mW/m2 for ∼100 kyr, which would melt snow <300 m
thick with the present average surface temperature or <100 m thick with average surface temperature of 230 K (Carr and Head, 2003).
Fig. 9. Typical small valleys toward the base of the north flank of Ceraunius
Tholus illustrating their generally constant width and lack of branching. The
white arrows indicate where small sediment fans that were deposited at the terminus of some of the small valleys. Note the relationship of the ∼6 km diameter
crater in the upper left with the valleys: although some valleys appear to have
been covered by the crater ejecta (mainly those near the center of the crater),
other valleys (immediately to its west and east) erode its ejecta. A smaller,
∼2 km degraded crater in the right part of the image was cut by a valley on
both its upslope and downslope rims, a common relationship for valleys in the
martian highlands. The large valley and fan in the lower right are part of the
valley and fan seen in Figs. 12B and 12D. Portion of HRSC image 1107.
Fig. 7. Illustrations of the nominal conductive thermal model for an intrusion
into Ceraunius Tholus at 8 km depth with parameters given in Table 1.
Viking frames (Fig. 10). The new images indicate that instead
of individual pits, the valley is connected and unusually sinuous
near the summit (Fig. 10). Down-flank it transitions into a style
quite similar to other (relatively large) valleys and near the base
of the volcano, it divides into two valleys, the deeper of which
truncates the shallower, and each of which has a fan-shaped deposit at its base (Fig. 10). Since this highly sinuous valley cuts
across and erodes the resurfaced flank to the west of the summit, and extends almost to the caldera rim, it is interpreted to be
younger than most of the other valleys observed on Ceraunius
Tholus. We believe its origin is likely similar to the other large
valleys discussed below.
126
C.I. Fassett, J.W. Head III / Icarus 189 (2007) 118–135
Fig. 11. Location of the origin of the major channels at the edge of the summit
caldera on the northwestern rim of the edifice. Portion of HRSC image 1107.
Fig. 10. West flank of Ceraunius Tholus. (A) HRSC images illustrate the continuity of the sinuous channel previously thought to be a disconnected chain of
pits. (B) Perspective view highlighting the fans at the end of the sinuous channel
at the base of the edifice.
4.3. Large valleys
Three large canyon-like valleys lie on the northwest flank
of Ceraunius Tholus (Figs. 2 and 3). These large valleys are
up to 2.5 km wide and 300 m deep. The biggest of these three
north-trending valleys is the westernmost, which also appears
to be the youngest based on cross-cutting relationships evident
near the summit of the volcano (Fig. 11). This valley is deepest
and most V-shaped near the summit of Ceraunius (Fig. 11), and
becomes more U-shaped near its base (Fig. 13A), below which
lies a substantial fan just inside the rim of the oblong crater
Rahe (Fig. 12). There is a clear areal association of the valley
and the fan; furthermore new data reveal an inner channel in
parts of this valley (Fig. 13) that can be seen exiting the valley at
its distal end, crossing the fan (just to the east of a 1 km diameter
crater in Fig. 13C), and extending down the distal slope of the
fan onto the floor of Rahe Crater. The valley and fan appear to
have a well-graded profile, with a near constant slope in MOLA
track data (Fig. 4, profile F–F ).
Rahe Crater is 35 × 18 km in diameter and up to 1 km
deep (Fig. 4), and it has been interpreted to have formed as
a highly oblique impact on the basis of its highly elongated
shape and unusual “butterfly” ejecta deposit (Nyquist, 1983;
Gault and Wedekind, 1978). Rahe has been cited as a candidate
for the impact event thought to have launched the SNC meteorites from Mars (Mouginis-Mark et al., 1992; Nyquist, 1983),
although there continues to be debate about their exact provenance. On the basis of superposition relationships (the eastern
valley modifies the ejecta deposit, and the western valley cuts
the crater rim deposit and the fan is emplaced in the crater interior) the crater predates the formation of at least two of the
large valleys on the northern flank of Ceraunius Tholus.
As seen in the perspective view in Fig. 12A, Rahe ejecta has
been emplaced up onto the lower flanks of the edifice, and the
easternmost of the large valleys has interacted with the edge of
the ejecta deposit, breached it, and deposited material in a small
fan (Figs. 12B and 12D). On this small fan, small channels and
erosional scarps are also observed (Fig. 12B, middle), and there
is some evidence for channel migration on the fan surface but
exact stratigraphic relationships are unclear at this resolution.
The third, central, valley is apparently the oldest of the three.
Near the base of Ceraunius, it becomes degraded and its stratigraphic relationship with Rahe’s ejecta is difficult to determine
(Fig. 12A); indeed, it could be older than Rahe Crater, with
the sediments at the channel end underlying the distal ejecta
rampart. The eastern and western of these large, north-trending
valleys unequivocally cut Rahe ejecta (Figs. 12A–12E), and because the westernmost valley is the largest and best exposed of
these valleys, our analysis explores it in detail (Section 5.2).
On the basis of their proximity, morphology, and stratigraphy,
it is likely that the other two valleys formed in a comparable
manner. For the reasons discussed below, we believe these large
canyons formed as a result of the drainage of water from a summit caldera lake.
5. Discussion
5.1. Small valleys
Other workers have made a compelling case that the small
valleys on Ceraunius Tholus (like those on Hecates Tholus)
were likely formed as a result of the action of water (e.g., Gulick
and Baker, 1990), so this will not be reviewed in detail here. Instead, only a few comments based on the new observations are
made here.
Valley formation on Ceraunius Tholus, Mars
127
Fig. 12. Large canyons on the north flank of Ceraunius Tholus. (A) Perspective view of three large valleys showing their source at the edge of the summit caldera,
their path down the flank of the edifice, and their relationship to the ejecta of the crater Rahe. The westernmost channel is largest and breaches the Rahe Crater rim,
forming the fan. (B) Base of the edifice and the edge of the Rahe Crater rim showing the easternmost large valley and the depositional fan at its base. This valley
eroded the Rahe Crater ejecta and formed small channels within the ejecta deposit (center and center-bottom). Portion of HRSC image 1107. (C) Rahe Crater and
the prominent fan-shaped deposit, which appears to have formed as a delta. Portion of HRSC image 1107. (D) Geological sketch map of (B). (E) Geological sketch
map of (C).
The relationships of valleys and small craters on the flanks
aid in constraining valley formation mechanisms and timing.
A number of small craters appear to have been temporarily
filled with water and then breached on the downslope rim (e.g.,
Fig. 9, at right). The fact that these craters have no evidence
of lava deposition supports the idea that the valleys formed
fluvially and not via lava erosion, because lava erosion is primarily a constructional process (e.g., Carr, 1974) and could not
breach the downslope rim of these craters without first filling
them with lava. Another apparently unique relationship is seen
in Fig. 9 (center-left), where the relationship of valleys with a
crater and its ejecta suggest that valleys were active both before
and after the impact event (the ejecta appears to be dissected
on the crater’s east and west edges but looks to be superposed
on the valleys north and south of the crater, closer to its central
axis). These relationships are consistent with the small valleys
128
C.I. Fassett, J.W. Head III / Icarus 189 (2007) 118–135
Fig. 13. Illustration of the small channel on the valley floor and fan surface. (A) HRSC image 1107; location for high resolution insets. (B) Geological sketch map
of (A). Note that the channel appears evident east of the prominent crater on the fan surface and near the valley mouth, but is hard to discern between these two
locations. (C) MOC image M1900701 superposed on HRSC image 1107. The ∼150-m wide channel segment is clearly visible to the east of the prominent crater.
Note also the serrated margin of the inner fan. (D) Sketch map of (C). (E) MOC image E0502375 and the ∼150-m wide inner valley near the canyon mouth.
(F) Sketch map of (E). North is to the top in all images.
forming in multiple episodes or over an extended period of
time.
We assessed more generally the relationship of the valleys
and craters to constrain their ages. For the 21 craters on Ceraunius Tholus having a diameter >1 km, it was possible to infer
with confidence the relative stratigraphic position (before or after the period of valley formation) of 14 craters. Nine craters
clearly post-date the period of valley formation, four pre-date
the valleys, and one example (Fig. 9) appears to have poten-
tially formed after valley activity began but before it ended. The
observation that ∼70% (9 out of 13) of the craters post-date
valley activity suggests a late Hesperian age [∼3.2 Gyr in the
Hartmann (2005) isochron system] for valley formation. However, the ambiguous stratigraphic relationship of seven craters
with the valleys implies that valley formation could be much
closer to the surface age of the volcano as a whole (if the ambiguous craters all post-date valley activity, formation of the
valleys may have ceased by ∼3.4 Gyr ago). Alternatively, val-
Valley formation on Ceraunius Tholus, Mars
129
Fig. 14. Topographic contour maps of the interior of Rahe Crater and a close-up of the fan deposit. The edge of the depositional fan appears to be consistently near
the ∼1600-m contour. See also Fig. 4, F–F .
ley formation may have taken place as recently as ∼3.0 Gyr
(earliest Amazonian; if all the ambiguous craters pre-date valley activity).
Gulick and Baker (1990) argue for a sequence of valley formation by runoff followed by groundwater sapping, based upon
the idea that some valleys appeared especially pristine and others more degraded. As we discuss above, this contrast is not as
striking in the new data, and in light of other observations that
suggest surface runoff (such as the relationships of valleys and
craters, e.g., Fig. 9) we believe the range of observations can
be explained by surface runoff alone. The small valleys on Ceraunius Tholus, like those on Hecates Tholus, share a variety of
characteristics with small valleys formed by surface runoff from
snowmelt on similarly steep slopes in the Antarctic Dry Valleys (Fassett and Head, 2006). The common characteristics include an abrupt beginning in the summit region, a near-constant
width and depth on the steep flanks, and subparallel, immature
drainage patterns. This abrupt sourcing of the small valleys in
the near-summit region is consistent with channelization at the
margin of a melting snowpack, though this is not the only possible interpretation.
Formation of these small valleys via snowmelt requires that
at least two conditions are met: snow or ice needs to be present
on the surface (see Section 1), and sufficient enough energy
needs to be transferred into this snowpack to drive melting (see
Section 3). One scenario that meets these conditions is the scenario we suggest: basal-melting of snowpack driven by heat flux
enhanced by volcanic activity. However, this is not the only
plausible scenario for melting snow on Mars (see, e.g., Clow,
1987). Further work is needed to establish what range of environments might trigger valley formation via this mechanism, as
well as whether we can further constrain the environment from
the observational record.
5.2. Origin of the largest valley and its associated fan deposit
There has been much speculation regarding the origin of the
large, westernmost north-trending valley and the nature of its
associated fan. Candidate models include: (1) fluvial action,
(2) volcanic density flows (e.g., Reimers and Komar, 1979),
(3) lava flow with the valley being a lava channel (e.g., Carr,
1974), (4) collapse (e.g., Caplinger, 2001), or (5) some combination of these processes.
New observations shed light on distinguishing among these
possible modes of origin. As described above, detailed examination of the topographic relationships of the valley and the
rim of the crater Rahe (Figs. 2, 4, and 12) shows that the valley
completely cuts the rim of the crater and extends into the crater,
forming a fan-shaped feature where it enters the crater. The nature of the fan-shaped feature provides potential evidence for its
origin. The fan is composed of three distinctive parts. The first,
uppermost section (Fig. 12C) begins at the mouth of the valley
and its boundaries extend laterally along the interior walls of
Rahe Crater for distances of 5 km (west) and 7 km (east). The
western boundary then turns north and forms a steep slightly
sinuous scarp that extends for about 5 km down onto the floor
of Rahe, where it turns ESE and extends for about 11 km to the
base of the crater wall. The morphology of the upper part of this
scarp varies, with the western segment appearing sharp, and the
eastern segment appearing serrated (Fig. 12C). The surface of
this portion of the fan is cratered, but appears generally smooth
(Figs. 12A and 12B); detailed examination (Fig. 12C) shows
that the eastern portion (east of the 1 km crater in the center)
is characterized by subtle linear scarps that appear to be related
to some of the serrations along the northern scarp. Indeed, the
westernmost of these scarps (adjacent to the crater) is channellike and appears to emerge from the valley, extend across the
fan, and then extend down the slope. The surface of this portion
of the fan, although relatively smooth, is not flat, and slopes
down from the mouth of the fan to the scarp at ∼4◦ (Fig. 2C;
Fig. 4, profile F–F ), only slightly shallower than the distal portions of the inside of the valley. The mouth of the channel (top
of the fan) lies at an elevation of about 2100 m, and the edge
of the top portion of the fan (the scarp) lies at about 1600 m
elevation (Fig. 14).
130
C.I. Fassett, J.W. Head III / Icarus 189 (2007) 118–135
The second section of the fan is an outward-facing scarp (up
to about 200 m high), and extends from the scarp lip 0.5 to 3 km
out onto the crater floor. Along its western half, the scarp is at
its steepest, with peak slopes of ∼13◦ (Figs. 2D, 11C, and 14),
but along its eastern half it is less steep (peak slopes ∼10◦ ) and
shows linear to sinuous small scarps extending down slope and
often correlated with the serrations in the scarp lip. The third
section of the fan deposit is formed of smooth plains that fill
parts of the floor of Rahe Crater (Fig. 12C). The distal portions of these deposits are clearly seen to extend 5–6 km out
onto the crater floor from the bottom of the scarp, but they then
tend to merge imperceptibly with the crater floor (Fig. 12C).
Some small lobe-like features are seen in this part of the deposit (Fig. 12C, lower middle), but the surface is otherwise
hummocky and relatively featureless besides superposed small,
clustered craters (presumably secondaries).
In summary, the three-part fan clearly begins at the mouth
of the large valley, and consists of an upper sloping partly
channeled surface, a medial scarp with a partly serrated crest,
and a distal flatter deposit, all superposed on the crater interior. The characteristics of this fan differ considerably from
many of the other recently described fans where highland valley networks debauched into preexisting impact craters (Malin
and Edgett, 2003; Moore et al., 2003; Fassett and Head, 2005;
Irwin et al., 2005 and references therein). Missing are the sinuous, looping channel deposits, and the multiple distributary
lobes. Those types of features likely represented the channelized flow of water and deposition of sediment on a delta undergoing significant channel-switching (avulsion). These differences could imply that the Rahe fan was not deposited by
aqueous processes, but might instead be related to lava or pyroclastic erosion/deposition processes. On the other hand, the
features we describe in the Rahe Crater fan are quite similar to
fans formed by intermittent aqueous activity with a heavy concentration of sediment, or fans building out into a standing body
of water, where the smoother portion of the fan is largely above
water, and the scarp and distal smooth portion of the deposits
are subaqueous.
Sedimentary fans and deltas form in a variety of environments, ranging from generally dry talus cones to pure, deep
submarine deltas. A common theme in most features involving water is the aqueous transport of sediment in confined
channels until some significant transition in stream power is
reached, such as: (1) alluvial fans, where channels open into
a much broader area, at which point the aqueous medium is
more widely dispersed and the sediment load is dropped to form
the fan, or (2) deltas: where the channel reaches a standing
body of water, the flow velocity decreases, and the sediment
load is largely deposited. Terrestrial deltas are characterized
by both constructional (depositional) and destructional (wave
and tidal erosion) processes and environments, and typically
consist of a delta plain (an upper predominantly subaerial surface), a delta front (submarine, where course grained sediments
are deposited), and a pro-delta plain (a lower submarine plain
where sediments are largely deposited from suspension). The
range of delta morphologies generally reflects the relationships
between depositional-dominated (constructive) and erosional-
Fig. 15. Results of a laboratory stream-table experiment forming a Gilbert-type
delta in a standing body of water. Note similarities in shape and morphology to
delta in Rahe Crater at the base of the largest channel (Figs. 12 and 14). This
delta is a few centimeters across. Image courtesy of Douglas Clark, from Clark
and Linneman (2005).
dominated (destructive) types. In many cases channels avulse
to create various overlapping channels and lobes; similar features have been recognized on Mars (Malin and Edgett, 2003;
Moore et al., 2003; Fassett and Head, 2005). In some deltas
known as Gilbert-type, relatively high outflow velocities, combined with a relatively steep drop at the channel mouth, cause
dominance by inertial forces in which the effluent spreads
and diffuses in the manner of a turbulent jet (e.g., Wright,
1977, 1985). These conditions are common where high gradient
streams enter deep bodies of water.
The feature seen on the floor of Rahe has a broad fan
shape (rather than the multiple birds-foot morphology typical
of avulsion-dominated deltas) and has three segments broadly
similar to the subaerial delta plain, delta front, and pro-delta
plain. It is also very similar in morphology to deltas that can
be produced experimentally on laboratory stream tables under specific conditions (Fig. 15). In these experimental cases,
a stream channel with a constant-flux of sediment-laden effluent produces a fan-shaped deposit built out into a standing
body of water. The final configuration is a Gilbert-type delta
characterized by a relatively steep delta front. The similarities
between this experimental configuration and the Rahe Crater
fan are striking (compare Figs. 14 and 15).
Another approach to assess the origin of this valley is to
compare the volume of the deposit and the volume of the valley itself. If the valley were formed by lava thermal erosion
processes (a process that requires very significant amounts of
lava to thermally erode the substrate to create the valley), then
the volume of the material in Rahe Crater should greatly exceed the volume of the valley (Reimers and Komar, 1979;
Carr, 1974). Modeling by Li and Robinson (2002) suggested
that erosion of the largest canyon on the north flank of Ceraunius might require eruption of ∼420 km3 of magma, which is
much greater than the volume (20×) than the volume of the observed fan on Rahe’s floor. If formed by pyroclastic flows, the
deposit volume should also be much greater than the channel.
Valley formation on Ceraunius Tholus, Mars
If formed solely by aqueous erosion, the volumes of the fan and
the channel should be comparable.
Using MOLA gridded data (1/128◦ ) and individual MOLA
profiles we measured the volume of the fan deposited at the
base of this valley in Rahe Crater to be 21.7 km3 and the volume of the valley itself to be 20.9 km3 (assuming a rectangular
cross section and an average depth of 220 m) or 20.6 km3 (using
a cross-sectional area from MOLA orbit 17623 as representative and multiplying by the total valley length). Each of these
volume estimates is uncertain at a level of ∼20%; the volume
of the valley is particular difficult to ascertain since there is a
12 km gap in MOLA orbit crossings. However, the very close
correspondence of these two measurements (erosional valley
and depositional fan) is consistent with the derivation of the
fan sediments entirely from the valley.
Further support for the idea that fluvial erosion formed the
large valleys comes from the lack of evidence for extensive
lava flooding, ponding, and buildup on the floor of Rahe Crater
itself, as would be expected if the valley formed by lava thermal erosion or if lava had subsequently utilized a valley created by other processes. Ash flows also appear unlikely to be
the dominant process in carving the channel and forming the
fan because this process, as with lava thermal erosion, would
require many multiples of the eroded channel volume to actually eroded the substrate, and this volume is not observed
in Rahe Crater. Finally, the existence of a small channel observed on the valley floor and fan surface (Fig. 13) supports
the idea that fluvial erosion formed the valley and fan. Thus,
on the basis of these observations and calculations, we pursue
the interpretation that the valley and the fan were formed by related aqueous erosional and depositional processes, to produce
a Gilbert-type delta at the mouth of the valley in Rahe Crater.
Other depositional fans on Mars that may be Gilbert-type
deltas have been studied in detail (Mangold and Ansan, 2006;
Ori et al., 2000). The delta deposit in SW Thaumasia shares
many of the characteristics of the fan we observe here, including a relatively shallow surface slope (2◦ ), a steep front (∼11◦ ),
and a comparable total volume (15 km3 ) (Mangold and Ansan,
2006).
The geomorphological evidence strongly suggests that the
channel and the delta formed in a standing body of water. We
can assess what contour level the water in this lake reached. In
other crater lakes, water commonly appears to have breached
the rim at a low point to create an exit channel (e.g., Fig. 9; see
also Irwin et al., 2005 and references therein). Examination of
the Viking, THEMIS, HRSC, and MOC images of Ceraunius,
the delta, and the interior of Rahe Crater, combined with topography derived from MOLA, shows that the lowest point on the
rim of Rahe is at its northeast end, where the exterior ejecta is
missing due to the bilaterally symmetrical ejecta pattern typical of very low angle impacts (Fig. 2, Fig. 4, profile C–C , and
Fig. 14). At this elevation (∼2100 m), no prominent erosional
feature is present in the rim crest in the present data, suggesting that water did not fill it to a level that would cause a lake
in Rahe Crater to overflow. Volume comparisons support this
interpretation. The total enclosed volume of the Rahe Crater is
∼233 km3 (below the 2100-m contour) and the volume of the
131
Table 2
Flooded volume of Rahe Crater to various contours
Contour (m)
1400
1500
1550
1600
1650
1700
1705
1710
Volumea (km3 )
2.4
15.0
25.9
38.8
53.2
68.9
70.5
72.2
Contour (m)
Volumea (km3 )
1715
1720
1725
1750
1800
1900
2000
2100
73.9
75.6
77.3
86.2
105.0
145.6
189.2
233.5
Note. The best estimate is that when the fan was deposited, Rahe was filled
to between the 1600- and 1700-m contours, implying water volumes of ∼35
to 70 km3 , both of which are less than the present enclosed volume of the
Ceraunius caldera.
a Upper limit volume of flood from the present Ceraunius caldera: ∼75 km3 .
current summit caldera is ∼75 km3 (below the current breach
elevation). On the basis of these comparative volumes, even if
the completely flooded caldera was to flood Rahe Crater, it still
would not have filled it sufficiently to breach the rim crest at its
lowest point. Indeed, the summit caldera volume would flood
Rahe Crater to approximately the 1700-m contour, on the upper
portion of the delta (Table 2).
A final relevant piece of evidence for the lake level reached
in Rahe is the elevation of the marginal scarp in the delta: from
terrestrial field and laboratory analogs, it is likely that the water
level inside Rahe Crater was in the vicinity of (or higher than)
this elevation when the fan was emplaced. Examination of a
MOLA topographic map of the interior of Rahe (Fig. 14) shows
that this elevation is consistently ∼1600 m (about 500 m below
the lowest point in the Rahe Crater rim). Therefore, our best
estimate is that Rahe Crater flooded to approximately the 1600
or 1700-m contour during the formation of the valley and delta;
we informally refer to this as Lake Rahe. The water volume at
various possible lake levels (contours) is given in Table 2.
On the basis of these observations, we outline the following
scenario for the formation of the large valleys, the delta, and
Lake Rahe. All three of the large canyons on the north flank
of Ceraunius Tholus have their source near the low point of the
summit caldera (Figs. 2–4 and 8). We interpret this to mean that
the summit caldera was the site of a lake that filled with water
from time to time related to the melting of summit snowpack
via conductive and advective heating, as outlined above. As the
caldera filled with water, drainage occurred from the low point
in the caldera rim crest (Figs. 6 and 11). This mechanism is directly analogous to the caldera-breaching event that triggered
a large flood at the volcano Aniakchak in Alaska (Waythomas
et al., 1996), with peak discharges (∼106 m3 /s) comparable or
larger than those inferred at Ceraunius Tholus (see Table 1). Initial drainage events formed the central and eastern of the three
large valleys, and at least for the eastern valley, deposition took
place at the base of the edifice interacting with the ejecta of
Rahe Crater. There is no evidence that the rim crest of Rahe
was breached in these first two events. The third and largest of
the valleys formed in a similar manner, cross-cutting the previously formed canyons (Fig. 11) and descending down the flanks
of the edifice to the edge of Rahe Crater, where the rim of Rahe
132
C.I. Fassett, J.W. Head III / Icarus 189 (2007) 118–135
Table 3
Discharge estimates for observed and inferred channel parameters from the
Darcy–Weisbach equation (Kleinhans, 2005; Wilson et al., 2004)
Observed channel width
Depth
Sediment size
Bed slope
Derived friction factor for range
of parameters (from Kleinhans,
2005, Eq. (13))
Darcy–Weisbach velocity
Discharge estimate
W = 250 m (inner channel, lower
canyon) to 700 m (upper canyon)
H = 20–60 m (highly uncertain)
D50 = 0.001–0.1 (m)
s = 9–14%
f = 0.8–1.5
u = 7–15 m/s
Q = 4 × 104 to 6 × 105 m3 /s
Note. These discharge estimates are uncertain beyond an order of magnitude,
primarily because of uncertainty in the appropriate channel depth, and the lack
of constraints of the entrained sediment, bed, and flow characteristics.
was breached and sediment-laden water entered the crater itself (Fig. 12). The flooding of Rahe was characterized by the
deposition of a fan-shaped deposit at the mouth of the channel
as water flowed into the crater and filled to a contour level of
∼1600–1700 m.
If we assume that the fan volume of ∼20 km3 transported
into Rahe was brought in a single large event, we can estimate a
range of sediment concentrations that this would imply. Flooding to the 1600- and 1700-m contours would presently require
∼40 or ∼70 km3 of water, respectively. These would imply
sediment to water ratios (by volume) of 1:2 or 1:3.5, respectively. For basaltic sediment, a ratio of ∼1:3 (by volume) is
near the upper limit of what is typically suggested for sediment
transport scenarios in flood events on Mars (Komar, 1980; but
see also Kleinhans, 2005). Achieving these high sediment concentrations might be favored for the large valleys on Ceraunius
Tholus due to their steep slopes (∼5◦ –8◦ ) compared to those of
the outflow channels (Komar, 1980). Thus, we anticipate two
possible scenarios: either (1) the fan in Rahe was built in a single hyper-concentrated, very high discharge event (perhaps with
peak fluid flux of order 105 m3 /s), or (2) it was built in multiple flood events with lower sediment concentration (still with
a substantial peak flux, perhaps Q ∼ 104 m3 /s) (see Table 3
for flux estimates). Even in this second scenario, the average
sediment to water ratio was likely 1:12 or greater (by volume),
since Rahe could not have filled beyond the 2100-m contour
(V ∼ 233 km3 ) without breaching its northeast rim.
Following this initial filling of Lake Rahe and deposition of
the majority of the fan, flow waned and the ∼200-m wide channel observed in the lower portion of the valley and on the fan
surface was carved (Fig. 13). Eventually, caldera lake drainage
and erosion ceased, Lake Rahe became input-starved, and eventually disappeared by evaporation, freezing and sublimation,
and/or subsurface drainage. The canyon source area (the low
point in the summit region) was then resurfaced by volcanic deposits (see the smooth nature of this area in the slope maps,
Figs. 2C–2F) and the immediate channel source depressions in
the lip of the caldera were embayed (Fig. 11). Further summit
activity locally lowered the caldera rim crest in a ∼7 km wide
segment in the caldera rim (Fig. 6). The final highly sinuous
valley was then focused by this breach and formed west of the
caldera, incising the volcano’s western flank of the edifice and
depositing small fans of material at its the base (Figs. 2, 3, and
10). Unlike the example to the north in Rahe Crater, the effluent
from this valley was not confined by a crater to form a lake, and
was dispersed into the surrounding plains, depositing its sediment at the break in slope.
Given this scenario for the formation of the large valleys,
it is worth asking the following question: why did these large
canyons form on Ceraunius Tholus but not on Hecates Tholus?
A primary difference between Hecates and Ceraunius is the difference in the edifices’ summit calderas. The Ceraunius caldera
encloses a volume of ∼75 km3 with a depth from its floor to
the peak of its rim of 2 km (which has probably been decreased
somewhat by late stage resurfacing). The Hecates caldera is
much less disrupted, and only encloses a volume of ∼18.5 km3 ,
with a depth from the caldera floor to rim of ∼500 m. Thus,
the deep, wide caldera at Ceraunius could trap a substantially
greater volume of material (ice and/or water) than the small
nested caldera at Hecates. Moreover, at the time that the largest
canyons on the north flank of Ceraunius formed, the western
rim of the caldera was potentially more intact, and the caldera
was less resurfaced, so it might easily have enclosed a much
greater volume (perhaps up to a factor of two more).
This difference in “trapping ability” might have helped not
only create conditions for large floods on Ceraunius Tholus
compared to Hecates Tholus, but may also have helped in the
accumulation of ice in the summit regions in the first place.
The steep eastern caldera rim wall on Ceraunius Tholus (Fig. 4,
profile A–A ) is an excellent potential focal point for accumulation of ice. It is similar in both relief and elevation to
the ∼2 km-high, west-facing scarp of a graben on the western
flank of Arsia Mons, where abundant ice-related features are
observed (Shean et al., 2007). On Ceraunius Tholus, it appears
that accumulated snow or ice was melted, trapped in a summit
caldera lake, and then escaped down the north flank in several
canyon-carving events. Further insight into the nature of this
process can be derived from an examination of the formation
and evolution of terrestrial caldera lakes formed by melting of
summit snowpack. In the following section we examine a welldocumented example from Iceland.
6. Icelandic terrestrial analog
Numerous areas of high topography in Iceland contain summit snowpack and many of these highs are active or recently
active volcanic edifices. The largest of these also contain summit calderas and sufficient snowpack to form glaciers. For example, the Myrdalsjokull ice cap (Fig. 16), the fourth largest
in Iceland (area of 600 km2 ), rises 1300–1500 m above the
surrounding lowlands, and is characterized by heavy winter
snowfall, high rates of summer melting, and is drained by
numerous rivers (Björnsson et al., 2000). The ice cap is underlain by an active volcano with a large central caldera,
which is part of the Katla volcanic system (Jakobsson, 1979;
Johannesson et al., 1990). Eruptions beneath the ice have recently averaged two per century, and each eruption rapidly
melts large volumes of ice leading to enormous jökulhaups
from the margins of the ice cap. These flooding events typi-
Valley formation on Ceraunius Tholus, Mars
133
Fig. 16. Myrdalsjokull, an Icelandic ice deposit capping a volcanic edifice. It is surrounded by numerous glacial streams and which is frequently drained by
jökulhaups. The perspective view shows the 14 km × 9 km Myrdalsjokull caldera, which is presently capped and filled by ice (blue line), with an ice volume inside
the caldera of ∼45 km3 . After Björnsson et al. (2000). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of
this article.)
cally follow one major drainage direction related to the summit
caldera topography, and often result in hyperconcentrated flow
over the adjacent outwash plains with velocities of 5–15 m/s,
peak discharges of 100,000–300,000 m3 /s, and total volumes
of 1–8 km3 (Björnsson et al., 2000). Björnsson et al. (2000)
have recently reported on an analysis of the surface and bedrock
topography of the icecap, and the implied structure and morphology of the sub-ice caldera. They show that the caldera is
∼9 × 13 km in diameter, covers ∼200 km2 , is 600–750 m deep,
and is characterized by significant floor structure and morphology, with several smaller recently active floor cauldrons
which are the location of lakes and the sources of the jökulhaups.
Bottom-up melting of the ice is due to general geothermal
flux aided by hydrothermal circulation and direct contact with
eruption products. Top-down melting is due to enhanced summer solar insolation, and this meltwater reaches the substrate
through moulins, crevasses and veins and joins with basal meltwater produced predominantly by geothermal sources. Sub-
glacial migration of meltwater causes coalescence into streams
that leave the edge of the ice through portals, forming a set of
radially arrayed streams around the edifice (Fig. 16). In the
caldera interior, migration paths and subglacial water-divide
locations are dictated by ice thickness and ice overburden pressure, and local gradients in ice overburden pressure drive water
out of the caldera through the deep passes in the caldera rim
(Björnsson et al., 2000).
Is there any evidence for jökulhaup-like catastrophic drainage of the summit caldera lake on Ceraunius Tholus, as seen
in the Icelandic examples? Discharge estimates for the largest
valley on Ceraunius Tholus are comparable (or greater) than
those in jökulhaup events that occur at Myrdalsjokull. The
enclosed volume of the summit calderas at Myrdalsjokull
(∼40 km3 ) and Ceraunius (∼75 km3 ), the inferred peak discharges (104 –105 m3 /s) and the total volume in flood events
observed at Myrdalsjokull (∼1–8 km3 ) and inferred at Ceraunius (∼40 to ∼70 km3 ) are all comparable (within an order of
magnitude). Based on these similarities, it seems plausible that
134
C.I. Fassett, J.W. Head III / Icarus 189 (2007) 118–135
the process which carved this valley is the rapid draining of a
caldera lake, as for the jökulhaups in the Icelandic examples.
We have interpreted the source of water for the small radial
valleys on Hecates Tholus (Fassett and Head, 2006) and Ceraunius Tholus to melting of summit snowpack by geothermal heat
and drainage of meltwater down the sides of the edifice. This
melting of snowpack at the Ceraunius Tholus summit would
likely cause substantial volumes of water to be trapped in the
caldera. Although any ice in the caldera would be expected to
exert overburden pressure and influence the drainage of sub-ice
meltwater, the pronounced asymmetry of the edifice summit
area and caldera (Fig. 4, profile A–A ; Fig. 6) strongly suggests that any meltwater residing in the summit caldera would
nonetheless drain through the low part of the rim to the west,
where the major valleys are located.
The characteristics of snow and ice-capped active Icelandic
volcanoes such as Myrdalsjokull thus help bolster our understanding of the origin of the valleys around Ceraunius Tholus. A primary difference of Ceraunius and Myrdalsjokull is
the lack of distinctive features on the flanks of Ceraunius
(and Hecates) related to wet-based glaciation (drumlins, scour,
moraines, etc.). This suggests that either snow and ice cover
was too thin for significant glacial flow to occur, or that resurfacing of the summit destroyed any evidence that might once
have existed. If melting occurred in relatively thin deposits of
snow or ice, the heat flux that led to melting may have been
higher than the conservative values calculated in Section 3.
7. Conclusions
Two major types of valleys are radially arrayed on the
Hesperian-aged flanks of Ceraunius Tholus, a volcanic edifice
in the Tharsis region. Small valleys are common on the flanks
of the volcano, form below the summit, and extend downslope
to the base with little branching or intersection, often ending in
small fans. These are quite similar to radial valleys on the flanks
of Hecates Tholus, another volcano with a Hesperian surface
age. Unlike Hecates, several larger valleys appear to emanate
from near the low point of the summit caldera rim. These larger
valleys extend down the flanks and formed large sedimentary
fans. The largest of the valleys breached the ejecta of Rahe
Crater at the base of the edifice and flowed into the crater interior, likely forming a temporary lake and depositing a ∼22 km3
delta. The morphology of both the small and large valleys on
Ceraunius, their geological setting, and terrestrial and laboratory analogs, all point to their erosion by water. The distribution
and characteristic of these valleys leads to the interpretation that
they formed by melting of summit snowpack.
On the basis of these observations, we conclude that like
Hecates Tholus, Ceraunius Tholus was the site of snow and
ice accumulation during climatic excursions, and that enhanced
heat flux led to basal melting of snowpack to form the observed
features. These observations underline the importance of a combination of circumstances (climate change to produce summit
snowpack and an active phase of magmatism in the edifice to
produce melting) to form the radial valley systems on some
volcanoes and not on others. If this interpretation is correct,
then other valley networks such as the apparently young valleys observed in the region of Valles Marineris by Mangold et
al. (2004) may have formed by a similar mechanism (bottomup geothermal melting of snowpack) rather than during a period of “warm and wet” climate; this mechanism may apply to
some Noachian valley networks as well (Carr and Head, 2003;
Clow, 1987). Ceraunius Tholus provides a basis for the further
analysis of this process and comparison to the array of valley
network morphologies and settings.
Finally, the stratigraphy, geological history, and superimposed impact crater population on the floor of Rahe Crater
suggests that it formed before the major canyons on Ceraunius,
likely in the Hesperian (or early Amazonian at the latest), and
is thus not a good candidate for the impact that ejected the SNC
meteorites, since their ejection ages are likely late Amazonian
(e.g., Nyquist et al., 2001).
Acknowledgments
This paper is dedicated to Jürgen Rahe, whose contributions to Solar System exploration and the individuals who participate in it are sorely missed. We gratefully acknowledge
financial support from the NASA Mars Data Analysis Program (NNG04GJ99G), the NASA Mars Express Participating
Scientist Program (JPL1237163), the NASA AISR program
(NNGO5GA61G), and the Los Alamos National Laboratory
(08337-001-05) (to J.W.H.), as well as the NASA Graduate Student Research Program (to C.I.F.). Thanks to Douglas Clark for
helpful communications related to his stream table experiments
and to Helgi Björnsson for communications related to Myrdalsjokull. We appreciate helpful reviews by Catherine Weitz and
Alan Howard. We also thank the HRSC Experiment Teams at
DLR and Freie Universitaet Berlin, as well as the Mars Express Project Teams at ESTEC and ESOC, for their successful
planning and acquisition of the HRSC data and for making
processed data available to the HRSC team. Finally, we would
like to acknowledge the efforts of the HRSC Co-Investigator
team who have contributed to this investigation in the mission
preparatory phase and via helpful scientific discussions.
Supplementary material
The online version of this article contains additional supplementary material.
Please visit DOI: 10.1016/j.icarus.2006.12.021.
References
Baker, V.R., 2001. Water and the martian landscape. Nature 412, 228–236.
Björnnson, H., Pálsson, F., Gudmundsson, M.T., 2000. Surface and bedrock
topography of the Myrdalsjokull ice cap, Iceland: The Katla caldera, eruption sites and routes of jökulhaups. Jokull 49, 29–46.
Caplinger, M.A., 2001. High-resolution imaging of Ceraunius Tholus, Mars.
Lunar Planet. Sci. 32. Abstract 1342.
Carr, M.H., 1974. The role of lava erosion in the formation of lunar rilles and
martian channels. Icarus 22, 1–23.
Carr, M.H., 1995. The martian drainage system and the origin of valley networks and fretted channels. J. Geophys. Res. 100, 7479–7508, doi:10.1029/
95JE00260.
Valley formation on Ceraunius Tholus, Mars
Carr, M.H., Head, J.W., 2003. Basal melting of snow on early Mars: A possible origin of some valley networks. Geophys. Res. Lett. 30, doi:10.1029/
2003GL018575. 2245.
Clark, D.H., Linneman, S.R., 2005. Combining stream table experiments, hightech particle analysis, and the web to help geomorphology students evaluate
landform evolution. J. Geosci. Educ. 53, 110–115.
Clow, G.D., 1987. Generation of liquid water on Mars through the melting of a
dusty snowpack. Icarus 72, 95–127.
Craddock, R.A., Howard, A.D., 2002. The case for rainfall on a warm, wet early
Mars. J. Geophys. Res. 107, doi:10.1029/2001JE001505. 5111.
Crumpler, L.S., Head, J.W., Aubele, J.C., 1996. Calderas on Mars: Characteristics, structural evolution, and associated flank structures. In: McGuire,
W.C., Jones, A.P., Neuberg, J. (Eds.), Volcano Instability on the Earth and
Other Planets. Geol. Soc. of London Special Publication 110, pp. 307–348.
Fassett, C.I., Head, J.W., 2005. Fluvial sedimentary deposits on Mars: Ancient
deltas in a crater lake in the Nili Fossae region. Geophys. Res. Lett. 32,
doi:10.1029/2005GL023456. L14201.
Fassett, C.I., Head, J.W., 2006. Valleys on Hecates Tholus, Mars: Origin by
basal melting of summit snowpack. Planet. Space Sci. 54, 370–378.
Forget, F., Haberle, R.M., Montmessin, F., Levrard, B., Head, J.W., 2006. Formation of glaciers on Mars by atmospheric precipitation at high obliquity.
Science 311, 368–371.
Gault, D.E., Wedekind, J.A., 1978. Experimental studies of oblique impact.
Proc. Lunar Sci. Conf. 9, 3843–3875.
Gulick, V.C., 1998. Magmatic intrusions and a hydrothermal origin for fluvial
valleys on Mars. J. Geophys. Res. 103, 19365–19387.
Gulick, V.C., 2001. Origin of the valley networks on Mars: A hydrological perspective. Geomorphology 37, 241–268.
Gulick, V.C., Baker, V.R., 1990. Origin and evolution of valleys on martian
volcanoes. J. Geophys. Res. 95, 14325–14344.
Gulick, V.C., Tyler, D., McKay, C.P., 1997. Episodic ocean-induced CO2
greenhouse on Mars: Implications for fluvial valley formation. Icarus 130,
68–86.
Harrison, K.H., Grimm, R.E., 2002. Controls on martian hydrothermal systems:
Application to valley network and magnetic anomaly formation. J. Geophys. Res. 107, doi:10.1029/2001JE001616. 5025.
Hartmann, W.K., 2005. Martian cratering. 8. Isochron refinement and the
chronology of Mars. Icarus 174, 294–320.
Hauber, E., van Gasselt, S., Ivanov, B., Werner, S., Head, J.W., Neukum, G.,
Jaumann, R., Greeley, R., Mitchell, K.L., Muller, P., and the HRSC CoInvestigator Team, 2005. Discovery of a flank caldera and very young
glacial activity at Hecates Tholus, Mars. Nature 434, 356–361.
Head, J.W., Marchant, D.R., 2003. Cold-based mountain glaciers on Mars:
Western Arsia Mons. Geology 31, 641–644.
Head, J.W., Neukum, G., Jaumann, R., Hiesinger, H., Hauber, E., Carr,
M., Masson, P., Foing, B., Hoffmann, H., Kreslavsky, M., Werner, S.,
Milkovich, S., van Gasselt, S., and the HRSC Co-Investigator Team, 2005.
Tropical to mid-latitude snow and ice accumulation, flow and glaciation on
Mars. Nature 434, 346–351.
Head, J.W., Marchant, D.R., Agnew, M.C., Fassett, C.I., Kreslavsky, M.A.,
2006a. Extensive valley glacier deposits in the northern mid-latitudes of
Mars: Evidence for late Amazonian obliquity driven climate change. Earth
Planet. Sci. Lett. 241, 663–671.
Head, J.W., Nahm, A.L., Marchant, D.R., Neukum, G., 2006b. Modification
of the dichotomy boundary on Mars by Amazonian mid-latitude regional
glaciation. Geophys. Res. Lett. 33, doi:10.1029/2005GL024360. L08S03.
Irwin, R.P., Howard, A.D., Craddock, R.A., Moore, J.M., 2005. An intense
terminal epoch of widespread fluvial activity on early Mars. 2. Increased
runoff and paleolake development. J. Geophys. Res. 110, doi:10.1029/
2005JE002460. E12S15.
Jakobsson, S.P., 1979. Petrology of recent basalts of the earstern volcanic zone,
Iceland. Acta Naturalia Islandica 26, 1–103.
Johannesson, H.K., Saemundsson, K., Jakobsson, S., 1990. Geological map of
Iceland. Icelandic Museum of Natural History and Iceland Geodetic Survey,
Reykjavik. Sheet 6, South Iceland.
135
Kleinhans, M.G., 2005. Flow discharge and sediment transport models for estimating a minimum timescale of hydrological activity and channel and
delta formation on Mars. J. Geophys. Res. 110, doi:10.1029/2005JE002521.
E12003.
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.
Laskar, J., Correia, A.C.M., Gastineau, M., Joutel, F., Levrard, B., Robutel, P.,
2004. Long term evolution and chaotic diffusion of the insolation quantities
of Mars. Icarus 170, 343–364.
Li, H., Robinson, M.S., 2002. Modeling channel formation on Ceraunius Tholus, Mars. Eos (Spring Suppl.) 83 (19). Abstract P31A-13.
Malin, M.C., Edgett, K.S., 2003. Evidence for persistent flow and aqueous sedimentation on early Mars. Science 302, 1931–1934.
Mangold, N., Ansan, V., 2006. Detailed study of an hydrological system of
valleys, a delta, and lakes in Southwest Thaumasia region, Mars. Icarus 180,
75–87.
Mangold, N., Quanitin, C., Ansan, V., Delacourt, C., Allemand, P., 2004.
Evidence for precipitation on Mars from dendritic valleys in the Valles
Marineris area. Science 305, 78–81.
Manning, C.E., Ingebritsen, S.E., 1999. Permeability of the continental crust:
Implications of geothermal data and metamorphic systems. Rev. Geophys. 37, 127–150, doi:10.1029/1998RG900002.
Moore, J.M., Howard, A.D., Dietrich, W.E., Schenk, P.M., 2003. Martian layered fluvial deposits: Implications for Noachian climate scenarios. Geophys. Res. Lett. 30, doi:10.1029/2003GL019002. 2292.
Mouginis-Mark, P.J., McCoy, T.J., Taylor, G.J., Keil, K., 1992. Martian parent
craters for the SNC meteorites. J. Geophys. Res. 97, 10213–10336.
Nyquist, L.E., 1983. Do oblique impacts produce martian meteorites? Lunar
Planet. Sci. 13 (2), J. Geophys. Res. (Suppl.) 88S, A785–A798.
Nyquist, L.E., Bogard, D.D., Shih, C.Y., Greshake, A., Stoffler, D., Eugster, O.,
2001. Ages and geologic histories of martian meteorites. Space Sci. Rev. 96,
105–164.
Ori, G.G., Marinangelli, L., Baliva, A., 2000. Terraces and Gilbert-type deltas
in rater lakes in Ismenius Lacus and Memnonia, Mars. J. Geophys. Res. 105,
17629–17641.
Plescia, J.B., 2000. Geology of the Uranius group volcanic constructs: Uranius Patera, Ceraunius Tholus, and Uranius Tholus. Icarus 143, 376–
396.
Reimers, C.E., Komar, P.D., 1979. Evidence for explosive volcanic density currents on certain martian volcanoes. Icarus 39, 88–110.
Scott, D.H., Tanaka, K.L., 1986. Geologic map of the western equatorial region
of Mars. U.S. Geol. Surv. Misc. Investigation Series Map I-1802A.
Shean, D.E., Head, J.W., Marchant, D.R., 2005. Origin and evolution of a coldbased tropical mountain glacier on Mars: The Pavonis Mons fan-shaped
deposit. J. Geophys. Res. 110, doi:10.1029/2004JE002360. E05001.
Shean, D.E., Head, J.W., Fastook, J.L., Marchant, D.M., 2007. Recent glaciation at high elevations on Arsia Mons, Mars: Implications for the formation and evolution of large tropical mountain glaciers. J. Geophys. Res.,
doi:10.1029/2006JE002761, in press.
Waythomas, C.F., Walder, J.S., McGimsey, R.G., Neal, C.A., 1996. A catastrophic flood caused by a caldera lake at Aniakchak Volcano, Alaska, and
implications for volcanic hazards assessment. Geol. Soc. Am. Bull. 108,
861–871.
Williams, R.M.E., Phillips, R.J., 2001. Morphometric measurements of martian valley networks from Mars Orbiter Laser Altimeter (MOLA) data. J.
Geophys. Res. 106, 23737–23752, doi:10.1029/2000JE001409.
Wilson, L., Ghatan, G.J., Head, J.W., Mitchell, K.L., 2004. Mars outflow channels: A reappraisal of the estimation of water flow velocities from water
depths, regional slopes, and channel floor properties. J. Geophys. Res. 109,
doi:10.1029/2004JE002281. E09003.
Wright, L.D., 1977. Sediment transport and deposition at river mouths: A synthesis. Geol. Soc. Am. Bull. 88, 837–868.
Wright, L.D., 1985. River deltas. In: Davis Jr., R.A. (Ed.), Coastal Sedimentary
Environments. Second ed. Springer-Verlag, New York, pp. 1–76.