Icarus 211 (2011) 316–329

Contents lists available at ScienceDirect

Icarus

journal homepage: www.elsevier .com/ locate/ icarus

Chaos formation by sublimation of volatile-rich substrate: Evidence fromGalaxias Chaos, Mars

G.B.M. Pedersen a,⇑, J.W. Head III b

a Department of Earth Sciences, University of Aarhus, Hoegh-Guldberggade 2, 8000 Aarhus C, Denmarkb Department of Geological Sciences, Brown University, 324 Brook Street, Box 1846, Providence, RI 02912, USA

a r t i c l e i n f o

Article history:Received 1 May 2010Revised 12 August 2010Accepted 3 September 2010Available online 16 September 2010

Keyword:MarsGalaxias ChaosElysium Volcanic ProvinceVastitas Borealis Formation

0019-1035/$ - see front matter � 2010 Elsevier Inc. Adoi:10.1016/j.icarus.2010.09.005

⇑ Corresponding author. Tel.: +45 89429539; fax: +E-mail address: gro.birkefeldt@geo.au.dk (G.B.M. PURL: http://www.marslab.dk/ (G.B.M. Pedersen).

a b s t r a c t

Galaxias Chaos deviates significantly from other chaotic regions due to the lack of associated outflowchannels, lack of big elevation differences between the chaos and the surrounding terrain and due togradual trough formation.

A sequence of troughs in different stages is observed, and examples of closed troughs within blockssuggest that the trough formation is governed by a local stress field rather than a regional stress field.Moreover, geomorphic evidence suggests that Galaxias Chaos is capped by Elysium lavas, which super-pose an unstable subsurface layer that causes chaotic tilting of blocks and trough formation.

Based on regional mapping we suggest a formation model, where Vastitas Borealis Formation embed-ded between Elysium lavas is the unstable subsurface material, because gradual volatile loss causesshrinkage and differential substrate movement. This process undermines the lava cap, depressions formand gradually troughs develop producing a jigsaw puzzle of blocks due to trough coalescence.

Observations of chaos west of Elysium Rise indicate that this process might have been widespreadalong the contact between Vastitas Borealis Formation and Elysium lavas. However, the chaotic regionshave probably been superposed by Elysium/Utopia flows to the NW of Elysium Rise, and partly sub-merged with younger lavas to the west.

� 2010 Elsevier Inc. All rights reserved.

1. Introduction collapse, the highland material is fractured into mesas making

Chaotic terrains are common on the martian surface, howeverthe origin of this landform is enigmatic and many different forma-tion models have been proposed.

Chaotic terrain was thoroughly described by Sharp (1973) asirregular jumbles of angular blocks of various size (kilometers totens of kilometers) many preserving remnants of the upland sur-face and sometimes controlled by linear features. Generally theblocks are flat topped, their sizes decrease away from the boundinghead scarp and the terrain consists of alcove-like depressions thatcommonly have a subcircular shape and can be divided into cells(Nummedal and Prior, 1981; Chapman and Tanaka, 2002; Rodri-guez et al., 2006). Usually, the chaotic terrain lies several hundredsof meters up to a couple of kilometers below the surrounding pla-teau and thus, the topographic characteristics are very distinctivefor chaotic terrain. Meresse et al. (2008) used topographic profilesfrom the Xanthe and Margaritifer Terra to resolve different stagesof chaos formation, showing an initial stage of shallow ground sub-sidence of a few tens to hundreds of meters, where the ancient pla-teau is heavily fractured and collapse is limited. During the

ll rights reserved.

45 89429406.edersen).

the chaos margins well defined and resulting in a vertical displace-ment of 1000–3000 m. Equally, Rodriguez et al. (2005) has delin-eated different stages of chaos formation based on morphologicand topographic observations of the Xanthe Terra region.

Chaotic terrain is found widespread and is generally geograph-ically associated with huge equatorial troughs like Valles Marine-ris, with outflow channels like Chryse outflow channels and withthe dichotomy boundary (e.g., Soderblom and Wenner, 1978).

Several different geologic scenarios have been proposed aschaos forming processes, often involving H2O as an importantagent. Sharp (1973) suggested both degradation of ground iceand excavation of magma as plausible formation processes andSoderblom and Wenner (1978) elaborated on a ground ice modelsuggesting that erosion of strata originating from different zonesof H2O stability would form chaotic regions. Recently, Zegerset al. (2010) proposed a new variant of this hypothesis suggestingthat water ice is a part of the rock sequence. As the rock overbur-den reaches a thickness of 1–3 km the ice begins to melt due to thethermal insulation caused by overlying sediment, and thus, theoverlying material destabilizes and water escapes creating chaosterrain.

The close spatial connection between chaotic terrain and out-flow channels has resulted in suggestions that chaotic terrainsare source regions for channel formation. Carr (1978, 1979) argued

G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329 317

that outflow channels emerging full born from chaotic regions arestrongly suggestive for catastrophic flooding by rapid release of aconfined aquifer under great pressure resulting in collapse in adja-cent areas. Triggering mechanisms could either be impact crateringor pore pressure reaching lithostatic pressure. Magma–ground iceinteractions is another possible triggering mechanism, both as aheat source for catastrophic melting of ground ice, as well as dis-rupting the cryosphere releasing a confined aquifer under pressure(Chapman and Tanaka, 2002; Head and Wilson, 2002; Meresseet al., 2008). This model has been supported by Glotch andChristensen (2005), who observed grey, crystalline hematite alongwith phyllosilicates and possible sulfate in Aram Chaos providingevidence for episodic pulses of water release as a chaos formingprocess. Cabrol et al. (1997) stress the importance of volcano tec-tonic strains as another triggering mechanism based on a surveyof Shalbatana Vallis, where crossing fault systems are suggestedto allow hydrothermal drainage of confined aquifers.

Rodriguez et al. (2005) suggest that the different stages of chaosformation result from multiple episodes of reactivation of thehydraulic head and propose a suite of processes such as: renewedmagmatic activity, topographic lowering with respect to thegroundwater table, subsiding cavern roofs or thickening of a per-mafrost seal, as mechanisms for increased hydrostatic pressure.Analysis by Wang et al. (2006) show that freezing-induced pres-surization is a possible mechanism for releasing groundwater,however not on the order of big outflow channel systems.

Debris flow mechanisms, triggered by large scale failures ofsubsurface material, have also been proposed to account for thedevelopment of chaos (Tanaka, 1999; Nummedal and Prior,1981). Based on observations of the Simud/Tiu deposits, Tanaka(1999) proposed that seismic activity could liquefy water-rich sed-iments causing a collapse and lowering of the chaos surface versusthe adjacent plateau, resulting from subsurface removal ratherthan rotational slumping or lobate landslide masses. Wang et al.(2005) support this idea by suggesting impacts as a quaking agentand find evidence for liquefaction in chaotic regions by checker-board patterns of gaps between blocks of chaotic terrain, whichis similar to liquefaction patterns on Earth.

Dissociation of clathrates as a mechanism of subsurface re-moval has also been put forward in several papers (e.g., Milton,1974; Hoffmann, 2000; Komatsu et al., 2000; Rodriguez et al.,2006) starting with Milton (1974) who proposed carbon dioxidehydrate as a possible phase, under which explosive dissociationwould result in a catastrophic dewatering. Komatsu et al. (2000)argue that the stress from decomposed clathrates releases a largequantity of CO2 and CH4, which would subject water-saturatedsediments to so much stress that the sediment would liquefy. Asimilar model, including a series of runaway degassing events forGanges Chaos, has been proposed by Rodriguez et al. (2006), whomention deep fracture propagation, magmatic intrusions and cli-matic thawing and thinning of the cryosphere as plausible triggermechanisms.

2. Geologic setting

Galaxias Chaos is situated in a highly complex region in thetransition zone between Elysium Rise to the south, Utopia Basinto the north, and is bounded by Hecates Tholus to the east andby Elysium/Utopia flows to the west (Fig. 1). Galaxias Chaos wasfirst recognized by Schaber and Carr (1977) and later mapped byScott and Carr (1978) (1:25,000,000) as a knobby material in a zonebetween the lowland surface of Vastitas Borealis and NoachianTerrain. Based on Viking imagery, Mouginis-Mark (1985) andMouginis-Mark et al. (1984) suggested that the upland remnantsrather were a part of the compound lava plains extending from

Elysium Mons than Noachian terrain and emphasized that struc-tural weakness seems to have played an important role. Likewise,Greeley and Guest (1987) suggested that the material of isolatedpatches of grooved terrain was lava and ascribed the formationto Late Hesperian. However, Tanaka et al. (1992) proposed an EarlyAmazonian origin related to Elysium lavas grooved due to flowageof subsurface material, possibly as a result of ground ice meltinginduced by volcanism, and Tanaka et al. (2005) later classified Gal-axias Chaos as a part of the Vastitas Borealis Marginal unit.

Since the first observations of the huge Chryse outflow channelsdraining into the northern lowlands, including Utopia Basin, stud-ies investigating a volatile-rich past have been carried out and en-hanced H2O content in Utopia Basin has been observed by the MarsOdyssey Neutron Spectrometer (Feldman et al., 2002). This resultconfirmed several proposals for a volatile-rich geologic historybased on a suite of morphologies (e.g., Cave, 1993), who observeda high frequency of fluidized ejecta in Utopia Planitia.

Other observations have suggested either an ocean hypothesisor a glaciation hypothesis and Kargel and Strom (1992) and Kargelet al. (1995) have been in favor of the latter. They suggested an an-cient continental glaciation in the martian northern plains basedon observations of thumbprint terrain in association with sinuoustroughs that contain medial ridges, resembling terrestrial glacialfeatures like moraines, tunnel channels and eskers.

On the other hand, Parker et al. (1993) argued for an oceanhypothesis and mapped two potential shorelines, contact 1 andcontact 2, based on a variety of features like cliffs, terraces andabrupt termination of outflow channels. Based on data from MOLAonboard Mars Global Surveyor, Head et al. (1999) checked whetherthe proposed shorelines were reflecting an equipotential line, find-ing that contact 2 was the best approximation to an equipotentialline having a surface elevation of�3760 m with a standard deviationof 0.56 km. The discrepancies vary nonrandomly and deviations oc-cur primarily in regions of Tharsis, Elysium and Isidis, where latermodification of topography is likely. However, as Carr and Head(2003) pointed out, though the standard deviation is small, individ-ual sections of contact 2 have significant ranges in elevation and byassessing the observational evidence for shorelines they found littleevidence for that interpretation. Chapman (1994) and Chapman andTanaka (2002) proposed that in absence of classical coastal morphol-ogies, volcano–ice interactions might provide morphological evi-dence of an ice sheet. Like Allen (1979), who found evidence foranalogs to terrestrial subglacial volcanism close to the Elysium vol-canic region, she found examples of three hyaloclastite-like ridgeson the NW flank of Elysium Rise indicating the existence of an icesheet with a height at least slightly higher than the ridges, whichreach an absolute elevation of �3753 m.

The preferred explanation by Carr and Head (2003) is that theVastitas Borealis Formation (VBF) supports the former presenceof water from large floods better than the proposed shorelines.The outer boundary of VBF lies in approximately �3660 m eleva-tion and VBF is generally smooth with gentle slopes and has a dis-tinctive 3 km scale background surface topography, whichsuggests a non-volcanic origin (Kreslavsky and Head, 2000). Thehomogeneous and fairly smooth nature of VBF at 100 m and longerwavelengths, its location in a basin and in the lower end of the out-flow channels as well as the similarity in age between the outflowchannels and VBF fit the hypothesis for VBF representing a subli-mation residue from a standing body of H2O created by the outflowchannel effluents. Modeling the evolution of outflow effluents, Kre-slavsky and Head (2002) show that freezing in a convection stateunder current martian conditions would occur over a period of104 years and hereafter sublime. Thus, the presence of both anocean and an ice sheet is a plausible result of a catastrophic waterrelease from Chryse and the fast freeze up of the ocean might ex-plain why a clear coastal morphology has not developed.

Fig. 1. A close up of the Elysium Volcanic Province, displaying Elysium Mons, the two flank volcanoes, Hecates Tholus and Albor Tholus, and the transition between ElysiumRise and the Utopia impact basin. Galaxias Chaos is situated in a highly complex region in the transition zone between Elysium Rise to the south, Utopia Basin to the north,and is bounded by Hecates Tholus to the east and by Elysium/Utopia flows to the west.

318 G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329

The Utopia/Elysium deposits west of Galaxias Chaos extendmore than 1500 km from the base of Elysium Rise forming a lobatetongue of relatively smooth and flat channelized materials associ-ated with broad levees (Christiansen and Greeley, 1981; Tanakaet al., 1992). Many of them originate from linear fracture or grabensystems, which are aligned NW–SE suggesting influence of a regio-nal stress field (Plescia, 1986; Hall et al., 1986).

The deposits have been interpreted as lahar flows based onobservations of their lobate outline, steep snouts, smooth medialchannels, their wet-looking nature and their association with vol-canism. Different models for generating lahars have been sug-gested, including volcano–ice interactions, volcano–ground iceinteractions and/or dike disruption of the cryosphere and ground-water release (Christiansen, 1989; Christiansen and Greeley, 1981;Christiansen and Hopler, 1986; Skinner and Tanaka, 2001; Russelland Head, 2001a,b, 2002, 2003; Tanaka et al., 1992).

Most recently, Russell and Head (2001a,b) investigated the Ely-sium/Utopia flows using MOLA data, including detrended and gra-dient topography, and they were able to resolve two main types offlows; lava flows and debris flows. They also interpreted the debrisflows as lahar deposits because of their morphology, stratigraphicrelationships and because of observed dendritic ridges, suggestingdewatering flow after the water-rich lahar deposits came to rest(Russell and Head, 2002). The distribution of the lahar deposits be-low elevations of �3100 m to �4300 m correlates with the distri-bution of fluvial channels, and led Russell and Head (2001b,2003) to suggest that the configuration of deep radial fossaes, lavaflows and lahar deposits can be explained by propagating dikesdisrupting a confined aquifer facilitating large amounts of waterresponsible for the lahar deposits. Thus, the clustering of channeldeposits may reflect the minimum elevation of a subsurface watertable at the time of dike emplacement. This evidence led Russelland Head (2007) to suggest that the global groundwater systemproposed by Clifford (1993) should be elaborated on to incorporatemultiple recharge centers at volcanic provinces including contribu-tion of snow melt to outflow channel activity.

Hrad Fossae is the easternmost outflow channel, and it modi-fies the western part of Galaxias Chaos and continues as Hrad Val-lis further to the west. Several investigations of this area havebeen carried out, some suggesting karst or thermokarst processesresponsible for depressions and channel incisions, implying thepresence of either carbonates, volcanoclastic materials or porousclastic material saturated with water/ice (De Hon, 1992). Mappingperformed by De Hon et al. (1999) strongly suggests interplay ofvolcanism, near-surface volatiles and surface run off. Wilson andMouginis-Mark (2003) propose that Hrad Vallis has a phreatomag-matic explosive origin generated by shallow sill intrusion interact-ing with ice-rich rock layers creating mudflows. This hypothesis issupported by Morris and Mouginis-Mark (2006) who outlineobservations of thermally distinct craters within the depositsand argue that the distribution of the craters and their morphol-ogy are a result of interactions between hot mud flows and groundice.

Hecates Tholus, situated east of Galaxias Chaos, is one of thefew martian volcanoes, which has been proposed to have an explo-sive volcanic origin, deviating from many other volcanic, Hesperianedifices by the presence of channel-incised flanks (e.g., Reimersand Komar, 1979; Fassett and Head, 2006). Recently, remarkable,smooth, lineated deposits with similar characteristics as lineatedvalley fill in a flank depressions at the base of Hecates Tholus weredocumented, and Hauber et al. (2005) interpreted them as evi-dence of young glacial deposits from recent climate changes. If thisinterpretation is correct, the close relation to Galaxias Chaos sug-gests that recent climate changes might have been a plausiblemodifying agent obscuring the primary processes that generatedthe chaos.

Though this example is the closest evidence of late Amazonianpermafrost and periglacial modification to the study area, lots ofmorphologies related to recent climate in Utopia have been re-ported including gullies, scalloped terrain, pingos and small sizepolygonal patterned ground (e.g., De Pablo and Komatsu, 2007;Soare and Osinski, 2007; Soare and Kargel, 2007).

G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329 319

In addition, Utopia Basin, among other regions, has been mutedby mantling material that has been proposed to be an ice-rich de-posit emplaced symmetrically down to latitudes around 30� duringperiods of high obliquity. At high obliquities water could migrateatmospherically from the poles to mid-latitudes where it wasdeposited as snow and frost. Today, this deposit is undergoingreworking, degradation and retreat in response to instability dueto climate changes (Carr and Schaber, 1977; Mustard et al., 2001;Head et al., 2003; Morgenstern et al., 2007).

3. Galaxias Chaos

Since the last investigations of Galaxias Chaos, a lot of new sa-tellite data has been provided by Mars Global Surveyor (MGS),Mars Odyssey, Mars Express and Mars Reconnaissance Orbiter

Fig. 2. (A) Topography of Galaxias Chaos displayed by MOLA digital elevation model drap(dark blue). (B) Geologic map of Galaxias Chaos and surroundings. The chaos itself consis(dark brown). (For interpretation of the references to color in this figure legend, the rea

(MRO) allowing more thorough and detailed studies of chaos form-ing processes.

3.1. Description of data

The region of Galaxias Chaos is covered by several different datasets allowing a description of the area on a regional scale as well asextracting information on small scale processes. Topographic datalike HRSC DEMs and MOLA data provide a new topographic per-spective on Galaxias Chaos. The resolution of HRSC DEM varies be-tween 75 m/pixel and �100 m/pixel and covers most of the region,with the westernmost part as an exception. On the other hand theMOLA DEM covers the whole region with a resolution of 460 m/pixel, but the distribution of MOLA tracks reveals that there is upto 13 km spacing across individual tracks.

ed over a THEMIS IR mosaic. The color code ranges from �3550 m (red) to �4000 mts of the grooved unit (dark orange), sub-grooved unit (light orange) and trough unitder is referred to the web version of this article.)

320 G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329

Several different data sets from cameras sensitive in the visibleto the infrared region have been released since the last examina-tion of Galaxias Chaos. Viking (230 m/pixel), THEMIS IR (100 m/pixel) and HRSC (12.5–25 m/pixel) provide a regional overview ofGalaxias Chaos and THEMIS VIS (18–80 m/pixel) covers a great ex-tent of especially the eastern part of Galaxias Chaos. MOC, CTX andHiRISE provide more sporadic, but detailed images with a resolu-tion of 1.5–7 m/pixel, 6 m/pixel and 0.3–0.6 m/pixel, respectively.

3.2. Geomorphology of Galaxias Chaos

The region of Galaxias Chaos (142–150�E, 33–38�N) wasmapped in 1:100,000, and in several places to greater detail wheredata allowed it. Galaxias Chaos is a mosaic of angular mesas, whichcovers an area of approximately 14,000 km2 and the elevation var-ies from approximately �3550 m to �4000 m (Fig. 2A). GalaxiasChaos is bounded to the east by Hecates Tholus and to the westby the Elysium/Utopia flows and it is situated at the foot alongthe edge of Elysium Rise as a complex jigsaw of polygonal blocks,which are mapped as the dark orange, grooved unit and as light or-ange, sub-grooved unit on the geologic map on Fig. 2B.

The grooved unit of Galaxias Chaos is characterized by well-de-fined blocks, which vary in size from 0.25 km2 to approximately100 km2 and there is no clear size progression away from ElysiumRise. Both rough hummocky knobs and smooth surface texture areobserved on top of the blocks, but no relationship between thetroughs and the surface texture is observed.

Some larger, topographically less pronounced blocks, which areseveral tens of kilometers across, are mapped as a sub-groovedunit. Due to their less distinct relief they are more difficult tomap, and they are mainly distributed in the eastern part of

Fig. 3. Close-up of four parts of the geologic map of Galaxias Chaos and surroundings. (Adifferent parts of Galaxias Chaos displayed in (B–D) and a zoom in on the lobate plateauseen from (A) this part of Galaxias Chaos consists of the grooved unit (polygonal blockslater NW–SE-wards fracturing and infilling of the trough unit. The chaos is bordered by ththe north. The image is a mosaic of CTX P14_006604_2140 and P15_006749_ 2149 and THcentral part of Galaxias Chaos. As displayed in (A), this part of Galaxias Chaos consist oftrough unit (dark brown) and borders the Elysium Rise unit to the south. Towards the nopronounced blocks are marked as the sub-grooved unit in light orange. Moreover, the blmaterial draping some of the blocks. CTX P16_007395_2148. (D) The easternmost part oforange and dark brown unit). Towards the north a big block partly incised by troughsremnant hills (small mesas in dark purple). The image consists of THEMIS VIS V0242material. The black arrows mark the distinct rims, which delineate the flat, lobate plateauthe plains material. THEMIS VIS V19191012. (For interpretation of the references to col

Galaxias Chaos further away from the edge of Elysium. Many ofthe troughs do not penetrate this unit, resulting in blocks with asize up to 40 km � 70 km and in some places the unit seems tobe submerged by material from the Utopia Basin. Like the groovedunit, rough and smooth surface textures are observed, which sug-gests that the surface material of both the grooved and sub-grooved unit is the same. The transition between the two units isabrupt and is displayed in Fig. 3A and D.

There is a huge morphological difference between the easternand western part of the grooved unit. To the east, the unit is welldefined and can be divided into subcircular cells, while the chaoticterrain in the western part has been modified by Galaxias Fossaeand related outflow activity, which seems to submerge existingtroughs (Fig. 3B).

Numerous smaller, angular to subangular blocks up to approx-imately 2–10 km across are observed in Utopia Basin and they existboth in the lobate plains material and plains material and aremarked as purple on Fig. 2B and in addition are displayed in thetop of Fig. 3D. At first sight it looks like these smaller blocks orig-inate from Galaxias Chaos due to their close spatial relationship,their shape and their elevation. However, these blocks can betraced all the way to Phlegra Montes more than 1000 km NE of Gal-axias Chaos as well as to the Isidis basin, more than 3000 km to thesoutheast. Furthermore, they do not display the same jigsaw-likepattern as the grooved and sub-grooved unit and are thereforenot considered to be a part of Galaxias Chaos, consistent with thegeologic map from Tanaka et al. (1992).

The main morphologic difference between the lobate plainsmaterial and the plains material is the characteristic lobate plateaufeatures with a distinct rim, which are distributed between theblocks (Fig. 3E) and both units contain unusual crater morphologies,

) Geologic map as displayed on Fig. 2B. The black boxes show the location of threematerial in (E). (B) Zoom in on the westernmost part of Galaxias Chaos. As it can bemarked with dark orange) and trough unit (dark brown), and has been modified bye Elysium Rise unit (dark red) to the south and by flood plain deposits (light blue) toEMIS VIS V025040061, V04389003, V11878007 and V14037006. (C) Zoom in on thethe well defined jigsaw puzzle of polygonal blocks (dark orange grooved unit) andrth the blocks are less distinct and the troughs are absent. The topographically lessocks to the north are surrounded by plains material (light blue), which is a smoothGalaxias Chaos has a cell-like nature consisting of distinct blocks and troughs (dark

is marked as a sub-grooved unit bordering lobate plateau material (dark blue) and9006 and V12427008. (E) Image displaying a typical morphology in lobate plainss and the white arrows mark subdued knobs, which occasionally seems to penetrateor in this figure legend, the reader is referred to the web version of this article.)

G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329 321

which have been described by Pedersen and Head (2009). The dis-tinct rim associated with the plateaus in lobate plains material isin some places draping the flanks of angular blocks, and within thelobate plains material subdued knobs are observed and they seemoccasionally to penetrate the lobate plains material (Fig. 3E).

For resolving the formation process of Galaxias Chaos, extraemphasis has been put on the data from the eastern and centralpart of Galaxias Chaos, where the chaos is best preserved.

After constraining the extent of the chaotic terrain (grooved,sub-grooved and trough unit) and the extent of later modifyingprocesses (deposits from outflow activity; flood plain deposits, le-vee deposits and smooth flow) an analysis of the surface materialwas carried out by tracing layering and rough surface texture.

3.3. Surface material of Galaxias Chaos

MOC data reveal a dark surface layer observed in trough walls inGalaxias Chaos that can be traced from troughs within ElysiumRise into Galaxias Chaos (Fig. 4). The two images are 80 km apart

Fig. 4. Close-up of the surface layer in the central and eastern part of Galaxias Chaos.respectively) are located 80 km apart and both images display a dark surface layer thaGalaxias Chaos (A, B and E–G) as shown by black arrows. This indicates that Galaxias ChaoMOC images are 3.8 km across.

and are from the eastern and central region of Galaxias Chaos,and they display a distinct dark, competent layer that can bemapped out across the troughs (marked by black arrows). Thisindicates that the same layer traced from Elysium Rise caps thepolygonal blocks within Galaxias Chaos.

Similarly, distinct rough texture and associated flow fronts canbe traced from Elysium Rise all the way to Galaxias Chaos (Fig. 5).Both these observations suggest that Galaxias Chaos is capped byElysium lavas implying that Galaxias Chaos was a part of ElysiumRise before trough formation.

3.4. Trough development

The trough formation itself can be studied by investigating thevariety of troughs, their geometry, and their spatial relationship tothe blocks.

The troughs are generally round-headed, with a width around600–800 m and they are approximately 100 m deep; they varybetween being flat-floored and v-shaped. As displayed in Fig. 6,

Two MOC images from the central and eastern part (m0304232 and r13044670,t can be traced across troughs from the Elysium Rise (C and H) into the blocks ofs is capped by the same material as the Elysium Rise unit that consists of lavas. Both

322 G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329

different stages of trough formation can be found. The early troughformation is displayed in Fig. 6A and B, and Fig. 6A shows a round-headed depression forming within a block marked by breakage ofthe competent surface layer. Likewise, Fig. 6B shows initial troughdevelopment forming within the block, but with a close relation tothe edge of the block, which indicates that the trough formed dueto instability of the slope of the block.

The troughs either continue to grow within the blocks, formingkilometers-long troughs or they incise the block from the edges.Fig. 6C shows an example of a trough which curves subparallel tothe edge of the block and which to the northwest is abutting an-other trough separated by less than 10 m of block material. More-over, the southwestern end of the trough seems to grow along alinear depression marked with black arrows. Similarly, two troughson each side of a block seems to incise a block along a 70 m widedepression, dividing the block into two parts, if the trough incisionprocess continues (Fig. 6D). An example of how the troughs coa-lesce and form cell-like trough systems making a jigsaw puzzleof mesas is displayed in Fig. 6E, and it is observed that minor orno rotation of the blocks has taken place. This indicates that chaosformation was not a catastrophic process, but a gradual develop-ment allowing different stages of trough formation to be preserved.Likewise the cell-like nature suggests that the trough formationhas been ongoing in several places along the contact as multipleevents of trough formation and collapse.

Fig. 5. Rough flow texture and flow fronts can be traced from Elysium Rise into the chaarrows point out flow fronts, which delineate the extent of the rough texture. North is

Moreover, the fact that the troughs form as depressions withinblocks also implies that the trough formation is not governed byregional stresses, but rather local stress within and along blockedges.

3.5. Topography of Galaxias Chaos

The mesas in Galaxias Chaos are developed at �3.6 km to�3.7 km elevation and individual blocks are 100–200 m high,reaching a maximum height around �3.5 km. In this manner, thetopographic characteristics of Galaxias Chaos deviate from otherdescribed chaotic regions, where the blocks lie 1–2 km below thesurrounding surface. In contrast, some of the blocks in GalaxiasChaos are higher than the adjacent slope (Carr, 1979; Rodriguezet al., 2005; Meresse et al., 2008). This is displayed in Fig. 7, wherefour selected MOLA profiles, going from the east to the west, crossthe eastern part of Galaxias Chaos and the black arrows delineatethe grooved unit. Profile 19875 (blue) and 13377 (red) show thatindividual blocks within Galaxias Chaos are slightly elevated withrespect to the surrounding plateau.

At some places a plateau is developed in front of the chaos, asshown in profile 10660 (yellow), but generally the elevation ofthe blocks decrease away from Elysium Rise.

From THEMIS- and CTX-images it is observed that some of theblocks display chaotic tilting, resulting in a well-defined relief in

os, displayed here in THEMIS image V11541008, which is 23 km across. The blackup.

Fig. 6. Different stages of trough development marked with black arrows can betraced within Galaxias Chaos indicating that chaos formation is a gradual process.(A) Incipient trough development within a block. Fractures are highlighted by blackarrows. MOC m0304232. (B) Trough formation along the edge of a block. Section ofCTX-image P04_002714_2144. (C) A closed trough within a block. In the bottom ofthe image a weakly developed linear depression indicates the direction of furthertrough propagation (marked by black arrows). To the left of the image a black arrowmarks a bridge of material, which bounds the closed trough from coalescing withother troughs. Section of CTX-image P04_002714_2144. (D) A block incised bytroughs from two sides. The arrow points to a linear depression, which indicates thedirection of trough propagation. Section of CTX-image P04_002714_2144. (E)Coalescing troughs creates polygonal and very well-defined blocks. Section of CTX-image P04_002714_2144.

G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329 323

some parts of a mesa, while other parts are more subdued indicat-ing that the surface material of Galaxias Chaos superposes anunstable material and that they have experienced differential ver-tical displacement (Fig. 8A and B).

Furthermore some blocks partly superpose a competent mate-rial causing breakage of the superposed blocks due to unequal sub-sidence. In the case of Fig. 8C it seems that the central part of thetwo yellow, highlighted blocks stalled on a stable subsurface mate-rial, while the flanks have experienced subsidence causing break-age of the central part of the block as the surface material couldnot sustain further differential stress. The two topographic profilesA–B and A0–B0 (Fig. 8) are based on HRSC DEM with a resolution of75 m/pixel and show clearly that the flanks are tilted in oppositedirections, having a local minimum, where the blocks break. If itis assumed that the blocks originally were horizontal and thatthe maximum elevation represents the initial elevation, an esti-mate of the subsidence can be calculated. Profile A–B reaches amaximum height of �3720 m, while the minimum elevation ofthe flanks is �3880 m resulting in an elevation difference of160 m. The equivalent calculation for A0–B0 gives an elevation

difference of 120 m, however this approximation is more uncertainsince the flanks of the block seem to have slid downhill, making thedistance between the flanks greater.

The calculated subsidence is within the same elevation range asthe depths of the troughs, and is similar to the topographic differ-ence of the elevated blocks compared to the adjacent slope. Thus, ifthese observations of differential subsidence up to 160 m can applyto the rest of Galaxias Chaos, elevated blocks can be explained bythe adjacent slope subsiding a bit more than some of the mesasin the chaos.

4. Model for formation of Galaxias Chaos

Galaxias Chaos is significantly different from other chaotic re-gions. First of all the topography is very different and in GalaxiasChaos the elevation of mesas is in the same range as the surround-ings, whereas other chaotic terrains have a significant elevationdifference up to several thousand meters between blocks and pla-teau. Moreover, Galaxias Chaos is not associated with huge outflowchannels, and Hrad Vallis is modifying rather than fed by the wes-tern part of Galaxias Chaos. The different stages of trough forma-tion clearly indicate a gradual trough formation with minorvertical movement, which contrasts to a catastrophic formationprocess. Thus, proposed formation models for chaotic regionsaround Chryse Planitia do not appear to be directly applicable toGalaxias Chaos.

The observations from Galaxias Chaos suggest that the surfacematerial is Elysium lavas based on the distinct surface layer andthe rough surface texture associated with flow fronts that havebeen traced from Elysium Rise into Galaxias Chaos. Moreover,the topography reveals that a differential amount of subsidencehas caused different stages of trough formation indicating anunstable subsurface layer. The tilting of mesas and the fact thatsome blocks have elevations equivalent or slightly higher thanthe surrounding plateau also support a differential vertical dis-placement. This implies that some parts of Elysium Rise have expe-rienced a slightly greater downward movement without causingbreakage in the lava cap than the few elevated mesas within Gal-axias Chaos. Additionally, the fact that troughs develop withinthe blocks indicate that the trough formation is not governed byregional stress fields, but by local stress fields, which fits an expla-nation of differential subsurface degradation and/or flow. Thus, aconsistent formation model for Galaxias Chaos should include asubstrate, which due to shrinkage/movement can cause gradualtrough formation induced by local stresses.

All these observations exclude proposed chaos formation mod-els such as magma excavation, clathrate dissociation, catastrophicgroundwater release and melting of ice due to thermal insulationcaused by overlying sediment. However, degradation of groundice is a plausible mechanism of differential downward movementof a substrate creating gradual trough formation, tilt of blocksand variable local stresses.

Considering the close spatial relationship with lobate plainsmaterial, which Tanaka et al. (2005) assigned to be a part of Vast-itas Borealis Formation, this unit would be a good candidate as asubsurface material from a stratigraphic point of view. Moreover,the observed morphologies of lobate plateau material drapingangular blocks and subdued knobs that occasionally seems to pen-etrate the lobate plains material indicate degradation and shrink-age revealing the subsurface topography (Fig. 3E).

Additionally, the VBF has been considered to be a residue from aH2O-rich material deposited by huge floods, as earlier mentioned(e.g., Kreslavsky and Head, 2002; Carr and Head, 2003). The thick-ness and the elevation of the outline of VBF fit both the depth andthe elevation of the troughs (Carr and Head, 2003) and the

Fig. 7. Four topographic profiles and their location in the eastern part of Galaxias Chaos. From the top the four profiles are selections of MOLA tracks; ap19875, ap10660,ap13377 and ap19863 starting from the east to the west. Black arrows point out the extent of the grooved unit.

324 G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329

Fig. 8. Chaotic tilting of blocks within Galaxias Chaos shows that Elysium lavas are emplaced on an unstable material making the blocks tilt. (A) Two blocks tilt towards thecenter of the image displaying well-defined relief to the east and to the north, while the outline of the blocks are weakly developed in the central part of the image. (B) The bigelongated block tilts towards the northwest resulting in well-defined relief in the southeastern part of the image compared to the northwestern part, where plains materialspartly surround the block. (C) Breakage of two blocks marked in yellow indicates that the area has subsided unequally making the blocks break, where they stalled on morestable subsurface material. The topographic profiles A–B and A0–B0 are based on HRSC DEM and show that the blocks break in the middle and slide to each side of atopographic high. If it is assumed that the blocks were horizontal before subsidence, a minimum estimate of the subsidence can be calculated. The image is a section ofTHEMIS VIS V124270081.

G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329 325

estimated age of VBF is Early Amazonian, compared to the latestactivity on the Elysium Rise (Tanaka et al., 1992, 2005), making ita plausible scenario that Vastitas Borealis Formation is sandwichedbetween Elysium lavas.

The suggested formation process for Galaxias Chaos is shown inFig. 9, where a block diagram shows the stratigraphic relationshipsbetween Elysium lavas and VBF. VBF is displayed in variable lightblue colors indicating variable volatile content and it is sand-wiched between Elysium lavas in grey. The lava cap on top ofVBF inhibits volatile sublimation, but in areas where there is nolava or where troughs have formed, volatiles sublime (indicatedby wavy, orange arrows). Sublimation of volatiles within the VBFresults in shrinkage and degradation of the unit and thereby facil-itates subsurface movement. Where the VBF wedges out betweenElysium lavas, VBF is so thin that mass movement and vertical dis-placement is so small that no trough formation occurs because ofthe strength of the lava cap itself. As the VBF increase in thicknessthe lateral transport of sediment and the potential vertical dis-placement increase, undermining the lava cap. Eventually the lavacap experiences higher stresses than it can support, creatingdepressions and later troughs as sublimation through the brokenlava cap occurs. The most intense chaos formation will thereforeoccur in the parts of Elysium Rise where the thickness of the VBFis sufficient, and where the slope is steep enough to facilitate andconcentrate subsurface movement. Where particularly intenseand continued mass movements occur, tilted blocks as well as sub-surface material submerging and superposing low lying blocksmight result.

This model explains several observations. First of all, thetroughs with highest elevation occur along an equipotential line

having an altitude around �3.6 km to �3.7 km. This fits with astratigraphy where the VBF was deposited with a surface elevationaround �3660 m and later capped by Elysium lava flows. Thus, ifthe VBF was deposited as an equipotential surface, it would be ex-pected that the occurrences of the most elevated troughs should beequipotential and it would have been a problem if chaos formedwell above the suggested altitude of the VBF.

Secondly, the extent of VBF may very well explain the cell-likenature of the chaos development. Before the VBF was deposited thetopography of Elysium Rise was controlled by its fingering flowscreating a wavy topography. Thus, as the VBF was emplaced asan equipotential surface its outline and its thickness were con-trolled by this pattern filling up lows between lava flows withthicker VBF deposits than elsewhere along Elysium Rise. Therefore,as the volatile loss initiated the movement of the substrate, it wascontrolled by local topography under the volatile-rich unit, facili-tating trough formation in the former lows between lava flows cre-ating the cell-like nature of Galaxias Chaos. This is illustrated inFig. 9 where a thicker deposit of the VBF occurs in the lows ofthe oldest Elysium lavas (dark grey).

Thirdly, this model explains the gradual development oftroughs creating a jigsaw pattern, because Galaxias Chaos basicallywas a part of Elysium Rise, which was incised by troughs causingwell preserved mesa surfaces, minor lateral movement of theblocks and no huge topographic difference between the mesasand the plateau itself.

Moreover, the fact that the mass movement is limited by thethickness of the VBF as well as by the dip of slope, might explainwhy the well defined grooved unit occurs within a belt. The sub-grooved unit is, like the grooved unit, capped by lava and the

Fig. 9. A schematic drawing of the chaos formation process, showing a bluish volatile-rich sediment (VBF) sandwiched between grey Elysium lavas and a deeper substrate.The lava cap on top of the VBF inhibits volatiles from sublimating, but sublimation and shrinkage occur outside the lava capped region causing lateral transport in the VBF,undermining the lava cap. Because of the wavy topography of the underlying Elysium lavas, the VBF varies in thickness. In areas with a thick volatile-rich substrate theshrinkage and lateral movement will be greater in areas with a thinner volatile-rich substrate, causing differential mass movement and variable local stresses. Thus, cells ofchaos will develop in areas with a thick VBF package, because the subsurface movement is so significant that the lava cap cannot sustain the extension. The lava cap breaksand creates a trough, from which volatiles can evaporate causing headwards growth of the troughs. Sometimes a linear depression indicates the direction of troughpropagation as a precursor of the headwards growth (e.g., Fig. 6C and D). The volatile content is illustrated by the intensity of the blue color showing a lowering in volatilecontent, where the sediment is in contact with the atmosphere allowing evaporation; the red arrows show the direction of mass movement. Because of differential substratemovement some blocks tilt (left block diagram), whereas in other areas blocks are elevated due to the existence of a competent subsurface layer (marked by the hatchedsignature in the right block diagram) and thus the lava cap stalls, whereas the surroundings subside. (For interpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)

326 G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329

deposits of the VBF must at least have been as thick as in the areaof the grooved unit. However, the region of the sub-grooved unithas a significantly smaller dip (Fig. 7) resulting in smaller subsur-face mass movement, which can explain less distinct trough devel-opment and mesa formation.

Finally, variable VBF thicknesses as well as heterogeneitieswithin VBF will cause differential vertical displacement duringthe downwasting of the VBF. This explains why some mesas withinGalaxias Chaos are higher than the plateau and why some blocksseem to have stalled due to a competent subsurface layer. Differen-tial downwasting will make the surface layer bend and tilt until thestress is so high that the surface layer breaks (Fig. 8C). This is illus-trated in the right part of the block diagram in Fig. 9, where thehatched signature symbolizes the competent subsurface layer.

The geologic history of Galaxias Chaos is summarized in Fig. 10.First Elysium lavas were emplaced in the Late Hesperian period instage A (Tanaka et al., 1992, 2005). In the Early Amazonian periodthe Vastitas Borealis Formation was deposited and was later super-posed by younger lavas due to continued volcanic activity (stageB). Because of sublimation of volatiles within VBF, either due to en-hanced geothermal activity or climate changes, the VBF degraded,causing differential movement and undermining of the lava cap,resulting in local extension and subsiding troughs. Along thetrough walls enhanced volatile evaporation caused the trough togrow headwards, producing a jigsaw puzzle of blocks as thetroughs coalesce (stage C).

Finally after formation of the chaos, continued volcanic activityalong the NW–SE trending fractures caused modification of the wes-tern part of Galaxias Chaos, producing outflow deposits (stage D).

5. Discussion

This proposed formation model raises several important ques-tions considering the nature of VBF; what mechanisms controlledthe sublimation of volatiles? And do other examples of similarstratigraphic relationships exist both in the region of Elysiumand elsewhere?

The role of the VBF in the history of the martian hydrologic cy-cle is crucial, especially with respect to resolving mechanisms ofdeposition and later modification, linking the present day mor-phology to its origin. This work stresses the importance in estimat-ing the amount of H2O deposited and sublimated as well asunderstanding the nature of the VBF. This is particularly importantin this study, because of the morphologies that developed duringvolatile loss, not only within the VBF, but also in the marginalareas, where the VBF might be superposed by other units. Kreslav-sky and Head (2002) modeled the evolution of outflow channeleffluents and proposed that after emplacement, intense sublima-tion and cooling, outflow effluents would freeze up solid, sublimeand in the end a residual ice-rich sediment producing VBF wouldbe left for later modification.

The porosity on Mars is considered to decay significantly slowerthan on Earth and Clifford (1993) calculated that a Moon-like crustscaled to martian conditions would have a porosity decay constantof 2.82 km. Thus, assuming that the lithostatic pressure is the onlycontrolling factor on the porosity in the VBF, the porosity would be97%, 93% and 90% of its initial porosity in 100 m, 200 m and 300 mdepth, respectively. Basically, this means that if the VBF is an oceandeposit containing lots of very fine grained material, and with H2O

Fig. 10. Formation model for Galaxias Chaos. (A) Emplacement of Elysium lava. (B)Deposition of Vastitas Borealis Formation (beige unit) superposed by new lava flowsfrom Elysium. (C) Substrate movement due to volatile loss causes local extension inthe overlaying lava cap, creating subsiding troughs. The trough formation enhancesthe volatile loss along the trough walls and therefore the troughs grow and coalesceproducing a jigsaw puzzle of blocks. (D) Later volcanic activity along the NW–SEtrending fractures caused modification of the western part of Galaxias Chaos.

G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329 327

filling up the pore space, immense amounts of volatiles would havebeen present in the deposits. Moreover, freezing such a finegrained material would certainly result in cryosuction, for whichreason ice segregation would produce massive ice lenses of groundice producing stratified or veined sequences of sediment and ice(Clifford and Parker, 2001). Therefore, it is reasonably to believethat the volatile content in the VBF would be sufficient for shrink-ing the VBF to a degree, where it would undermine a lava cap pro-ducing troughs; moreover, the heterogeneous distribution of icelenses would facilitate differential subsidence.

Another interesting question is: How was the degradation of theVBF below the lava cap initiated? Obviously, where the VBF was incontact with the atmosphere, an ice-rich sediment under currentmartian condition would sublime and cause shrinkage. The ques-tion, however, is whether this downwasting was significant en-ough to cause downward and lateral movement within the lava-capped part of the VBF initiating the trough formation, whichwould facilitate further volatile escape, or whether enhanced geo-thermal flux was essential. Gulick (1998) showed that in volcanicregions with intrusions greater than several hundreds of cubic kilo-meters, hydrothermally derived fluids can melt through 1–2 kmthick permafrost in several thousand years, and since GalaxiasChaos is associated with Elysium Mons and Hecates Tholus, it isclear that a heat source for melting ground ice is available in theregion. However, it is very likely that there will be no geomorphicdifference, whether Galaxias Chaos originates purely due to climat-ically sublimed volatiles or whether enhanced geothermal flux ex-isted, and it is questionable if mineralogical data from orbit wouldprovide any evidence. Thus, the most plausible way to constrain

this problem would be to estimate the amount of volatiles depos-ited in VBF and the amount of sublimed volatiles estimating thatthe vertical displacement of the exposed VBF would be sufficientto cause a gravitational subsurface flow in the region of GalaxiasChaos.

If one suggests that the stratigraphic relationship between theVBF and Elysium lavas is controlling the formation of GalaxiasChaos, an import question to raise is whether chaos formation oc-curs elsewhere on the edge of Elysium Rise. From the proposed for-mation model, one would expect to find chaos, where the VBF andElysium lavas are interbedded, and from the geologic map made byTanaka et al. (1992, 2005) this contact between VBF and Elysiumlavas can be traced from west of Hecates Tholus along the bound-ary of Elysium Rise to the western part of Elysium Rise. However,the later Amazonian outflow channels have modified this contactheavily to the northwest of Elysium Rise, and therefore the westernpart of Elysium Rise is the only area where there might be rem-nants of chaos formation.

Fig. 11 displays chaotic terrains found on the western flank ofElysium and the similarities with Galaxias Chaos are striking. Skin-ner and Tanaka (2007) ascribed these terrains to fractured risesoriginating from laccolith-like intrusions, but because of the simi-lar morphology and because they are situated approximately at thesame elevation, they are here suggested to have formed the sameway as Galaxias Chaos. The chaotic terrains can be seen onFig. 11A as red areas, where the contour lines �3500 m, �3600 mand �3700 m are marked by orange, blue and green, respectively.Some chaotic terrains on the western flank are isolated featuresand seem to have been partly submerged by later lava flows(Fig. 11B and C), which might explain why some of the chaotic ter-rains are situated at �3400 m height. A similar stratigraphic rela-tionship has been reported by Ivanov and Head (2003), where azone of plateau breakup of young lava flows that have been super-imposed onto the surface of a volatile-rich substratum has been re-ported along the contact between Syrtis Major and VBF in thetransition zone between Syrtis to Isidis. Therefore, it reasonableto believe that the formation process of Galaxias Chaos is relevantto several transition zones producing chaotic terrains significantlydifferent from the chaos regions in Valles Marineris and in the re-gion of Chryse outflow channels.

6. Conclusion

Galaxias Chaos deviates significantly from other chaotic regionsdue to the lack of associated outflow channels, lack of large eleva-tion differences between the chaos and the surrounding terrain,and due to gradual trough formation. A sequence of troughs in dif-ferent stages is detected, and examples of closed troughs withinblocks suggest that the trough formation is governed by a localstress field rather than a regional stress field. The geomorphic evi-dence suggests that Galaxias Chaos is capped by Elysium lavas,which superpose an unstable subsurface layer that causes chaotictilting of the blocks as well as trough formation. Based on regionalmapping we here suggest a formation model where the unstablesubsurface material, which is interbedded with Elysium lavas, isthe Vastitas Borealis Formation, and which due to gradual volatileloss, causes shrinkage and differential substrate movement. Thisprocess would result in local extension and depression formation,ending with trough development and trough coalescence produc-ing a jigsaw puzzle of blocks. Observations of chaos regions tothe west of Elysium Rise indicate that this process might have beenwidespread along the contact between the Vastitas Borealis Forma-tion and Elysium lavas, but probably has been covered to the NWof the Elysium Rise by Elysium/Utopia flows, and has partly beensubmerged by younger lavas to the west.

Fig. 11. (A) Distribution of chaos terrain in the Elysium Volcanic Province. Chaos is marked as red areas and the elevations; �3700 m, �3600 m and �3500 m are marked bythe green, blue and orange contour lines, respectively. Chaotic terrain is observed approximately in the same elevation along the northern transition zone and the westerntransition zone between Elysium Rise and the Utopia Basin. (B) Zoom in on the chaotic terrain on the western flank of Elysium, which seems to be partly submerged by laterlava flows. THEMIS IR mosaic. (C) Chaotic terrain in the western part of Elysium with a similar polygonal jigsaw pattern as observed in Galaxias Chaos. The rough surfacetexture is associated with flow fronts and the troughs show rounded head scarps like Galaxias Chaos. CTX-image P010013431967. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article.)

328 G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329

Acknowledgments

Thanks are extended to James Dickson and Caleb I. Fassett fromBrown Planetary Geoscience Group for their very valuable help inGIS mapping and data processing. Moreover, we thankfullyacknowledge Per Noernberg and the rest of the Mars SimulationLaboratory staff at Aarhus University as well as Sandra Schumacherand an anonymous reviewer for their improvement of the manu-script. Furthermore, we appreciate the efforts of the HiRISE, CTX,HRSC, MOLA, MOC and THEMIS teams to make their data available.

References

Allen, C.C., 1979. Volcano–ice interactions on Mars. J. Geophys. Res. 84, 8048–8059.Cabrol, N.A., Grin, E.A., Dawidowicz, G., 1997. A model of outflow generation by

hydrothermal underpressure drainage in volcano-tectonic environment,Shalbatana Vallis (Mars). Icarus 125, 455–464.

Carr, M., 1978. Formation of martian flood features by release of water fromconfined aquifers. NASA Technical Memorandum 79729, 260–262.

Carr, M., 1979. Formation of martian flood features by release of water fromconfined aquifers. J. Geophys. Res. 84, 2995–3007.

Carr, M., Schaber, G.G., 1977. Martian permafrost features. J. Geophys. Res. 82 (28),4039–4054.

Carr, M., Head, J.W., 2003. Oceans on Mars: An assessment of the observationalevidence and possible fate. J. Geophys. Res. 108 (E5). doi:10.1029/2002JE001963.

Cave, J.A., 1993. Ice in the northern lowlands and southern highlands of Mars and itsenrichments beneath the Elysium lavas. J. Geophys. Res. 98, 11079–11097.

Chapman, M.G., 1994. Evidence, age, and thickness of a frozen Paleolake in UtopiaPlanitia, Mars. Icarus 109, 393–406.

Chapman, M.G., Tanaka, K.L., 2002. Related magma–ice interactions: Possibleorigins of Chasmata Chaos, and surface materials in Xanthe, Margaritifer, andMerdiani Terrae, Mars. Icarus 155, 324–339. doi:10.1006/icar.2001.6735.

Christiansen, E.H., 1989. Lahars in the Elysium region of Mars. Icarus 17, 203–206.Christiansen, E.H., Greeley, R., 1981. Mega-lahars in the Elysium region, Mars. Lunar

Planet. Sci. XII, 138–140 (abstract).

Christiansen, E.H., Hopler, J.A., 1986. Geomorphic evidence for subsurface volatilereservoirs in the Elysium Region of Mars. Lunar Planet. Sci. XVII, 125–126(abstract).

Clifford, S.M., 1993. A model for the hydrologic and climatic behavior of water onMars. J. Geophys. Res. 98 (E6), 10973–11016.

Clifford, S.M., Parker, T.J., 2001. The evolution of the martian hydrosphere and itsimplications for the fate of a primordial ocean. Icarus 154, 40–79. doi:10.1006/icar.2001.6671.

De Hon, R.A., 1992. Polygenetic origin of Hrad Vallis region of Mars. Lunar Planet.Sci. XXII, 45–51 (abstract).

De Hon, R.A., Mouginis-Mark, P.J., Brick, E.E., 1999. Geologic map of theGalaxias Quadrangle, MTM 35217 of Mars. Atlas of Mars, US GeologicalSurvey.

De Pablo, M.A., Komatsu, G., 2007. Pingo fields in the Utopia Basin, Mars: Geologicaland climatic implications. Lunar Planet. Sci. XXXVIII, 1278 (abstract).

Fassett, C.I., Head, J.W., 2006. Valleys on Hecates Tholus, Mars: Origin by basalmelting of summit snowpack. Planet. Space Sci. 54, 370–378. doi:10.1016/j.pss.2005.12.011.

Feldman, W.C. et al., 2002. Global distribution of neutrons from Mars: Results fromMars Odyssey. Science 297, 75–78. doi:10.1126/science.1073541.

Glotch, T.D., Christensen, P.R., 2005. Geologic and mineralogical mapping of AramChaos: Evidence for water-rich history. J. Geophys. Res. 110. doi:10.1029/2004JE002389.

Greeley, R., Guest, J.E., 1987. Geologic map of the eastern equatorial region of Mars.US Geol. Surv. Misc. Invest. Ser. Map I-1802-B.

Gulick, V.C., 1998. Magmatic intrusions and a hydrothermal origin for fluvial valleyson Mars. J. Geophys. Res. 103, 19365–19387.

Hall, J.L., Solomon, S.C., Head, J.W., 1986. Elysium region, Mars: Tests of lithosphericloading models for the formation of tectonic features. J. Geophys. Res. 91 (B11),11377–11392.

Hauber, E., Gasselt, S.V., Ivanov, B.A., Werner, S.C., Head, J.W., Neukum, G., Jaumann,R., Greeley, R., Mitchell, K.L., Muller, J.P., and The HRSC Co-Investigator Team,2005. Discovery of flank caldera and very young glacial activity at HecatesTholus, Mars. Nature 434, 356–361. doi:10.1038/nature03423.

Head, J.W., Wilson, L., 2002. Mars: A review and synthesis of general environmentsand geological settings of magma–H2O interactions. In: Smellie, J.L., Chapman,M.G. (Eds.), Volcano–Ice Interaction on Earth and Mars. Geological Society,London, pp. 27–57.

Head, J.W., Hiesinger, H., Ivanov, M.A., Kreslavsky, M., Prat, S., Thomson, B.J., 1999.Possible ancient oceans on Mars: Evidence from Mars Orbiter Laser AltimeterData. Science 286, 2134–2137. doi:10.1126/science.286.5447.2134.

G.B.M. Pedersen, J.W. Head III / Icarus 211 (2011) 316–329 329

Head, J.W., Mustard, J.W., Kreslavsky, M., Milliken, R.E., Marchant, D.R., 2003. Recentice ages on Mars. Nature 426, 797–802. doi:10.1038/nature02114.

Hoffmann, H., 2000. White Mars: A new model for Mars’ surface and atmospherebased CO2. Icarus 146, 326–342. doi:10.1006/icar.2000.6398.

Ivanov, A.M., Head, J.W., 2003. Syrtis Major and Isidis Basin contact: Morphologicaland topographic characteristics of Syrtis Major lava flows and material of theVastitas Borealis Formation. J. Geophys. Res. 108 (E6), 5063. doi:10.1029/2002JE001994.

Kargel, J.S., Strom, R.G., 1992. Ancient glaciation on Mars. Geology 20, 3–7.Kargel, J.S., Baker, V.R., Begét, J.E., Lockwood, J.F., Péwé, T.L., Shaw, J.S., Strom, R.G.,

1995. Evidence of ancient continental glaciation in the martian northern plains.J. Geophys. Res. 100 (E3), 5351–5368.

Komatsu, G., Kargel, J.S., Baker, V.R., Strom, R.G., Ori, G.G., Mosangini, C., Tanaka, K.L.,2000. A chaotic terrain formation hypothesis: Explosive outgas and outflow bydissociation of clathrate on Mars. Lunar Planet. Sci. XXXI, 1434 (abstract).

Kreslavsky, M., Head, J.W., 2000. Kilometer-scale roughness of Mars: Results fromMOLA data analysis. J. Geophys. Res. 105 (E11), 26695–26712. doi:10.1029/2000JE001259.

Kreslavsky, M., Head, J.W., 2002. Fate of outflow channel effluent in the northernlowlands of Mars: The Vastitas Borealis Formation as a sublimation residuefrom frozen pounded bodies of water. J. Geophys. Res. 107 (E12). doi:10.1029/2001JE001831.

Meresse, S., Costard, F., Mangold, N., Masson, P., Neukum, G., 2008. Formation andevolution of the chaotic terrains by subsidence and magmatism: HydraotesChaos, Mars. Icarus 194 (2), 487–500. doi:10.1016/j.icarus.2007.10.023.

Milton, D.J., 1974. Carbon dioxide hydrate and floods on Mars. Science 183, 654–656.Morgenstern, A., Hauber, E., Reiss, D., Gasselt, S.V., Grosse, G., Schirrmeister, L., 2007.

Deposition and degradation of a volatile-rich layer in Utopia Planitia andimplications for climate history on Mars. J. Geophys. Res. 112. doi:10.1029/2006JE002869.

Morris, A.R., Mouginis-Mark, P.J., 2006. Thermally distinct craters near Hrad Vallis,Elysium Planitia, Mars. Icarus 180, 335–347. doi:10.1016/j.icarus.2005.09.015.

Mouginis-Mark, P.J., 1985. Volcano/ground ice interactions in Elysium Planitia,Mars. Icarus 64, 265–284.

Mouginis-Mark, P.J., Wilson, L., Head, J.W., Brown, S.H., Hall, J.L., Sullivan, K.D., 1984.Elysium Planitia, Mars: Regional geology, volcanology and evidence forvolcano–ground ice interactions. Earth Planet. Sci. Lett. 30, 149–173.

Mustard, J.W., Cooper, C.D., Rifkin, M.K., 2001. Evidence for recent climate changeon Mars from the identification of youthful near-surface ground ice. Nature 412,411–414. doi:10.1038/35086515.

Nummedal, D., Prior, D.B., 1981. Generation of martian chaos and channels bydebris flows. Icarus 45, 77–86.

Parker, T.J., Gorsline, D.S., Saunders, R.S., Pieri, D.C., Schneeberger, D.M., 1993.Coastal geomorphology of the martian northern plains. J. Geophys. Res. 98,11061–11078.

Pedersen, G.B.M., Head, J.W., 2009. Overview of possible ice-related morphologies inthe transition zone between Elysium and Utopia Basin, Mars. Lunar Planet. Sci.XXXX, 2081 (abstract).

Plescia, J.B., 1986. Tectonics of the Elysium region, Mars. Lunar Planet. Sci. XVII,672–673 (abstract).

Reimers, C.E., Komar, P.D., 1979. Evidence for explosive volcanic density currents oncertain martian volcanoes. Icarus 39, 88–110.

Rodriguez, J.A.P., Sasaki, S., Kuzmin, R.O., Dohm, J.M., Tanaka, K.L., Miyamoto, H.,Kurita, K., Komatsu, G., Fairén, A.G., Ferris, J.C., 2005. Outflow channel sources,reactivation, and chaos formation, Xanthe Terra, Mars. Icarus 175, 36–57.doi:10.1016/j.icarus.2004.10.025.

Rodriguez, J.A.P., Kargel, J.S., Crown, D.A., Bleamaster III, L.F., Tanaka, K.L., Baker,V.R., Miyamoto, H., Dohm, J.M., Sasaki, S., Komatsu, G., 2006. Headwardgrowth of chasmata by volatile outburst, collapse, and drainage: Evidencefrom Ganges chaos, Mars. Geophys. Res. Lett. 33. doi:10.1029/2006GL026275.

Russell, P.S., Head, J.W., 2001a. The Elysium/Utopia flows: Characteristics fromtopography and a model of emplacement. Lunar Planet. Sci. XXXII. Abstract1040.

Russell, P.S., Head, J.W., 2001b. The Elysium/Utopia flows: Evidence for release ofconfined groundwater in the context of a global cryosphere–hydrospheresystem. In: Conference on the Geophysical Detection of Subsurface Water onMars. Abstract 7052.

Russell, P.S., Head, J.W., 2002. Unusual dendritic Ridged Terrain in Utopia Planitia:Dewatering of Megalahars? Lunar Planet. Sci. XXXIII. Abstract 2032.

Russell, P.S., Head, J.W., 2003. Elysium–Utopia flows as mega-lahars: A model ofdike intrusion, cryosphere cracking, and water-sediment release. J. Geophys.Res. 108, 18-11–18-28. doi:10.1029/2002JE001995.

Russell, P.S., Head, J.W., 2007. The martian hydrologic system: Multiple rechargecenters at the large volcanic provinces and the contribution of snowmelt tooutflow channel activity. Planet. Space Sci. 55, 315–332. doi:10.1016/j.pss.2006.03.010.

Schaber, G.G., Carr, M., 1977. Martian permafrost features. J. Geophys. Res. 82,4039–4054.

Scott, D.H., Carr, M., 1978. Geologic map of Mars. US Geol. Surv. Misc. Invest. Ser., M.I-1083.

Sharp, R.P., 1973. Mars: Fretted and chaotic terrains. J. Geophys. Res. 78, 4073–4083.

Skinner, J.A., Tanaka, K.L., 2001. Regional emplacement history of the Utopia andElysium Plains deposits, Mars. Lunar Planet. Sci. XXXII. Abstract 2154.

Skinner, J.A., Tanaka, K.L., 2007. Evidence for implications of sedimentary diapirismand mud volcanism in the southern Utopia highland–lowland boundary. Icarus186, 41–59. doi:10.1016/j.icarus.2006.08.013.

Soare, R.J., Kargel, J.S., 2007. Thermokarst processes and the origin of crater-rimgullies in Utopia and western Elysium Planitia. Icarus 191 (1), 95–112.doi:10.1016/j.icarus.2007.04.018.

Soare, R.J., Osinski, G.R., 2007. Periglacial evidence (Using HiRISE, MOC and THEMISImagery) in Utopia and western Elysium Planitia, for a recent wet and warmMars. Lunar Planet. Sci. XXXVIII. Abstract 1440.

Soderblom, L.A., Wenner, D.B., 1978. Possible fossil H2O liquid–ice interfaces in themartian crust. Icarus 34, 622–637.

Tanaka, K.L., 1999. Debris-flow origin for Simud/Tiu deposits on Mars. J. Geophys.Res. 104, 8637–8652. doi:10.1029/98JE02552.

Tanaka, K.L., Chapman, M.G., Scott, D.H., 1992. Geologic map of the Elysium regionof Mars. US Geol. Surv. Misc. Invest. Ser., Map I-2147.

Tanaka, K.L., Skinner, J.A., Hare, T.M., 2005. Geologic map of the northern plains ofMars. US Geol. Surv. Misc. Invest. Ser. Map I-2811.

Wang, C.Y., Manga, M., Wong, A., 2005. Floods on Mars released from groundwaterby impact. Icarus 175, 551–555. doi:10.1016/j.icarus.2004.12.003.

Wang, C.Y., Manga, M., Hanna, J.C., 2006. Can freezing cause floods on Mars?Geophys. Res. Lett. 33, L20202. doi:10.1029/2006GL027471.

Wilson, L., Mouginis-Mark, P.J., 2003. Phreatomagmatic explosive origin of HradVallis, Mars. J. Geophys. Res. 108 (E8), 1–16. doi:10.1029/2002JE001927. 1-1.

Zegers, T.E., Oosthoek, J.H.P., Rossi, A.P., Blom, J.K., Schumacher, S., 2010. Melt andcollapse of buried water ice: An alternative hypothesis for the formation ofchaotic terrains on Mars. Earth Planet Sci. Lett. 297 (3–4), 496–504.doi:10.1016/j.epsl.2010.06.049. ISSN 0012-821X.