Icarus 214 (2011) 413–432

Contents lists available at ScienceDirect

Icarus

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

Constraints on the origin and evolution of the layered mound in Gale Crater,Mars using Mars Reconnaissance Orbiter data

B.J. Thomson a,⇑, N.T. Bridges a, R. Milliken b, A. Baldridge c, S.J. Hook d, J.K. Crowley e, G.M. Marion f,C.R. de Souza Filho g, A.J. Brown h, C.M. Weitz c

a The Johns Hopkins Applied Physics Laboratory, 11100 John Hopkins Rd., Laurel, MD 20723, United Statesb University of Notre Dame, Notre Dame, IN 46556, United Statesc Planetary Science Institute, 1700 East Fort Lowell, Suite 106, Tucson, AZ 85719, United Statesd Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Dr., Pasadena, CA 91109, United Statese PO Box 344, Lovettsville, VA 20180, United Statesf Desert Research Institute, 2215 Raggio Parkway, Reno, NV 89512, United Statesg University of Campinas, PO Box 6152, 13083-970 Campinas, São Paulo, Brazilh SETI Institute, 189 Bernardo Ave., Mountain View, CA 94043, United States

a r t i c l e i n f o a b s t r a c t

Article history:Received 15 August 2010Revised 8 April 2011Accepted 2 May 2011Available online 14 May 2011

Keywords:Mars, SurfaceGeological processesCrateringInfrared observations

0019-1035/$ - see front matter � 2011 Elsevier Inc. Adoi:10.1016/j.icarus.2011.05.002

⇑ Corresponding author. Present address: Boston USensing, 725 Commonwealth Ave., Rm. 433, Boston, M

E-mail address: bjt@bu.edu (B.J. Thomson).

Gale Crater contains a 5.2 km-high central mound of layered material that is largely sedimentary in originand has been considered as a potential landing site for both the MER (Mars Exploration Rover) and MSL(Mars Science Laboratory) missions. We have analyzed recent data from Mars Reconnaissance Orbiter tohelp unravel the complex geologic history evidenced by these layered deposits and other landforms inthe crater. Results from imaging data from the High Resolution Imaging Science Experiment (HiRISE)and Context Camera (CTX) confirm geomorphic evidence for fluvial activity and may indicate an earlylacustrine phase. Analysis of spectral data from the CRISM (Compact Reconnaissance Imaging Spectrom-eter for Mars) instrument shows clay-bearing units interstratified with sulfate-bearing strata in the lowermember of the layered mound, again indicative of aqueous activity. The formation age of the layeredmound, derived from crater counts and superposition relationships, is �3.6–3.8 Ga and straddles theNoachian–Hesperian time-stratigraphic boundary. Thus Gale provides a unique opportunity to investi-gate global environmental change on Mars during a period of transition from an environment thatfavored phyllosilicate deposition to a later one that was dominated by sulfate formation.

� 2011 Elsevier Inc. All rights reserved.

1. Introduction

Layered sedimentary sequences with repetitious bedding are ofparticular interest in planetary exploration because they constituteone of the key differences between bodies that posses an atmo-sphere or hydrosphere and those that lack them. Specifically, thesetypes of sedimentary deposits require suitable transport mediasuch as wind or liquid water to form. Thick, laterally extensive sed-imentary sequences with finely stratified materials may require anextended period of time for formation, particularly if sediment fluxand accumulation rates were low during deposition. Therefore,such thick sequences have the potential to capture secular, epi-sodic, or cyclical environmental changes that may have occurredin their depositional settings. The overall objective in studyingsuch deposits is to determine (or more broadly, place constraints

ll rights reserved.

niversity Center for RemoteA 02155, United States.

upon) the environmental conditions that prevailed when the sedi-ments were laid down.

Gale Crater hosts a sequence of layered deposits that exceedsseveral kilometers in thickness. Analysis of visible to near-infraredreflectance spectra (0.4–4 lm) from the Compact ReconnaissanceImaging Spectrometer for Mars (CRISM) instrument indicates stra-ta in the lowermost section of the mound exhibit an upward tran-sition from clay- and clay/sulfate-bearing beds to predominantlysulfate-bearing beds (Milliken et al., 2010). Such a transition inmineralogy is broadly consistent with the proposed global environ-mental shift from circum-neutral/alkaline conditions to moreacidic conditions inferred from analysis of OMEGA (Bibring et al.,2006) and CRISM spectral data (Murchie et al., 2009).

In this paper, we analyze the impact crater population andstratigraphic relationships to better constrain the time frame with-in which the layered strata in Gale Crater were deposited. Imagesfrom the High Resolution Imaging Science Experiment (HiRISE) ta-ken at 25 cm/pixel scale and Context Camera (CTX) taken at 6 m/pixel scale onboard the Mars Reconnaissance Orbiter (MRO) are

414 B.J. Thomson et al. / Icarus 214 (2011) 413–432

used to map and characterize the layered deposit within GaleCrater in detail to help unravel the complex geologic history of thisregion. By placing the layered deposits in Gale in a more tightlyconfined temporal context, this analysis will help establishscientific objectives and testable hypotheses that may help guidepotential future landed exploration.

2. Background and regional context

2.1. Physiography

Gale Crater is a large (152 km diameter) impact crater centeredat 5.3�S latitude, 222.3�W longitude in the Aeolis Mensae region(Fig. 1). Within the crater lies a slightly off-centered mound of lay-ered material that measures approximately 45 km by 90 km inareal extent. This deposit has a maximum relief—meaning the ele-vation difference between the highest point on the mound and thelowest portion of the surrounding crater floor—of �5.2 km and anaverage height of 3.8 km. This corresponds to a total mound vol-ume of about 1.7 � 104 km3 of sedimentary material. By compari-son, the average depth of the Grand Canyon on Earth (�1.6 km)is less than half the average exposed stratigraphic thickness ofthe Gale mound (e.g., Webb et al., 2008).

2.2. Relationship to dichotomy boundary

Gale Crater lies on the southern, upland margin of the martianhemispheric dichotomy boundary, which is the topographic dis-continuity that divides the heavily cratered uplands from thesmoother and lower-elevation northern lowlands. Along thedichotomy is a zone of disruption dubbed fretted terrain where in-tact cratered terrain has been partially disrupted into mesas, hum-mocks, and knobs (Sharp, 1973). As pointed out by Irwin et al.(2004), fretted terrain in the equatorial zone near Aeolis Menase(�10�N–10�S, 240�W–210�W) differs from fretted terrain in themidlatitudes north of Arabia Terra (�25�N–50�N, 350�W–260�W). The former lacks indicators of ice-facilitated mass wastingthat are present in the latter, such as lineated valley fill and lobate

Fig. 1. Topographic map of Mars (Robinson projection). Location of Gale Crater is ininterpretation of the references to color in this figure legend, the reader is referred to th

debris aprons. Gale crater straddles the transition from crateredhighland surfaces to disrupted surfaces of Aeolis Mensae. Whereasthe southern rim of the crater is largely intact, the terrain immedi-ately adjacent to the northern rim has been heavily dissected inconjunction with the formation of fretted terrain (Fig. 2). Due tothe downward pre-existing slope of the planetary dichotomyboundary, the northern rim lies several km lower than the south-ern rim (e.g., see Irwin and Watters, 2010 Fig. 6B and D). Thecentral layered mound is also topographically higher than thelow-lying northern rim, giving the appearance from certainperspectives that the mound elevation exceeds the crater rim ele-vation. In fact, there are a few points along the southern edge of thecrater rim that slightly exceed the current maximum mound eleva-tion. Regardless, the mound height and volume suggest that theentire crater was once largely filled with sediment, and it is evenpossible that the entire crater was previously buried and has sincebeen exhumed.

2.3. Other large crater-hosted sedimentary mounds

While Gale Crater is noteworthy in terms of the size and volumeof its interior layered deposit, it is not an isolated example: severalother large impact craters in the vicinity also possess central lay-ered material of a similar nature. Examples include Reuyl Crater(86 km in diameter, 9.8�S, 193.2�W) and Nicholson Crater(103 km in diameter, 0.2�N, 164.6�W). In each of these cases, thecraters host massive central mounds of light-toned, layered sedi-mentary that may have once been part of more extensive deposits,though the original maximum extent and heights of those depositsremains unknown. Given the proximity and similarity of characterof these remnant layered mounds (of which Gale Crater is the larg-est) to the Medusae Fossae Formation (MFF), it is possible that GaleCrater may be a remnant of a formerly more extensive MFF (e.g.,Schultz and Lutz, 1988). Craters act as sedimentary traps, and inaddition to being depositional foci, the crater walls offer a measureof protection against subsequent erosion. Based on detailed geo-logic mapping and the distribution of pedestal craters around thepresent extent of the MFF, Schultz and Lutz (1988) inferred thatthese deposits were formerly more extensive and have undergone

dicated with red star; prior Mars landing sites are labeled with black dots. (Fore web version of this article.)

Fig. 2. MOLA shaded relief map of Gale Crater and surrounding terrain. Inset is perspective view of Gale looking toward the southeast (vertical exaggeration � 4�).

B.J. Thomson et al. / Icarus 214 (2011) 413–432 415

significant erosion and deflation. In addition to physical proximity,many of these layered deposits are also similar in their surficialspectral properties (i.e., they are dust-dominated), low thermalinertia, number and spacing of layers, and presence of yardangs(wind-carved ridges) (e.g., Bradley et al., 2002).

2.4. Previously proposed origins

Numerous interpretations of the layered deposits in Gale Craterand the potentially similar units of the Medusae Fossae Formationhave been advanced. Given the low thermal inertia of these depos-its and thin bedding, many have suggested eolian-related forma-tive processes, including volcanic ash (Scott and Tanaka, 1982;Greeley and Guest, 1987; Bradley et al., 2002; Hynek et al., 2003)or eolian dust plus ice (Schultz and Lutz, 1988; Watters et al.,2007). A lacustrine origin was proposed on the basis of horizontalto subhorizontal layering and terraces, the channel system thatcuts the southern rim of the crater and debouches onto the wes-tern crater floor, and numerous other small, inward-draining chan-nels identified in Viking Orbiter images (Cabrol et al., 1999). Usinghigher-resolution MOC images, a subaqueous origin of this andother layered material was reasserted based on the affinity of thelayered materials for impact craters, their thickness, regular layer-ing, bed orientations, and apparent lack of cross beds (Malin andEdgett, 2000). Conversely, a stated lack of evidence for classiclacustrine features such as terraces, deltas, and fans in Gale was ci-ted as evidence in support of formation of the layered centralmound instead as a hydrothermal spring deposit (Rossi et al.,2008). Although not addressed to Gale Crater specifically, in a sim-ilar vein it has been proposed that many crater-hosted and intra-crater sedimentary deposits such as Meridani may be due togroundwater upwelling and evaporation, resulting in evaporiteprecipitation and cementation of eolian sediment (Andrews-Hannaet al., 2007). Additionally, it has been suggested that there may be agenetic link between the mineralogy of light-toned deposits andaqueous processes. Specifically, accumulations of buried evaporitic

hydrated salts can create thermal anomalies (due to their low ther-mal conductivities) that may have sustained hydrologic activity inthe more recent past (Kargel et al., 2007).

2.5. Previous age determinations

Establishing the formation ages of the layered deposits in GaleCrater and elsewhere has proved challenging. In planet-wide geo-logic mapping efforts using Mariner 9 data, Gale Crater and its inte-rior deposits were mapped as unit Nplc (Noachian cratered plateaumaterial), implying the crater’s maximum age is Noachian (Scottand Carr, 1978). A unit of HNpd (deflation plains material) abutsthe eastern edge of the crater, interpreted as older (Noachian)surfaces exhumed in wind erosion of relatively soft, younger (ca.Hesperian) materials. Later mapping with higher-resolution Vikingimages placed the central layered mound in the Hesperian(grouped with other isolated exposures of smooth plains materialin unit Hpl3) (Greeley and Guest, 1987). Some workers have in-ferred that the mound is much younger, perhaps as young as theLate Amazonian (Scott and Chapman, 1995; Cabrol et al., 1999).Others have contended that on the basis of evidence for wide-spread crater exhumation, Gale and other layered sedimentarystrata were originally deposited in the Noachian Period (Malinand Edgett, 2000). Finally, an analysis of the regional geology ofGale and its contributing watershed suggested that the Gale im-pact occurred near the Noachian/Hesperian boundary after theperiod of widespread crater degradation (based on its relativelypristine morphology) but before the development of adjacent fret-ted terrain (Irwin et al., 2005). The crater interior deposits were in-ferred to have formed soon thereafter but prior to the terminalphase of highland valley network activity.

In global to regional scale geologic maps, some of the conflictingages reported for Gale and its interior deposits may have been dueto the fact that its time-stratigraphic position was largely inferredfrom its general morphology rather than assigned via direct mea-surement of its crater size–frequency distribution (due to the

Fig. 3. Geologic map of Gale central layered mound based on mapping from CTX and HiRISE images. See Appendix A for additional layer descriptions.

416 B.J. Thomson et al. / Icarus 214 (2011) 413–432

relatively small area and lack of large craters on the mound). Cra-ters were measured directly using Viking images of the floor andinterior mound (Cabrol et al., 1999), but both the crater floor mate-rial and layered mound material were combined into a single countarea in that study, thus lowering the overall apparent crater den-sity and inferred age. In this study, we use revised crater countsand newly revealed details about the stratigraphic relationshipsbetween mapped units to place firmer constraints upon the periodof time in which this layered sequence was deposited.

3. Methodology

3.1. Mapping technique

Using all available HiRISE images of Gale Crater as well as amosaic of CTX images acquired through July 2010, we identifiedand delineated geologic units at a scale of 1:50,000. Unit identi-fications were made on the basis of geomorphic expression,surface texture, tone (approximation of albedo), and contact

Fig. 4. (A and B) Portions of HiRISE image PSP_001488_1750 with polygonal ridges interpreted as zones of preferential cementation along fractures. (C) Exhumed impactcrater located at erosional contact between the Upper and Lower Members of the layered mound (subset of HiRISE image PSP_009149_1750). (D) Example surface texture inUpper Member of layered mound (subset of PSP_001620_1750). Insets show reticulate bedforms influenced by layer erosional expression (top inset) and more traditionallinear bedforms in local topographic low (bottom inset).

B.J. Thomson et al. / Icarus 214 (2011) 413–432 417

relationships with other mapped units (e.g., Wilhelms, 1990). Be-cause the type areas of beds in the Gale layered stack lack formalor even provisional names, we have assigned them names basedon physical characteristics (e.g., layered material). This becomespotentially problematic due to the fact that some unit namesmay reflect both primary characteristics (those that occurred dur-ing deposition) and secondary features due to post-emplacementprocesses (e.g., weathering, diagenesis; see further discussion inHansen (2000)). Thus we have limited our discussion of relativeage inferences to those units for which we infer a parallelism be-

tween rock and time units, particularly those in the NW quadrantof the mound. Despite these limitations, the central mound isbounded on all sides by erosional scarps and therefore Gale Cra-ter presents an opportunity to assess the full three-dimensionalgeometry of the beds.

3.2. Geometric measurements

The geometry of selected geologic contacts was measured usingdata extracted from individual Mars Orbiter Laser Altimeter

Fig. 5. (A) HiRISE image PSP_006855_1750 centered over large chasm on the western margin of Gale Crater’s layered mound (image centered at �5.09�N lat, 222.77�W lon).White boxes give locations of B–D. (B–D) Subsets of (A) centered on inverted channel-like features cut into chasm walls. Arrows indicate down gradient direction.

418 B.J. Thomson et al. / Icarus 214 (2011) 413–432

(MOLA) profiles. First, a unit’s traceable extent was mapped usingHiRISE and CTX images. Using GIS software, we identified pointswhere the mapped contact intersect the MOLA profiles. The eleva-tion of these intersection points was estimated by assuming a lin-ear slope between the two adjacent MOLA elevation points oneither side of the intersection point along a given profile (e.g., Be-yer and McEwen, 2005). MOLA observations have a footprint�168 m in diameter and are spaced �300 m along the spacecraftground track (Smith et al., 2001). Uncertainty associated withMOLA elevation measurements has been assessed through an anal-ysis of crossing profiles (Neumann et al., 2001), which yield a RMSerror of 1.8 m. For purposes of fitting a surface to the network ofxyz points, the x and y locations were given as meters in a simplecylindrical coordinate system with (0, 0) given at an arbitrary localreference point. z-Values were also given in meters.

In addition to determining the attitude of mapped contacts, wealso examined characteristics of individual beds within ourmapped units. The high resolution and high signal-to-noise ratioof HiRISE images provide unprecedented detail about the expres-

sion of individual beds and have revealed that several units previ-ously characterized as massive in lower spatial resolution imagesactually consist of finely stratified beds with minimal erosionalcontrast. Stereo pairs of HiRISE images have been analyzed usingphotogrammetric software by the USGS to produce extremely highquality topographic maps with 1 m grid spacing (e.g., Kirk et al.,2008), although labor-intensive manual editing of the resultingDigital Terrain Model (DTM) is necessary to remove processingartifacts. Here, we use topographic profiles extracted from singlelines of stereo pairs of HiRISE images to infer the thickness andapparent dip of individual layers.

The HiRISE instrument uses Time Delay Integration (TDI) tointegrate the exposures of each line of ground observed up to128 times to increase the signal-to-noise ratio (Delamere et al.,2003). A linear array of 10 overlapping CCDs provides a �6 kmswath width at red wavelengths (center wavelength is 694 nm).Each line of a compiled HiRISE image is essentially like a separateimage, and two HiRISE observations of the same ground targetwith different look angles can be used to compute the relative

Fig. 6. (A) Mapped outline of a distinct unit (ume3) of the Upper mound formation. Red circles are intersection points between mapped outline and MOLA ground tracks.(B) Best-fit planar surface to erosional contact between Upper and Lower formations of the layered mound. (For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.)

B.J. Thomson et al. / Icarus 214 (2011) 413–432 419

heights of features in a given line of pixels using a variation of thebasic stereo equation:

h ¼ pðtan h1B þ tan h2BÞ

: ð1Þ

Here, h is the height of the feature of interest, p is the measuredparallax, and h1B and h2B are the respective emission angles of thetwo HiRISE images that constitute the stereo pair (accounting fortheir position in the CCD array). This single-line stereo techniqueis only applicable near the martian equator where the projectionof the MRO spacecraft’s orbital tracks onto the surface are near-parallel (further details are given in Appendix A). Fortunately, GaleCrater is centered within about 5� of the equator. Other obviouslimitations of this technique are that the resulting small-baseline

heights are known only in a relative sense, e.g., the height of a fea-ture such as scarp above a designated point in its surroundings.Also, no compensation has been made for spacecraft jitter or po-tential small misregistrations of the CCD array. Repeated measure-ments of beds using multiple pairs of line profiles suggests that theresulting profile data is accurate to within a few meters (�3–5 m)in a relative sense.

3.3. Measuring crater size–frequency distributions

Determining the impact crater size–frequency distribution of asurface remains the principal means of assessing relative age(specifically crater retention age) in the absence of returned sam-ples. Within the study area, all craters not grouped into obvious

Fig. 7. (A) Single-line stereo profile location on HiRISE image PSP_001488_1750 (2nd image used for stereo is PSP_001752_1750). (B) Topographic profile derived from single-line stereo. Elevations are referenced to arbitrary zero value at point g.

420 B.J. Thomson et al. / Icarus 214 (2011) 413–432

secondary clusters or chains were visually identified and recorded.HiRISE, CTX, and THEMIS (Thermal Emission Imaging System) vis-ible and daytime infrared images were used to analyze the craterpopulation within the study area. We also obtained nested cratercounts for crater counts contained within a single mapped unit.In nested counts, all craters greater than a certain cutoff diameterare recorded using images with a range of resolutions. In small rep-resentative sub-region(s), smaller craters are recorded down to thelimit of confident identification (which varies depending on sur-face roughness, lighting conditions, etc. but is generally �5 pixels(Schultz et al., 1977)). These nested counts from different imagedata sets can be combined on a single plot if the surface areas overwhich the counts were performed are taken into account.

Summary plots of the incremental crater size–frequency distri-bution were made according to established techniques and conven-tions and were overlain on modeled production functions(Hartmann and Neukum, 2001; Hartmann, 2005). The measuredcrater densities of the martian time-stratigraphic boundaries (Tana-ka, 1986) were also included on the summary plots for comparison.

4. Results

4.1. Preliminary geologic mapping results

In the central layered mound, we have identified 22 distinctgeologic units (Fig. 3), not including the crater’s walls and rela-tively flat crater floor. This sequence can be divided into two majorcomponents: a Lower mound (Lm) formation and an Upper mound(Um) formation (consistent with the provisional names assigned inMilliken et al. (2010)). Layered strata in the Lower mound forma-tion are generally horizontal to sub-horizontal (as measured fromsingle-line stereo profiles). The Upper mound formation, in con-trast, consists of more finely layered units that have higher-anglebounding surfaces, at least within certain units. An erosionalunconformity, initially recognized by Malin and Edgett (2000), sep-arates these two formations in the NW quadrant of the mound(Fig. 4C). The interpretation of contact relationships elsewhere issomewhat complicated by the unconsolidated mantling units thatare prevalent throughout the mapped area. Eolian bedforms areespecially abundant on units in the Upper mound formation(Fig. 4D). The more informal and general term ‘‘unit’’ is used over‘‘member’’ to describe subdivisions of these provisional formationsas strict lithostratigraphic definitions cannot be uniformly assignedto all mapped units (e.g. Tanaka et al., 2005).

Abundant geomorphic evidence for aqueous activity is presentin multiple units in the Lower formation of the layered mound.Polygonal ridge networks (Fig. 4A and B) are observed to emerge

from beneath light-toned layers with fractured surface textures.We interpret these ridge networks as evidence for subsurface fluidflow and preferential cementation along fracture zones. Similarevidence for fracture-controlled fluid flow has been observed inlayered deposits in Candor Chasma (Okubo and McEwen, 2007).Numerous channels and inverted channels are also present innumerous locations in the Gale deposits. Inverted channels appearto be limited to the Lower mound formation (Fig. 5). These positiverelief structures are erosional remnants of indurated channel-filldeposits that are often located in larger incised (negative relief)channels. Recent comprehensive mapping efforts in Gale byAnderson and Bell (2010) also identified large numbers of sinuouschannels on the crater floor interpreted as inverted channels.

4.2. Geometry of layers and layer contacts

The attitude of a portion of the erosional contact between theUpper and Lower formations of the layered mound was deter-mined by attempting to minimize the residual offset between amodeled planar surface and the observed xyz points as interpolatedfrom individual MOLA topographic profiles. As depicted in Fig. 6,the contact at this location is best described by a planar surfacewith a strike and dip of N45E, 12.1�NW (i.e., inclined to the NW).

Using the single-line stereo technique, the geometry of individ-ual beds was derived. In Figs. 7 and 8, two examples of single-linestereo profiles are plotted: one from the Lower formation of thecentral mound (Fig. 8) and another from the Upper formation(Fig. 8). The six layers in the profile given in Fig. 7B are between10 and 30 m thick and have an apparent dip of �2�. Individual lay-ers have variable geomorphic expressions—many appear to have acoarse zone of rubble at their base and fine upwards, while othersmay be topped with coarse rubble zones (depending on where theboundaries between layers are drawn). Note these rubble zonesmay just be the result of in place fracturing rather than being rep-resentative of a coarsening-upward sequence. Layers in the Upperformation, in contrast, are between 3 and 7 m thick and haveapparent dips >5� and <10� (Fig. 8). No differences in constituentgrain size (as observed at HiRISE image resolutions) are apparentin the weathering characteristics across bed boundaries, and thebeds appear to be near-uniform in tone and texture.

4.3. Alteration mineralogy in the near-IR: Evidence of phyllosilicatesand sulfates

Visible and near-infrared spectral data from the CRISM instru-ment have revealed the presence of a diverse suite of alterationminerals in the Lower formation of the layered mound where

Fig. 8. (A) Single-line stereo profile location on HiRISE image PSP_009927_1750 (2nd image used for stereo is PSP_008002_1750). (B) Topographic profile derived from single-line stereo. Elevations are referenced to arbitrary zero value at point d0 .

Fig. 9. Comparison of Mars CRISM spectra and laboratory spectra showing the presence of hydrated sulfates and clay minerals in Gale Crater. All CRISM spectral ratios are7 � 7 pixel averages. (A) CRISM spectral ratios for an Fe-smectite (nontronite) deposit near the base of the Gale Crater mound compared to a lab reflectance spectrum ofnontronite. (B) CRISM RGB composite (RGB = 0.749, 0.501, and 0.449 lm, respectively) image FRT0000B6F1 with locations of spectra presented in (A) marked with coloredcircles. The red CRISM spectrum in (A) represents a ratio of the region marked by the green circle divided by the region marked by the red circle. The black CRISM spectrum in(A) represents a ratio of the green circled region divided the black circled region. The choice of the denominator affects the spectral slope for wavelengths less than 1.5 lm butdoes not affect the presence or position of absorption features at longer wavelengths. (C) CRISM spectral ratios for sulfate deposits in Gale Crater. Some regions are consistentwith monohydrated sulfate (kieserite; black spectra) whereas others are consistent with polyhydrated sulfate (likely MgSO4�nH2O; red spectra). (D) CRISM RGB composite(RGB = 0.749, 0.501, and 0.449 lm, respectively) image FRT000095EE with locations of spectra presented in (C) marked with colored circles. The red CRISM spectrum in (C) isa ratio of the region marked by the red circle divided by the region marked by the white circle. Black CRISM spectrum in (C) corresponds to the black circled region divided bythe white circled region.’’ (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

B.J. Thomson et al. / Icarus 214 (2011) 413–432 421

Fig. 10. (A) CTX mosaic of the NW portion of Gale Crater’s central layered mound. (B) CRISM spectral composite map overlain on CRX mosaic. Colors indicate band strengthand spatial coherence of specific absorptions; mineral identifications given in white. (For interpretation of the references to color in this figure legend, the reader is referred tothe web version of this article.)

422 B.J. Thomson et al. / Icarus 214 (2011) 413–432

the moderate dust cover permits identification of the surfacecomposition (Milliken et al., 2010). Following standardized atmo-spheric correction procedures (Murchie et al., 2009), spectra ofindividual geologic units were divided by the spectrum of a near-by spectrally neutral (dusty) region. These spectral ratios werethen compared with laboratory mineral spectra (Fig. 9). Movingfrom the lowermost stratigraphic units upwards in Fig. 10A and

B, there are sulfate-bearing rocks possibly mixed with smectite,a thin but distinct smectite-bearing unit (specifically the iron-richphyllosilicate nontronite), a layer with a mixture of smectite andolivine in which the latter is associated with unconsolidated eo-lian material, a unit with a mixture of sulfate and smectite, sul-fate-bearing beds, and finally the absence of hydrated mineralsin the layers of the upper member of the central mound (i.e.,

Fig. 11. Mosaic of THEMIS nighttime temperatures overlain on THEMIS daytimeband 9 images. Nighttime THEMIS images are dominated by thermal inertia and aretherefore a proxy for particle size and degree of induration of the surface.

Fig. 12. (A) CTX mosaic of Gale layered mound showing impact craters identified with fill(C) CTX mosaic of subset of layered mound showing unit with high concentration offrequency distribution for the exhumed crater unit on layered mound. (E) HiRISE, CTX, awith filled circles. (F) Incremental crater size–frequency distribution for the northern cridentified with filled circles; unfilled circles indicate craters interpreted to have predatedfor Gale ejecta.

B.J. Thomson et al. / Icarus 214 (2011) 413–432 423

the beds are spectrally similar to dust). Thus, the mineralogy tran-sitions from mixed sulfate-clay assemblages to a dominantly sul-fate alteration assemblage upsection. The sulfates do not exhibitspectral characteristics common of Ca and Fe-sulfates and are in-stead most consistent with Mg-bearing sulfates (e.g., kieserite,hexahydrite, etc.).

4.4. Mid-infrared results and thermal inertia

The surficial geology of Gale Crater has been mapped in detailby Pelkey et al. (2004) using 100 m/pixel daytime and nighttimethermal data from the Mars Odyssey THEMIS (Thermal EmissionImaging System) instrument supplemented with THEMIS visibleand MOC (Mars Orbital Camera) images. Seven surface types wereidentified and mapped on the basis of their thermal characteristics,including 3 units of floor materials, central mound material, sandsheets, crater wall material, and material in the surrounding undi-vided terrain (dubbed plateau material). Material on the northernfloor of Gale Crater, including the proposed MSL landing ellipse,were found to have high temperatures and corresponding highthermal inertias. This thermal signature was attributed to bedrockor extensively indurated material (Pelkey and Jakosky, 2002).

Fig. 11 is a plot of THEMIS nighttime temperatures. The dataindicate that temperatures decrease upslope on the central mound.Regions with the lowest temperatures are found at the highest ele-vations, and are consistent with dust cover in excess of a few cmthick (i.e., exceeding a diurnal thermal skin depth in thickness).The dust cover appears to taper downslope to reveal more indu-rated units, which is consistent with stronger near-infraredabsorption features being evident in the lower, less dust-coveredunits in the Lower formation compared to the units of the Upperformation.

ed circles. (B) Incremental crater size–frequency distribution for the layered mound.exhumed craters. Crater identified with filled circles. (D) Incremental crater size–nd THEMIS daytime IR images of Gale floor units showing impact craters identifiedater floor. (G) THEMIS daytime IR mosaic of Gale Crater ejecta. Superposed cratersthe Gale crater-forming impact. (H) Incremental crater size–frequency distribution

Fig. 12 (continued)

424 B.J. Thomson et al. / Icarus 214 (2011) 413–432

4.5. Crater size–frequency distribution results

4.5.1. Direct age measurement of the layered moundThe most straightforward method to determine the age of the

layered units in Gale is to determine the crater retention age ofthe mound itself. This is problematic given the apparent materialproperties of these units. Most of the exposed surfaces do not ap-pear to retain craters, and overall measured crater distribution(Fig. 12A and B) yields a young age near the Late Hesperian/EarlyAmazonian boundary. However, much like the layered depositsin the north and south polar regions (e.g., Herkenhoff and Plaut,

2000), this young age likely represents an exposure age or resurfac-ing age that reflects recent erosion, and it is not necessarily indic-ative of the depositional age of the material. Unit lm3 of the Lowerformation contains a significant number of craters that appear tohave been exhumed from underneath subsequently deposited lay-ers that are now being stripped back (see example given in Fig. 13).

4.5.2. Minimum age constraint: Superposed floor units and valleynetwork deposits

A more accurate constraint upon the minimum formation age ofthe layered units in Gale can be obtained by using basic superposi-

Fig. 13. Exhumed impact structures in unit lm3 of Lower Member of layered mound (portion of CTX image P01_001488_1751_XI_04S222W_061120). Inset shows irregularsurface texture of partially eroded crater fill (portion of HiRISE image PSP_006855_1750).

B.J. Thomson et al. / Icarus 214 (2011) 413–432 425

tion relationships. As evidenced by the geologic cross-section givenin Fig. 3B, it is clear that the units on the floor of Gale Crater overlapand embay the lower units of the layered deposits. Thus, thesefloor units are topographically lower but stratigraphically higher,and therefore younger, than the Lower formation of the Galemound. By determining the age of this floor unit we can place afirm minimum age constraint upon the layered mound. The mea-sured crater size–frequency distribution of the floor unit is givenin Fig. 12C and D, indicating an age of Early Hesperian. Therefore,the lower portions of the mound can be no younger than the EarlyHesperian. Evidence supporting this result was noted by Irwinet al. (2005), who observed that the inlet channel breaching thesouthern rim diverges into two small terminal deposits on eitherside of the eastern lobe of the central mound. Such a configurationrequires that the lower portions of the mound have been in placeprior to the terminal phase of activity of this channel.

4.5.3. Maximum age constraint: Gale ejecta crater distributionIn order to bracket the age of the deposits in Gale, it is necessary

to determine an upper age boundary that marks the maximum po-tential age of the deposits. For this study, we used the age of GaleCrater itself as a time-stratigraphic marker. Fig. 12F is a plot of thecrater size–frequency distribution of craters superposed on Gale’sejecta. The outer terminus of Gale ejecta is gradational in natureand its exact extent is difficult to pinpoint, so we have measuredthe crater distribution within the estimated continuous ejectablanket using the relation Rce = (2.348 ± 0.5)R1.006 (Moore et al.,1974). Here, Rce is the radius of continuous ejecta and R is theapparent crater radius. Although this relationship was empiricallydetermined for lunar craters, it has been used to provide first-orderestimates of martian ejecta (e.g., Grant et al., 2008). An outline ofthe area counted is given in Fig. 12E and the count data are givenin Fig. 12F. Data from the two largest crater size bins suggest thatthe Gale impact occurred in the Late Noachian, although the den-sity of smaller-sized craters (2–16 km in diameter) are more con-sistent with an age that straddles the Late Noachian/EarlyHesperian boundary. Error bars in a few diameter bins (e.g., cen-tered at 9.51 and 13.45 km diameter) subtend epoch boundaries,adding some degree of uncertainty to the age assignments givenabove. Two potential factors may account for the apparent discrep-ancy between the larger and smaller crater size bins. First, thismight be due to differential erosion such that smaller superposedcraters were degraded more efficiently than larger ones (e.g.,Chapman and Jones, 1977; Craddock and Howard, 2002). Second,although we have endeavored to exclude craters that predate the

Gale impact from the count (mapped with unfilled circles inFig. 12E), it is possible that a few older craters were included,resulting in a slightly older apparent age in the largest size bins.

5. Discussion and interpretation

5.1. Summary of the nature of the layered material

Images and data from observations of Gale Crater by the MROspacecraft have verified and expanded previously recognized char-acteristics and also revealed a wealth of previously unknown de-tails. The Upper and Lower formations of the mound have distinctmorphologic and mineralogic characteristics, and they may havebeen formed by separate processes (e.g., Milliken et al., 2010). Theerosional surface between these two members is non-horizontal,but this does not necessarily constrain its origin. Such a surfacecould be the result of a landslide (perhaps an exposed footwall fol-lowing normal faulting) or alternatively could be the result of eolianerosion. Given the lack of debris directly attributable to mass wast-ing processes downslope of this contact, this suggests that eolianerosion may control this and other surfaces of omission.

CRISM data have revealed an intriguing suite of alteration min-erals in the Lower formation of the mound, including an upwardtransition from a series of phyllosilicate-bearing to a group of sul-fate-bearing layers (Milliken et al., 2010). If formed in situ, thesealteration minerals hint at the former presence of water in Gale,and the morphology of numerous inverted channels are also evi-dence a significant role for local aqueous processes.

5.2. Formation time constraints

Key considerations of any geologic site are determining whenthe processes inferred occurred and their duration. In the case ofGale Crater, salient questions are when was the layered materialdeposited, and when were the channel systems active? We haveattempted to answer these questions by measuring the crater pop-ulation in combination with superposition relationships to estab-lish an upper and lower time boundary for the formation of thelayered material. Direct measurement of the crater size–frequencydistribution on the entire layered mound yields a Late Hesperianage (Fig. 12A and B), but the widespread occurrence of exhumedcraters indicates that this age only represents an exposure or resur-facing age. Exhumed craters represent only a fraction of the totalcrater population given that an indeterminate number of impactstructures must have been lost with removal of the overburden.

Fig. 14. Martian chronology after Tanaka (1986) with two different model agesfrom Hartmann and Neukum (2001). Arrows represent maximum and minimumtime constraints on deposition of sedimentary layers in Gale Crater; inferredgeologic history given in bulleted form.

426 B.J. Thomson et al. / Icarus 214 (2011) 413–432

A more accurate minimum formation age in the Early HesperianPeriod is obtained by an analysis of the crater size–frequency dis-tribution on the floor units that overlap the lowermost units of thelayered mound (Fig. 12C and D). It is also true, however, that theerosional unconformity that separates the Upper and Lower forma-tions represents an unknown hiatus in the depositional history re-corded in Gale. There is also a distinct change in materialproperties above and below this boundary. Below this unconfor-mity, many units contain evidence of exhumed craters (e.g., unitlm3 in Fig. 3), while above this boundary the material appears tobe more friable and retains craters poorly. The age of the units inthe Upper formation are therefore unclear based on both strati-graphic relationships or their measured crater distribution. Thisuncertainty parallels the disparate ages estimated for the MedusaeFossae Formation, which has been alternatively placed in the Ama-zonian due to on the basis of its low crater retention age (Greeleyand Guest, 1987; Tanaka, 2000; Bradley et al., 2002) or the Hespe-rian based on an analysis of the number density of surroundinglarge pedestal craters (Schultz and Lutz, 1988). Units in the Upperformation, if connected to the MFF, are plausibly consistent witheither of these proposed chronologies.

An additional constraint can be obtained from the age of the lastdated activity in valley networks around Gale. Studies of valleynetworks using a buffered crater count technique (Fassett andHead, 2008) indicate that no valley networks in the cratered south-ern highlands are younger than the Late Hesperian (younger valleynetworks exist, but they are exclusively confined to volcanic ter-rains). Numerous small valley networks debouch onto the craterfloor from all sides, and a sizable percentage of the floor materialappears to have been fluvially transported. It is no coincidence,then, that the age of the floor units matches the age of last activityof the surrounding valley networks. The small inverted channel onthe mound itself may also be a valley network-like feature. Becausethis feature superposes all of the units in the Lower formation ofthe mound, if its age is comparable to the rest of the more tradi-tional valley networks, it would place firmer constraints uponthe relative age of the Lower formation.

A maximum formation age for the layered deposits near theNoachian/Hesperian boundary (Fig. 12E and F) is provided by theage of Gale itself as dated by superposed craters on its ejecta. Thisage is consistent with the general morphology of Gale, which is lessdegraded than many other craters of comparable size in theNoachian highland terrain. Using computer models of terraindegradation by a variety of processes (Forsberg-Taylor et al.,2004), it has been shown that craters follow a predictable sequenceof morphologies as erosion proceeds. Many of the craters inNoachian terrain have flat, infilled floors relative to their diametersand exhibit shallowed wall slopes. Gale Crater, despite containing alarge mound of layered material, still retains some elements ofmore pristine wall morphology (excluding the degraded terrainadjacent to the northern crater rim).

We are therefore able to constrain these layered deposits ashaving formed sometime between near the end of the LateNoachian Period and Early Hesperian Period (Fig. 14). This generaltime frame still represents a possible �0.2–0.3 Ga window withinwhich these deposits were lain down. But even these looseconstraints pin down the Gale deposits to an interesting and littleexplored time-stratigraphic boundary in martian history. Resultsfrom the OMEGA instrument suggest a potentially global-scaletransition from environmental conditions that favored phyllosili-cate formation to conditions that favored sulfate formation(Bibring et al., 2006), potentially due to the extensive outgassingof volcanogenic sulfur coupled with a rapid, global drop in atmo-spheric pressure and loss of water. Alternatively, co-deposition ofboth phyllosilicates and sulfates are known from terrestrialanalogs, for example where diverse mineralogies occur in close

proximity in evaporation-dominated, ephemerally wet playa sys-tems (e.g., Baldridge et al., 2009). Both morphologic and mineral-ogic evidence indicates that Gale Crater captured a stratigraphicrecord of this apparent transition (Milliken et al., 2010), makingthis locale one of the few recognized sites that would allowin situ exploration of this transition at a scale accessible to a rover.

5.3. Origin(s) of layered materials

Of the myriad proposed origins for the layered deposits, a fewcan be set aside or deemed unlikely. It is clear, for example, thatthe Gale mound consists of a stratified sequence of sedimentarylayers. Additionally, as evidenced by the thermophysical data(Fig. 11), the lack of a lag deposit despite significant evidence foreolian deflation necessitates that the layered material consist offine-grained sediments with little to no coarse fraction. Whilestratified sedimentary deposits can form in high-energy environ-ments such as those near impact craters or explosive volcanicvents, a thick accumulation of a finely layered, repetitive sequenceof sedimentary strata generally requires a lower-energy deposi-tional environments such as an eolian sand sheet or a lacustrine/marine basin (turbidite deposits are one notable exception to thisgeneralization, however). Therefore we consider it unlikely thatvolcanoclastic processes alone are responsible for the observedcharacteristics of the deposits, particularly the fine regular layeringrevealed in HiRISE images. The diverse alteration mineralogy of thelayered stack is also difficult to explain in a volcanic scenario with-out evoking radically different fluid chemistries affecting only cer-tain portions of the mound.

An intriguing possibility is that Gale may have once hosted alacustrine phase of activity. Evidence of aqueous processes areabundant, including: widespread exposure of alteration minerals,evidence of subsurface fluid flow in the form of polygonal ridgenetworks, and evidence for fluvial activity in the form of both po-sitive and negative relief channels. But it seems unlikely that lacus-trine evolution alone is solely responsible for the entirety of thelayered mound. First, many of the benches and terraces inferredto be wave-cut platforms in lower resolution data (Cabrol et al.,

B.J. Thomson et al. / Icarus 214 (2011) 413–432 427

1999) turn out to be stratigraphic contacts between distinct layers,some of which are clearly non-horizontal (e.g., Fig. 6). Second, gi-ven that the height of the mound exceeds the rim height exceptin all but a few small locations, Gale would have to have been filledto a level above the rim, which also necessitates complete inunda-tion of the surrounding terrain. Indeed, Gale’s location along thedichotomy boundary means that the entire northern plains wouldhave to be filled with water were Gale to be filled (e.g., Parker et al.,1989; Head et al., 1999). On the other hand, if only the lowermound units were completely submerged, the implied inundationextent of the surrounding terrain would be much less.

A variant of a lacustrine formation scenario is that the watermay have been sourced by groundwater rather than overland flow.Motivated by the interpretation of Meridiani as a former playa de-posit containing eolian-reworked evaporites (Grotzinger et al.,2005), a geophysical model was proposed in which an upwellinggroundwater table intersects the surface, evaporates, and cementseolian sediments into place (Andrews-Hanna et al., 2007, 2010).Requiring an extended time period with surface temperaturesabove freezing, the proposed model envisions repeated wettingand evaporative episodes that build up a substantial thickness ofsedimentary material in localities with favorable hydrologic condi-tions (such as deep craters). A potential obstacle to this hypothesisis that the limited degree of weathering seen at Meridiani is moreconsistent with closed system alteration (Niles and Michalski,2009) rather than the open system origin required by an intercon-nected groundwater system. But as the full mineralogic diversity atGale is only beginning to be unraveled (e.g., Milliken et al., 2010), itremains to be seen if additional geochemical data will support ordisfavor an open system model.

In their analysis of sedimentary deposits on Mars using MOCimages, Malin and Edgett (2000) concluded that the sedimentswere most likely deposited in a quiescent environment; specifi-cally, they favored subaqueous, lacustrine depositional processesover subaerial, eolian processes. Part of their reasoning was thatthe inferred age of the deposits implies that the formative pro-cesses are no longer operative today, at least with the same vigor.While it is true that these deposits do not seem to be forming con-temporaneously in the equatorial zone, one need not look too farafield to find a modern day martian process that produces depositspotentially analogous to those lain down in the past. Both polar re-gions contain sequences of stratified deposits that consist of frozenvolatiles mixed with a small percentage of dust (e.g., up to 10% inthe south polar layered deposits (Plaut et al., 2007). These depositsunconformably overlie the basement terrain, contain layers trace-able for tens to hundreds of kilometers in horizontal extent, andshow an affinity for craters (e.g., Byrne and Ivanov, 2004; Fishb-augh and Hvidberg, 2006; Milkovich and Plaut, 2008). The locationof Gale Crater and other layered deposits in non-polar regions thatstraddle the equator obviously require further explanation. Dueprimarily to the lack of a stabilizing satellite like the Earth’s Moonand the larger influence of nearby Jupiter, the obliquity of Mars hasbeen shown to exhibit chaotic behavior and may frequently reachvalues as extreme as 60� or greater (Laskar et al., 2004). Duringperiods of high obliquity, the increase in insolation at polar regionsmay initiate a planet-wide redistribution of polar surface and sub-surface volatiles to lower latitudes. It is possible that Gale is a rem-nant of one of these periods of high obliquity. A related possibilityis that the rotation axis of Mars may have migrated to its presentposition (i.e., through true polar wander) (Murray and Malin,1973; Schultz and Lutz, 1988). Numerous studies of Mars’ remnantmagnetic field have inferred past polar wander (e.g., Sprenke andBaker, 2000; Frawley and Taylor, 2004; Boutin and Arkani-Hamed,2006), and the long-wavelength topography of northern plains-encircling geologic contacts have been calculated to better fit anequipotential surface if �30–60� of true polar wander has occurred

(Perron et al., 2007). However, the occurrence of polar wander hasbeen challenged on the basis of expected tectonic features (e.g.,Grimm and Solomon, 1986) and disputed geologic ages of sedi-mentary units (Tanaka, 2000). Despite these potential objectionsto polar wander, either high obliquity or polar wander may yet ex-plain the equatorial concentrations of sedimentary rock and theunexpected concentrations of near-surface hydrogen in non-polarlocations; in this case the mineral signatures would likely repre-sent diagenetic processes (Niles and Michalski, 2009).

One remaining possibility is that the entire stack of sedimentscould be the result of dry eolian deposition. If this were the case,then the diverse suite of alteration minerals detected would haveto be formed elsewhere, transported, and deposited within Gale.It is known, for example, that soil deposits in the extremely aridAtacama Desert on Earth contain salts accumulated from atmo-spheric influx (Ewing et al., 2006). Yet given the mineralogicaldiversity at Gale, a single source region seems unlikely. It wouldappear that multiple source regions (or an evolving source) are re-quired to supply and sequester these altered minerals in distinctlayered units. The potential source region(s) remain unidentified.Furthermore, the geomorphic evidence of fluid flow on and withinthe mound (channels, inverted channels, and filled fractures)seems at odds with a completely anhydrous depositional environ-ment. Lithification of individual sedimentary layers is also difficultto accomplish in a completely anhydrous environment.

In sum, we are left with two overarching alternative explana-tions: an aqueous chemical and clastic deposit (consistent withmoderately deep inundation), or a stratified dusty or high obliquity(polar-like) deposit; or some combination of the two. For example,the mound may have a polygenetic origin in which the Lower for-mation of the mound formed by one set of processes that includeda lacustrine phase whereas the Upper formation was dominated byeolian airfall. Regardless of the exact formation scenario, however,the mound captures an important mineralogic and potentiallyclimate-related transition in martian history near the Noachian–Hesperian boundary.

5.4. Implications for future landed exploration

The layered sedimentary material in Gale Crater presents anopportunity to explore one of the largest and best-exposed strati-graphic sections accessible by the MSL landing system and pay-load. The flat, northern crater floor presents a relatively safelanding ellipse on a terrain similar to that landed on by MER Spiritin Gusev Crater. Within the nominal landing ellipse is an alluvialfan from a small valley network incised into the crater’s northernrim. This fan presents the opportunity to assess some basic pa-leo-hydrogeologic questions about valley networks, including thesediment load and constraining the water-to-rock ratio of valleynetworks, which remain unconstrained. At the southern edge ofthe ellipse is a traversable route with a moderate grade up throughthe layered stack of sediments in the central mound (e.g., Bridges,2001, 2006; Bell et al., 2006; Thomson et al., 2007; Anderson andBell, 2010). These deposits appear to have been lain down acrossthe Noachian–Hesperian boundary, and as such capture anenvironmental record of the conditions that initially favored phyl-losilicate formation and deposition and later shifted to a sulfate-dominated system. If the phyllosilicates were formed by surfaceprocesses, this transition may reflect a loss of an initially denserand more clement early martian atmosphere.

Another, more philosophical issue is whether it is better to directfuture landed exploration to a site that represents typical Mars (orperhaps a site that typifies a process) versus directing explorationto a unique site that may be atypical of Mars. But because recentlylanded exploration by the MER rovers have revealed an increasingcomplex picture of Mars’ evolution, it has become more difficult

Fig. A.1. Measurement of layer height as defined by difference in elevation betweenpoints A and B using single-line stereo technique.

428 B.J. Thomson et al. / Icarus 214 (2011) 413–432

to state definitively what ‘‘typical’’ really means. Furthermore, themost unambiguous evidence of ancient life on the Earth often re-quires somewhat anomalous preservation conditions, for example,‘‘Konservat-Lagerstätten,’’ rare deposits known for the exceptionalpreservation of fossilized organisms such as the Jurassic SolnhofenLimestone (e.g., Barthel et al., 1990), Cambrian Burgess Shale(e.g., Gould, 1989; Briggs and Fortey, 2005), or earlier life tracesin Archean Warrawoona Group (e.g., Hofmann et al., 1999; Brownet al., 2004). Thus, Gale Crater is perhaps a type example — one ofa class of craters containing interior layered deposits — that bearwitness to an extended sedimentary record that will likely be vitalto understanding the history of past habitability on Mars.

6. Conclusions

From crater count data and superposition relationships, units inthe Lower formation of Gale Crater layered materials can be reli-ably constrained to have been deposited near the Early Hespe-rian/Late Noachian boundary. Spectral analyses of layers in thelower mound indicate an upward transition from phyllosilicate-bearing to sulfate-bearing layers, albeit with some evidence ofinterfingering (Milliken et al., 2010). Formative scenarios consis-

Table A.1Comparison of elevation values derived from single-line stereo technique and elevation valand PSP_001752_1750.

Point ID 1488 nomap x-pixel 1488 nomap y-pixel Lon (�E) Lat (�N)

B0 842 15,444 137.344260 �4.868115A 739 15,444 137.344726 �4.868061B 613 15,444 137.345297 �4.867996C 542 15,444 137.345616 �4.867959D 507 15,444 137.345774 �4.867940E 460 15,444 137.345988 �4.867917F 428 15,444 137.346132 �4.867900G 380 15,444 137.346350 �4.867875

a Elevation values normalized to elevation of point A obtained using single-line stere

tent with the observed mound properties include lacustrine(sourced from either groundwater or overland flow) or eoliandeposition in a paleo-polar or high obliquity environment. It is alsopossible that units in the Upper and Lower mound formations donot share the same mode of origin. Nevertheless, an important rolefor aqueous processes is demonstrated by positive and negative-relief channels and valley network-like deposits, evidence for fluidmovement and cementation along fracture zones, and the simplepresence of mineralogic signatures of alteration. The lithificationof layers may also point to a role for water, although this criterionalone is not definitive.

An erosional surface or surface of omission separates the upperand lower portions of the mound, and this inclined interface repre-sents a fundamentally unknown length of time. It is possible that itis the result of a brief depositional interlude, or alternatively it mayrepresent a significant gulf of time. If the Upper formation of theGale layered deposits is genetically related to the Medusae FossaeFormation (MFF), then Gale presents an opportunity to explore aunique martian puzzle. Specifically, what are the MFF layers com-posed of, and how did they form? Are they eolian dust deposits,volcanic ash deposits, paleo-polar deposits, or something else? Re-cent radar sounding data indicates that they have an anomalouslylow density and may even be ice-rich (Watters et al., 2007). Irre-spective of any potential connection to the MFF, spectral evidencefor phyllosilicates and geomorphic evidence for fluvial activity andsubsurface fluid flow in the Lower formation of the mound suggesta prior history that involves water in Gale Crater. In situ investiga-tion of the nature and origin of these alteration minerals and hostsediments will allow us to determine the role of aqueous activityand how it fits into the larger evolution of layered deposits of Mars.

Acknowledgments

This manuscript was improved by thoughtful reviews fromRossman Irwin and an anonymous reviewer. This research wassupported in part by a NASA Interdisciplinary Exploration Sciencegrant to Simon J. Hook.

Appendix A. Single-line stereo method

In the text, we presented a method for rapidly measuringterrain heights (specifically the thickness of layers exposed instair-step fashion) using manual measurements on a single line ex-tracted from each image of a stereo pair of HiRISE images. Here weexplain this methodology in greater detail and compare the resultswith topographic points extracted from a DTM created by the USGS(Kirk et al., 2009) using advanced stereo-matching algorithmscontained in SOCET SET (�BAE Systems).

The fundamental stereoscopic parallax equation is well knownand is used to derive terrain heights from aerial photography(e.g., Jensen, 2000).

ues extracted from DTM. HiRISE images used in both methods were PSP_001488_1750

Single-line Elev (m) DTM Elev (m) Norm. DTM Eleva (m) Elev diff (m)

51.3 �3342.92 48.0 3.358.3 �3332.65 58.3 0.050.0 �3349.79 41.2 8.831.4 �3364.23 26.7 4.720.5 �3366.21 24.7 �4.216.0 �3372.64 18.3 �2.3

9.4 �3375.56 15.4 �6.00.0 �3383.69 7.3 �7.3

o method.

Fig. A.2. Elevation values obtained using single-line stereo method (solid red line; given reversed in Fig. 7) compared with those extracted from DTM (dashed green line). Theprofiles are given with no vertical exaggeration. Elevation values were normalized at the highest point in both datasets. RMS error between the two methods is 5.3 m; ratioand difference values between the two profiles are plotted with blue and tan lines, respectively. (For interpretation of the references to color in this figure legend, the reader isreferred to the web version of this article.)

B.J. Thomson et al. / Icarus 214 (2011) 413–432 429

h ¼ H � dPPb þ dP

ðA:1Þ

In Eq. (A.1), h is the height of the object being measured, H is thealtitude of the sensor, Pb is the absolute stereoscopic parallax be-tween the two stereo images, and dP is the differential parallax be-tween two reference points of interest. The single-line stereotechnique is a variation on this formulation (Eqs. (A.2) and (1) inmain body of the text):

h ¼ pðtan h1B þ tan h2BÞ

: ðA:2Þ

Here, h is the height of the feature of interest, p is the measuredparallax, and h1B and h2B are the respective viewing angles of thetwo HiRISE images that constitute the stereo pair. The viewing an-gle of an image is the angle between a surface normal vector and avector from the sensor to the intercept point to (i.e., it is 0� in a na-dir viewing image). Use of this equation is predicted on a numberof assumptions, including that the images used in the stereo pairare close enough to the equator such the projection of the MROspacecraft’s orbital tracks onto the surface are essentially parallel.Fig. A.1 gives a schematic cross-sectional view of the observationgeometry. From basic trigonometry,

tan h1B ¼DX2

h; ðA:3Þ

and

tan h2B ¼DX1

h; ðA:4Þ

where Dx1 and Dx2 are the distances between the top and bottom ofpoint B as projected into the base level plane (Fig. A.1). The differ-ential parallax, dP, is the difference between line segments A1B1

and A2B2 (Eq. (A.5)).

p ¼ A1B1 � A2B2 ¼ DX1 þ DX2 ¼ h ðtan h1B þ tan h2BÞ ðA:5Þ

Re-arranging (A.5) yields Eq. (A.2). Elevation values obtained withthe single-line stereo technique compared to those obtained di-rectly from a full stereo DTM product (produced from the same HiR-ISE images PSP_001488_1750 and PSP_001752_1750) are given inTable A.1 and plotted in Fig. A.2. The RMS error of the normalizedelevation values is 5.3 m. This result indicates that to a first-order,the single-line technique yields reasonable results.

Appendix B. Preliminary geologic unit descriptions

Unit name

Abbrev. Description

Lower mound 1unit

lm1

Occurs on northern edge ofcentral mound. Forms light-toned layered knobs

Lower mound 2unit

lm2

Occurs below lm3 and abovelm1; intermediate in tone.Layers exposed in placessuggesting conformal beddingwith surrounding units, butalso some evidence of drapedmaterial. CRISM data indicatesthe presence of phyllosilicates

Lower mound 3unit

lm3

Occurs immediately belowcontact with Upper moundunits in NW quadrant ofmound. Exposures in west arerelatively flat-lying andpossess numerous craters,some of which appear to havebeen exhumed. CRISM datasuggests weak sulfatesignature in upper portion,possibly interbedded with

(continued on next page)

430 B.J. Thomson et al. / Icarus 214 (2011) 413–432

Preliminary geologic unit descriptions (continued)

Unit name

Abbrev. Description

sulfates in lower portion

Lower mound

subdued unit 3

lm3s Occurs along margin of

western mound lobe. Containssteep-sided canyon exposurecutting through sequence ofLower mound units

Lower moundembayed unit

lmeb

Occurs in NE margin of moundat transition from mound tosurrounding plains. Formsridges of intermediate tolight-toned material that areembayed by dark-tonedmaterial of floor units

Lower moundfractured unit

lmf

Characterized by steep-sidedmargins exposing layers onboth upslope and downslopemargins. Interpretation:Erosional fenster exposingunderlying layered material

Lower moundknobby unit

lmk

Occurs in northern portion ofmound. Forms irregularblocky, fractured terrain withlocally reworked materialdistributed between moreresistant knobs. Many knobshave streamlined shapessuggestive of eolian erosivecontrol

Lower moundsubdued knobbyunit

lmks

Occurs in northern portion ofmound. Similar to unit lmk-forms irregular blocky,fractured terrain with locallyreworked material distributedbetween more resistantknobs. Surface appearsmantled with a materialsimilar in tone to underlyingmaterial, muting thetopographic relief of the knobs

Lower moundeastern layeredunit

lml

Occurs on eastern lobe ofmound. Forms broad, planarsurface with variable texture.Some portions smooth andsubdued, some portionsetched, some eroded;gradational internal contacts

Lower moundmantled unit

lmm

Occurs on eastern margin ofmound. Forms chaoticdistribution of ridges andknobs with interveningsmooth fill. Northern portionsof unit have more mutedtopography, appears to haveintermediate-toned mantlepartially overlain with dark-toned sand

Lower moundmarginal unit 1

lmmr1

Occurs along northern andwestern edge of moundforming contact betweenlayered units and crater floor.Crater floor material clearly

Preliminary geologic unit descriptions (continued)

Unit name

Abbrev. Description

onlaps and partially embayunit margin. Numerous dark-toned tunes also overlie muchof the unit

Lower moundmarginal unit 2

lmmr2

Occurs in thin band innorthern margin of mound.Occurs as streamlined ridgesof resistant material,presumably yardangs. Arepartially embayed by materialof crater floor

Lower moundmarginal unit 3

lmmr3

Occurs along southern marginof mound. Formsintermediate-toned, subduedhummocky material withoutevident layered, possiblydominated by local mass-wasting and reworking.

Upper moundchaotic unit

umc

Occurs adjacent to unit ump,forms smaller (1–2 km across)ridges and knobs sheddingdebris into local valleys

Upper moundetched units 1–4

ume1–4

Forms tightly packed array ofstreamlined erosional ridgesand valleys. Valleys are partlyfilled debris that ispresumably locally sourced,but relief remains sharpindicating minimal mantling.Interpreted as erosionalyardang-like features

Upper moundlayered units 1–4

uml1–4

Forms broad, bench-likeoutcrops of light-tonedlayered material. The edges ofresistant layers are commonlyeroded into irregular, saw-tooth like patterns

Upper moundlayered etchedunit

umle

Occurs in western portion ofmound. Forms steep-sidedlinear ridges interpreted aserosional yardangs, orientedpredominantly in a N–Sdirection. Ridges areintermediate to light-toned.Material in between ridgesappears darker in tone, locallysimilar in character to unitlm3

Upper moundsubdued layeredunits 1–3

umls1–3

Forms varitextured surfacewith exposures of layering.Little to no high-angleexposures of erosional knobsare observed as evidenced inother Upper mound layers.Subdued nature of topographyimplies substantial thicknessof light-toned mantling unitthat is presumably locallyderived

Upper moundmantled units

umm1–2

Forms light-toned, low-reliefpattern of streamlined ridges

B.J. Thomson et al. / Icarus 214 (2011) 413–432 431

Preliminary geologic unit descriptions (continued)

Unit name

Abbrev. Description

1–2

with muted topographicexpression, consistent withmantle material of sufficientthickness to obscurecharacteristics of underlyingsurface. No layering is evident

Upper moundmountainousunit

ump

Forms cluster of three high-standing mountains �5 kmacross each with prominentNE-SW trending ridges nearmound center. No layering isevident. Interpretation:possibly related to cratercentral peak zone

Upper moundsmooth plainsunits 1–2

umsp1–2

Forms planar surface ofrelatively flat-lying, light-toned material. Some cratersfilled with dark-tonedindurated material, possiblypreviously buried andexhumed

References

Anderson, R.C., Bell III, J.F., 2010. Geologic mapping and characterization of GaleCrater and implications for its potential as a Mars Science Laboratory landingsite. Mars 5, 76–128. doi:10.1555/mars.2010.0004.

Andrews-Hanna, J.C., Phillips, R.J., Zuber, M.T., 2007. Meridiani Planum and theglobal hydrology of Mars. Nature 446, 163–166.

Andrews-Hanna, J.C., Zuber, M.T., Arvidson, R.E., Wiseman, S.M., 2010. Early Marshydrology: Meridiani playa deposits and the sedimentary record of ArabiaTerra. J. Geophys. Res. 115, E06002. doi:10.1029/2009JE003485.

Baldridge, A.M., Hook, S.J., Crowley, J.K., Marion, G.M., Kargel, J.S., Michalski, J.L.,Thomson, B.J., de Souza Filho, C.R., Bridges, N.T., Brown, A.J., 2009.Contemporaneous deposition of phyllosilicates and sulfates: Using Australianacidic saline lake deposits to describe geochemical variability on Mars.Geophys. Res. Lett. 36, 19201. doi:10.1029/2009GL040069.

Barthel, K.W., Swinburne, N.H.M., Morris, S.C., 1990. Solnhofen: A Study in MesozoicPalaeontology. Cambridge University Press, 245pp.

Bell III, J.F., Edgett, K.S., Rowland, S., Malin, M.C., 2006. The Gale Crater Mound: Astrong candidate landing site for the 2009 Mars Science Laboratory. FirstLanding Site Workshop for the Mars Science Laboratory, Pasadena, CA.

Beyer, R.A., McEwen, A.S., 2005. Layering stratigraphy of eastern Coprates andnorthern Capri Chasmata, Mars. Icarus 179, 1–23.

Bibring, J.-P., Langevin, Y., Mustard, J.F., Poulet, F., Arvidson, R., Gendrin, A., Gondet,B., Mangold, N., Pinet, P., Forget, F., 2006. Global mineralogical and aqueousMars history derived from OMEGA/Mars Express data. Science 312, 400–404.

Boutin, D., Arkani-Hamed, J., 2006. Pole wandering of Mars: Evidence frompaleomagnetic poles. Icarus 181, 13–25.

Bradley, B.A., Sakimoto, S.E.H., Frey, H., Zimbelman, J.R., 2002. Medusae FossaeFormation: New perspectives from Mars Global Surveyor. J. Geophys. Res. 107,5058. doi:10.1029/2001JE001537.

Bridges, N.T., 2001. Assessing layered materials in Gale Crater. First Landing SiteWorkshop for the 2003 Mars Exploration Rovers, NASA Ames, Mountain View,CA.

Bridges, N.T., 2006. Studies of martian sedimentological history through in-situstudy of Gale and Oudemans Craters: Two landing site proposals for the MarsScience Laboratory. First Landing Site Workshop for the Mars ScienceLaboratory, Pasadena, CA.

Briggs, D.E.G., Fortey, R.A., 2005. Wonderful strife: Systematics, stem groups, andthe phylogenetic signal of the Cambrian radiation. Paleobiology 31, 94–112.

Brown, A., Walter, M., Cudahy, T., 2004. Short-wave infrared reflectanceinvestigation of sites of paleobiological interest: Applications for Marsexploration. Astrobiology 4, 359–376.

Byrne, S., Ivanov, A.B., 2004. Internal structure of the martian south polar layereddeposits. J. Geophys. Res. 109, E11001. doi:10.1029/2004JE002267.

Cabrol, N.A., Grin, E.A., Newsom, H.E., Landheim, R., McKay, C.P., 1999.Hydrogeologic evolution of Gale Crater and its relevance to the exobiologicalexploration of Mars. Icarus 139, 235–245.

Chapman, C.R., Jones, K.L., 1977. Cratering and obliteration history of Mars. Ann.Rev. Earth Planet. Sci. 5, 515–540.

Craddock, R.A., Howard, A.D., 2002. The case for rainfall on a warm, wet early Mars.J. Geophys. Res. 107, 5111. doi:10.1029/2001JE001505.

Delamere, A. et al., 2003. MRO High Resolution Imaging Science Experiment(HiRISE): Instrument development. Lunar Planet. Sci. 6. Abstract #3287.

Ewing, S.A., Sutter, B., Owen, J., Nishiizumi, K., Sharp, W., Cliff, S.S., Perry, K., Dietrich,W., McKay, C.P., Amundson, R., 2006. A threshold in soil formation at Earth’sarid hyperarid transition. Geochim. Cosmochim. Acta 70, 5293–5322.

Fassett, C.I., Head, J.W., 2008. The timing of martian valley network activity:Constraints from buffered crater counting. Icarus 195, 61–89.

Fishbaugh, K.E., Hvidberg, C.S., 2006. Martian north polar layered depositsstratigraphy: Implications for accumulation rates and flow. J. Geophys. Res.111, 06012. doi:10.1029/2005JE002571.

Forsberg-Taylor, N.K., Howard, A.D., Craddock, R.A., 2004. Crater degradation in themartian highlands: Morphometric analysis of the Sinus Sabaeus region andsimulation modeling suggest fluvial processes. J. Geophys. Res. 109, E05002.doi:10.1029/2004JE002242.

Frawley, J.J., Taylor, P.T., 2004. Paleo-pole positions from martian magnetic anomalydata. Icarus 172, 316–327.

Gould, S.J., 1989. Wonderful Life: The Burgess Shale and the Nature of History. W.W.Norton & Company, New York, 352pp.

Grant, J.A., Irwin, R.P., Grotzinger, J.P., Milliken, R.E., Tornabene, L.L., McEwen, A.S.,Weitz, C.M., Squyres, S.W., Glotch, T.D., Thomson, B.J., 2008. HiRISE imaging ofimpact megabreccia and sub-meter aqueous strata in Holden Crater, Mars.Geology 36, 195–198.

Greeley, R., Guest, J.E., 1987. Geologic map of the eastern equatorial region of Mars.USGS Misc. Map, I-1802-B, scale 1:15,000,000.

Grimm, R.E., Solomon, S.C., 1986. Tectonic tests of proposed polar wander paths forMars and the Moon. Icarus 65, 110–121.

Grotzinger, J.P. et al., 2005. Stratigraphy and sedimentology of a dry to wet eoliandepositional system, Burns formation, Meridiani Planum, Mars. Earth Planet.Sci. Lett. 240, 11–72.

Hansen, V.L., 2000. Geologic mapping of tectonic planets. Earth Planet. Sci. Lett. 176,527–542.

Hartmann, W.K., 2005. Martian cratering 8: Isochron refinement and thechronology of Mars. Icarus 174, 294–320.

Hartmann, W.K., Neukum, G., 2001. Cratering chronology and the evolution of Mars.Space Sci. Rev. 96, 165–194.

Head III, J.W., Hiesinger, H., Ivanov, M.A., Kreslavsky, M.A., Pratt, S., Thomson, B.J.,1999. Possible ancient oceans on Mars: Evidence from Mars Orbiter LaserAltimeter data. Science 286, 2134–2137.

Herkenhoff, K.E., Plaut, J.J., 2000. Surface ages and resurfacing rates of the polarlayered deposits on Mars. Icarus 144, 243–253.

Hofmann, H.J., Grey, K., Hickman, A.H., Thorpe, R.I., 1999. Origin of 3.45 Ga coniformstromatolites in Warrawoona Group, Western Australia. Geol. Soc. Am. Bull.111, 1256–1262.

Hynek, B.M., Phillips, R.J., Arvidson, R.E., 2003. Explosive volcanism in the Tharsisregion: Global evidence in the martian geologic record. J. Geophys. Res. 108,5111. doi:10.1029/2003JE002062.

Irwin, R.P., Watters, T.R., 2010. Geology of the martian crustal dichotomy boundary:Age, modifications, and implications for modeling efforts. J. Geophys. Res. 115,E12S15. doi:10.1029/2010JE003658.

Irwin, R.P., Watters, T.R., Howard, A.D., Zimbelman, J.R., 2004. Sedimentaryresurfacing and fretted terrain development along the crustal dichotomyboundary, Aeolis Mensae, Mars. J. Geophys. Res. 109, E09011. doi:10.1029/2004JE002248.

Irwin, R.P., Howard, A.D., Craddock, R.A., Moore, J.M., 2005. An intense terminalepoch of widespread fluvial activity on early Mars: 2. Increased runoff andpaleolake development. J. Geophys. Res. 110, E12S15. doi:10.1029/2005JE002460.

Jensen, J.R., 2000. Remote Sensing of the Environment. Prentice-Hall Inc., 544pp.Kargel, J.S., Furfaro, R., Prieto-Ballesteros, O., Rodriguez, J.A.P., Montgomery, D.R.,

Gillespie, A.R., Marion, G.M., Wood, S.E., 2007. Martian hydrogeology sustainedby thermally insulating gas and salt hydrates. Geology 35, 975–978.

Kirk, R.L. et al., 2008. Ultrahigh resolution topographic mapping of Mars with MROHiRISE stereo images: Meter-scale slopes of candidate Phoenix landing sites. J.Geophys. Res. 113, E00A24. doi:10.1029/2007JE003000.

Kirk, R.L., Howington-Kraus, E., Rosiek, M.R., 2009. Build your own topographicmodel: A photogrammetry guest facility for planetary researchers. Lunar Planet.Sci. 40. Abstract 1414.

Laskar, J., Correia, A.C.M., Gastineau, M., Joutel, F., Levrard, B., Robutel, P., 2004. Longterm evolution and chaotic diffusion of the insolation quantities of Mars. Icarus170, 343–364.

Malin, M.C., Edgett, K.S., 2000. Sedimentary rocks of early Mars. Science 290, 1927–1937.

Milkovich, S.M., Plaut, J.J., 2008. Martian South Polar Layered Deposit stratigraphyand implications for accumulation history. J. Geophys. Res. 113, 06007.doi:10.1029/2007JE002987.

Milliken, R.E., Grotzinger, J.P., Thomson, B.J., 2010. Paleoclimate of Mars as capturedby the stratigraphic record in Gale Crater. Geophys. Res. Lett. 37, L04201.doi:10.1029/2009GL041870.

Moore, H.J., Hodges, C.A., Scott, D.H., 1974. Multiringed basins – Illustrated byOrientale and associated features. Proc. Lunar Sci. Conf. 5, 71–100.

Murchie, S. et al., 2009. Evidence for the origin of layered deposits in CandorChasma, Mars, from mineral composition and hydrologic modeling. J. Geophys.Res. 114, E00D05. doi:10.1029/2009JE003343.

Murray, B.C., Malin, M.C., 1973. Polar wandering on Mars? Science 179, 997–1000.

432 B.J. Thomson et al. / Icarus 214 (2011) 413–432

Neumann, G.A., Rowlands, D.D., Lemoine, F.G., Smith, D.E., Zuber, M.T., 2001.Crossover analysis of Mars Orbiter Laser Altimeter data. J. Geophys. Res. 106,23753–23768.

Niles, P.B., Michalski, J., 2009. Meridiani Planum sediments on Mars formed throughweathering in massive ice deposits. Nat. Geosci. 2, 215–220.

Okubo, C.H., McEwen, A.S., 2007. Fracture-controlled paleo-fluid flow in CandorChasma, Mars. Science 315, 983–985.

Parker, T.J., Saunders, R.S., Schneeberger, D., 1989. Transitional morphology in WestDeuteronilus Mensae, Mars: Implications for modification of the lowland/upland boundary. Icarus 82, 111–145.

Pelkey, S.M., Jakosky, B.M., 2002. Surficial geologic surveys of Gale Crater and MelasChasma, Mars: Integration of remote-sensing data. Icarus 160, 228–257.

Pelkey, S.M., Jakosky, B.M., Christensen, P.R., 2004. Surficial properties in GaleCrater, Mars, from Mars Odyssey THEMIS data. Icarus 167, 244–270.

Perron, J.T., Mitrovica, J.X., Manga, M., Matsuyama, I., Richards, M.A., 2007. Evidencefor an ancient martian ocean in the topography of deformed shorelines. Nature447, 840–843.

Plaut, J.J. et al., 2007. Subsurface radar sounding of the south polar layered depositsof Mars. Science 316, 92–95.

Rossi, A.P., Neukum, G., Pondrelli, M., van Gasselt, S., Zegers, T., Hauber, E., Chicarro,A., Foing, B., 2008. Large-scale spring deposits on Mars? J. Geophys. Res. 113,E08016. doi:10.1029/2007JE003062.

Schultz, P.H., Lutz, A.B., 1988. Polar wandering of Mars. Icarus 73, 91–141.Schultz, P.H., Greeley, R., Gault, D., 1977. Interpreting statistics of small lunar

craters. Proc. Lunar Sci. Conf. 8, 3539–3564.Scott, D.H., Chapman, M.G., 1995. Geologic and topographic maps of the Elysium

paleolake basin, Mars. USGS Misc. Map, I-2397, scale 1:5,000,000.

Scott, D.H., Tanaka, K.L., 1982. Ignimbrites of Amazonis Planitia region of Mars. J.Geophys. Res. 87, 1179–1190.

Scott, D.H., Carr, M.H., 1978. Geologic Map of Mars. USGS Map, I-1083, scale1:25,000,000.

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

Smith, D.E. et al., 2001. Mars Orbiter Laser Altimeter: Experiment summary after thefirst year of global mapping of Mars. J. Geophys. Res. 106, 23689–23722.

Sprenke, K.F., Baker, L.L., 2000. Magnetization, paleomagnetic poles, and polarwander on Mars. Icarus 147, 26–34.

Tanaka, K.L., 1986. The stratigraphy of Mars. J. Geophys. Res. 91, E139–E158.Tanaka, K.L., 2000. Dust and ice deposition in the martian geologic record. Icarus

144, 254–266.Tanaka, K.L., Skinner Jr., J.A., Hare, T., 2005. Geologic Map of the Northern Plains of

Mars. USGS Sci. Inv. Map, 2888, scale 1:15,000,000.Thomson, B., Bridges, N., Bell, J., Milliken, R., Calvin, W., 2007. Gale Crater layered

mound: A closed hydrologic system. Second Mars Science Laboratory LandingSite Workshop, Pasadena, CA.

Watters, T.R. et al., 2007. Radar sounding of the Medusae Fossae Formation Mars:Equatorial ice or dry, low-density deposits? Science 318, 1125–1128.

Webb, R.H., Griffiths, P.G., Rudd, L.P., 2008. Holocene debris flows on the ColoradoPlateau: The influence of clay mineralogy and chemistry. Geol. Soc. Am. Bull.120, 1010–1020.

Wilhelms, D.E., 1990. Geologic mapping. In: Greeley, R., Batson, R.M. (Eds.),Planetary Mapping. Cambridge University Press, Cambridge, pp. 208–260.