channel slope reversal near the martian dichotomy boundary ...fnimmo/website/lefort.pdfchannel slope...

Geomorphology 240 (2015) 121–136

Contents lists available at ScienceDirect

Geomorphology

j ourna l homepage: www.e lsev ie r .com/ locate /geomorph

Channel slope reversal near the Martian dichotomy boundary: Testingtectonic hypotheses

Alexandra Lefort a,⁎, Devon M. Burr a, Francis Nimmo b, Robert E. Jacobsen a

a Department of Earth and Planetary Sciences, University of Tennessee, 1412 Circle Dr., Knoxville, TN 37996-1410, United Statesb Department of Earth and Planetary Sciences, University of California Santa Cruz, Santa Cruz, CA 95064, United States

⁎ Corresponding author at: 2414-9357 Simpson Dr.Canada. Tel.: +1 604 619 5779.

E-mail address: [email protected] (A. Lefort).

http://dx.doi.org/10.1016/j.geomorph.2014.09.0280169-555X/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t
a r t i c l e i n f o
Article history:Received 20 November 2013Received in revised form 16 September 2014Accepted 25 September 2014Available online 20 December 2014

Keywords:MarsFluvialGeomorphologyTectonicsDichotomy boundary

Faults along theMartian dichotomy boundary are evidence of tectonic activity, which by analogy with terrestrialtectonism may cause changes in or even reversal of fluvial longitudinal profiles. In the eastern hemisphere be-tween 30° E and 150° E, this tectonic activity has beenhypothesized to result from lower crustalflowor from lith-ospheric flexure, for which loading (e.g., by material deposition) of the northern lowlands is a possible cause(Watters, 2003a; Nimmo, 2005). The topographic (slope) changes resulting from these two different mecha-nisms are distinct and can provide ameans for distinguishing between them, although other processesmay com-plicate interpretations of reversed longitudinal profiles. Two fan-shaped networks of inverted channels arelocated near 150° E in the Aeolis Dorsa region just north of the dichotomy boundary. Their original flow direc-tions, inferred from planformmorphology, suggest flow to the northeast in contrast to their current longitudinalprofiles sloping down to the southwest. This contrast indicates slope reversal. We investigate the lower crustalflow and flexure mechanisms for slope reversal by testing three different hypotheses: 1) lower crustal flow,2) flexure caused by material erosion from the highlands and deposition in the lowlands, and 3) flexure causedby highland erosion and deposition in the lowlands plus deposition of theMedusae Fossae Formation in the low-lands.We test these three hypotheses by comparing the inferredmagnitudes of the slope reversals with predict-ed slope changes from geophysical models for these processes. Taking the possibility of non-tectonic(i.e., collapse) processes into account, our results suggest that, among these three models, the slope reversal ismost consistent with the predicted tectonic response to erosion and deposition of highlandmaterial in conjunc-tion with deposition of the Medusae Fossae Formation. Contrary to previous findings, our results do not supportthemechanismof lateral crustalflow, although they donot rule it out because lower crustalflowand erosionmayboth have been operating, but on different time scales. The result of this hypothesis-testing provides insight intothe evolution of the dichotomy boundary and Martian crust in this location.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

One the most prominent features of the geology of Mars is the di-chotomy boundary, the global topographic and geologic contact sepa-rating the older and heavily cratered southern highlands from themore recent and relatively smoother northern lowlands. The boundaryis a narrow region that occupies an ~700 kmwide belt around the plan-et and is characterized by a distinct change in elevation, with the high-lands standing 2 to 5 km above the lowlands (e.g., Watters et al., 2007).The dichotomy boundary is expressed both as a distinctive topographicchange and as a change in crustal thickness, with an average crustalthickness of 32 km in the northern lowlands, 58 km in the southernhighlands (e.g., Zuber et al., 2000; Watters et al., 2007), and 55 ±

NW, Edmonton, AB T6R 0N3,

20 km near the dichotomy (Nimmo, 2002; Wieczorek and Zuber,2004). The crustal dichotomy boundary is spatially correlated with thetopographic dichotomy boundary only between 110° and 190° E(e.g., Zuber et al., 2000).

The dichotomy boundary is a complex and enigmatic structure. Hy-pothesized to have formed early in the history of Mars (e.g., Watters,2003a; Frey, 2004; Watters et al., 2007; Andrews-Hanna et al., 2008),it has been modified by tectonic, mass wasting, fluvial, aeolian and gla-cial processes (McGill and Dimitriou, 1990; Watters and Robinson,1999; Watters, 2003a,b; Irwin et al., 2005; Tanaka et al., 2005), whosetiming is poorly constrained. Between 30° E and 150° E in the vicinityof the dichotomy boundary, the presence of large-scale faults is evi-dence of tectonic deformation (Guest and Smrekar, 2005). On Mars,large-scale processes that may generate tectonic stresses includelower crustal flow. This horizontal flow, which occurs in a ductilelower crust, is generated by lateral pressure gradients caused by differ-ences in crustal thickness, and over time leads to reduction in crustalthickness variations (e.g., Zuber et al., 2000; Nimmo and Stevenson,

http://crossmark.crossref.org/dialog/?doi=10.1016/j.geomorph.2014.09.028&domain=pdf

http://dx.doi.org/10.1016/j.geomorph.2014.09.028

mailto:[email protected]

http://dx.doi.org/10.1016/j.geomorph.2014.09.028

http://www.sciencedirect.com/science/journal/0169555X

www.elsevier.com/locate/geomorph

122 A. Lefort et al. / Geomorphology 240 (2015) 121–136

2001; Nimmo, 2005). Alternatively, large-scale tectonic stresses may begenerated by lithospheric flexure that occurs in response to erosion andredeposition of surficial material (e.g., Tanaka et al., 2001; Searls andPhilips, 2004; Nimmo, 2005). On Mars, lower crustal flow and litho-spheric flexure are hypothesized to have contributed to modifying thedichotomy boundary (Watters, 2003a; Nimmo, 2005).

Comparing these mechanisms with observations of tectonic defor-mation may help constrain the evolution of the dichotomy boundary(Nimmo, 2005) and the conditions of earlyMars. For instance, reductionof variation in local or regional crustal thickness caused by lower crustalflow is temperature-dependent and provides, therefore, a constraint onthe thermal evolution of the crust (e.g., Parmentier and Zuber, 2007). Iflower crustal flow occurred, then it places constraints on the thermalstate of the crust at that time, which has implications for our under-standing of the internal evolution of Mars. Likewise, determination ofthe elastic thickness of the lithosphere also constrains the thermal gra-dient, because the depth of the lower limit at which ductile responsereplaces brittle response is essentially governed by temperature(e.g., Solomon and Head, 1990). Determination of the elastic thicknessis, therefore, important to reconstruct the thermal history of Mars.

Lower crustal flow and lithospheric flexure may cause significantsurface deformations, as observed on Earth where, for example,lower-crustal flow contributes to the evolution of the Tibetan plateau(e.g., Royden et al., 1997; Bendick and Flesch, 2007),material depositionin sedimentary basins is a cause of lithospheric flexure (e.g.,Watts et al.,1982;Watts, 1989), and flexural isostasy of the lithosphere has contrib-uted to the drainage reversal of the Amazon river (Sacek, 2014). OnMars, each of these mechanisms (lower crustal flow and flexure causedby erosion/deposition) would produce different topographic modifica-tions, namely, different amounts or directions of land surface tilting.Thus, a test for the cause of the tectonic deformation is the amount ordirection of change or even reversal of local or regional slopes. Slopechange or reversal may be inferred through discernment that the cur-rent slopes of features do not match the original slopes, such as thetilting of impact craters with originally horizontal floors (e.g., Irwinand Watters, 2010) or modification of longitudinal profiles of fluvialforms. Analysis of the longitudinal profiles of fluvial channels is a tech-nique that has been widely used on Earth to identify and characterizetectonic deformations (e.g., Seeber and Gornitz, 1983; Demoulin,1998; Holbrook and Schunun, 1999; Snyder et al., 2000; Jain andSinha, 2005; Chen et al., 2006; Larue, 2008; Lahiri and Sinha, 2012;Roberts et al., 2012). The availability of high-resolution Martian geo-morphic and topographic data allows us to apply the same techniqueto Mars.

Close to the dichotomy boundary north of Terra Cimmeria, inverted(positive-relief) fluvial features are observed in the Aeolis Dorsa region(formerly referred to as the Aeolis–Zephyria Plana) in the westernMedusae Fossae Formation (MFF) (e.g., Pain et al., 2007; Burr et al.,2009, 2010; Zimbelman and Griffin, 2010). Previous analysis hasshown that instead of sloping monotonically downward, the longitudi-nal profiles of some of these features undulate, indicating some post-flow modification (Lefort et al., 2012). Such modification may havebeen caused by multiple and possibly concurrent processes, such assediment compaction or collapse and tectonic deformation. In addition,the original flow direction and channel slope for many of these fluvialfeatures are not conclusively known, making slope or direction changesdifficult to determine. Where sedimentary causes may be ruled out andthe original flow conditions inferred from the geomorphology, howev-er, changes in slope can be inferred.

Of the multiple branching networks of inverted fluvial features inthe Aeolis Dorsa, two networks have large-scale, fan-shaped plan viewmorphologies. These morphologies provide constraints on the originalflow direction, and allow the original slopes to be estimated from com-parison to other similar fan-shaped networks on Earth andMars. Undu-lations in the longitudinal profiles of some of these features suggest thatmodification by sedimentary processes (e.g., compaction, collapse)may

have occurred, but these undulations are small (b100 m amplitude,~b15 km in wavelength; Lefort et al., 2012) compared to the scale ofthe fluvial networks. Thus, these fluvial features with regional-scaleslopes that are reversed from the estimated original slopes may beused to differentiate between the possible mechanisms for tectonicdeformation. Tectonic activity in this location is implied by the presenceof an inferred extensional fault along the southwestern boundary of thewestern MFF (McGill and Dimitriou, 1990; McGill et al., 2004) in theAeolis Dorsa region (Fig. 1).

In this work, we use multiple datasets to derive the current slopesalong the best-preserved and coeval individual fluvial features inthese two fan-shaped networks. Based on morphological analysis andcomparison to slopes found in morphologically similar terrestrial andMartian branching networks, we infer themagnitude of the slope rever-sals. We then compare these magnitudes to model slope changes foreach tectonic deformation mechanism to test among three proposeddeformation scenarios: 1) a lower-crustal flow model, 2) a lithosphericflexuremodel caused bymaterial erosion from the highlands and depo-sition in the lowlands, and 3) a lithospheric flexure model includinghighlands erosion and deposition in the lowlands as well as MFF depo-sition in the lowlands. Our results are complicated by the possibility oflocal deformation caused by collapse in the south, between these net-works and the dichotomy boundary. Geospatial relationships, however,suggest that only one of the two networks may have been affected bycollapse and that this processmost likely did not have a significant influ-ence on the slope reversal. For both networks, the results are conclu-sively more consistent with the erosion–deposition model includingdeposition of the MFF than with the lower crustal flow model. Fromthose results, we derive implications for the type and timing of thecrustal processes that occurred at this location along theMartiandichot-omy boundary.

2. Regional background: the Medusae Fossae Formation and theAeolis dorsa

The MFF is an extensive, light-toned, friable, layered deposit like-ly composed of volcanic ash deposits (e.g., Scott and Tanaka, 1982,1986; Bradley et al., 2002; Hynek et al., 2003; Mandt et al., 2008)and up to 3.5 km thick in places (Scott and Tanaka, 1986; Greeleyand Guest, 1987; Hynek et al., 2003; Watters et al., 2007). It is situat-ed along the dichotomy boundary between 130° and 240° E. Its pres-ent extent is estimated to be 2.1–2.5 × 106 km2 and its presentvolume to 1–1.9 × 106 km3, although it may have once covered upto 5.7 × 106 km2 (Bradley et al., 2002; Harrison et al., 2010). Al-though the MFF was previously dated to the Amazonian epoch(e.g., Scott and Tanaka, 1986; Tanaka, 1986; Head and Kreslavsky,2001, 2004; Werner, 2006), recent work suggests that it was largelyin place by the (late) Hesperian (Burr et al., 2009; Kerber and Head,2010; Zimbelman and Scheidt, 2012).

The western MFF corresponds to the lowest and, therefore, oldestmember of the MFF, and contains the Aeolis Dorsa, an extensive andnumerous population of sinuous ridges. These ridges have beeninterpreted as inverted fluvial paleochannels or meander belts (Burret al., 2009; Zimbelman and Griffin, 2010), formed by surface runoff,chemical cementation of the fluvial sediments, burial by subsequentdeposition, and finally exhumation and inversion by aeolian abrasion(Pain et al., 2007; Burr et al., 2009, 2010). Whereas another positive-relief flow network has been noted in the central part of the MFF(Harrison et al., 2013), the Aeolis Dorsa contains the highest density ofinverted surface flow features identified in these deposits (Burr et al.,2009).

The two large fan-shaped networks of paleochannels that are thefocus of this study are located ~275 km north of the dichotomy bound-ary (Fig. 1). These networks display a complex stratigraphy, as shownby relative elevations, overlapping relationships and surface textures.For example, in Fig. 2, the surface of the fan-shaped plateau, marked

Fig. 1. Top: context image modified from Nimmo (2005) and Burr et al. (2009). AP: Aeolis Planum, CP: Cerberus Palus, AV: Athabasca Valles. Black arrows: compression zones, white ar-rows: extension zones. Black delineation:westernMFF. Black square: study area. GrayN–S line: example of location of profile in Fig. 8. Bottom: study area, covering 148.9° E to 152° E, and−5.7° to−7.3°, showingwestern and eastern networks. For the eastern network, black lines indicate most recent paleochannels, gray lines indicate intermediate age paleochannels, andwhite lines indicate oldest paleochannels. For western network, gray areas indicate the surface of the plateau under which paleochannels may still be buried. Large white rectangle: lo-cation of Fig. 2. Large black squares (A and B): location of Fig. 4(a) and (b). Small black squares (a, b, c, d): locations of close-up images showing cross-cutting and stratigraphic relationshipsbetween the different paleochannels (Fig. 3).

123A. Lefort et al. / Geomorphology 240 (2015) 121–136

as “A”, is characterized by large elongated N–S oriented yardangs,whereas the surfaces marked as “B”, located at a lower elevation than“A”, are characterized by a relatively smooth texture. The surface at C,located at a lower elevation than surface “B”, is knobby in texture. We,therefore, interpret A, B and C as the surfaces of three different deposi-tional strata, from youngest to oldest, respectively. The inset in Fig. 2shows a yardang overlying a paleochannel, suggesting the existence ofyet another stratum, deposited after formation of the paleochannels,and now almost entirely eroded (Jacobsen and Burr, 2013).

Of the two networks, the more prominent network to the east hasbeen inferred in relatively low resolution (THEMIS IR, 100m/px) imagesto be distributarywith a paleoflowdirection to the northeast (e.g., Irwinet al., 2005; McMenamin andMcGill, 2005;Williams and Edgett, 2005).Although less well exposed, the similar fan-shaped morphology andorientation for the western network suggests that it was likewise dis-tributary and also flowed to the northeast. Previous analysis, however,showed that the current slopes of the best-exposed paleochannels inthe eastern network are undulatory and overall down to the southwest(Lefort et al., 2012). Based on the inference from the fan shape of an

original down-to-the-northeast flow direction (e.g., Irwin et al., 2005;McMenamin andMcGill, 2005; Williams and Edgett, 2005), the currentdown-to-the-southwest slope indicates slope reversal, and provides apromising location for additional network analyses to investigate thetectonic deformation mechanisms discussed above.

Crater retention ages for the lower andmiddlemembers of thewest-ern MFF indicate emplacement during the Hesperian (Zimbelman andScheidt, 2012), consistent with evidence from embayment and cross-cutting relationships (Kerber and Head, 2010). These lower and middlemembers either include or post-date many of the Aeolis Dorsa fluvialfeatures (Jacobsen and Burr, 2013). Mapping (Zimbelman and Scheidt,2012) places the two networks discussed here in a unit with a crater-retention age at the Hesperian–Amazonian boundary, indicatingemplacement during or before the late Hesperian (Zimbelman andScheidt, 2012). The Aeolis Mensae, immediately west of and possiblycoeval with the lower strata of the Aeolis Dorsa, has been dated to thelate Noachian/Early Hesperian (Irwin et al., 2004). On this basis, theeastern network discussed in this work has been posited to correlateto the Noachian–Hesperian transition (Kite et al., 2013).

Fig. 2. Stratigraphy of the eastern network. CTX image with CTX DTM overlay. Dark areas: lower elevations, light areas: higher elevations. A: plateau with N–S yardangs. B: smoother sur-face at intermediate elevation. C: lower knobby surface. Inset: yardang overlaying inverted channel. Largewhite circle: possible avulsion node. Small black circles: possible scattered avul-sions. The small black and white arrows mark the beginning and end of the profile.


3. Methods and data

3.1. Mapping of fan units and derivation of fluvial feature slopes

3.1.1. Data processing for plan view mappingWe identified the best-preserved and coeval individual fluvial

deposits in these networks through mapping. For our mapping, weused the infrared (IR) day mosaic from the Thermal Mapper ImagerSpectrometer (THEMIS) (Christensen et al., 2001) with a resolutionof 592.75 pixels per degree (ppd) Christensen et al., 2004) as abasemap, overlain by images from the Context Camera (CTX,[Malin et al., 2007], 6 m/pixel, ~95% area coverage) and from theHigh Resolution Imaging Science Experiment (HiRISE, [McEwenet al., 2007], 25 cm/pixel, ~10% area coverage). The CTX and HiRISEimages were acquired from the Planetary Data System (PDS,http://pds.nasa.gov/) and radiometrically corrected and projectedusing the Integrated Software for Imagers and Spectrometers (ISIS;[Gaddis et al., 1997; Torson and Becker, 1997; Anderson et al.,2004]). All images were then imported into ArcGIS for mapping inequirectangular projection.

3.1.2. Topographic data processingTo derive the current slopes of these linear fluvial deposits, we

used two digital terrain models (DTMs) produced from stereo pairimages, and individual data points from the Mars Orbiter LaserAltimeter (MOLA; Smith et al., 2001); we also used the MOLAgridded data during dataset co-registration. One DTM was producedby a team member of the High Resolution Stereo Camera (HRSC,Neukum et al., 2004) using HRSC stereo images (100 m/pixel) with~30 m vertical accuracy (Gwinner et al., 2010). We produced a sec-ond DTM using CTX images in the Ames Stereo Pipeline (A.S.P.,Moratto et al., 2010), giving 6 m/pixel resolution and 10–20 m verti-cal accuracy. The lower precision of the Mars datum (size of the ref-erence spheroid) used in ASP compared to the precision used byMOLA results in an ~500 m offset between the MOLA data and theCTX DTM. As explained in Lefort et al. (2012), this offset wascorrected by creating a difference map between the MOLA and CTXDTMs, deriving a histogram of this difference, and adding themedianvalue of the histogram to the CTX DTM values. The result is generalalignment between the CTX and the MOLA DTM. The HRSC DTMcovers all of the western network and most of the eastern network.

http://pds.nasa.gov/


The higher-resolution CTX DTM coversmost of the eastern network.Wehave, therefore, better topographic data on the eastern network, whichis also better preserved and so can provide better topographic profiles.

All topographic data were imported into ArcGIS and overlaid on theimages in order of increasing resolution (Lefort et al., 2012).We first co-registered the THEMIS day IR 512 ppd mosaic to the MOLA DTM. Thenthe HRSC images and DTMs, which are already aligned to the MOLADTM, were overlaid onto the THEMIS day IR 512 ppd mosaic. The CTXimages were then overlaid on the HRSC images, and finally the HiRISEimages were overlaid to the CTX images. Any misalignments betweenCTX, HiRISE data and the THEMIS basemap were manually correctedin ArcGIS using the georeferencing tool. Because of their high resolution,the CTX andHRSC DTMswere the preferred datasets for the eastern andwestern networks, respectively, for deriving paleochannel topographicprofiles. Paleochannel profiles were acquired from these data in ArcGIS.Additional independent topographic information was provided by indi-vidual data points from the Mars Orbiter Laser Altimeter (MOLA, ~150-m-footprint and ~300-m-along-track spacing) (Neumann et al., 2001;Smith et al., 2001). Data points centrally located on top of thepaleochannels were manually selected and then plotted in ArcGIS ontop of the CTX and HRSC profiles.

3.1.3. Error minimizationErrors in our results may arise from data co-registration errors. The

technique (Section 3.1.2) that we employed to co-register the differentdatasets, however, minimizes mis-registration errors between the vari-ous DTMs, the individual MOLA data points and the HRSC, CTX andHiRISE images.

For the profiles derived from individual MOLA data points, the largefootprints of the MOLA data points are another possible source of error.The MOLA footprint may overlap the top and sides of narrowpaleochannels, as well as the surrounding terrain, thus causing inaccu-rate topographic elevation estimates for the top surface of thepaleochannels. To minimize errors, we manually selected MOLA datapoints that are located as centrally as possible on top of thepaleochannels as seen in the CTX andHiRISE images (Lefort et al., 2012).

3.1.4. Geomorphic mapping techniques and slope derivationTo select the best-preserved and coeval examples of inverted fluvial

features for assessment of slope change, we first delineated all the fluvi-al features, using the CTX images in transparency with the topographyvisible through the images for additional guidance. Consistent with ter-restrial analogs (e.g., Williams et al., 2007), the flat tops of several of thepaleochannels are interpreted as a capping layer of indurated fluvialsediments, overlying non-indurated material. Various degrees of ero-sion of this capping layer by fracturing and fragmentation may be ob-served in CTX and HiRISE images (see Fig. 2 in Lefort et al., 2012).Where the capping layer has been entirely removed, the underlyingless consolidated material is exposed. Whereas channels with a pre-served capping layer have a flat surface, channels without a cappinglayer are much thinner (narrower) in plan-view, because of the pro-gressive erosion of the exposed unconsolidated material (e.g., Burret al., 2009, Fig. 10). The presence and degree of preservation of a cap-ping layer were two of the main criteria used to discriminate betweendifferent generations of paleochannels. Other criteria include the heightand width of the paleochannels, elevation with respect to otherpaleochannels (based on topographic information from CTX and HRSCDTMs), cross-cutting and superposition relationships, and sinuosity.

We also identified the different geomorphic units that contain or areadjacent to the paleochannels and derived their stratigraphic relation-ship. Geomorphic units were identified first by considering the plan-view surface textures, e.g., smooth, knobby, presence of elongated hillsinterpreted as yardangs (streamlined remnants of erosional abrasion).For yardang surface textures, the orientation, size, and aspect ratio ofthe yardangs were also distinguishing characteristics. The relative strat-igraphic relationships among the different geomorphic units were

determined by assessing their relative elevations with the MOLA, CTXand HRSC DTMs (e.g., Fig. 2).

These methods allowed us to distinguish different strata and differ-ent generations of paleochannels (Fig. 1) and to classify themby relativeage (most recent, intermediate, oldest, Fig. 3). Using the techniques de-scribed above, we then derived longitudinal profiles (Fig. 4) using CTX,HRSC andMOLAdata on themost pristine of the recent examples (threein the eastern network, four in thewestern network) to obtain themostreliable results. The slopes in Table 2 were calculated using distance-along-channel. They were derived by fitting a straight line with linearregression to the data points over the paleochannels. The slopes werefitted to the three datasets separately and averaged together.

3.2. Modeling of proposed tectonic deformation mechanisms

Flow of lower crustal material from the highlands to the lowlandsand flexure caused by the erosion of highlands material and itsredeposition in the lowlands have been proposed as mechanisms bywhich the dichotomy boundary has been deformed (Watters, 2003a;Nimmo, 2005). In this location, the MFF must have imposed a loadnorth of the boundary. Thus, for the two possible tectonic deformationmechanisms, we consider and model three possible scenarios: 1) alower-crustal flow model, 2) an erosion/deposition model in whichthe same amount of surficial material is removed from the highlandsand deposited in the lowlands, and 3) the same erosion/depositionmodel with the addition of MFF deposition. For each model, we followNimmo (2005) in considering elastic thicknesses of T = 15 km, T =30 km and T=60 km. Thesemodels calculate the lithospheric deforma-tion and the change in slope across the dichotomy boundary referencedto the center of the boundary, for which we use the inflection point ofthe long-wavelength topography (see Section 3.3 below). The modelswere previously developed and described by Nimmo and Stevenson(2001) and Nimmo (2005).

3.2.1. Lower crustal flowIn our scenario for lower crustal flow, crustal material would flow

from beneath the thicker southern highland to beneath the thinnernorthern lowlands (e.g., Nimmo, 2005). Lower crustal flow leads to a re-duction in elevation of the thick crust and increase in elevation of thethin crust (e.g., Bird, 1991; Nimmo and Stevenson, 2001; Nimmo,2005). Thus, in our scenario, the southern highlands should undergosubsidence, and the northern lowlands experience uplift, leading to rel-ative tilting of regional slopes down to the south.

The change in slope caused by lower crustal flow is derived directlyfrom the output of the numerical simulations presented in Nimmo(2005). Lower crustal flow of a non-Newtonianmaterial is modeled fol-lowing the approach described inNimmoand Stevenson (2001), using afinite-difference representation, and the parameters given in Nimmo(2005) (Table 1). At the end of the model run (duration 100 My), thecalculated change in crustal thickness is converted into an effectiveload L(x) acting on the base of an elastic plate, where x is the horizontalcoordinate. The resulting surface deflection w(x) is calculated usingequation (1) of Nimmo (2005) for a particular assumed elastic thicknessT. The change in slope Δs is then calculated using

Δs ¼ dwdx

:

Because relaxation is initially rapid and then slows down, most ofthe evolution happens within the first 30 Myr. Therefore, differentmodel times (e.g., 30, 100 and 300 My in Fig. 5) produce only minimaldifferences in the amount of slope change.

3.2.2. Erosion and deposition of surficial materialsSignificant erosion in the highlands during the Noachian is sug-

gested by observations of crater degradation, whereas deposition of

Fig. 4. Examples of topographic profiles along paleochannels, showing agreement regarding overall slope direction amongmultiple datasets. (a) and (c): Longitudinal profile (A–B) alongthe paleochannel in thewestern network. (b) and (d): Longitudinal profile (C–D) along paleochannel in the eastern network. Red dots= elevation data from theHRSCDTM, green dots=elevation data from the CTX DTM, blue dots = MOLA data points. The location of the other profiles is shown as black lines in (a) and (b).

Fig. 3. Lithostratigraphic units, paleochannel morphologies and superposition relationships in paleochannel network (close-up on black rectangles a, b, c and d in Fig. 1). White arrow:paleochannel superposition relationship. Black arrows: locations where partially exhumed paleochannels connect with plateaus. We postulate that these paleochannels continueunder the surface of these plateaus. Inferred paleochannel chronology within these individual networks: 3 — most recent paleochannels, 2 — paleochannels of intermediate age, and1 — oldest observed paleochannels. In c, the intermediate-age paleochannels are not seen within the area shown in this sub-figure. The inferred relative ages pertain to each individualnetwork; correspondence of ages between the eastern and western networks is not necessarily implied.


Table 1Values of parameters used in the models (Nimmo and Stevenson, 2001;Nimmo, 2005).

Parameter Value

Surface gravity 3.72 m s−2

Density contrast 500 kg m−3

Surface temperature 220 KCrustal enrichment factor 4Mantle enrichment factor 0.9Thermal conductivity 3.2 W m−1 K−1

Mean crustal thickness 70 kmCrustal rheology Dry diabaseCrust density 2900 kg m−3

Mantle density 3400 kg m−3

Young's modulus 100 GPaTime step 1010 sGrid spacing 10 km

Fig. 6. Material removed and deposited in the erosion models, in a 1000 km profile.1: Wedge of material eroded from the highlands. 2: Equivalent wedge of material depos-ited in the lowlands. 3: Wedge representing MFF deposition in the lowlands.


that material in the lowlands is suggested by highland remnants foundwithin sedimentary deposits (e.g., Craddock andMaxwell, 1993; Hynekand Phillips, 2001; Tanaka et al., 2005). The amount of material erodedfrom the highlands and deposited in the lowlands, however, is contro-versial. Irwin et al. (2011) suggest that relatively little of the highlandshad a clear transport pathway to the lowlands. Other studies(e.g., Head et al., 2002; Tanaka et al., 2003) suggest that the northernlowland plains are mostly Hesperian volcanic fill with only thinnersedimentary veneers in places. Although redeposition of highlandsmaterial may not have been as extensive as previously thought, itremains the most probable origin for the northern lowlands deposits.Thus, erosion of southern highlands and its deposition in the northernserves as our baseline scenario for flexure.

To simulate erosion and deposition in this location, we modeled theremoval of awedge of sediment 200 km long and 1 kmmaximumdepthat the dichotomy boundary from the highlands and added a same-sizedwedge ofmaterialwith a reversed orientation to the lowlands (Fig. 6). Asurface load L(x) is calculated based on the assumed load topographyand a density of 2900 kg m−3. Although the rate of response to thisload depends on the viscosity structure of the crust, we are most inter-ested in the final deflection, so we calculate the long-term (elastic)deflection and slope change along the profile in exactly the same man-ner as for the lower crustal case described above.

3.2.3. Erosion and deposition of surficial materials, plus deposition of theMFF

Although the amount of redeposited highlands material is some-what uncertain, the MFF is clearly a substantial deposit. In this location,the MFF, by virtue of its presence and extent, evidently loaded thesurface north of the dichotomy boundary. Thus, we considered anothererosion model in which the material deposited is equivalent in amountto the combined load of the eroded highlands material and the MFFdeposits. Based on an estimated current maximum thickness of thewestern MFF of 1.2 km and a maximum N–S extent of 600 km, weadded to the modeled amount of material deposited in the lowlands a

Fig. 5. Change in slope for the baseline lower crustal flow case. The results after three dif-ferent times are plotted; most of the crustal flow happens within 30 Myr.

wedge of material 1.2 km thick at the dichotomy boundary andstretching 600 km to the northeast, where it pinches out (Fig. 6).

3.3. Identification of the midpoint of the dichotomy boundary

Accurate comparison of the observations with the modeling resultsrequires knowing the distance of networks from the dichotomy bound-ary. To determine the position of themidpoint of the dichotomy bound-ary scarp at this location, we generated 15 MOLA profiles across thedichotomy boundary perpendicular to the scarp, with a spacing of~20 km, south of the western Medusae Fossae (for location, seeFig. 1A inNimmo, 2005).We thenfit a polynomial function to these pro-files and calculated the location of its inflection point, which was takenas the midpoint of the dichotomy boundary scarp for this work.

Transfer of material from the highlands to the lowlands likely ledto the retreat of the boundary scarp, as suggested by the presence ofoutliers of the highlands north of the present scarp, for example, flat-topped mesas in Aeolis Mensae (Irwin et al., 2005). Because of theuncertainties in timing and rate of scarp retreat, we chose to neglectthis process in our model. We infer that erosion happened on thesouthern side of the original dichotomy boundary and depositionon the northern side. As long as the total amount of scarp retreat issmaller than the widths of the erosion/deposition zones beingmodeled, this simplification is unlikely to significantly affect ourmodeling results.

4. Results

4.1. Mapping of fan units and topographic profiles

4.1.1. Eastern networkThe better exposed of the two networks, the eastern network, which

has been previously analyzed in Lefort et al. (2012), consists ofbranching paleochannels displaying various morphologies and super-position relationships. Surface morphology, texture, superposition andelevation differences suggest the presence of three different fluvialstrata (Fig. 3).

We label the different types of paleochannels in this network fromthe oldest (type 1) to the most recent (type 3) on the basis of the in-ferred stratigraphy. We use the morphological categories of Burr et al.(2009) in which “thin” paleochannels refer to inverted channels lessthan a few tens of meters in width, with a constant width and sharpcrests, and “flat” paleochannels refer to inverted channels with flatand wide upper surfaces, sharp lateral edges and steeply dippingsides. We discuss them in reverse order, from stratigraphically highestto stratigraphically lowest, based on their discernibility and exposure.

Paleochannels in the northeastern part of the network belong to thecategory “thin”, with widths of a few tens of meters, relief of about 10–


15 m, and a low sinuosity (average sinuosity = 1.09). Their nearlycontinuous planar upper surfaces, interpreted as capping layers, indi-cate that these paleochannels are consistently well-preserved. Thesepaleochannels are located on top of the fan-shaped plateau, whosesurface is characterized by N–S elongated yardangs (surface A inFig. 2). Stratigraphic positions relative to the plateau surface and dis-cernable superposition relationships show that this type ofpaleochannel (Fig. 3a, b, c) is located at the highest stratigraphic levelrelative to the other types of paleochannels and is, therefore, labeledtype 3. In someplaces, these paleochannels are superposed by yardangs,suggesting that they were overlain by another MFF stratum that hasbeen mostly eroded (inset in Fig. 2) (Jacobsen and Burr, 2013). Someof these paleochannels are only partially exhumed (e.g., Fig. 3a, and c).These channels have a similar morphology (narrow width, low sinuos-ity, smooth surface) as the fully exhumed type 3 paleochannels, buttheir surfaces are raised only slightly above the surface of theplateau and large sections are still overlaid by yardangs. Theseexamples of poor exposure suggest that additional type 3 unexhumedpaleochannels may be contained within the volume of this plateau.Some of the type 3 paleochannels are characterized by multiple, lowsinuosity, interconnected channel belts, and show little evidence of lat-eral migration, which has been preserved in some Aeolis Dorsa deposits(Howard, 2009; Burr et al., 2010).

The western and north-western part of the network displays thinpaleochannels about 10–15 m in relief, with a similarly low sinuosity(average sinuosity= 1.1). The surfaces of these paleochannels, with-out the N – S striations (Fig. 3a and b), are smoother than those of theplateau. They also, however, appear qualitatively less continuousthan the type 3 paleochannels, indicating a more eroded state. Su-perposition relationships show that these paleochannels are locatedat a lower stratigraphic level than the type 3 paleochannels. Thus,these paleochannels are inferred to be older than the type 3paleochannels, and, based on direct superposition, are denotedtype 2 paleochannels. Like the type 3 paleochannels, they are also“thin” paleochannels (Burr et al., 2009).

The southern part of the network displays paleochannels (Fig. 3c)that are wider (a few tens of meters) and more sinuous (average sinu-osity = 1.3) than type 3 or type 2 paleochannels. These paleochannelssometimes consist only of lines of knobs and elongated buttes withflat or rounded tops, attesting to a much greater degree of erosionthan is shown by type 3 or type 2 paleochannels. These buttes have asimilar orientation as the yardangs located on top of the fan-shapedplateau and, where preserved, their flat surfaces display N – S striationssimilar to those yardangs. Based on their greater width, these examplesare classified as ‘flat’ (Burr et al., 2009). They are at a lower elevationthan the type 3 paleochannels, located on top of a knobby surface (sug-gested on the basis of texture and elevation to be continuous with sur-face C in Fig. 2). Some of them connect to the edge of the fan-shapedplateau on which the type 3 paleochannels are located, and, wherethey connect with the plateau, their surfaces are at a similar elevationas terrain B (Fig. 2). We postulate that they continue within the volumeof material beneath the plateau surface, so that they are stra-tigraphically lower and, therefore, older than the type 3 paleochannels.In addition, they are at a lower elevation than the type 2 paleochannels,and, although we do not observed direct superposition relationship be-tween these two paleochannel types, we infer them to be most likelyolder than the type 2 paleochannels, and denote them as type 1paleochannels.

Thus, we distinguish three distinct episodes of paleochannel forma-tion in the eastern fan-shaped network. If the eastern network wasdistributary, as previously interpreted (e.g., Irwin et al., 2005;McMenamin and McGill, 2005; Williams and Edgett, 2005) thesethree types of paleochannels, located in three areas and strata of thefan, point to three separate periods of flow and lobe switching. Channelsinuosity decreases with time. The fact that the stratigraphically lower,wider, type 1 paleochannels are more sinuous than the type 2 and 3

paleochannels suggests either amore shallow slope or a longer durationof flow and more mature network than the stratigraphically higher,straighter type 2 and 3 paleochannels.

4.1.2. Western networkLess well exposed than the eastern network, the western network

consists of two discernable types of paleochannels at two stratigraphiclevels (Figs. 1 and 3d) associated with a second fan-shaped flat plateau.

At the higher stratigraphic level, thin, low-relief paleochannels (afew meters in relief) are observed on the northern part of the plateau(Fig. 3d), as well as on an adjacent plateau to the southeast (Fig. 1).These thin paleochannels have a relatively low sinuosity (average sinu-osity = 1.15). Because they are located on the plateau, we infer thatthese paleochannels must be more recent than the stratigraphicallylower paleochannels, and denote them as type 2.

At the lower stratigraphic level, flat paleochannels with a greateraverage sinuosity (average sinuosity = 1.5) connect to the fan-shapedplateau around its edges. As in the southern area of the eastern network,some of these paleochannels, particularly south of the plateau (Fig. 1),have been eroded into individual elongated knobs. Paleochannels tothe north of the plateau have reliefs of 100 to 250 m; those to thesouth of the plateau have reliefs of 50 to 100m. Their average sinuosityismoderate (average sinuosity=1.4, close to the sinuosity of the type 1paleochannels of the eastern network). Owing to their stratigraphic lo-cation, these type 1 paleochannels must be older than the type 2paleochannels and must correspond to a different period of flow. Forthis largely unexhumed system, it is however difficult to infer if addi-tional distinct periods of flow occurred. Likewise, it is also difficult tostate whether specific types of paleochannels in the two different net-works correspond stratigraphically or temporally.

4.1.3. Inferred paleoflow directions

4.1.3.1. Eastern network. A variety of characteristics observed on theeastern network are indicative of distributary flow in a depositionalenvironment, with paleoflowdirections to the northeast. The distinctivemorphologies of the three types of paleochannels indicate at least threeepisodes of formation.We interpret the three main lobate network sec-tions as sedimentary lobes. The different morphologies and levels ofpreservation suggest lobe switching by avulsion, such as occurs onchannelized fans or deltas (e.g., Stouthamer et al., 2005).

The different degrees of exhumation of the type 3 paleochannels, thevisible overlapping relationships of some type 3 and type 2paleochannels, and the connection of the type 1 paleochannels to theedge of the plateau suggest that channel stacking occurs within the vol-umeof the plateau. Channel stacking is indicative of aggradation, causedby a surplus of sediments and/or changes in level of base level(e.g., Klingera et al., 2003; Salcher et al., 2010). Aggradation is one ofthe causes of avulsion, the abandonment of oneflowpath and formationof a new flow path, which is a characteristic of channels on deltas andfans. It occurs notably in the presence of local high rates of sedimenta-tion near the fixed entry point or apex, which may become an avulsionnode, i.e., the location of frequently repeated switching between chan-nels (Makaske, 2001). The bifurcation of the trunk channel in the south-western part of the eastern network suggests the presence of anavulsion node (Fig. 2). The convex profile of the network, which is un-likely to be primarily the product of erosion because the paleochannels'capping layer is preserved over both higher and lower sections of thenetwork, is also highly favorable to random avulsions. Over time, asthe delta or fan develops, avulsion nodes may become scattered overthe fan-shaped deposit (Makasaka, 2001), and numerous such potentialavulsion sites can be observed on the eastern network (Fig. 2).

The arcuate distalmargin of the plateau is outlined by a frontal scarp,a morphology observed in Gilbert-type deltas on Earth (e.g., Gilbert,1885) andMars (e.g., Hauber et al., 2013) that reflects base-level controlin a depositional system. We infer that the scarp, like the plateau, was

terrestrialfan

san

dde

ltas.N

egativeva

lues

represen

tado

wn-to-the

-north

tilt,p

ositiveva

lues

represen

tado

wn-to-the

-sou

thtilt.

geMod

eled

slop

ech

ange

(atne

tworklocation

andt=

15km

)

edLo

wer-crustal

flow

Erosion/de

position

mod

elErosion/de

position

mod

elwith

MFF

depo

sition

T=

15km

T=

30km

T=

60km

T=

15km

T=

30km

T=

60km

T=

15km

T=

30km

T=

60km

−0.03

°−

0.09

°−

0.05

°0.01

°0.05

°0.03

°0.11

°0.12

°0.06

°


exposed by erosion of an overburden, the remnants of which are visiblein the fewyardangs superposed on the type 3 paleochannels. Thebroad-ly arcuate planform of the scarp extending over 90°, however, is incon-sistent with a primarily erosional origin, e.g., of the headwaters of adendritic (tributary) system. The slightly serrated planform morpholo-gy of the scarp appears to be caused by erosion of locally less indurated(interchannel) sediments. This interpretation is consistent with inter-pretation of other similar features on Mars as Gilbert deltas (Gilbert,1885; Postma, 1990).

Whereas the longitudinal profiles of the channels slope mostlytoward the SW, we observe a change in slope direction at a distance of~35 km along the profiles (Fig. 4). From this point northward, the chan-nels are inclined toward the northeast with a slope of ~0.6°. The slopedirection of the channels at the distal part of the network is similar tothe slope direction of foreset beds in a delta. Foreset beds consist of sed-iments deposited on a sloping surface along the delta front, and dip inthe direction of flow (e.g., Gilbert, 1890; Rafferty, 2011). The dip angleof foreset beds varies from less than 1° in large marine deltas to morethan 20° in deltaswith coarse bedload (Rafferty, 2011). Because the sur-face of the channels is well-preserved along the entire profiles, it isunlikely that differential surface erosion caused the northward sloping.Potential collapse or compaction effects are also possible, but wouldlikely be evidence by features such as depressions or ‘windows’ intothe subsurface. Because such features are not observed, we infer thatcollapse was not a significant contributor to the change in slopedirection.

Thus, although some of the characteristics of the eastern network,such as avulsions, might be found in a tributary system, we find theevidence of aggradation, terminal scarp and putative foreset beds tobemost consistentwith a distributary origin. This conclusion is in agree-ment with previous interpretations, based on morphology and context,of paleoflow to the northeast (Irwin et al., 2005; McMenamin andMcGill, 2005; Williams and Edgett, 2005). A paleoflow direction to thenortheast requires that the original slope was down to the northeast.

erang

eof

chan

gesin

slop

eforea

chne

tworkba

sedon

slop

eva

lues

for

Inferred

original

slop

e(=

slop

efor

terrestrialfl

uvial

fans

andde

ltas)

Inferred

avera

slop

ech

ange

(dow

n-to-

the-south)

eMax

slop

eAve

rage

slop

eStan

dard

deviation

=Mea

sured

slop

e+

inferr

original

slop

e

°0.2°

0.14

°0.04

°Min:0

.01

Max

:2.5

Min:~

0.15

°Max

:~2.64

°°

0.38

°0.25

°0.09

°Min:~

0.26

°Max

:~2.75

°

4.1.3.2. Western network. Because of the poor degree of exhumation ofthewestern network, less evidence exists to infer its original flow direc-tion than for the eastern network. Whereas at least two episodes ofchannelized fluvial activity are inferred for this network (seeSection 4.1.2, above), the plateau displays multiple discrete segmentsthat we interpret as different depositional lobes (Fig. 1). These multiplelobes suggest flowwith lobe-switching during sedimentation in a fan ordelta. For these reasons, we consider this network, like the easternnetwork, is also more likely to be distributary than tributary (althougha tributary network with multiple discrete watersheds could be an al-ternative possibility). For a distributary network, the original slopewould have been down to the northeast.

On the basis of the fan-shaped plan form of these two networks, thepresence ofmultiple inferred avulsion nodes among channels and chan-nel stacking and overlapping relationships in the eastern network, weinfer fluvial deposition in a distributary network, with original flowdirections and slopes down to the northeast.

Table2

Paleocha

nnel

slop

emea

suremen

tsan

dpo

ssibl

Mea

suredslop

e

Num

berof

most

recent

paleo-

chan

nels

mea

sured

Min

slop

Easternne

twork

30.12

Western

netw

ork

40.13

4.1.4. Results of paleochannel slope analysis

4.1.4.1. Eastern network. For the better-exposed eastern network, com-bined analysis of the HRSC DTM, CTX DTM and MOLA data points overthree paleochannels (Fig. 4) shows that the current slope is down tothe southwest. The paleochannels for which slopes were derived werelimited to the best examples of the most recent channels (type 3 as de-scribed in Section 4.1.1). Measurements from the three independentdatasets agreewell, and give an average slope over these three channelsof 0.14° (number of channels N= 3, Table 2), with a standard deviationequal to 0.04.


4.1.4.2. Western network. The current slopes of the paleochannels of thewestern network are also down to the southwest (Fig. 4). Because thewestern network is not fully exhumed from beneath the fan-shapedplateau, we cannot measure the entire paleochannel slopes. However,the morphology (sinuosity, width and texture) and slope of thepaleochannels in the northern part of the plateau is similar to the mor-phology and slope of the paleochannels on the southern side of theplateau, suggesting that these segments form a single continuouspaleochannel. The approximate pathway of an unexhumed section ofpaleochannels under the surface of the plateau was drawn in planview (Fig. 4a) based on continuity with the northern paleochannelsand using a smooth arc to connect it to the southwestern paleochannels.The profile for these paleochannels was derived fromMOLA data pointsand HRSC DTM data over the paleochannels to the north and south ofthe plateau and using the approximate pathway of the unexhumedsection to give a minimum linear distance between the paleochannels.We calculate for these channels an average slope of 0.25° (number ofchannels N = 4, Table 2), with a standard deviation equal to 0.1. Be-cause the approximate distance between the northern and southernpaleochannels on either side of the fan shape is drawn as theminimumreasonable distance, the slope is likely a maximum value. The slope foronly the exposed paleochannels to the southwest of the fan-shaped pla-teau, however, is similar to the value for this full length. The result of theslopemeasurements shows that the elevation is higher over the plateauthan either to the north or the southwest of the plateau.

4.2. Estimate of the change in slope

Determining the change in slope for these networks requires know-ing their original slopes. Given their inferred formation as deltas or allu-vial fans, we use the range of slopes for these landforms on Earth as theoriginal slopes for these networks. On Earth, the average slope varieswith the mode of deposition, fans dominated by debris flows beingsteeper than those dominated by normal fluvial deposition (e.g., Wellsand Harvey, 1987; Blair and McPherson, 1994a, b; Parker et al., 1998;Harvey et al., 2005; Moore and Howard, 2005). In these two Martiannetworks, the morphological evidence points to fluvial deposition. We,therefore, consider the original slopes of these networks as being similarto those of either fluvial fans dominated by channelized flow or deltas.

To derive minimum and maximum slope changes for each network,we added the average measured slope from the horizontal to the mini-mum and maximum slope estimates for terrestrial deltas and alluvialfans dominated by channelized flow (= inferred original slope, Fig. 7).

Fig. 7. Calculation of the change in slope. Min/max change in slope (c) =measured slopefrom the horizontal (a) + min/max slope estimates for terrestrial deltas and fluvialfans (b).

The average measured down-to-the-south slopes for the eastern andwestern networks are 0.14° and 0.25°, respectively. Terrestrial fluvialfans and deltas generally have slopes within the 0.01 – 2.5° range(Coleman and Wright;, 1975; Blair and McPherson, 1994a; Harvey,1997; McCarthy et al., 1997; Moore and Howard, 2005; Hashimotoet al., 2008). With these gradients, the change in slope ranges between0.15° and 2.64° to the south for the eastern network, and between 0.26°and 2.75° for the western network (Table 2).

Because Mars has a lower gravity than Earth, the transported sedi-ments would have a lower settling velocity on Mars and, thus, may becarried further in Martian channels than they are in comparable terres-trial channels (Komar, 1980; Burr et al., 2006), possibly resulting in adecrease in gradient up to 40% for equivalent water and sedimentsupply, depending on bedload, threshold and suspended load condi-tions (Craddock and Howard, 2002). Therefore, the slopes of delta andalluvial fans in Aeolis Dorsamay be less steep than the slope of terrestri-al alluvial fans and deltas. In that case, our estimated slope changes arelikely to be maximum values.

4.3. Comparison of modeling results with slopes measurements

4.3.1. Lower crustal flow modelWith the lower crustal flow model (Fig. 8), we obtain a central

region ~200 km wide (i.e. ~100 km on each side of the inflectionpoint of the dichotomy boundary scarp) in which the slopes are tiltedsouthwards, whereasmore distally, tilting is to the north. At the locationof the networks, slopes modified by lower crustal flow would be modi-fiedby tilting to the north. This direction of slope change does notmatchour data, which show a down-to-the-south tilt for both networks(Fig. 8).

4.3.2. Erosion/deposition modelsThe erosion/deposition models show the ~200 km wide central re-

gion to be characterized by northerly tilt (Fig. 8) whereas fartheraway, including the network locations, the tilt is down to the south, con-sistent with the inferred slope changes. The erosion/deposition modelwith MFF deposition and with an elastic thickness of T = 30 km(Table 2), consistent with Watters (2003a), shows the best fit with themeasurements (~0.12° tilt at the location of the networks, Fig. 8 andTable 2). The result obtained with an elastic thickness of T = 15 km,however, is very close to the result obtained for T = 30 km (~0.11° tiltwith MFF deposition). With an elastic thickness of T = 60 km, the tiltis ~0.03° (without MFF deposition) and 0.06° (with MFF deposition)at the location of the networks.

The modeled slope change is for the original surface, not of the sur-face incorporating the deposited MFF material. Based on stratigraphicstudies and age estimates (Burr et al., 2009; Kerber and Head, 2010;Zimbelman and Scheidt, 2012; Jacobsen and Burr, 2013), the two net-works are located near the base of the MFF, so that the change inslope at this location should be comparable with the change in slopeof the original surface.

If redeposition of highlands materials in the northern lowlands wasnot extensive, as discussed in the Methods and data section, then theloading by the MFF would play a more significant role than the loadingby the eroded highlandsmaterial. In either case, the slope changewouldbe down-to-the-south and, therefore, consistent with the direction ofourmeasured change in slope. The estimated amount ofmaterial, corre-sponding to MFF deposition, is greater than the estimated amount ofmaterial deposited from the highlands alone (Fig. 6). Because themodeled change in slope resulting frombothMFF andhighlandmaterialdeposition is similar at the location of the networks to the modeledchange in slope resulting from highland material deposition alone(Fig. 8), a model that would considerMFF deposition as the only surfaceload on the lowlands is likely to yield very similar results.


4.4. Response timescale of the deformation

In the erosion/deposition model, the response is assumed to beelastic (i.e. instantaneous). Therefore, the rate of slope change is deter-mined by the rate of erosion and deposition. In the lower crustal flow

model, the response timescale depends on the rate at which crustalmaterial can flow, which depends on temperature structure, crustalthickness, etc. For our baseline lower crustal flow model, Fig. 5 showshow the slope change varies with time. Most of the slope changehappens early (during the first 30 Myr), because lower crustal flow isinitially fast and then slows down, as the stresses driving flow arereduced.

5. Discussion

5.1. Inferred vs modeled slope change, and potential causes fordiscrepancies

The erosion models for all elastic thicknesses match the direction ofinferred down-to-the south slope change for the paleochannels in bothnetworks (Fig. 8). In addition, the erosionmodelswith an elastic thicknessof 30 km (consistent with previous work; Watters, 2003a) are the bestmatch for the magnitude of inferred slope change of the paleochannels.

For the better exposed (eastern) network (Table 2), we were ableto derive elevations for fully exhumed paleochannels. Moreover,because of their similar morphology and preserved capping layer(Section 4.1.1), we are confident that the inverted channels overwhich the slopesweremeasured formed during the same episode of de-position. Therefore, because of the higher level of preservation of thepaleochannels in the eastern network, we consider the slope measure-ments on the eastern network to be more reliable than those on thewestern network. Although both networks match the direction ofslope change for the erosion/deposition model (for all elastic thick-nesses), the modeled magnitude of the change was a better match(0.03°minimumdifference betweenmodeled and inferred slope chang-es) for the better exposed eastern network (Table 2). For the less ex-posed western network, the modeled and inferred magnitudes of thechange differed by a minimum of 0.14°. Therefore, the network whoseslope measurements we are most confident about is the best matchand the western network may be a poorer fit simply because our dataover this network are less reliable. That is, because the original averageslope for the western paleochannels may be a maximum value (seeSections 4.1.4.2 and 4.2), the difference between the data and themodel may also be a maximum.

In both cases, our modeled slope change is less than the measuredslope change at the location of the network. This underestimation bythe models may be even more significant if we consider the timescaleof the erosion/deposition and lower crustal flow process compared tothe timescale of formation of the networks. In the lower crustal flowmodel, equilibrium is reached within ~30Myr (see Fig. 5) and probablystarted during the Noachian, possibly before the formation of the net-works. For the erosion/deposition model, the response is effectivelyinstantaneous and, therefore, depends on the rate and timescale of ero-sion/deposition. Loading of the lowlands and MFF deposition may havestarted before formation of the networks and definitely continued

Fig. 8. The plots show, for elastic thickness T = 15 km, T = 30 km and T = 60 km, thechange in slope (in degrees) according to the lower-crustal flow model (light gray), andthe erosion model, with (black) and without (medium gray) MFF deposition. North is tothe right and the center of the dichotomy boundary (i.e. inflection point of the scarp) is lo-cated at 750 km (shown by the X marker). Positive slope values represent slopes to thesouth,whereas negative values represent slopes to the north. The data points representingthe range of inferred change in slope (white for the eastern network, black for thewesternnetwork) are located at the average distance of the networks from the center of the di-chotomy (~275 km) and are showed slightly apart for better readability. At the locationof the networks, for Te = 15 km, the change in slope is 0.01° for the erosion model with-out MFF deposition, 0.11 for the erosion model with MFF deposition, and −0.03° for thelower crustal flow model. For Te = 30 km, the change in slope is 0.12° for the erosionmodel with MFF deposition, 0.05° for the erosion model without the MFF depositionand −0.09° for the lower crustal flow model. For Te = 60 km, the change in slope is0.03° for the erosion model without MFF deposition, 0.06° for the erosion model withMFF deposition, and −0.05° for the lower crustal flow model.


afterwards, as evidenced by the burial of the networks beneath MFF. Bycalculating the net rate of sedimentation in the Aeolis Dorsa region, Kiteet al. (2013) estimate at 1–20Myr theminimum timescale of fluvial ac-tivity in this area. For the erosion/deposition model, the tilting wasprobably already underway by the time the networks formed and be-came inverted. For lower crustalflow, the tiltingwas likely already com-pleted by the time the network formed. Therefore, the deformationrecorded by the networks may only be a fraction of the actual deforma-tion. Because the slope change estimated by our models is already lessthan the slope change recorded by the networks, ourmodelsmay highlyunderestimate the slope change.

This discrepancymay be explained by the sensitivity of themodel toinput parameters, and particularly in the case of the erosion/depositionmodel, to the amplitude and geometry of the erodedmaterial. Modifica-tions in the estimated amount of eroded/deposited material, width of

Fig. 9. Sensitivity of results to varying input parameters. Here Te=30 kmand the baselineassumes symmetrical triangular erosion/deposition. (a) Effect of changing amplitude(thickness) of eroded/deposited region. (b) Effect of changing half-width. (c) Effect ofload geometry.

the eroded region and load geometry lead to variations in the amplitudeof slope change. A larger amount of eroded material results in largerslope changes (Fig. 9a), as does an increase in the width of the erodedregion (Fig. 9b). Similarly, a rectangular region of erosion also resultsin greater slope changes because the total amount of material removedincreases (Fig. 9c). In addition to issues relative to the amount andgeometry of the eroded material, neglecting the boundary scarp retreatis another possible source of error in our results. In all cases, however,the qualitative results are very similar, as the direction of slope changeremains the same. Future mapping (e.g., Zimbelman and Scheidt,2012) and analysis could help to determine more precisely the amountof eroded and deposited materials, as well as the amount and timing ofslope dichotomy boundary retreat at this location.

Some of the discrepancy between modeled slope change and in-ferred slope change based on current measurements may also resultfrom the effect of the lower gravity on Mars. The original slopes ofthese fan-shaped deposits were likely shallower under Martian gravity(Komar, 1980; Craddock and Howard, 2002). Shallower original slopeswould give a better fit to the modeled slope change.

Errors arising during data acquisition and registrationmay also part-ly explain some of the discrepancy (Fig. 8, Table 2) between modeledand measured slope change, although we consider that the techniqueused to co-register the different datasets significantly reduces the possi-bility of co-registration errors. Other possible sources of errors in themeasurements are data artifacts, whichmay be caused by various issuesincluding spacecraft motion, camera noise or errors in the stereo corre-lation process itself. In push-broom cameras like CTX and HiRISE (Malinet al., 2007; McEwen et al., 2007), distortions and waviness may occuralong the scan direction. Because the paleochannels trend roughlyeast–west, perpendicular to the flight path/image direction, push-broom distortions are unlikely to be a major issue. HiRISE and CTXDTMs can have other geometric distortions caused by a number ofinstrument and spacecraft effects (e.g., Mattson et al., 2009; Sheanet al., 2011), as well as mismatches and offsets in the stereo correlationprocess itself. The topographic profiles, however, are consistent acrossall topographic datasets, which suggests that data artifacts do not signif-icantly affect them. We therefore consider that data artifacts and regis-tration errors are not a likely explanation for the discrepancy betweenmodeled and measured slope change.

The discrepancy betweenmodeled andmeasured slope change mayalso result from collapse processes.

5.2. Possible alternative processes: collapse caused by mass wasting andvolatile sublimation

As described in Lefort et al. (2012), multiple potential causes existfor the observed undulations in some of the inverted fluvial featuresin Aeolis Dorsa. These causes, including removal of underlying materialby sublimation of volatiles or mass wasting, may also have contributedto the change in slope. However, inspection of the spatial relationshipsin this region suggests that any contribution by these other causes tothe inferred slope reversals was less important than tectonic deforma-tion caused by erosion.

An approximately east–west-oriented trough, located south of thetwo fan-shaped networks (Fig. 10), may be evidence for collapse. Onthe southern side of the through, adjacent to the highlands, is a relative-ly flat bench that is at a similar elevation as both networks. This benchshows unexhumed paleochannels (Burr et al., 2010). This observationsuggests that this area may have been a continuous surface before de-veloping into a trough. Tanaka et al. (2005) mapped a geologic unitalong and below the dichotomy boundary called theNepenthesMensaeunit, which consists of rectangular knobs and mesas of highland rocks,spaced several kilometers apart and surrounded by plain material.This unit is interpreted as “highland rocks fractured and collapsedduring Late Noachian and extending into Early Hesperian due to basalsapping of volatiles and mass wasting” (Tanaka et al., 2005). Although

Fig. 10. Location of the trough south of the study area. Dotted line:−2100m topographic contour delineating the trough, circles: location of thewestern and eastern networks. North is tothe top.


the map does not extend south of the equator, an extension of this unitmay occur, embayed and buried by the MFF, along the dichotomyboundary in the study area. The ~ E–W-oriented trough south of thetwo networks may be part of this equivalent unit and have formedsimilarly by collapse caused by sapping andmasswasting. If the inferredcollapse of this region occurred after formation of the paleochannels,then this collapse could have contributed to or caused the paleochannelslope reversals. That is, if collapse occurred, then the total inferred slopechange (as plotted in Fig. 8) would be either a combination of collapse-induced reversal and erosion/deposition-induced reversal or possiblyentirely caused by collapse. The possible effects of collapse could be anexplanation for why the inferred slope change is larger than themodeled erosion/deposition slope change. If the trough area corre-sponds indeed to the Nepenthes Mensae unit and collapse contributedto the slope reversal, then this implies that the paleochannels formedduring or before the late Noachian–early Hesperian.

The geographic relationships, however, do not support a strong con-tribution to the inferred slope reversals by a possible collapse of terrainin the location of the trough. To the south of the eastern fan-shapednetworks, the trough contains knobs and mesas, as does the NepenthesMensae unit, but these knobs and mesas contrast with those in theNepenthes Mensae by being very closely juxtaposed or packed directlyadjacent to each other and tending to be elongated in onedirection. Thisdifference in morphology could indicate that the putative collapseprocess did not proceed as far within our study area as it did elsewhere

Fig. 11. Timeline of events in the study area from the forma

in the northern plains region mapped by Tanaka et al. (2005). Alterna-tively, this difference in morphology might indicate that the troughformed by some other mechanism (e.g. deep tectonic processes) orthat this area has been affected by different secondary processes (e.g.different erosion processes) than those that occurred in the NepenthesMensae. The trough lacks morphological evidence for wide-spreadlandslides or rockfalls, which are commonly associated with collapse.Neither does any morphological evidence exist for collapse south ofthe western network: the terrain between the network and the dichot-omy boundary is a scoured shelf of planar material eroded by the wind,with one largemesa on this shelf showing evidence for landslides off itssides. Whereas the gridded MOLA topography for the eastern networkshows a decrease in elevation southward to this trough, the griddedtopography for the western network shows a consistent increase inelevation from the base of the network southward to the dichotomyboundary. Thus, whereas a putative collapse could have had an effecton the eastern network, its effect on the western network, if any,would have been much less significant.

In summary, we do not see conclusive evidence for the possibilitythat collapse was a significant factor in the inferred slope reversal,particularly for the western network. We conclude that, although col-lapse or loss of volatiles is a possible (contributing) mechanism for theinferred slope reversals of our two study networks, it is likely not suffi-cient for causing themagnitude of the reversed slopes. The possibility ofadditional reversal caused by collapse could account for the difference

tion of the dichotomy boundary to the slope reversal.


between the slope changes inferred from morphology and the slopechanges derived from models of tectonic processes alone.

5.3. Comparison with previous results

Our results are inconsistent with the conclusions of Nimmo (2005),who found that the predicted locations of compression and extensionfrom the lower crustal flow model matched the locations of previouslymapped thrust faults (Watters, 2003a,b) and extensional faults(McGill and Dimitriou, 1990; McGill et al., 2004), whereas the predic-tions of the erosion model did not match those locations. Stresses caus-ing such thrust faults, however, may result frommechanisms other thanlithospheric deformation, such as long-term thermal contraction (as onthe Moon and Mercury, Solomon and Chaiken, 1976; Kirk andStevenson, 1989; Watters et al., 2004; Nahm and Schultz, 2011;Watters et al., 2012). It is also possible that lower crustal flow and ero-sionwere both operating, but on different time scales. The paleochannelnetworks used in this study may be as young as late Hesperian in age,based on age-dating of the MFF in which they are located (Burr et al.,2009; Kerber and Head, 2010; Zimbelman and Scheidt, 2012) andthus may be younger than the mapped thrust and extension faults, in-ferred to be late Noachian–early Hesperian (McGill and Dimitriou,1990; Watters and Robinson, 1999; Watters, 2003b; McGill et al.,2004; Tanaka, 2004). Because lower crustal flow is most rapid earlyon, when the crust is hot and the thickness gradients are large(Nimmo and Stevenson, 2001; Nimmo, 2005), this process may havecaused early formation of the thrust faults and may have later beenfollowed by a response of the crust to highlands erosion, and/or possiblyto the deposition of theMFF, causing the inferred slope reversals. Volca-nic loading has caused crustal deformations in other areas of Mars, no-tably on Tharsis, where graben systems are interpreted as the result oftectonic deformation of the Tharsis rise associated with either upwell-ing or loading (Sleep and Phillips, 1985; Tanaka and Davis, 1988;Banerdt et al., 1992; Lowry and Zhong, 2003). Aeolis Dorsa, along withthe northern plains (Tanaka et al., 2001) and Utopia (e.g., Searls andPhilips, 2004), may be an example of tectonic deformation caused byloading of sedimentary deposits.

6. Implications

The results of this study suggest that near Aeolis Dorsa and TerraCimmeria, loading of the northern plains, probably by a combinationof erosion/deposition andMFF deposition was a major contributing fac-tor to the slope reversal. The location of these inverted channels withinthe lower member of MFF, whose emplacement began no later than theHesperian (Kerber andHead, 2010; Zimbelman and Scheidt, 2012), sug-gests that this topographicmodification of the dichotomy boundary andadjacent terrains might have continued throughout at least part of theearly to mid-Hesperian.

The timing of the events that modified the dichotomy boundaryis largely uncertain. From this study, we propose a timeline for thegeologic events that occurred at this location (Fig. 11). First, thedichotomy boundary was formed during the pre-Noachian or earlyNoachian (e.g., Watters, 2003a; Frey, 2004; Watters et al., 2007;Andrews-Hanna et al., 2008). Then lower crustal flow led to thrustand extensional faults (Nimmo, 2005). This deformation likely occurredshortly after the dichotomy formed, when pressure and thermal gradi-ents were highest. Any such flow must have been largely completeprior to emplacement of the channels, because the effect of flow onslope change is opposite to that observed (see Fig. 8). Fluvial erosionof material from the adjacent highlands and redeposition in the low-lands likely started during the late Noachian and progressively de-creased throughout the Hesperian (Craddock and Maxwell, 1993;Hynek and Phillips, 2001), although see Irwin et al. (2011) for an alter-native view on the magnitude of this process. During or prior to themid-Hesperian, the MFF deposition began in the lowlands (Burr et al.,

2009; Kerber and Head, 2010; Zimbelman and Scheidt, 2012), progres-sively adding to any load of material eroded from the highlands. Mag-matic surface deposition and intrusion during that period may alsohave contributed to the loading of the northern plains (Head et al.,2002; Tanaka et al., 2005; Irwin andWatters, 2010), although itsmagni-tude and timing at this location are uncertain. Fluvial networks formedprior to or during the earliest period of MFF deposition, as indicated bytheir stratigraphic position at the base of theMFF. Timing constrains forthe onset on these fluvial episodes are currently lacking, although Kiteet al. (2013) posit a late Noachian/early Hesperian timing for some ofthe paleochannels deposits. The networks underwent multiple genera-tions of flow and lobe switching, as well as variation of flow duration,deduced from the variations in paleochannel sinuosity. The networkswere then indurated, and subsequently inverted (Pain et al., 2007;Burr et al., 2009, 2010; Zimbelman and Griffin, 2010). Finally, surficialloading resulted in lithospheric flexure that, possibly assisted bycollapse, resulted in the slope reversals of the inverted fluvial featuresdocumented here.

7. Conclusion and future work

This work illustrates the potential utility, and the potential issues, ofusing fluvial analyses to investigate large scale landscape modificationby tectonic processes on Mars. We conclude that longitudinal slopereversals occurred in two fan-shaped networks along the hemisphericaldichotomy boundary. For our modeling, we favor tectonic deformationresulting from loading of the crust by sedimentary erosion and possiblydeposition as the most significant factor in these slope reversals. Possi-ble collapse processes south of the networks and potential magmaticloading may also have contributed to the magnitude of the slopereversals.

Future work is necessary to untangle the relative timing and con-tributions of these possible causes. Refinements of our geophysicalmodeling should be possible with regional mapping, which wouldallow for more accurate estimates of sediment volumes and loading.If supported by mapping, additional loading mechanisms, such asglacial (Kite et al., 2011) or magmatic loading (Head et al., 2002;Tanaka et al., 2003; Irwin and Watters, 2010), may also improveour modeling. Our geophysical model suggests that the tectonic pro-cesses leading to slope reversals should have affected an area 400 to700 km wide on each side of the dichotomy boundary in that region.Therefore, we expect that slope change has affected other landformsin the area, so that identification and analysis of other inverted chan-nel profiles and/or tilted crater floors may provide data to test ourresult. Identification of slope modification, however, is dependenton the accurate knowledge of the original slope, which is difficultto assess in this region because of the complex stratigraphy, exten-sive erosion, and partial exhumation of the landscape.

Channelized fans and deltas are ideal for the study of slope reversal,as their original slope directionmay be determined, and in certain cases,a range of original gradientsmay be estimated. Future identification andtopographic analysis of other channelized fans and deltas along thedichotomy boundary, as well as extended regional investigation (nota-bly regional structural mapping, documentation of extended deforma-tion, and paleochannel stratigraphy) may help to further test ourconclusions for this localized study.

Acknowledgments

This work was supported by NASA MDAP grant NNX08BA25G toDMB. We thank Ross Beyer for his help with the Ames Stereo Pipelineand Dominik Neu for creating the HRSC DTM.We also thank Alan How-ard, Caleb Fassett and Susan Conway for their helpful reviews, aswell asJoe Levy, Ross Irwin, and Ken Tanaka for helpful comments on an earlyversion of this work.


References

Anderson, J.A., Sides, S.C., Soltesz, D.L., Sucharski, T.L., Becker, K.J., 2004. Moderniza-tion of the integrated software for imagers and spectrometers. Lun. Planet. Sci.XXXV, 2039.

Andrews-Hanna, J.C., Zuber, M.T., Banerdt, B., 2008. The Borealis basin and the origin ofthe Martian crustal dichotomy. Nature 453, 1212–1215. http://dx.doi.org/10.1038/nature07011.

Banerdt,W.B., Golombek, M.P., Tanaka, K.L., 1992. Stress and tectonics onMars. In: Kieffer,H.H., et al. (Eds.), Mars. Univ. of Ariz. Press, Tucson, pp. 249–297 chap. 8.

Bendick, R., Flesch, L., 2007. Reconciling lithospheric deformation and lower crustal flowbeneath central Tibet. Geology 35, 895–898.

Bird, P., 1991. Lateral extrusion of lower crust from under high topography, in the isostaticlimit. J. Geophys. Res. 96, 10,275–10,286.

Blair, T.C., McPherson, J.G., 1994a. Alluvial fans and their natural distinction from riversbased on morphology and, hydraulic processes, sedimentary processes and facies as-semblages. J. Sediment. Res. A64, 450–489.

Blair, T.C., McPherson, J.G., 1994b. Alluvial fan processes and forms. In: Abrahams, A.D.,Parsons, A.J. (Eds.), Geomorphology of Desert Environments. Chapman & Hall,London, pp. 354–402.

Bradley, B.A., Sakimoto, S.E.H., Frey, H., Zimbelman, J.R., 2002. Medusae Fossae Formation:new perspectives from Mars Global Surveyor. J. Geophys. Res. 107 (E8).

Burr, D.M., Emery, J.P., Lorenz, R.D., Collins, G.C., Carling, P.A., 2006. Sediment transport byliquid overland flow: application to Titan. Icarus 181, 235–242. http://dx.doi.org/10.1016/j.icarus.2005.11.012.

Burr, D.M., Enga, M.-T., Williams, R.M.E., Zimbelman, J.R., Howard, A.D., Brennand, T.A.,2009. Pervasive aqueous paleoflow features in the Aeolis/Zephyria Plana region,Mars. Icarus 200, 52–76. http://dx.doi.org/10.1016/j.icarus.2008.10.014.

Burr, D.M., Williams, R.M.E., Wendell, K.D., Chojnacki, M., Emery, J.P., 2010. Inverted fluvi-al features in the Aeolis/Zephyria Plana region, Mars: formation mechanism and ini-tial paleodischarge estimates. J. Geophys. Res. 115 (E07011). http://dx.doi.org/10.1029/2009JE003496.

Chen, Y.-C., Sung, Q., Chen, C.-N., Jean, J.-S., 2006. Variations in tectonic activities of thecentral and southwestern foothills, Taiwan, inferred from river hack profiles. Terr.Atmos. Ocean. Sci. 17 (3), 563–578.

Christensen, P.R., et al., 2001. Mars Global Surveyor thermal emission spectrometer ex-periment: investigation description and surface science results. J. Geophys. Res. 106(p. 23,823–23,871).

Christensen, P.R., Jakosky, B.M., Kieffer, H.H., Malin, M.C., McSween Jr., H.Y., Nealson, K.,Mehall, G.L., Silverman, S.H., Ferry, S., Caplinger, M., Ravine, M., 2004. The ThermalEmission Imaging System (THEMIS) for the Mars 2001 Odyssey Mission. Space Sci.Rev. 110, 85–130 http://dx.doi.org/10.1023/B:SPAC.0000021008.16305.94.

Coleman, J.M., Wright, L.D., 1975. Modern river deltas. Variability of processes and sandbodies. In: Broussard, M.L. (Ed.), Deltas, Models for Exploration. Houston GeologicalSociety, Houston, TX, pp. 99–149.

Craddock, R.A., Howard, A.D., 2002. The case for rainfall on a warm, wet early Mars.J. Geophys. Res. 107 (E11), 5111. http://dx.doi.org/10.1029/2001JE001505.

Craddock, R.A., Maxwell, T.A., 1993. Geomorphic evolution of the Martian highlandsthrough ancient fluvial processes. J. Geophys. Res. 98, 3453–3468.

Demoulin, A., 1998. Testing the tectonic significance of some parameters of longitudinalriver profiles: the case of the Ardenne (Belgium, NW Europe). Geomorphology 24(2–3), 189–208.

Frey, H.V., 2004. Impact Constraints on the Age and Origin of the Crustal Dichotomy onMars. Workshop on Martian, Hemispheres, # 4012.

Gaddis, L., et al., 1997. An overview of the Integrated Software for Imaging Spectrometers(ISIS). Lun. Planet. Sci. XXVIII, 1226.

Gilbert, G.K., 1885. The topographic features of lake shores. U.S. Geol. Surv. Annu. Rep. 5,75–123.

Gilbert, G.K., 1890. Lake Bonneville. USGS Monogr. 1 (438 pp.).Greeley, R., Guest, J.E., 1987. Geological Map of the Eastern Equatorial Region of Mars. U.S.

Geol. Surv. Misc. Invest. Mapp. I-102B.Guest, A., Smrekar, S.E., 2005. Relaxation of the Martian dichotomy boundary: faulting in

the Ismenius Region and constraints on the early evolution of Mars. J. Geophys. Res.110 (E12S25). http://dx.doi.org/10.1029/2005JE002504.

Gwinner, K., Scholten, F., Preusker, F., Elgner, S., Roatsch, T., Spiegel, M., Schmidt, R.,Oberst, J., Jaumann, R., Heipke, C., 2010. Topography of Mars from global mappingby HRSC high-resolution digital terrain models and orthoimages: characteristicsand performance. Earth Planet. Sci. Lett. 294, 506–519.

Harrison, S.K., Balme, M.R., Hagermann, A., Murray, J.B., Muller, J.-P., 2010. Mapping Me-dusae Fossae Formation materials in the southern highlands of Mars. Icarus 209(2), 405–415.

Harrison, S.K., Balme, M.R., Hagermann, A., Murray, J.B., Muller, J.-P., Wilson, A., 2013.A branching, positive relief network in the middle member of the Medusae Fos-sae Formation, Equatorial Mars — evidence for sapping? Planet. Space Sci. 85(142).

Harvey, A.M., 1997. The role of alluvial fans in arid-zone fluvial systems. In: Thomas,D.S.G. (Ed.), Arid Zone Geomorphology; Process, Form and Change in Drylands, 2ndedn. Wiley, Chichester, pp. 231–259.

Harvey, A.M., Mather, A.E., Stokes, M., 2005. Alluvial fans: geomorphology, sedimentolo-gy, dynamics. Geol. Soc. Lond. Spec. Publ. 251, 1–7.

Hashimoto, A., Oguchi, T., Hayakawa, Y., Lin, Z., Saito, K., Wasklewicz, T.A., 2008. GISanalysis of depositional slope change at alluvial-fan toes in Japan and the AmericanSouthwest. Geomorphology 100, 120–130.

Hauber, E., Platz, T., Reiss, D., Le Deit, L., Kleinhans, M.G., Marra, W.A., de Haas, T.,Carbonneau, P., 2013. Asynchronous formation of Hesperian and Amazonian-aged

deltas on Mars and implications for climate. J. Geophys. Res. Planets 118,1529–1544. http://dx.doi.org/10.1002/jgre.20107.

Head, J.W., Kreslavsky, M.A., 2001. Medusae Fossae Formation as volatile-rich sedimentsdeposited during high obliquity: an hypothesis and tests. Conference on the Geo-physical Detection of Subsurface. Water on Mars, Lunar and Planet. Inst., Houston,Tex., # 7053.

Head, J.W., Kreslavsky, M.A., 2004. Medusae Fossae Formation: icerich airborne dust de-posited during periods of high obliquity? Lun. Planet. Sci. 35, 1635.

Head III, J.W., Kreslavsky, M.A., Pratt, S., 2002. Northern lowlands of Mars: evidence forwidespread volcanic flooding and tectonic deformation in the Hesperian Period.J. Geophys. Res. 107 (E1), 5004. http://dx.doi.org/10.1029/2000JE001445.

Holbrook, J., Schunun, S.A., 1999. Geomorphic and sedimentary response of rivers totectonic deformation: a brief review and critique of a tool for recognizing subtleepeirogenic deformation in modern and ancient settings. Tectonophysics 305,287–306.

Howard, A.D., 2009. How to make a meandering river. Proc. Natl. Acad. Sci. 106 (41),17245–17246.

Hynek, B.M., Phillips, R.J., 2001. Evidence for extensive denudation of the Martian high-lands. Geology 29, 407–410.

Hynek, B.M., Phillips, R.J., Arvidson, R.E., 2003. Explosive volcanism in the Tharsis region:global evidence in the Martian geologic record. J. Geophys. Res. 108 (E9), 5111.http://dx.doi.org/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 (E11006).http://dx.doi.org/10.1029/2010JE003658.

Irwin, R.P., Watters, T.R., Howard, A.D., Zimbelman, J.R., 2004. Sedimentary resurfacingand fretted terrain development along the crustal dichotomy boundary, AeolisMensae, Mars. J. Geophys. Res. 109. http://dx.doi.org/10.1029/2004JE002248 (issn:0148-0227).

Irwin, R.P., Maxwell III, T.A., Howard, A.D., Craddock, R.A., Moore, J.M., 2005. An intenseterminal epoch of widespread fluvial activity on early Mars: 2. Increased runoff andpaleolake development. J. Geophys. Res. 110 (E12S15). http://dx.doi.org/10.1029/2005JE002460.

Irwin, R.P., Craddock III, R.A., Howard, A.D., Flemming, H.L., 2011. Topographic influenceson development ofMartian valley networks. J. Geophys. Res. 116 (E02005). http://dx.doi.org/10.1029/2010JE003620.

Jacobsen, R.E., Burr, D.M., 2013. Local-scale Stratigraphy of Inverted Fluvial Features in Ae-olis Dorsa, Western Medusae Fossae Formation, Mars. 44th Lun. Planet. Sc. Conf., #2165.

Jain, V., Sinha, T.R., 2005. Response of active tectonics on the alluvial Baghmati River, Hi-malayan foreland basin, eastern India. Geomorphology 70, 339–356.

Kerber, L., Head, J.W., 2010. The age of the Medusae Fossae Formation: evidence ofHesperian emplacement from crater morphology, stratigraphy, and ancient lavacontacts. Icarus 206, 669–684. http://dx.doi.org/10.1016/j.icarus.2009.10.001.

Kirk, R.L., Stevenson, D.J., 1989. The competition between thermal contraction anddifferentiation in the stress history of the moon. J. Geophys. Res. 94,12133–12144.

Kite, et al., 2011. Evidence for Melt-fed Meandering Rivers in the Gale–Aeolis–ZephyriaRegion. Mars. 43th Lun. Planet. Sc. Conf., #2778.

Kite, E.S., Lucas, A., Fassett, C.I., 2013. Pacing Early Mars river activity: embedded craters inthe Aeolis Dorsa region imply river activity spanned ≥ (1–20) Myr. Icarus 225,850–855.

Klingera, Y., Avouacb, J.P., Bourlesc, D., Tisneratd, N., 2003. Alluvial deposition and lake-level fluctuations forced by Late Quaternary climate change: the Dead Sea case exam-ple. Sediment. Geol. 162, 119–139.

Komar, P., 1980. Modes of sediment transport in channelized water flows with ramifica-tions to the erosion of the Martian outflow channels. Icarus 42, 317–329.

Lahiri, S.K., Sinha, R., 2012. Tectonic controls on the morphodynamics of the Brahma-putra River system in the upper Assam valley, India. Geomorphology 169–170,74–85.

Larue, J.-P., 2008. Effects of tectonics and lithology on long profiles of 16 rivers of thesouthern Central Massif border between the Aude and the Orb (France). Geomor-phology 93 (3–4), 343–367.

Lefort, A., Burr, D.M., Beyer, R.A., Howard, A.D., 2012. Inverted fluvial features in theAeolis–Zephyria Plana, western Medusae Fossae Formation, Mars: evidence forpost-formation modification. J. Geophys. Res. 117 (E03007). http://dx.doi.org/10.1029/2011JE004008.

Lowry, A.R., Zhong, S., 2003. Surface versus internal loading of the Tharsis rise, Mars.J. Geophys. Res. 108 (E9), 5099. http://dx.doi.org/10.1029/2003JE002111.

Makaske, B., 2001. Anastomosing rivers: a review of their classification, origin and sedi-mentary products. Earth Sci. Rev. 53, 149–196.

Malin, M.C., et al., 2007. Context Camera Investigation on board the Mars ReconnaissanceOrbiter. J. Geophys. Res. 112 (E05S04). http://dx.doi.org/10.1029/2006JE002808.

Mandt, K.E., de Silva, S.L., Zimbelman, J.R., Crown, D.A., 2008. The origin of the MedusaeFossae Formation, Mars: insights from a synoptic approach. J. Geophys. Res. 113(E12011). http://dx.doi.org/10.1029/2008JE003076.

Mattson, S., et al., 2009. HIJACK: correcting spacecraft jitter in HiRISE images of MARS. Eur.Planet. Sci. Congr. 4 (Abstract EPCS2009-604-1).

McCarthy, T.S., Barry, M., Bloem, A., et al., 1997. The gradient of the Okavango fan,Botswana, and its sedimentological and tectonic implications. J. Afr. Earth Sci. 24,65–78.

McEwen, A.S., et al., 2007. MRO's High Resolution Imaging Science Experiment (HiRISE).J. Geophys. Res. 112 (E05S02). http://dx.doi.org/10.1029/2005JE002605.

McGill, G.E., Dimitriou, A.M., 1990. Origin of theMartian global dichotomy by crustal thin-ning in the Late Noachian or Early Hesperian. J. Geophys. Res. 95 (B8). http://dx.doi.org/10.1029/JB095iB08p12595 (p. 12,595–12,605).

http://refhub.elsevier.com/S0169-555X(14)00624-2/rf0370



http://dx.doi.org/10.1038/nature07011

http://dx.doi.org/10.1038/nature07011















http://dx.doi.org/10.1016/j.icarus.2005.11.012



http://dx.doi.org/10.1029/2009JE003496

http://dx.doi.org/10.1029/2009JE003496










http://dx.doi.org/10.1029/2001JE001505















http://dx.doi.org/10.1029/2005JE002504


















http://dx.doi.org/10.1002/jgre.20107







http://dx.doi.org/10.1029/2000JE001445









http://dx.doi.org/10.1029/2003JE002062

http://dx.doi.org/10.1029/2010JE003658

http://dx.doi.org/10.1029/2004JE002248

http://dx.doi.org/10.1029/2005JE002460

http://dx.doi.org/10.1029/2005JE002460

http://dx.doi.org/10.1029/2010JE003620


























http://dx.doi.org/10.1029/2011JE004008

http://dx.doi.org/10.1029/2011JE004008

http://dx.doi.org/10.1029/2003JE002111



http://dx.doi.org/10.1029/2006JE002808

http://dx.doi.org/10.1029/2008JE003076






http://dx.doi.org/10.1029/2005JE002605

http://dx.doi.org/10.1029/JB095iB08p12595


McGill, G.E., Smrekar, S.E., Dimitriou, A.M., Raymond, C.A., 2004. Crustal Evolution of theProtonilus Mensae Area, Mars. Workshop onMartian Hemispheres. Lunar and Plane-tary Institute Contribution 1213, Houston, Texas (# 4003).

McMenamin, D.S., McGill, G.E., 2005. Processes involved in the formation of Martian fan-shaped deposits. Lun. Planet. Sci. Conf. XLI (CD-ROM, # 1732).

Moore, J.M., Howard, A.D., 2005. Large alluvial fans on Mars. J. Geophys. Res. 110, E04005.http://dx.doi.org/10.1029/2004JE002352.

Moratto, Z.M., Broxton, M.J., Beyer, R.A., Lundy, M., Husmann, K., 2010. Ames Stereo Pipe-line, NASA's Open Source Automated Stereogrammetry Software. 41st Lun. Plan. Sc.Conf., # 2354.

Nahm, A.L., Schultz, R.A., 2011. Magnitude of global contraction on Mars from analy-sis of surface faults: implications for Martian thermal history. Icarus 211 (1),389–400.

Neukum, G., Jaumann, R., the HRSC Co-Investigator Team, 2004. HRSC: The High Resolu-tion Stereo Camera of Mars Express. ESA Special Publications SP-1240.

Neumann, G.A., Rowlands, D.D., Lemoine, F.G., Smith, D.E., Zuber, M.T., 2001. Crossoveranalysis of Mars orbiter laser altimeter data. J. Geophys. Res. 106 (E10). http://dx.doi.org/10.1029/2000JE001381 (p. 23,753–23,768).

Nimmo, F., 2002. Admittance estimates of mean crustal thickness and density at the Mar-tian hemispheric dichotomy. J. Geophys. Res. 107 (E11), 5117. http://dx.doi.org/10.1029/2000JE001488.

Nimmo, F., 2005. Tectonic consequences of Martian dichotomy modification by lower-crustal flow and erosion. Geology 33 (7), 533–536. http://dx.doi.org/10.1130/G21342.1.

Nimmo, F., Stevenson, D.J., 2001. Estimates of Martian crustal thickness from viscous re-laxation of topography. J. Geophys. Res. 106, 5085–5098.

Pain, C.F., Clarke, J.D.A., Thomas, M., 2007. Inversion of relief on Mars. Icarus 190 (2),478–491. http://dx.doi.org/10.1016/j.icarus.2007.03.017.

Parker, G., Paola, C., Whipple, K.X., Mohrig, D., 1998. Alluvial fans formed by channelizedfluvial and sheet flow. I Theory J. Hydraul. Eng. 124 (10), 985–995.

Parmentier, E.M., Zuber, M.T., 2007. Early evolution of Mars with mantle compositionalstratification or hydrothermal crustal cooling. J. Geophys. Res. 112. http://dx.doi.org/10.1029/2005JE002626 (01/2007).

Postma, G., 1990. Depositional architecture and facies of rivers and fan deltas: a synthesis.Spec. Publ. Inl. Ass. Sedimentol. 10, 13–27.

Rafferty, J.P., 2011. Rivers and Streams, Eds Rosen Education Service (282 pp.).Roberts, G.G., White, N.J., Martin-Brandis, G.L., Crosby, A.G., 2012. An uplift history of the

Colorado Plateau and its surroundings from inverse modeling of longitudinal riverprofiles. Tectonics 31 (4). http://dx.doi.org/10.1029/2012TC003107.

Royden, L.H., Burchfiel, B.C., King, R.W., Wang, E., Chen, Z., Shen, F., Liu, Yuping, 1997. Sur-face deformation and lower crustal flow in Eastern Tibet. Science 276 (5313),788–790. http://dx.doi.org/10.1126/science.276.5313.788.

Sacek, V., 2014. Drainage reversal of the Amazon River due to the coupling of surface andlithospheric processes. Earth Planet. Sci. Lett. 401, 301–312.

Salcher, B.C., Faber, R., Wagreichc, M., 2010. Climate as main factor controlling the se-quence development of two Pleistocene alluvial fans in the Vienna Basin (easternAustria) — a numerical modelling approach. Geomorphology 115, 215–227.

Scott, D.H., Tanaka, K.L., 1982. Ignimbrites of Amazonis Planitia region of Mars. J. Geophys.Res. 87 (B2), 1179–1190. http://dx.doi.org/10.1029/JB087iB02p01179.

Scott, D.H., Tanaka, K.L., 1986. Geologic map of the western equatorial region of Mars. U.S.Geol. Surv. Misc. Invest. Map I–1802–A.

Searls, M.L., Phillips, R.J., 2004. Mechanics of Utopia Basin on Mars. Lun. Planet. Sci. XXXV(# 1822).

Seeber, L., Gornitz, V., 1983. River profiles along the Himalayan arc as indicators of activetectonics. Tectonophysics 92, 335–467. http://dx.doi.org/10.1016/0040-1951(83)90201-9.

Shean, D.E., Fahle, J., Malin, M.C., Edwards, L.J., Posiolova, L., 2011. MRO CTX stereo imageprocessing and preliminary DEM quality assessment. Lunar Planet. Sci. XVII (# 2646).

Sleep, N.H., Phillips, R.J., 1985. Gravity and lithospheric stress on the terrestrial planetswith emphasis on the Tharsis region of Mars. J. Geophys. Res. 90, 4469–4489.

Smith, D.E., et al., 2001. Mars Orbiter Laser Altimeter: experiment summary after the firstyear of global mapping ofMars. J. Geophys. Res. 106 (E10). http://dx.doi.org/10.1029/2000JE001364 (p. 23,689–23,722).

Snyder, N.P., Whipple, K.X., Tucker, G.E., Merritts, D.J., 2000. Landscape response to tec-tonic forcing: digital elevationmodel analysis of stream profiles in theMendocino tri-ple junction region, northern California. Geol. Soc. Am. Bull. 112 (8), 1250–1263.

Solomon, S.C., Chaiken, J., 1976. Thermal Expansion and Thermal Stress in the Moon andTerrestrial Planets — Clues to Early Thermal History. Lunar Science Conference, 7th,Houston, Tex., March 15–19, 1976, Proceedings. v 3. (A77-34651 15-91). PergamonPress, Inc., New York, pp. 3229–3243.

Solomon, S.C., Head, J.W., 1990. Heterogeneities in the thickness of the elastic lithosphereof Mars: constraints on heat flow and internal dynamics. J. Geophys. Res. 07. http://dx.doi.org/10.1029/JB095iB07p11073.

Stouthamer, E., Cohen, K.M., Gouw, M.J.P., 2005. Avulsion and Its Implications for Fluvial–Deltaic Architecture: Insights from the Holocene Rhine–Meuse Delta. SEPM SpecialPublication 97 978-1-56576-305-0, pp. 215–232 (p. 215–231).

Tanaka, K.L., 1986. The stratigraphy of Mars. J. Geophys. Res. 91. http://dx.doi.org/10.1029/0JGREA000091000B1300E139000001.

Tanaka, K.L., 2004. Topographic and Geomorphic Modification History of the Highland/Lowland Dichotomy Boundary of Mars: 1. Noachian Period. Workshop on MartianHemispheres. Lunar and Planetary Institute Contribution 1213, Houston, Texas (#4023).

Tanaka, K.L., Davis, P.A., 1988. Tectonic history of the Syria Planum province of Mars.J. Geophys. Res. 93 (p. 14,893–14,917).

Tanaka, K.L., Banerdt, W.B., Kargel, J.S., Hoffman, N., 2001. Huge, CO2-charged debris-flowdeposits and tectonic sagging in the northern plains of Mars. Geology 29, 427–430.

Tanaka, K.L., Skinner Jr., J.A., Hare, T.M., Joyal, T., Wenker, A., 2003. Resurfacing history ofthe northern plains of Mars based on geologic mapping of Mars Global Surveyor data.J. Geophys. Res. 108 (E4). http://dx.doi.org/10.1029/2002JE001908 (p. 24-1–24-32,8043).

Tanaka, K.L., Skinner Jr., J.A., Hare, T.M., 2005. Geologic Map of the Northern Plains ofMars. U.S. Geological Survey Scientific Investigations Map 2888.

Torson, J.M., Becker, K.J., 1997. ISIS: a software architecture for processing planetary im-ages. Lun. Plan. Sci. Conf. XXVIII (CD-ROM, abstract 1443).

Watters, T.R., 2003a. Lithospheric flexure and the origin of the dichotomy boundary onMars. Geology 31, 271–274. http://dx.doi.org/10.1130/0091-7613(2003) 031b0271:LFATOON2.0.CO;2.

Watters, T.R., 2003b. Thrust faults along the dichotomy boundary in the eastern hemi-sphere of Mars. J. Geophys. Res. 108 (E6), 5054. http://dx.doi.org/10.1029/2002JE001934.

Watters, T.R., Robinson, M.S., 1999. Lobate scarps and the Martian crustal dichotomy.J. Geophys. Res. 104 (E8), 18981–18990. http://dx.doi.org/10.1029/1998JE001007.

Watters, T.R., Robinson, M.S., Bina, C.R., Spudis, P.D., 2004. Thrust faults and the globalcontraction of Mercury. Geophys. Res. Lett. 31, L04701. http://dx.doi.org/10.1029/2003GL019171 (2004).

Watters, T.R., McGovern, P.J., Irwin III, R.P., 2007. Hemispheres apart: the crustal dichoto-my on Mars. Annu. Rev. Earth Planet. Sci. 35, 621–652. http://dx.doi.org/10.1146/annurev.earth.35.031306.140220.

Watters, T., Solomon, S.C., Klimczak, C., Freed, A.M., Head, J.W., Ernst, C.M., Blair, D.M.,Goudge, T.A., Byrne, P.K., 2012. Extension and contraction within volcanically buriedimpact craters and basins on Mercury. Geology 40, 1123–1126.

Watts, A.B., 1989. Lithospheric flexure due to prograding sediment loads: implications forthe origin of offlap/onlap patterns in sedimentary basins. Basin Res. 2 (3), 133–144.http://dx.doi.org/10.1111/j.1365-2117.1989.tb00031.x.

Watts, A.B., Karner, G.D., Steckler, M.S., 1982. Lithospheric flexure and the evolution ofsedimentary basins. In: Kent, P.S., Bott, M.H.P., McKenzic, D.P., Williams, C.A. (Eds.),The Evolution of Sedimentary Basins. Phil, Trans. Roy. Soc. London 305A, pp. 249–281.

Wells, S.G., Harvey, A.M., 1987. Sedimentologic and geomorphic variations in storm-generated alluvial fans, Howgill Fells, northwest England. GSA Bull. 98, 182–198.

Werner, S.C., 2006. Major Aspects of the Chronostratigraphy and Geologic EvolutionaryHistory of Mars. Ph.D. thesis, Freie Universitat, Berlin, p. 165 (http://www.diss.fuberlin.de/2006/33/indexe.html).

Wieczorek, M.A., Zuber, M.T., 2004. Thickness of the Martian crust: improved constraintsfrom geoid-to-topography ratios. J. Geophys. Res. 109, E01009. http://dx.doi.org/10.1029/2003JE002153.

Williams, R.M., Edgett, K.S., 2005. Valleys in theMartian rock record. Lun. Planet. Sci. Conf.XXXVI (# 1099).

Williams, R.M., Chidsey Jr., E., T.C., Eby, D.E., 2007. Exhumed paleochannels in centralUtah: analogs for raised curvilinear features on Mars, in Central Utah. In: Willis,G.C., et al. (Eds.), Diverse Geology of a Dynamic Landscape: Salt Lake City. Utah Geo-logical Association Publication 36, Utah, pp. 220–235.

Zimbelman, J.R., Griffin, L.J., 2010. HiRISE images of yardangs and sinuous ridges in thelower member of the Medusae Fossae Formation, Mars. Icarus 205, 198–210.

Zimbelman, J.R., Scheidt, S.P., 2012. Hesperian Age for Western Medusae FossaeFormation, Mars. Science 336 (6089), 1683. http://dx.doi.org/10.1126/science.1221094.

Zuber, M.T., Solomon, S.C., Phillips, R.J., Smith, D.E., Tyler, G.L., Aharonson, O., Balmino, G.,Banerdt, W.B., Head, J.W., Johnson, C.L., Lemoine, F.G., McGovern, P.J., Neumann, G.A.,Rowlands, D.D., Zhong, S.J., 2000. Internal structure and early thermal evolution ofMars from Mars Global Surveyor topography and gravity. Science 287, 1788–1793.






http://dx.doi.org/10.1029/2004JE002352









http://dx.doi.org/10.1029/2000JE001381

http://dx.doi.org/10.1029/2000JE001488

http://dx.doi.org/10.1029/2000JE001488

http://dx.doi.org/10.1130/G21342.1

http://dx.doi.org/10.1130/G21342.1






http://dx.doi.org/10.1029/2005JE002626




http://dx.doi.org/10.1029/2012TC003107

http://dx.doi.org/10.1126/science.276.5313.788











http://dx.doi.org/10.1016/0040-1951(83)90201-9

http://dx.doi.org/10.1016/0040-1951(83)90201-9





http://dx.doi.org/10.1029/2000JE001364

http://dx.doi.org/10.1029/2000JE001364












http://dx.doi.org/10.1029/0JGREA000091000B1300E139000001

http://dx.doi.org/10.1029/0JGREA000091000B1300E139000001










http://dx.doi.org/10.1029/2002JE001908





http://dx.doi.org/10.1130/0091-7613(2003) 031<0271:LFATOO>2.0.CO;2




http://dx.doi.org/10.1029/2002JE001934

http://dx.doi.org/10.1029/2002JE001934

http://dx.doi.org/10.1029/1998JE001007

http://dx.doi.org/10.1029/2003GL019171

http://dx.doi.org/10.1029/2003GL019171

http://dx.doi.org/10.1146/annurev.earth.35.031306.140220

http://dx.doi.org/10.1146/annurev.earth.35.031306.140220



http://dx.doi.org/10.1111/j.1365-2117.1989.tb00031.x









http://dx.doi.org/10.1029/2003JE002153

http://dx.doi.org/10.1029/2003JE002153









http://dx.doi.org/10.1126/science.1221094

http://dx.doi.org/10.1126/science.1221094



channel slope reversal near the martian dichotomy boundary ...fnimmo/website/lefort.pdfchannel slope...

Documents