Icarus 264 (2016) 213–219

Contents lists available at ScienceDirect

Icarus

journal homepage: www.journa ls .e lsevier .com/icarus

Enhanced erosion rates on Mars during Amazonian glaciation

http://dx.doi.org/10.1016/j.icarus.2015.09.0370019-1035/� 2015 Elsevier Inc. All rights reserved.

⇑ Corresponding author.E-mail address: joe.levy@utexas.edu (J.S. Levy).

Joseph S. Levy a,⇑, Caleb I. Fassett b, James W. Head c

aUniversity of Texas Institute for Geophysics, Austin, TX 78758, USAbMount Holyoke College, South Hadley, MA 01075, USAcBrown University, Providence, RI 02906, USA

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

Article history:Received 27 May 2015Revised 23 September 2015Accepted 23 September 2015Available online 9 October 2015

Keyword:Mars, surfaceMars, climateGeological processes

Observations of Mars from the surface and from orbit suggest that erosion rates over the last �3 Gyr (theAmazonian) have been as slow as 10�5 m/Myr and have been dominated by aeolian processes, whileancient (Noachian) erosion rates may have been orders of magnitude higher due to impact bombardmentand fluvial activity. Amazonian-aged glacial deposits are widespread on Mars, but rates of erosionresponsible for contributing debris to these remnant glacial deposits have not been constrained. Here,we calculate erosion rates during Amazonian glaciations using a catalog of mid-latitude glacial landformscoupled with observational and theoretical constraints on the duration of glaciation. These calculationssuggest that erosion rates for scarps that contributed debris to glacial landforms are 4–7 orders of mag-nitude higher than average Amazonian rates in non-glaciated, low-slope regions. These erosion rates aresimilar to terrestrial cold-based glacier erosion and entrainment rates, consistent with cold-based glaciermodification of parts of Mars.

� 2015 Elsevier Inc. All rights reserved.

1. Introduction

The surface of Mars contains crustal units and surface depositsthat range in age from >4 billion years to less than a few hundredthousand years (Tanaka et al., 2014). Accordingly, erosion ratesestimated for Mars are highly variable as a function of locationand the window of time over which erosion occurs (Golombekand Bridges, 2000; Golombek et al., in press). Observations of aeo-lian deflation at the Pathfinder landing site suggested erosion ratesover 1.8–3.5 Gyr of 1–4 � 10�5 m/Myr (Golombek and Bridges,2000). Measurements of crater degradation along the traverse ofthe Opportunity rover suggest that erosion rates may be up to1 m/Myr for the freshest craters (0.1–1 Myr), but fall off to<0.1 m/Myr for 10–20 Myr old craters, before decreasing to theslower average rates for craters from the middle to early Amazo-nian (Golombek et al., in press). Recent measurements oflandslide-driven scarp retreat in the interior layered deposits(ILD) of Valles Marineris suggest rapid erosion of these friable,potentially ice-rich, and steeply sloped deposits, on the order of1.2–2.3 m/Myr over the past 400 Myr (Grindrod and Warner,2014). In contrast, orbital observations of large, Noachian andHesperian craters suggest erosion rates in the distant past mayhave reached up to 1–100 m/Myr (Carr, 1996; Craddock and

Maxwell, 1993; Craddock et al., 1997). Here, we calculate erosionrates from debris-bearing glacial landforms that are distributedwidely across the martian mid-latitudes and that formed over anextended period of time during the Amazonian.

Erosion associated with cold-based glaciation on Earth is knownto be very different and to have much lower rates than wet-basedglacial erosion (Hallet et al., 1996). Erosion and sediment forma-tion in wet-based glacial environments occurs primarily due to(1) basal melting, rapid glacial movement, and associated abrasionand transport, as well as (2) freeze–thaw cycling of adjacent out-crops that promotes generation of supraglacial debris. In cold-based glaciation, glacial movement is extremely slow, deformationoccurs overwhelmingly within the ice, and generation of meltwa-ter is limited to thin films. These factors combine to result in extre-mely low basal erosion rates (�9 � 10�1 to 3 � 100 m/Myr) (Cuffeyet al., 2000). Extremely low surface temperatures (thermal cyclingbut little to no freeze–thaw activity) mean that debris shed fromadjacent, steep-walled outcrops is also correspondingly less inenvironments where cold-based glacial activity dominates thanin the case of wet-based glaciers, e.g., Mackay et al. (2014).

For example, the vast majority of debris in and on cold-basedglaciers in Antarctica is derived from rockfall and erosion fromthe steep cliffs that surround the accumulation zones of these gla-ciers. It is deposited supraglacially, and often becomes englacial bythe continued deposition of snow and ice in the accumulationzone. It is returned to the surface debris blanket via glacier flow

Fig. 2. Example of LDA emanating from sheltered alcoves along a massif. Note thatdebris-bearing flow lineations (arrows) can be mapped back to glacier accumula-tion areas in these alcoves, as noted by Head et al. (2006b), implying erosion ofsediment in the accumulation zone and transport of the debris down-slope byglacier flow. Portion of CTX image P18_008019_2227.

214 J.S. Levy et al. / Icarus 264 (2016) 213–219

and sublimation (Marchant et al., 2002; Marchant and Head, 2007;Mackay et al., 2014). These Antarctic Dry Valley cold-based glaciersare generally protective rather than erosional at their bases (e.g.,Denton et al., 1993), although minor basal entrainment of siltand fine sand occur (Cuffey et al., 2000).

Debris-covered glacial deposits on Mars (Figs. 1 and 2), includ-ing lineated valley fill (LVF), lobate debris aprons (LDA), and con-centric crater fill (CCF), cover 7 � 105 km2 of the martian surfacebetween ±�30–50� latitude to a thickness of several hundredmeters (Levy et al., 2014). LDA, LVF, and CCF are exceptionallycommon along the martian dichotomy boundary (e.g., the valley-and-mesa fretted terrain in Deuteronilus and Protonilus mensae,Sharp, 1973) and east of the Hellas impact basin (Squyres, 1978).Initial geomorphic analyses of these features interpreted them asindicators of ground ice emplaced by vapor diffusion in a differentclimate epoch, and the mobilization and flow of ice-cemented deb-ris (boulders and finer-grained sediments) eroded from the escarp-ments (Sharp, 1973; Squyres, 1978, 1979; Squyres and Carr, 1986).Other studies (Lucchitta, 1984; Head et al., 2006a, 2006b, 2010;Holt et al., 2008; Plaut et al., 2009; Levy et al., 2010; Karlssonet al., 2015) have noted the similarities between these featuresand terrestrial debris-covered glaciers, both in terms of morphol-ogy and radar properties, and have interpreted these features asdeposits formed by the accumulation and flow of glacial ice(Li et al., 2005) that are covered by a debris lag derived fromerosion and entrainment of debris from the accumulation zone ofthe glacier.

Morphological criteria use to recognize martian debris-coveredglaciers including LVF, LDA, and CCF, include: (1) the presence ofalcoves in mesas, valley walls, or crater walls that could haveserved as ice accumulation zones; (2) parallel or arcuate ridgesemerging from these alcoves that are interpreted as flow-derivedlineations; (3) tightening or folding of ridges where flow was con-fined by obstacles such as bedrock ridges; (4) broadening or expan-sion of lineations in unconfined regions; (5) merging of lineationswhere two features meet; (6) arête-like morphology of rock spursbetween glacial features; (7) merging of LVF, LDA, and/or CCF toform integrated valley glacier systems; and (8) convex-up topogra-phy with a parabolic down-slope profile (Head et al., 2010). Radarobservations interpreted to indicate a debris-covered glacier origininclude (1) radar velocity determinations consistent with water ice

Fig. 1. Map showing the distribution of lineated valley fill (LVF), lobate debris aprondichotomy boundary (see location inset). The erodible contributing area that could havemethod is marked in red. Base map is MOLA 128 ppd gridded topography rendered astopography rendered in hillshade and color-coded elevation (red = high, blue = low) in acolor in this figure legend, the reader is referred to the web version of this article.)

and (2) a lack of internal reflectors including layers or point reflec-tors, which is interpreted to indicate clean (low debris content) ice(Holt et al., 2008; Plaut et al., 2009). Lineated debris bands on LDA,LVF, and CCF surfaces extend from the steep scarps bounding theiraccumulation zones to the termini of these features. This is evi-dence that headwall material in and around the glacial accumula-tion zone has been eroded and transported downslope by glacialflow in a process analogous to that observed on Earth (Headet al., 2006b; Fastook et al., 2014) (Fig. 2).

Based on the rarity of glacial outwash (Fassett et al., 2010) ormelt features (e.g., kettles, push moraines), LDA, LVF, and CCF havebeen interpreted as evidence of cold-based glaciation (Head et al.,2010), with glacier ice preserved by a thin debris cover similar tothe protective lag that overlies cold-based debris-covered glaciersin Antarctica (Marchant et al., 2002; Marchant and Head, 2007;Fastook et al., 2014; Mackay et al., 2014).

Early workers noted that the rocky debris transported by LVF,LDA, and CCF likely was sourced by erosion from the valley walls,isolated massifs, and craters in and around which the deposits

s (LDA), and concentric crater fill (CCF) along an example portion of the martianacted as a source for debris cover sediment based on our contributing area defininga hillshade centered on 41.5�N, 26.5�E. Inset location map is global MOLA 128 ppdMollweide projection centered on 0�N, 0�E. (For interpretation of the references to

Fig. 4. A ‘‘ghost LDA” deposit from Hauber et al. (2008) consisting of a depressedregion that surrounds a massif, and which, in turn, is surrounded by elevated lavaflows. Such landforms are interpreted to be features marking the former extent ofLDA deposits that blocked the emplacement of the lavas, but from which the ice hassubsequently sublimated away, leaving behind lineated, pit-and-butte patternedmaterial interpreted as remnant glacial debris. Portion of CTX imageP19_008616_2076.

J.S. Levy et al. / Icarus 264 (2016) 213–219 215

formed (Squyres, 1978; Lucchitta, 1984; Squyres and Carr, 1986).However, in the absence of radar or global topography data, itwas impossible to determine the volume and debris content ofthe LVF, LDA, and CCF. Accordingly, it was unclear how to trace thisdebris to possible source outcrops. Based on recent mappingresults (Levy et al., 2014) that show the extent of martian remnantglacial landforms, coupled with MOLA digital elevation model(DEM) data, the goal of this manuscript is to determine the magni-tude of erosion from surfaces proximal to preserved glacial fea-tures on Mars by relating the volume of supraglacial debris tothe surface area of the eroded slopes from which this lithic mate-rial is derived.

Several compositional and morphological aspects of LDA, LVF,and CCF suggest that they formed primarily through entrainmentand/or erosion of debris that was mobilized by ice, rather than,for example, through mobilization of fine-grained aeolian sedi-ments (e.g., dust). For example, flow lineations on LDA, LVF, andCCF typically can be traced back to sheltered alcoves on the valleywalls or massifs, that have been interpreted as glacier accumula-tion zones (Head et al., 2006b) (Fig. 2). Subsequent geomorphicobservations noted that LVF and LDA appeared to erode primarilyalong valley and massif slopes, leaving mesa tops and inter-valley plateaus largely intact—a process termed ‘‘mesa-ization”(Head et al., 2006a) (Fig. 1).

Martian debris-covered glacier surfaces contain a range of sed-iment sizes, from boulders visible in HiRISE images to fine-grained,polygonally patterned sediments (Fig. 3). This is consistent withhigh thermal inertia values for LDA, spanning 90–320 J m�2 K�1 s�1/2,indicating a surface composed of sand, cobbles, and boulders(Parsons et al., 2011). Although some heavily degraded LDA depos-its appear at HRSC resolution to have been nearly entirely removed(Hauber et al., 2008), suggesting a dominance of dust as the lithicmaterial in LDA, CTX images of these deposits show that remnantdebris, including lineated and pit-and-butte patterned materialremains at these sites, consistent with a thin debris blanket overthe ice-dominated part of the LDA that is blocky enough to resisterosion (Fig. 4). This fines-to-boulders sedimentology spans thesize range of debris entrained by terrestrial, cold-based, debris-covered glaciers, which is characterized by silt, sand, and boulders(Cuffey et al., 2000; Marchant et al., 2002; Mackay et al., 2014).Accordingly, this raises the possibility that terrestrial and martiandebris-covered glaciers share a sediment cover grain size distribu-tion driven by common physical processes, such as entrainment ofrock fall and finer-grained debris in the glacier accumulation zoneand possible entrainment of silt and sand from the glacier bed.

Fig. 3. Example of boulder-sized and fine-grained sediments on the surface of alobate debris apron located at 49.0�N, 50.7�E. Portion of HiRISE imageESP_034720_2295.

Despite these morphological observations of martian debris-covered glacier surface properties, the thickness of the debris covercannot be directly measured because of limitations on the verticalresolution of SHARAD and because it is not clear if the ice/debriscontact is sharp or gradational (Holt et al., 2008). However, radarobservations indicate that the debris thickness is on the order ofthe wavelength of SHARAD (�5–10 m in rock) (Holt et al., 2008;Plaut et al., 2009). This interpretation is supported by a lack ofGRS/NS detections of water-equivalent hydrogen in these deposits(Boynton et al., 2002; Feldman et al., 2004), indicating that thedebris must be greater than �1 m thick. This is consistent withmodels indicating that water ice is stable on modern Mars atdepths >1 m below the surface at these latitudes (Mellon andJakosky, 1993). Together, these geophysical observations and mod-els bound the possible thickness range for debris cover in martianremnant glaciers: the debris must be thicker than several meters topreserve the underlying ice observed by SHARAD, but not muchthicker than the SHARAD wavelength, in order to explain the lackof a debris/ice reflector (Holt et al., 2008; Plaut et al., 2009).

Together, these observations indicate a radar-limited debristhickness of no more than �10 m in the observed lobatedebris aprons (Holt et al., 2008; Plaut et al., 2009). Supraglacialdebris thickness may be somewhat higher in CCF than LDA dueto the more prominent topography of the crater rim crest(Fastook and Head, 2014) than the LDA plateau rims, and due tothe concentration of debris within the crater interior, rather thandispersal of debris from mesas and massifs. Measuring the size dis-tribution impact craters in CCF that appear to have been modifiedby ice interactions (‘‘ring-mold” craters) versus those that appearto have not penetrated through the debris into ice, Kress andHead (2008) estimated the supraglacial debris thickness for CCFdeposits to be �15 m. We use an estimated debris thickness of10 m for all landforms in this study, as established for LDAs bySHARAD and GRS/NS observations, and as supported by calcula-tions of the thermal stability of ground ice on Mars. CCF debriscover may be 50% thicker, a value that would make the erosionrates calculated here a lower bound, but which would not changethe order of magnitude of erosion rates calculated in this study.Likewise, we do not include the debris dispersed englacially withinthe icy portions of the landforms—only the debris that is present inthe near-surface lag deposit—which also implies that our erosion

216 J.S. Levy et al. / Icarus 264 (2016) 213–219

rates are a lower bound. However, debris content in terrestrial,cold-based, debris-covered glaciers is typically small, e.g., 3.1% byvolume in Mullins Glacier, Antarctic (Marchant et al., 2002), andno debris-band reflectors are observed in LDA (Holt et al., 2008),suggesting that englacial debris content in LDA, LVF, and CCFmay be comparably low.

2. Methods

In order to calculate the magnitude of erosion resulting fromthe formation of martian glacial deposits, we made use of a spatialcatalog of LVF, LDA, and CCF that contains landform area, location,and volume derived from MOLA global gridded data and computedlandform geometry (Levy et al., 2014). The catalog consists of GISshapefiles outlining the locations and volumes of 11,241 LVF,LDA, and CCF landforms, mapped at �1:250 k scale, between30–50� latitude and across all longitudes. Where mapping led toidentification of remnant glacial landforms that extended outsidethese latitude bands, such features were included in the databasefor completeness (Levy et al., 2014). The landforms mapped in thiscatalog cover 7.1 � 105 km2 of the martian surface—3.9 � 105 km2

of LDA, 2.4 � 105 km2 of CCF, and 8.2 � 104 km2 of LVF. Thecalculated volumes of glacial ice preserved in these features are2.6 � 105 km3 in LDA deposits versus 8.8 � 104 km3 in CCF, and6.5 � 104 km3 in LVF (Levy et al., 2014).

Surface areas of CCF, LVF, and LDA units were extracted fromthe catalog and were multiplied by 10 m to produce a volume esti-mate for surficial debris. In order to constrain the erodible area sur-rounding the CCF, LVF, and LDA deposits, a 5 km spatial buffer wasgenerated around the glacial landform shapefiles from the catalog.The spatial extents of the glacial deposits were deleted from thisglacially-influenced zone to produce a mask for extracting the areawithin 5 km of LDA and LVF deposits (grouped together) and CCFdeposits (treated separately). Grouping LDA with LVF and treatingCCF separately facilitates subsequent processing steps (i.e., concav-ity filtering). Using the LDA and LVFmask, and the CCF mask, MOLAgridded topography and slope data were extracted from the global128 ppd MOLA global dataset. The slope data raster was resampledsuch that slopes >5� were retained and slopes <5� were removed(Fig. 1). Because of its sparse sampling of martian topography,MOLA gridded data underestimates short baseline slopes (Phillipset al., 1998), so this cutoff value likely captures all of the steepslopes over which rapid transport of sediment would be expected(Fig. 1).

For LVF and LDA, the processed slope dataset was convertedpixel-by-pixel into a shapefile and the area of the shapefile (con-tributing slopes steeper than 5�) was calculated. For CCF deposits,an additional processing step was required. Using MOLA griddedtopography data extracted using the CCF contributing area mask,the surface planform curvature was calculated from the MOLAgridded topography data. Crater interior slopes have negative plan-form curvature, while crater exterior rim slopes have positive plan-form curvature. Since only slopes within the craters are likely tohave contributed to CCF debris that is preserved on the inside ofthe crater, the slopes of pixels with positive planform curvaturewere removed. The CCF slope raster was then processed to extractslopes steeper than 5�, so that surfaces that would have shed debrisoutside of the crater were excluded, leaving only the area capableof shedding debris inside of the crater where it could contribute toCCF debris cover. This processed slope dataset was convertedpixel-by-pixel into a shapefile and the area of the shapefile wascalculated to determine the total erodible area for crater interiorsthought to have sourced CCF deposit debris. An example of howthese calculations map out spatially to highlight scarps upslopeof LDA, LVF, and CCF is shown in Figs. 1 and 6.

In order to verify that the global MOLA dataset processingdescribed above captures the detailed structure of LDA depositsand their eroded zones, we compare the global values for supragla-cial debris volume, erodible area, and average eroded thicknessdescribed above, to local-scale, manual measurements of the char-acteristics of massifs associated with LDA deposits. Isolated massifssurrounded by LDA (Figs. 1 and 4) were selected from the catalogand were manually mapped. Using the original 128 ppd MOLAgridded topography dataset, we measured scarp height and scarpperimeter for massif slopes upslope of 50 LDA (25 in the northernhemisphere and 25 in the southern hemisphere). As in the globalcalculation, LDA deposit surface areas were extracted from theLevy et al. (2014) catalog and were multiplied by 10 m, to producea debris volume estimate. Erodible scarp area was calculated bymultiplying average scarp height (determined by averaging theelevation range from two orthogonal transects across the massif)by the perimeter of the massif. Erosional scarp retreat caused bytransfer of massif bedrock to the LDA debris cover was calculatedby dividing the supraglacial debris volume atop the LDA by themeasured scarp area.

In order to transform these eroded thickness and scarp retreatcalculations into an erosion rate, we divided the eroded thick-nesses by the minimum and maximum durations of glacial activitydetermined from observational and theoretical studies (Fassettet al., 2014; Fastook et al., 2014) in order to generate an averagerate of erosion in meters per million years (m/Myr).

3. Results

The volume of debris was estimated by multiplying the area ofthe LVF, LDA, and CCF deposits by 10 m, producing an integratedsurface lag debris volume of 7.2 � 103 km3. The total erodible areasurrounding the deposits calculated using our method is1.1 � 106 km2. Dividing debris volume by erodible area producesan average total eroded depth of �7 m removed in order to pro-duce the observed debris covers.

In order to formulate an erosion rate from the eroded thicknessvalue, we divided the eroded thickness by the range of durations ofAmazonian glacial activity determined from observational and the-oretical studies. Measurement of the frequency of the craterssuperposed upon the glacial landforms that we mapped in thenorthern hemisphere, as well as the frequency of craters formedduring their period of activity, indicate that the glaciation in themartian northern hemisphere occurred during a (possibly non-continuous) �600 Myr period that ended �100 Myr ago (Fassettet al., 2014). These age estimates from crater statistics are unableto constrain the continuity or episodicity of glacial activity duringthis period of time. However, if erosion was continuous, then theaverage erosion rate could be no slower than �1 � 10�2 m/Myr(Fig. 5). Conversely, the minimum time required for formation oftypical LDA in the northern hemisphere of Mars based on glacialflow modeling is �500 kyr (Fastook et al., 2014). If all LVF, LDA,and CCF formed in a single, 500 kyr episode, then the maximumerosion rate needed to produce the observed debris cover is�1 � 101 m/Myr (Fig. 5). Our estimate for the erosion rate thatled to the formation and preservation of the observed LVF, LDA,and CCF deposits is bounded by these two endmember erosionrates, and lies somewhere between 10�2 m/Myr and 101 m/Myr.

Calculated idealized scarp retreat distances for 50 LDA in thenorthern and southern hemispheres of Mars span �15–130 m withan average of 40 m and a standard deviation of 26 m. Taking theminimum and maximum formation times noted above, thisaverage eroded distance produces average, scarp-integratedretreat rates between 8 � 101 m/Myr and 7 � 10�2 m/Myr for

Fig. 5. Range of glacial erosion rates calculated in this study (black bar) comparedto erosion rates for martian surfaces and landforms and rates of erosion interrestrial glacial environments. The erosion rate ranges are indicated by theheights of the boxes and bars should be read off the y-axis at left (i.e., erosion ratescalculated in this study span 10�2–101 m/Myr, while erosion rates in Beacon Valley,Antarctica, on Earth, span 0.2–2.2 � 100 m/Myr). Fresh martian craters formed overthe past 1–20 Myr and Amazonian values are from observations along the MEROpportunity traverse (Golombek et al., in press); erosion rates in the VallesMarineris interior layered deposits (ILD) are based on CTX observations oflandslides dated to �400 Myr (Grindrod and Warner, 2014); and Pathfinder erosionrate range is derived from measurements of aeolian deflation (Golombek andBridges, 2000). Beacon Valley erosion rate is derived from cosmogenic nuclidemeasurements (Balco and Shuster, 2009); Meserve glacier erosion rate is deter-mined from ejection rates of entrained sediment (Cuffey et al., 2000); and wet-based glacier erosion rates are from a global catalog of glacial erosion rates (Halletet al., 1996).

J.S. Levy et al. / Icarus 264 (2016) 213–219 217

LDA producing scarps. The full table of measured scarp parametersis included in the supplementary material.

4. Discussion

The erosion rate calculated here is 4–7 orders of magnitude fas-ter than the �1–4 � 10�6 m/Myr erosion rates observed for low-latitude, low-slope, and unglaciated parts of Mars at the Pathfinderlanding site (Golombek and Bridges, 2000) that is commonly heldas a representative, landscape-scale erosion rate for AmazonianMars. Indeed, our calculated glacial erosion rate meets or exceedsthe erosion rates determined for the most rapid degradation offresh impact craters at the MER Opportunity site (which is alsoat low latitude and on relatively flat terrain) (Golombek et al., inpress). Our erosion rates span the measured erosion rates recentlydetermined for the interior layered deposits in Valles Marineris(Grindrod and Warner, 2014)—another high-slope (and potentiallyice-rich) region of Mars, however, one characterized by erosion offriable material, rather than bedrock mesas. These results indicatethat specific regions of Mars, including the hemisphere-scale mar-tian dichotomy boundary, may have experienced comparativelyrapid erosion during Amazonian time.

Note that our calculations of the amount of surficial debris donot include the contribution of the latitude-dependent mantle(LDM, Head et al., 2003), which covers glacial deposits in somelocations (Levy et al., 2009) and is interpreted as a mixture of iceand dust. However, because the debris-covered glacier catalog(Levy et al., 2014) does not include landforms polewards of �50�latitude where LDM is most prominent, the contribution of the

mantle to glacial debris thickness is likely small. Moreover,because the LDM is generally interpreted to be derived from airfallrather than from transport of debris off of nearby slopes, its exclu-sion from our erosion calculations is appropriate.

The erosion that formed the debris cover of martian glaciallandforms did so at rates (10�2–101 m/Myr) that are similar to ter-restrial erosion rates in the coldest, driest portions of Antarctica.Basal erosion by entrainment into the cold-based Meserve Glacierin Wright Valley, Southern Victoria Land is 9 � 10�1 m/Myr (Cuffeyet al., 2000), while sandstone bedrock erosion rates by cold-basedglacial activity and aeolian abrasion in Beacon Valley, SouthernVictoria Land, span 0.2–2.2 � 100 m/Myr (Balco and Shuster,2009). In contrast, glacial erosion by wet-based glaciers spans101–103 m/Myr (Hallet et al., 1996), supporting the interpretationthat wet-based glaciation on Mars was not a major agent in the for-mation of LVF, LDA, and CCF (Head et al., 2010).

The calculated erosion rates vary linearly with the inferred deb-ris layer overlying martian glacial landforms. Accordingly, if debristhickness were twice as thick as indicated by SHARAD observa-tions, the required erosion rates needed to generate that debrislayer would be twice as fast. However, as shown above, there areorders of magnitude differences between the erosion rates gener-ated by wet-based and cold-based glacier environments, as wellas between the erosion rates measured in this study and erosionrates measured at other, non-glaciated Amazonian-aged sites. Thisindicates that small heterogeneities or uncertainties in debris-layerthickness are not sufficient to change the overall interpretationthat martian remnant glacial landforms are most consistent withdebris erosion and entrainment through cold-based glacialprocesses.

Scarp erosion rates measured on isolated massifs surrounded byLDA suggest possible hemisphere-scale differences in erosion pat-terns. Average eroded distance in the northern hemisphere is 48 m,versus 34 m in the southern hemisphere. This difference is moder-ately significant (P = 0.055), and may result from either the greaternumber of steep, erodible slopes associated with northern hemi-sphere fretted terrain, or possible circulation-driven differencesin ice accumulation between hemispheres.

For both hemispheres, average scarp erosion rates for LDA-producing massifs are similar to the spatially-averaged erosionrates calculated using the global catalog, but are slightly higher,than calculated erosion rates that include CCF and LVF. LDA depositand scarp geometry particularly lend themselves to backwastinganalysis because individual LDA deposits can be traced to particu-lar regions of the upslope landscape and because LDA depositscommonly completely surround the isolated massif or mesaaround which they form. LVF and CCF commonly emerge frommultiple alcoves along the feature flowpath or crater rim, makingdebris contributions from individual scarps impossible to disen-tangle (Fig. 6). The elevated erosion rates along LDA-forming scarpsmay be a result of bedrock weakness associated with the formationof isolated buttes and massifs, or with the larger volumes of icepreserved in LDA deposits than in LVF and CCF deposits (e.g.,2.6 � 105 km3 in LDA deposits versus 8.8 � 104 km3 in CCF, and6.5 � 104 km3 in LVF, Levy et al., 2014), suggesting LDA may havebeen larger, thicker, or more active glaciers.

Interestingly, typical mesa spacing and valley width in the fret-ted terrain (Sharp, 1973) is several tens of kilometers. If cold-basedglacial erosion is the primary agent that carved these valleys andmesas, it suggests that thousands of glaciations on the scale ofthe most recent glacier-forming event have removed and trans-ported debris from the massifs and valley walls. Alternatively,cold-based glaciation may be the most recent major driver of ero-sion along the dichotomy boundary and has modified topographythat formed earlier through regional structural and/or fluvial pro-cesses (e.g., Sharp, 1973).

Fig. 6. Examples of LVF (a) and CCF (c), showing multiple alcoves and input pathways for ice and debris flow. As in Fig. 1, mapped landform geometry is highlighted in yellow(LVF) and blue (CCF), and computationally-generated erosional contributing areas are highlighted in red in (b) and (d). (a) is a portion of CTX image P03_002375_2209 and (c)is a portion of CTX image P02_001926_2186. North is to image top in all panels. (For interpretation of the references to color in this figure legend, the reader is referred to theweb version of this article.)

218 J.S. Levy et al. / Icarus 264 (2016) 213–219

5. Conclusions

Taken together, the amount of erosion needed to produce thedebris covers present on LDA, LVF, and CCF suggest that the mar-tian mid-latitudes have experienced erosion rates that were locallyseveral orders of magnitude higher than average Amazonian ero-sion rates in non-glaciated regions. The erosion of valley walls,massifs, and craters proximal to the ice surface, led to most debrisbeing entrained from rockfall in glacier accumulation zones. Boththe processes involved, and their rates, are similar on Mars toobservations of cold-based glaciers on Earth.

Acknowledgments

This work was supported by NASA award NNX13AN50G to JSLand CIF. Support from the NASA Mars Data Analysis Program(NNX11AI81G)and theMarsExpressHighResolutionStereoCamera(HRSC) Team (JPL 1488322) to JWH is gratefully acknowledged. AllArcGIS shapefiles used in this study are available for download in azipped folder from the BrownUniversity Planetary Group data page.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.icarus.2015.09.037.

References

Balco, G., Shuster, D.L., 2009. Production rate of cosmogenic 21Ne in quartzestimated from 10Be, 26Al, and 21Ne concentrations in slowly eroding Antarctic

bedrock surfaces. Earth Planet. Sci. Lett. 281, 48–58. http://dx.doi.org/10.1016/j.epsl.2009.02.006.

Boynton, W.V. et al., 2002. Distribution of hydrogen in the near surface of Mars:Evidence for subsurface ice deposits. Science 297, 81–85. http://dx.doi.org/10.1126/science.1073722.

Carr, M.H., 1996. Water on Mars. Oxford University Press, New York.Craddock, R.A., Maxwell, T.A., 1993. Geomorphic evolution of the martian highlands

through ancient fluvial processes. J. Geophys. Res. Earth Surf. 98, 3453–3468.

Craddock, R.A., Maxwell, T.A., Howard, A.D., 1997. Crater morphometry andmodification in the Sinus Sabaeus and Margaritifer Sinus regions of Mars. J.Geophys. Res. Earth Surf. 102, 13321–13340.

Cuffey, K.M. et al., 2000. Entrainment at cold glacier beds. Geology 28, 351–354.Denton, G.H. et al., 1993. East Antarcitc Ice Sheet sensitivity to Pliocene climatic

change from a dry valleys perspective. Geograph. Annaler, Ser. A, Phys. Geogr.75, 155–204.

Fassett, C.I. et al., 2010. Supraglacial and proglacial valleys on Amazonian Mars.Icarus 208, 86–100. http://dx.doi.org/10.1016/j.icarus.2010.02.021.

Fassett, C.I. et al., 2014. An extended period of episodic northern mid-latitudeglaciation on Mars during the Middle to Late Amazonian: Implications for long-term obliquity history. Geology. http://dx.doi.org/10.1130/G35798.1.

Fastook, J.L., Head, J.W., 2014. Amazonian mid- to high-latitude glaciation on Mars:Supply-limited ice sources, ice accumulation patterns, and concentric crater fillglacial flow and ice sequestration. Planet. Space Sci. 91, 60–76. http://dx.doi.org/10.1016/j.pss.2013.12.002.

Fastook, J.L., Head, J.W., Marchant, D.R., 2014. Formation of lobate debris aprons onMars: Assessment of regional ice sheet collapse and debris-cover armoring.Icarus 228, 54–63. http://dx.doi.org/10.1016/j.icarus.2013.09.025.

Feldman, W.C. et al., 2004. Global distribution of near-surface hydrogen on Mars. J.Geophys. Res.: Planets (1991–2012), 109. http://dx.doi.org/10.1029/2003JE002160.

Golombek, M.P., Bridges, N.T., 2000. Erosion rates on Mars and implications forclimate change: Constraints from the Pathfinder landing site. J. Geophys. Res.Earth Surf. 105, 1841–1853.

Golombek, M.P. et al., 2014. Small crater modification on Meridiani Planum andimplications for erosion rates and climate change on Mars. J. Geophys. Res.Earth Surf. (in press). http://dx.doi.org/10.1002/2014JE004658.

Grindrod, P.M., Warner, N.H., 2014. Erosion rate and previous extent of interiorlayered deposits on Mars revealed by obstructed landslides. Geology 42, 795–798. http://dx.doi.org/10.1130/G35790.1.

Hallet, B., Hunter, L., Bogen, J., 1996. Rates of erosion and sediment evacuation byglaciers: A review of field data and their implications. Global Planet. Change 12,213–235.

J.S. Levy et al. / Icarus 264 (2016) 213–219 219

Hauber, E. et al., 2008. Geomorphic evidence for former lobate debris aprons at lowlatitudes on Mars: Indicators of the martian paleoclimate. J. Geophys. Res. 113(E2), E02007. http://dx.doi.org/10.1029/2007JE002897.

Head, J.W., Mustard, J.F., Kreslavsky, M.A., Milliken, R.E., Marchant, D.R., 2003.Recent ice ages on Mars. Nature 426 (6968), 797–802.

Head, J.W. et al., 2006a. Extensive valley glacier deposits in the northern mid-latitudes of Mars: Evidence for Late Amazonian obliquity-driven climatechange. Earth Planet. Sci. Lett. 241, 663–671. http://dx.doi.org/10.1016/j.epsl.2005.11.016.

Head, J.W. et al., 2006b. Modification of the dichotomy boundary on Mars byAmazonian mid-latitude regional glaciation. Geophys. Res. Lett. 33, 10.1029-2005GL024360.

Head, J.W. et al., 2010. Northern mid-latitude glaciation in the Late Amazonianperiod of Mars: Criteria for the recognition of debris-covered glacier and valleyglacier landsystem deposits. Earth Planet. Sci. Lett. 294, 306–320. http://dx.doi.org/10.1016/j.epsl.2009.06.041.

Holt, J.W. et al., 2008. Radar sounding evidence for buried glaciers in the southernmid-latitudes of Mars. Science 322, 1235–1238.

Karlsson, N.B., Schmidt, L.S., Hvidberg, C.S., 2015. Volume of martian mid-latitudeglaciers from radar observations and ice-flow modeling. Geophys. Res. Lett. 42(8), 2627–2633.

Kress, A.M., Head, J.W., 2008. Ring-mold craters in lineated valley fill and lobatedebris aprons on Mars: Evidence for subsurface glacial ice. Geophys. Res. Lett.35, L23206. http://dx.doi.org/10.1029/2008GL035501.

Levy, J.S., Head, J.W., Marchant, D.R., 2009. Concentric crater fill in UtopiaPlanitia: History and interaction between glacial ‘‘brain terrain” andperiglacial mantle processes. Icarus 202, 462–476. http://dx.doi.org/10.1016/j.icarus.2009.02.018.

Levy, J., Head, J.W., Marchant, D.R., 2010. Concentric crater fill in the northern mid-latitudes of Mars: Formation processes and relationships to similar landforms ofglacial origin. Icarus 209, 390–404. http://dx.doi.org/10.1016/j.icarus.2010.03.036.

Levy, J.S., Fassett, C.I., Head, J.W., Schwartz, C., Watters, J.L., 2014. Sequesteredglacial ice contribution to the global martian water budget: Geometricconstraints on the volume of remnant, midlatitude debris-covered glaciers. J.Geophys. Res.: Planets 119 (10), 2188–2196.

Li, H., Robinson, M., Jurdy, D., 2005. Origin of martian northern hemisphere mid-latitude lobate debris aprons. Icarus 176, 382–394. http://dx.doi.org/10.1016/j.icarus.2005.02.011.

Lucchitta, B.K., 1984. Ice and debris in the fretted terrain, Mars. In: Lunar andplanetary science conference proceedings, vol. 14, pp. B409–B418.

Mackay, S.L. et al., 2014. Cold-based debris-covered glaciers: Evaluating theirpotential as climate archives through studies of ground-penetrating radar andsurface morphology. J. Geophys. Res. Earth Surf. 119, 2505–2540. http://dx.doi.org/10.1002/2014JF003178.

Marchant, D.R., Head III, J.W., 2007. Antarctic dry valleys: Microclimate zonation,variable geomorphic processes, and implications for assessing climate changeon Mars. Icarus 192, 187–222. http://dx.doi.org/10.1016/j.icarus.2007.06.018.

Marchant, D.R. et al., 2002. Formation of patterned ground and sublimation till overMiocene glacier ice in Beacon Valley, southern Victoria Land, Antarctica. Geol.Soc. Am. Bull. 114, 718–730.

Mellon, M.T., Jakosky, B.M., 1993. Geographic variations in the thermal and diffusivestability of ground ice on Mars. J. Geophys. Res. Earth Surf. 98, 3345–3364.

Parsons, R.A., Nimmo, F., Miyamoto, H., 2011. Constraints on martian lobate debrisapron evolution and rheology from numerical modeling of ice flow. Icarus 214,246–257. http://dx.doi.org/10.1016/j.icarus.2011.04.014.

Phillips, R.J., Brown, C.D., Hauck, S.A., Harrington, B.W., Wieczorek, M.A., andHead, J.W., III, 1998. Preliminary geomorphology results from the MGS MOLAexperiment. In: Lunar and Planetary Science Conference XXIX, Abstract #1503.

Plaut, J.J. et al., 2009. Radar evidence for ice in lobate debris aprons in the mid-northern latitudes of Mars. Geophys. Res. Lett. 36, L02203. http://dx.doi.org/10.1029/2008GL036379.

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

Squyres, S.W., 1978. Martian fretted terrain: Flow of erosional debris. Icarus 34,600–613.

Squyres, S.W., 1979. The distribution of lobate debris aprons on Mars. J. Geophys.Res. Earth Surf. 84, 8087–8096.

Squyres, S.W., Carr, M.H., 1986. Geomorphic evidence for the distribution of groundice on Mars. Science 231, 249–252.

Tanaka, K.L. et al., 2014. Geologic map of Mars. U.S. Geological Survey ScientificInvestigation Maps, Map 3292.