An alternative interpretation of late Amazonian ice flow: Protonilus Mensae, Mars

Icarus 225 (2013) 495–505

Contents lists available at SciVerse ScienceDirect

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

An alternative interpretation of late Amazonian ice flow: Protonilus Mensae, Mars Colin J. Souness, Bryn Hubbard ⇑ Centre for Glaciology, Institute of Geography and Earth Sciences, Aberystwyth University, Aberystwyth, Wales SY23 3DB, UK

a r t i c l e

i n f o

Article history: Received 7 May 2012 Revised 27 March 2013 Accepted 28 March 2013 Available online 28 April 2013 Keywords: Mars Ices Mars, Surface Mars, Climate

a b s t r a c t The morphological properties of glacial deposits located within a cirque-like alcove in eastern Protonilus Mensae, Mars (Mars grid reference: 54.55° lon, 40.80° lat), have been interpreted as indicative of an early phase of glaciation involving ice flowing out from the main glacierized valley and into a higher-elevation box canyon or alcove. This interpretation implies that the elevation of the glacial deposits currently located within the alcove mark a minimum former ice elevation in this catchment, providing a regional datum for former ice thicknesses with implications for martian climate reconstruction. Here, we raise the possibility of an alternative interpretation. Detailed geomorphological mapping and evaluation of the deposits remaining within the alcove suggest that they may have been formed by the more conventional mechanism of ice flowing out from the alcove and not into it. Both small-scale and catchment-scale morphologies, allied to the site’s location and orientation within an otherwise conventionally glacierized catchment, are consistent with outbound flow. This alternative hypothesis has important implications for ice thicknesses both in the local area and, by extension, regionally. Ó 2013 Elsevier Inc. All rights reserved.

1. Introduction 1.1. Mars’ integrated glacial landsystems Landforms of glacial origin have been observed and mapped in both mid-latitude regions of present-day Mars (e.g. Squyres, 1978, 1979; Milliken et al., 2003; Souness et al., 2012). This ice appears to have flowed in recent geological times (e.g. Forget et al., 2006; Head et al., 2005; Marchant and Head, 2003), leading to the formation of what Head et al. (2010) described as ‘integrated glacial landsystems’. These ice deposits are possibly the remnants of a larger glacial ice mass that formed during a past martian ice age (Head et al., 2003). This martian ice age is thought to have occurred under climatic conditions markedly different from those that prevail on present-day Mars. The transition from this past climatic regime to that of the present-day is believed to have been driven by shifts in the obliquity and eccentricity of Mars’ orbital rotation (Read and Lewis, 2004; Ward, 1992), the most recent and substantial example of which is thought to have occurred approximately 5  106 years BP (Touma and Wisdom, 1993). Investigations using high-resolution imagery have shown that flowing ice on Mars often forms integrated patterns, representing mass transportation that begins in multiple discrete cirque-like alcoves and converges before commonly traveling for considerable distances down-valley (Dickson et al., 2008). These landsystems are composed of several distinctive flow-induced features, includ⇑ Corresponding author. E-mail address: [email protected] (B. Hubbard). 0019-1035/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.icarus.2013.03.030

ing glacier-like forms (GLFs) (Hubbard et al., 2011; Souness et al., 2012), lobate debris aprons (LDAs), and lineated valley-fill (LVF) (Squyres, 1978, 1979; Lucchitta, 1981). GLFs represent the firstorder component of this glacial system, originating in small valleys or cirque-like alcoves, whilst LDA (broad, rampart-like flows) are often formed where multiple adjacent GLFs converge, or icy material originates from a long escarpment. Where opposing LDA converge or coalesce, complex flow patterns and contortions are often observed; this surface being known as LVF (Squyres, 1978, 1979; Lucchitta, 1981; Souness and Hubbard, 2012b). The surface patterning produced by ice flow within these systems is diverse and complicated, exhibiting many textures without obvious counterparts on Earth. As such, flow histories are often difficult to extrapolate and the mechanisms of ice flow are still poorly understood. 1.2. Ice loss since the last martian glacial maximum (LMGM) Although substantial areas of present-day Mars’ mid-latitudes are characterised by glacial features and the action of flowing ice (e.g. Milliken et al., 2003; Souness et al., 2012; Squyres, 1978) (Section 1.1.), the flow of these formations does not appear to result from the gravity-driven creep of accumulated snow, such as is the case in Earth’s glaciers. In the absence of evidence for elevation-dependent snow accumulation on Mars, we infer the presence of a pre-existing mass of deformable ice-rich material (Souness et al., 2012). This interpretation, along with evidence from other observations, such as raised glacial trimlines (Dickson et al., 2008, 2010) and the degraded remains of tropical mountain

496

C.J. Souness, B. Hubbard / Icarus 225 (2013) 495–505

glaciers (Head and Marchant, 2003), points to these icy flow features being the remnants of a once much larger ice deposit that is thought to have existed under previously warmer conditions on Mars, perhaps associated with a shift in orbital obliquity. However, the precise extent and volume of this hypothesised last martian glacial maximum (LMGM) ice mass remains unknown. Some estimates of possible down-wasting and ice mass loss between the LMGM and the present day have been made. Dickson et al. (2008) proposed a minimum surface deflation of 920 m, suggesting that Mars’ mid-latitudes once hosted ice deposits that were >1 km thick. These calculations were based on the identification of what are thought to be trimlines and glacial highstands observed in the Protonilus Mensae region of Mars’ northern mid-latitudes (Dickson et al., 2008), a well as in craters in Phlegra Montes (Dickson et al., 2010). Protonilus Mensae is part of the dichotomy boundary (which separates Mars’ raised southern highlands from its northern lowlands) and is topographically complex, characterised by long, incised valleys and ‘fretted terrain’ consisting of isolated buttes separated by wide mesas (Sharp, 1973). Protonilus Mensae, along with the similar and neighbouring regions of Deuteronilus Mensae and Coloe Fossae, also contains a very high proportion of Mars’ GLFs and associated glacial features (e.g. Souness et al., 2012). Therefore, care must be taken in the interpretation of what is a very complex, varied, and often somewhat ambiguous assemblage of landforms. 1.3. Existing hypotheses relating to former ice flow directions in Protonilus Mensae Dickson et al. (2008) interpreted an ice mass located in a large box canyon or cirque-like alcove in eastern Protonilus Mensae (hereafter referred to as ‘Protonilus Mensae 1’ [PM1]) as having flowed into that alcove rather than out from it (Fig. 1). Although Dickson et al. (2008) described mass sourcing and subsequent flow patterns for the Protonilus Mensae area as generally being characterised by ‘‘accumulation. . . in fretted valleys, in alcoves along the bounding scarp, and in alcoves surrounding the massifs, with flow extending away from these regions into local lows. . .’’, they invoke a different history for PM1. The authors suggest that, in this specific case, mass flowed during an early, large-scale glacial phase (equivalent to the LMGM) into the alcove instead of out from it. This interpretation is based on the presence of a large, loop-like ridge that encompasses a lobe of ice-rich material and which, when viewed in planform, is oriented into alcove PM1 rather than out of it as is more conventionally the case (Fig. 1). Dickson et al. (2008) go on to suggest that general flow pattern was then reversed to the more normal condition of down-slope flow (i.e. out from the raised alcove) following substantial down-wasting of the main trunk glacier, possibly during a later, smaller-scale phase of valley glaciation. Although these later events at least partially altered the detailed geomorphology of the ice-rich material present within the alcove, they did not obscure the orientation of the primary loop-like lobe. Dickson et al. (2008) attributed the positive relief of the ridge forming the boundary of this lobe to compression and local folding resulting from the inward flow of material abutting the alcove’s walls. Dickson et al. (2008) equated the ( 800 m) elevation of this compressional lobe with other glacial deposits and possible trimlines elsewhere in the area. The  800 m-high terminus of this hypothesised invasive ice tongue is currently raised well above the ice-rich surface of the adjacent valley floor and has been proposed as representing the elevation of an LMGM high-stand and a datum for estimations of regional post-LMGM ice loss (Dickson et al., 2008). Although this interpretation has been widely adopted (e.g. Dickson et al., 2010; Parsons et al., 2011; Pearce et al., 2011; Soare et al., 2012; Head, 2012), a case may be made for revisiting

the evidence for the initial ice invasion of PM1 because (i) the terrain in this locality is complex, (ii) morphologies observed on the surface of viscous flows in this area are locally variable and somewhat ambiguous, and (iii) ice invasion into a high-elevation cirque is rare on Earth, requiring the surface elevation of the ice in the main trunk valley to be higher than that in the cirque. In this paper we re-examine the morphological evidence of former flow patterns at the cirque-like alcove PM1, and raise the possibility of an alternative interpretation to that proposed by Dickson et al. (2008). Such an investigation is important because the reconstructed glacial geomorphology of PM1 has implications for the interpretation of similar features elsewhere on Mars. The datum at  800 m also has implications for reconstructing Mars’ LMGM ice volume and for constraining models of Mars’ glaciation and climate (although other, independent, evidence does exist pointing to the possibility of massive martian glaciation [e.g. Madeleine et al., 2009]). 2. Study site 2.1. Location The flow feature PM1 forms a north-east-facing alcove located on the southern edge of an enclosed glacial catchment in eastern Protonilus Mensae (Mars grid reference 54.55° lon, 40.80° lat). This broader glacierized catchment appears to be filled with ice-rich material that has flowed from multiple sources. Numerous GLFs occupy the walls of the catchment’s escarpments, several of which form the southeast face of a raised butte, typical of those that characterise the ‘fretted terrains’ (Sharp, 1973) in this area. These GLFs appear to contribute a substantial proportion (although by no means all) of the icy material that has historically flowed into the catchment. This mass then merges into a number of LDAs that appear to flow away from their accumulation areas before finally converging as a complex LVF in the catchment’s centre. A portion of this contorted material appears to be leaving the catchment by flowing through two narrow topographic channels, one of which lies to the north of the basin and the other to the west. 2.2. Description of PM1 Flow feature PM1 (Fig. 1) is 6.5 km wide and 11.8 km long, indicating a surface area of 77 km2. The feature fills a broad, amphitheatre-shaped cirque-like alcove on the southern wall of the catchment and is distinctive for the wide, ‘‘loop-like lobe’’ (Dickson et al., 2008) of raised material that can be seen to occupy its floor. This lobe is aligned parallel to, and at an approximately uniform distance (radially) of some 2300 m from, the alcove’s lateral walls and headwall. Detailed mapping reveals that the lobe is slightly further from the headwall than the valley walls, and that the width of the peripheral surface decreases towards the open end of PM1. This loop-like lobe of material encloses a tongueshaped area of smoothed appearance that gives a clear overall impression of an invasive ice body that has flowed into the alcove, as interpreted by Dickson et al. (2008). The topography of PM1 is typical of terrestrial cirques and of cirque-like alcoves observed elsewhere on Mars in that the mass occupying the valley floor slopes downhill away from the headwall (Dickson et al., 2008; their Fig. 2). Thus, the loop-like lobe is currently sloping out from the alcove and, under present-day conditions, gravity would be driving flow in that direction. Dickson et al. (2008) acknowledged this fundamental geometrical configuration and argued that since the initial large-scale invasion of PM1, the regional ice mass has now thinned and accords more closely with local bed slope – driving more recent and smaller-scale flow

C.J. Souness, B. Hubbard / Icarus 225 (2013) 495–505

497

Fig. 1. The location and detail of the cirque-like alcove PM1. The upper left inset (a) shows Protonilus Mensae: an area of the ‘fretted terrains’ in Mars’ northern mid-latitudes. This context map is gridded in decimal degrees and has been shaded using Mars Orbiter Laser Altimeter (MOLA) elevation data. Blue indicates lower elevations and yellows higher elevations. Expansion (b) illustrates CTX image P22_009455_2214_XN_41N305W (georeferenced to a mosaic of Thermal Emission Imaging System (THEMIS) daytime infra-red images). The dashed box (i) indicates the cirque-like alcove PM1. This dashed box is expanded in (c) which shows a detailed view of the PM1 feature. All of (a–c) are oriented north-up and (b and c) are illuminated from the west. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

out from the alcove. We fully agree with the authors’ interpretation of present-day flow being directed out from the alcove, in accordance with ice surface slopes. In this paper, we focus solely on the glacial and glacial–geomorphological evidence for the initial large-scale phase of invasive ice flow, seeking to evaluate whether an alternative, more conventional, interpretation may be plausible. 3. Methods 3.1. Image sources We use Context Camera (CTX) (6 m per pixel resolution) and High Resolution Imaging Science Experiment (HiRISE) camera (30 cm per pixel resolution) imagery to examine, map and inter-

pret the evidence for the directionality of geologically-recent flow within the ice deposits located at PM1 (see Section 2). 3.2. Surface mapping Images were initially viewed using the Mars Image Explorer (MIE) interface, hosted by Arizona State University (ASU) (viewer.mars.asu.edu). These data were subsequently downloaded from the MIE and from the HiRISE website (hirise.lpl.arizona.edu/amazitisi.php), before being uploaded into ESRI’s ArcMap 9.3 geographic information system (GIS) program. In ArcMap, the images were georeferenced to the 128 pixels per degree Mars Orbiter Laser Altimeter (MOLA) gridded digital elevation model (DEM) and the 256 pixels per degree global mosaic of Thermal Emission Imaging

498

C.J. Souness, B. Hubbard / Icarus 225 (2013) 495–505

Fig. 2. Geomorphological maps showing the landform and surface texture assemblages observed and classified on the surface of PM1 and its parent catchment. (a) Landforms and surfaces classified at a catchment scale, accompanied by a context image (CTX image P22_009455_2214_XN_41N305W) in (i)). (b) An expanded detail of the PM1 alcove (identified in ‘a’ by a dashed box). The small-scale streamlined mounds shown in Fig. 3 have been classified and mapped in (b). At the time of writing this manuscript only two HiRISE images were available for this locality (ESP_018725_2215 and PSP_001834_2215), providing incomplete coverage. Thus, mapping of smaller-scale features was limited to the coverage shown in Fig. 3a. (c) Illustrates reconstructed flow directions, interpreted from the landforms, slopes and textures mapped in (a and b). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

System (THEMIS) daytime infra-red images. This raster dataset was then used to build geomorphological maps. 3.2.1. Interpretation of PM1 surface textures and landforms The geomorphological maps of PM1 were drawn from interpretations made through manual inspection. These interpretations were informed using terrestrial glaciological principles and published literature as a source of analogs. 4. Results 4.1. Feature-scale geomorphological mapping Detailed geomorphological mapping of PM1 (Fig. 2) reveals three distinct primary surface types: (i) glacial material; (ii) sub-

dued glacial surfaces, and; (iii) raised, textured areas. These three surfaces are associated with numerous smaller-scale surface textures and features, including: (iv) a suite of flow-parallel and flow-transverse lineations; (v) pits and troughs; (vi) moraine like ridges (MLRs, following Arfstrom and Hartmann, 2005); (vii) degraded (and in some cases distorted) craters, and (viii) widely-distributed small-scale flow-parallel streamlined mounds (Fig. 3), similar to the ‘mound-and-tail terrain’ described by Hubbard et al. (2011). The division of PM1’s surface into three textural types reveals a horseshoe-shaped band of raised, textured ground running around the amphitheatre, lying approximately 2300 m distal from the headwall and, on its concave side, enclosing a central, lobe-shaped area of subdued glacial material. This is the ‘loop-like lobe’ described by Dickson et al. (2008), which we here classify under

C.J. Souness, B. Hubbard / Icarus 225 (2013) 495–505

Fig. 3. Images illustrating the small-scale streamlined mounds present on the surface of the flow feature PM1 (Section 4.1). (a) A section of CTX image P22_009455_2214_XN_41N305W provides the context of the expansions (b and c) by small white boxed areas. Panel (a) also shows as larger, slightly oblique white boxes the extent of HiRISE images ESP_018725_2215 and PSP_001834_2215 which cover only a portion of the scene. The enlargements (b–d) are sections of HiRISE image ESP_018725_2215_RED, the last of which shows a close-up of the ‘moundand-tail’ morphology of several streamlined mounds located within PM1’s central, enclosed lobe. Note the stoss and lee morphology of the mounds in (d), interpreted to indicate flow to the NNW. All images are oriented north-up and illumination is from the west.

‘raised, textured surfaces’. Along the centreline of this enclosed lobe lies a second area of raised and somewhat textured terrain which, combined with the principal band of textured surface (above), gives PM1 a concentric appearance. The concentric surface of PM1 is associated with a sequence of MLRs that, for the most part, are nested on both sides of, and run parallel to, the principal loop-like lobe of raised and textured material (above). However, at the open (northern) end of PM1, these MLRs assume a different spatial layout, defining two separate lobes that protrude from each opposing side of the PM1 alcove, specifically from the space enclosed between the alcove’s valley walls and the loop-like lobe of raised material (Fig. 4). In stark contrast to the primary (outward-opening lobe) these lateral lobes are orientated opening inwards, towards the PM1 alcove – consistent with outwards flow towards the catchment’s main valley floor. The easternmost of these lobes is particularly well-defined (Fig. 4) and in this case, beyond the last well-defined MLR, arcuate chevron-like formations (flow transverse lineations) occur intermittently, but with diminishing clarity, in a northward and then westward direction, at which point the two separate lobes begin to converge. The western lobe is fundamentally similar but less well-defined (Fig. 5). Where these eastern and western lateral lobes converge, the central part of PM1’s surface, enclosed by the primary lobe,

499

appears to pinch out somewhat, giving this central zone a teardrop-like shape, the tailing end of which gradually diminishes in diameter and points progressively westwards with distance (northward) from the centre of the PM1 alcove. Those MLRs and chevrons located on the eastern side of the PM1 alcove were identified in earlier work by Dickson et al. (2008) – and were attributed to recent conventional flow out from PM1. However those to the west (Fig. 5) were not. The small-scale ‘streamlined mounds’ (Fig. 3) mapped within PM1 (Figs. 2b, 4 and 5) exhibit a complex layout. Within the headwall and valley-wall-proximal zone of subdued glacial material, the mounds appear to originate from the valley sides (with the steeper stoss-side being proximal to the valley wall), proceeding down-slope towards the centre of PM1, enclosed by the loop-like lobe of raised terrain. As the mounds approach this raised area they gradually begin to change orientation, assuming a trajectory aligned less adjacent, and more parallel, to the valley walls. In the south-eastern half of PM1 this turn is to the right (the mounds rotating clockwise) (Figs. 2b, 3 and 4b), whereas in the northwestern half of PM1 the turn is to the left (the mounds rotating anti-clockwise) (Figs. 2b, 3, 4b and 5). In both cases, the mounds continue down-slope, flanking the enclosed area and intervening loop-like lobe, their narrower, lee-ends, pointing towards the alcove entrance and the two apparently outbound lobes which emerge proximal to the opposing valley sides (see above and Fig. 4). In all of these observed instances, streamlined mounds are oriented with their wider and steeper ‘stoss’ ends located at the uphill, valley-wall end (Fig. 3). The development of streamlined mounds within PM1’s central enclosed lobe is less widespread and less well defined than on the headwall proximal surface. However, those that could be mapped (located predominantly on the headwall–proximal edge of the lobe, but with some examples located towards the central raised mass and in the intervening space [Figs. 3 and 4]) are arranged roughly adjacent to the loop-like lobe’s margin, appearing to converge towards the raised and textured area in the centre of the lobe before turning towards a more flow-parallel orientation in the direction of the alcove opening (Figs. 2b and 4b). Again, the stoss side of the observed mounds consistently faces uphill, towards the valley wall (Fig. 3). It is also interesting to note that these raised streamlined mounds, observed across PM1, appear in some cases to have a lumpy surface (e.g. Fig. 3d). 4.2. Local catchment-scale flow mapping Surface structures were mapped for the wider catchment centred to the north east of the PM1 alcove, as shown in Fig. 2. Inspection of these mapped features and textures, combined with the extrapolation of apparent spatial associations, makes it possible to construct an interpretive map of former mass transport pathways. The distribution of raised moraine-like ridges (MLRs), elongated pits and troughs, and in some cases deformed craters, provided indicators for past flow vectors, especially where flow could be traced directly from the terminus of a GLF or similar mass-source alcove (e.g. PM1). Overall, it appears that mass was formerly sourced in several individual sub-catchments, the ice from each originating within or adjacent to one or more GLFs or LDAs. Here, near the alcoves providing the mass source, the surface expression of each flow unit is generally broad and laterally extensive. However, these mass surfaces can be traced away from the source area, where they gradually become laterally constricted, resulting in a considerable reduction in surface area and the production of clear lateral surface structures (Fig. 6). Clear examples are visible in the north west of the catchment and along its southern flank, including the area proximal to and apparently extending directly from alcove PM1.

500

C.J. Souness, B. Hubbard / Icarus 225 (2013) 495–505

Fig. 4. Detailed surface geomorphological maps of the two lobes that protrude from the lateral sides of the PM1 alcove. (a) A detail of the PM1 alcove featured in CTX image P22_009455_2214_XN_41N305W. (b) A geomorphological map of the PM1 alcove (also shown in Fig. 2b). The extruding portion of each lobe is highlighted in both (a and b) by rectangular boxes labelled ‘(i)’ and ‘(ii)’, corresponding to the western and eastern lobes respectively. The highlighted lobes protrude from the space enclosed between the alcove’s valley walls and the horseshoe-shaped band of raised material (see the attached legend), and are delineated by moraine-like ridges (MLRs) and arcuately-shaped flow-transverse lineations, shown in brown and red respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The map in Fig. 6, coloured to indicate mass sourced from various individual sub-catchments, indicates that ice originating in the PM1 alcove can be traced northwards before making a sharp turn to the west (following local gradients) where it merges with neighbouring ice flows, is compressed into a narrow band of material, and is extruded from the wider catchment system through the topographic channel to the west. 5. Discussion 5.1. Feature-scale geomorphological mapping Small-scale streamlined mounds were observed and mapped within PM1. These mounds displayed a stoss (broad and steep) and lee (narrow and shallow) morphology, with the stoss ends all facing uphill towards the valley headwall and sides. They also appeared lumpy, suggesting that they may be compositionally distinctive from the adjacent glacial surfaces. Although insufficient evidence is available to evaluate this suggestion, their raised morphology would also be consistent with a distinctive composition being responsible for making them more resistant to erosion than the adjacent glacial material. The morphology of these stoss and lee features is similar to at least two glacigenic forms with terrestrial analogs. First, they appear similar to the mound-and-tail terrain observed on a GLF located in eastern Hellas and described

by Hubbard et al. (2011). These authors compared these forms to terrestrial drumlins, which have a similar stoss and lee morphology. On Earth, drumlins are classified as being depositional bedforms of subglacial origin. In the case of PM1, however, we interpret the streamlined mounds as being supraglacial features, so the drumlin analog is valuable only in that it links the observed stoss-and-lee morphology to directional mass movement. Second, several types of longitudinal structures, including foliation, flow stripes and debris stripes (e.g., Glasser et al., 1998), on Earth’s glaciers and ice streams are visually similar to the small-scale streamlined mounds located within PM1. In contrast to (subglacially-formed) drumlins, longitudinal foliation occurs on the surface of glaciers (Fig. 7) and is created by ice flow and deformation (e.g. Hambrey, 1975). This deformation can, through the folding and re-organisation of adjacent ice layers (with different compositional characteristics), create flow-parallel streaks or relief features similar to those observed on the surface of PM1 (Fig. 3). Such foliation can also be compositionally variable along its length as a consequence of variations in, for example, bubble content (Hambrey, 1975) and the concentration of debris sourced from the glacier bed entrained within it (e.g. Glasser et al., 1998; Hubbard et al., 2004) (Fig. 7b). Longitudinal foliation can also occur at very different scales, from 10 1 m across on the surface of valley glaciers to 104 m across on the surface of large glaciers and ice streams, where the morphology is generally reported as ‘flow

C.J. Souness, B. Hubbard / Icarus 225 (2013) 495–505

501

Fig. 5. A detailed view of the western lateral lobe complex, represented as HiRISE image PSP_001834_2215 in (a) and as a geomorphological map in (b). The area shown corresponds approximately to (‘i’) in Fig. 4b. Panel (a) is illuminated from the left. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6. Interpreted flow map for PM1 and its immediate surroundings. Our interpreted flow directions are shown in (a). Note that arrows of differing lengths do not imply corresponding flow velocity. Individual and distinct flow units have been delineated and separately colourized to aid in the interpretation of local flow characteristics. A similarly colourized map with mapped landforms and surface textures is shown in the inset ‘(i)’ for reference. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

stripes’ (e.g., Glasser and Gudmundsson, 2012). Fahnestock et al. (2000) described these features as ‘‘surface topographic ridges. . . with metre-scale relief, hundreds of metres to km in width, (and) tens to hundreds of km in length’’. Campbell et al. (2008) similarly described flow stripes as ‘‘surface undulations with kilometre-scale spacing and metre-scale relief’’. These forms are consistently characterised as being curvilinear (e.g. Casassa et al., 1991), and they

are recognised in the vast majority of cases as being aligned parallel to flow. One interpretation of the initial formation of flow stripes involves lateral compression, for example induced by the convergence of two or more individual flow units. Importantly, this model involves the formation of a broader ‘stoss’ end in the upstream position of the newly-formed flow stripe (Glasser and Gudmundsson, 2012). The raised nature of PM1’s streamlined

502

C.J. Souness, B. Hubbard / Icarus 225 (2013) 495–505

Fig. 7. Images of longitudinal foliation on the surface of glaciers located in Svalbard. Panel (a) shows tight foliation on the surface of Comfortlessbreen. The direction of ice flow is towards the camera. Image courtesy of ‘glaciers online’, available at http://www.swisseduc.ch/glaciers/glossary/foliation-en.html). Panel (b) shows more open foliation of greater relief (up to 1 m high and several metres across [see two rucksacks and rifle case in middle right of image for scale]) present on the surface of Austre Brøggerbreen. The direction of ice flow is away from the camera. In this case the foliation is associated with a high concentration of incorporated debris sourced from the glacier’s bed (Hubbard et al., 2004). This flow-parallel foliation is similar to the streamlined mounds observed on PM1 (Fig. 3), particularly the closely-packed example of streamlined mounds shown in Fig. 3b.

mounds, and their occasionally lumpy appearance (Section 3.1), are consistent with spatially variable surface composition, and could be explained by spatial heterogeneity in ice characteristics and/or debris content. We therefore suggest that, although tentative, the most suitable analog we currently have for these mounds is as broken-up longitudinal foliation and we therefore provisionally interpret the mounds observed on PM1 as being a product of ice flow and deformation. Consequently, we take their orientation as being indicative of local motion, indicating a down-slope and outwards direction. This interpretation is based primarily on the observation that the wider stoss sides of the observed mounds are in all cases located at the uphill, valley-wall-proximal end of the features. Our interpretation of outbound flow in PM1, based partly on the orientation of the observed streamlined mounds, is supported by the pattern observed in the overall layout of those mounds (Figs. 2b, 4 and 5) which in both the centrally enclosed and peripheral arc sections of the PM1 alcove shows convergence from both sides of the alcove and subsequent outward (northward) flow (Figs. 4 and 5). Therefore we propose that the location and orientation of the observed PM1 mounds, both in the valley-wall-proximal arc of subdued terrain and within the enclosed central lobe (Section 4.1; Fig. 3), indicate that material was universally sourced from the PM1 alcove’s periphery, most likely from the valley walls and alcove headwall, and that this mass flowed towards, and ultimately converged in, the valley centre. It should be noted that these streamlined mounds are observed only on the mapped ‘subdued glacial surfaces’, and not on the adjacent ‘raised and textured areas’ mapped in Figs. 2 and 4. One such raised, textured area is the primary lobe originally identified by Dickson et al. (2008) and interpreted by those authors as being a compressional feature caused by the collision and subsequent deformation of two mass units. Both our interpretation and that of Dickson et al. (2008) attribute the formation of this ridge to large-scale compression (although the directions of the deformational forces involved are opposite), and we suggest that any small-scale pre-existing supraglacial forms, including foliation, would almost certainly have been destroyed or obscured by compressive forces. Thus, it is not surprising that streamlined mounds are absent from the raised, textured areas. Unfortunately, due to their small size, it was only possible to map the distribution of streamlined mounds using the high resolution imagery provided by the HiRISE camera. Since only two HiRISE images were available for this location at the time of writing the streamlined mounds mapped in Figs. 2, 4 and 5 are confined to

these imaged areas – and therefore appear to be present only in the central portion of the PM1 alcove. However, the fact that these mounds appear across the area of subdued glacial surfaces captured in HiRISE images implies that, were the rest of PM1 to be imaged at as high a resolution, streamlined mounds would likely be observed there as well. Further to the streamlined mounds discussed above, two distinctively convex lobes of material were observed to extrude from the two opposing ends of PM1’s north-facing open end (Figs. 4 and 5). These lateral lobes are evidenced by the presence of various arcuate chevrons and MLRs in the geomorphological record (Figs. 4i, ii and 5), suggesting the expulsion of mass. The more easterly of these lobes was identified by Dickson et al. (2008), and their interpretation of outward flow (during a smaller-scale glacial phase subsequent to the regional glaciation responsible for driving flow into PM1) is entirely consistent with our own. In fact, this outward-facing, convex down-slope morphology not only suggests outward flow, but in this context appears to preclude the possibility of inbound flow. This observation, combined with those summarised above, suggests that flow within (and immediately adjacent to) the PM1 alcove was of a conventional outbound type. The westward propagation of the eastern outbound lobe series and its eventual encroachment upon the more westerly lobe (Figs. 4i and 5) also ‘cuts off’ the primary central lobe from the rest of the (northern) catchment. Therefore, if an early phase of ice flow was directed inwards, into the alcove of PM1, the smaller-scale local evidence for that flow appears to have been over-printed by subsequent conventional outward-directed flow. This is entirely possible, but the principle of parsimony may well suggest a more conventional interpretation: that all flow was directed outwards and an explanation therefore needs to be proposed for the orientation of the principal PM1 lobe (Section 6). Finally, Figs. 3 and 4 also reveal the presence of a previouslyunreported strip of ‘raised textured area’ orientated longitudinally, approximately north–south, along PM1’s centreline. The shape and orientation of this feature are consistent with transverse compression, i.e. acting in an approximately east–west direction inwards from the lateral valley-side walls of PM1. 5.2. Catchment-scale flow mapping Catchment-scale mapping of the area, reconstructing former flow directions (Fig. 6), is also consistent with an interpretation involving outbound ice flow from PM1. Here, the reconstructed flow direction for PM1 compares well with examples from

C.J. Souness, B. Hubbard / Icarus 225 (2013) 495–505

similarly-configured glacial catchments on Earth: the ice appears to leave PM1 and turn immediately westwards, merging with ice masses sourced from adjacent GLFs and LDAs. After merging, the combined mass continues to flow down-slope and narrows as it is constrained by the local topography, exiting the catchment to the west. The layout of the surface structures illustrated in Figs. 2, 4 and 5, and which delineate the flow units presented in Fig. 6, are entirely compatible with consistent ice flow directions out from alcoves into the main trunk valley glacier which flows north-east to south-west. Again, while this is not incompatible with an early phase of inbound ice flow invading PM1, followed and overprinted by outbound flow from the alcove, it is simpler and consistent with the detailed evidence revealed by our mapping. 5.3. Testing the ‘inbound flow’ hypothesis According to the hypothesis presented by Dickson et al. (2008) the mass enclosed within cirque-like alcove PM1 in eastern Protonilus Mensae represents ice from an early-stage large-scale glacial (LMGM) phase that flowed into the alcove, rather than out of it. Subsequently, ice in the main feeder valley thinned substantially, allowing at least ice located around the principal lobe in that alcove to reverse direction and flow back out from the alcove, more conventionally following local bedslope. The remnant ice in the invaded alcove thereby represents a preserved glacial ‘highstand’ from the LMGM. On Earth, and following basic physical principles, the main context in which ice flowing along a trunk valley could invade a higher-elevation cirque is when the ice surface in that cirque is lower than that in the trunk valley. Such conditions are rare on Earth (and apparently on Mars) and are most likely to occur when a lobe from a regionally-generated ice mass advances into an area of substantially lower accumulation, i.e. under conditions of a steep lateral gradient of declining mass balance. Refinements to this scenario might include conditions whereby the invaded alcove is over-deepened by prior glacial or nival erosion, or when the surface elevation of ice in the trunk valley suddenly rises (or that in the alcove suddenly lowers) due to a flow instability such as occurs during surge-type activity (Kargel, personal communication). It is possible for regional ice to invade an area if the lowest extent of that ice mass’ accumulation zone is located a considerable distance away from the invaded region (otherwise ice would also develop locally at higher elevations). An example of just such a mountain range on Earth is the now de-glaciated Khibiny Mountains, Russia, which the geomorphological record suggests were overridden from the west by the Fennoscandian Ice Sheet during the last glacial (Hättestrand et al., 2007). According to this model, for a mountain range to be overridden by allochthonous ice, or for individual cirques within a range to be invaded, it is fundamentally necessary that local conditions are not conducive to the build-up of ice. On Earth, this might be due to locally warmer temperatures and/or lower precipitation, for example in the McMurdo Dry Valleys, Antarctica (Dickson, personal communication, 2012). Here, local aridity may well allow intermittent invasion by outlet glaciers from the East Antarctic Ice Sheet. On Mars, however, geologically recent conditions in Protonilus Mensae appear to have been well suited to the widespread accumulation and flow of glacial material. One possibility is that the earlystage invasion of the region occurred by a far larger ice mass, but there is currently no specific evidence pointing to such conditions. Further, if local conditions were conducive to invasion by allochthonous ice then it would be anticipated that many if not all of the high-level alcoves similar to PM1 would have been similarly invaded and consequently now characterised by similar geomorphic evidence. This is not the case. Indeed, the analysis of Souness et al. (2012) shows that GLFs are predominantly poleward-facing in both of Mars’ hemispheres, suggesting that ice accumulates more

503

readily and/or is preferentially preserved at these orientations. Since PM1 has a generally northerly aspect it is, if anything, more likely to have accumulated and/or retained local ice than similarlyelevated south-facing alcoves in the area.

6. An alternative hypothesis The interpretation of Dickson et al. (2008) involves early-stage ice advancing uphill and flowing into the PM1 alcove. The main line of evidence supporting this interpretation is the primary loop-like lobe of material that has the unusual orientation of opening outwards from the alcove (Fig. 1). As an alternative hypothesis we raise the possibility that all of the geomorphological evidence available can be explained by more conventional outwards, down-slope flow. Numerous glacial events have previously been proposed for this region (e.g. Dickson et al., 2008, 2010). We propose that the older of these events – rather than invading PM1 – left a residual deposit that occupied the floor of the alcove prior to a subsequent re-advance of ice during a more recent smallerscale event. Evidence of just such intermittent glacial pulses (possibly climatically-driven) has been observed elsewhere in Mars’ northern mid-latitudes (e.g. Grindrod and Fawcett (2011), who observed regularly spaced textural bands oriented transverse to the direction of flow in LDA in Mars’ northern hemisphere). During this hypothesised recent phase of valley glaciation we propose that mass entered the system from the alcove’s perimeter headwall and valley sides rather than as elevation-controlled precipitation as is common on Earth. This perimeter-based supply is consistent with the orientation of small-scale streamlined mounds and the relatively unambiguous directionality indicated by the morphology of the outbound lateral lobes that enter the catchment system from each side of the PM1 alcove (Section 5.1), consistent with the interpretations made by Dickson et al. (2008). We propose that this material advanced down-slope from all of the alcove’s bounding walls and converged on the remnant icy material surviving in the alcove’s centre (Fig. 8). This convergence caused the compression of both ice units along their shared boundary, resulting in the formation of the main loop-like lobe of raised, textured material, as well as the various nested MLRs that occupy the area (Figs. 2 and 4). Such compression could also explain the formation of the raised and slightly textured central strip (Fig. 1) that is present along the centreline of the enclosed lobe (Figs. 2 and 4). Mapping and inspection of small-scale streamlined mounds within this centrally enclosed lobe also suggest an outbound flow component existed, or indeed currently exists, here also. Several of the small-scale morphologies reported herein are also consistent with flow that is outbound from alcove PM1. These include the orientation and morphology of small streamlined mounds, the orientation of the flanking outbound lobes, the layout of flow structures at the entrance to the PM1 alcove which currently appear to cut it off from the wider catchment’s central mesa, and the longitudinal north–south orientation of compressional terrain within the loop-like lobe itself, suggesting an arcuate compressional regime orientated orthogonal to the valley walls. While this alternative interpretation can explain the unusual primary loop-like lobe reported by Dickson et al. (2008) it does additionally require (i) the presence of an pre-existing ice-rich mass within the alcove at the onset of valley glaciation, (ii) the radially-uniform supply of new mass to the alcove during that valley glaciation, and (iii) some mechanism to explain the survival of the principal lobe opening outwards from the alcove. The survival of a remnant ice mass from an earlier large-scale glacial phase is not disputed, is common on Mars (e.g., Hubbard et al., 2011) and is required by both the inbound and outbound hypotheses. Further, the relative scarcity of streamlined mounds in the central area of

504

C.J. Souness, B. Hubbard / Icarus 225 (2013) 495–505

whereas PM1 had a diameter of 6.5 km. This relatively large size means that, in order to expel the pre-existing older ice out from the alcove and into the principal trunk valley, the late-stage valley glaciation phase needed to advect that ice further from alcove PM1 than from other, smaller hollows. Instead, this small-scale valley glaciation phase only resulted in the partial advection and compression of older deposits rather than their collective expulsion from PM1, creating the parabolic morphology seen at the present-day. Second, a local change in bed conditions such as the presence of a bedrock bump (reducing overlying ice thickness and therefore the local driving stress) could reduce the flow speed of the ice in the centre of the alcove, effectively holding that central lobe back relative to the ice immediately adjacent to it. 7. Conclusion We believe that the present-day suite of landforms in and around the cirque-like alcove PM1, observed at the catchmentscale, feature-scale and small-scale, may be explained by conventional outbound flow, both during a larger-scale glacial phase and a later more local phase of valley glaciation. We place this interpretation side-by-side with that of Dickson et al. (2008), who interpreted the evidence in terms of invasive flow into the hollow during the earlier glacial phase, overprinted by later outbound flow. Room exists for multiple hypotheses at this stage, and both models require some degree of special pleading in terms of the primary lobe being preserved only in this particular alcove. Nonetheless, the more conventional model we propose has the benefit of simplicity and does not require that medium-scale ice flow into or out from this hollow reverses during two phases of glaciation. Which of these competing hypotheses is correct is potentially important because of their influence on the area’s glacial reconstruction: the Dickson et al. (2008) flow-reversal model points to a regional glacial high-stand at an elevation of  800 m while the conventional model advanced as a possible alternative herein simply points to the a rather distinctive highelevation source area essentially similar to other alcoves in the area. Acknowledgments

Fig. 8. Schematic illustration of a potential mechanism for the formation of the surface morphologies present in the cirque-like alcove PM1. We hypothesise that a body of residual ice survived from a major previous phase of regional glaciation and occupied the PM1 alcove (a). During a later smaller-scale glacial phase or stadial, mass entered PM1 from the bounding valley sides, indicated by the superimposed bold arrows in panel (b). This mass converged on the pre-existing residual ice mass, caused compression and thickening, indicated by the superimposed bold arrows in panel (c).

We thank three anonymous reviewers and the Editor for their very helpful comments, which have helped to improve the manuscript greatly. We acknowledge the assistance of Duncan Quincey for commenting on an earlier version of the manuscript and Neil Glasser for assisting with the mapping methods. Finally, we thank the National Environmental Research Council (NERC) for supporting CS and the Climate Change Consortium of Wales (C3W) for supporting BH. References

the principal lobe (Section 5.1; Fig. 2) is consistent with this surface being older than the flanking material, as smaller-scale landforms may have been progressively erased or overprinted through time. In terms of requirement (ii), the supply of new material from the walls of the alcove is as likely as any other spatial arrangement of mass supply on Mars, where this ice was likely emplaced throughout the mid-latitudes during periods of relatively high obliquity during late Amazonian time (e.g., Forget et al., 2006). Finally, the preservation of this uncommonly orientated primary lobe can be explained by some combination of at least two processes that may be particular to this alcove. First, PM1 is abnormally large compared to most first-order glacial hollows present in Mars’ mid-latitudes; these are generally only 1–2 km in diameter

Arfstrom, J., Hartmann, W.K., 2005. Martian flow features, moraine-like ridges, and gullies: Terrestrial analogs and interrelationships. Icarus 174, 321–335. Campbell, I., Jacobel, R., Welch, B., Pettersson, R., 2008. The evolution of surface flow stripes and stratigraphic folds within Kamb Ice Stream: Why don’t they match? J. Glaciol. 54, 421–427. Casassa, G., Jezek, K.C., Turner, J., Whillans, I.M., 1991. Relict flow stripes on the Ross Ice Shelf. Ann. Glaciol. 15, 132–138. Dickson, J., Head, J.W., Marchant, D.R., 2008. Late Amazonian glaciation at the dichotomy boundary on Mars: Evidence for glacial thickness maxima and multiple glacial phases. Geology 36, 411–414. Dickson, J.L., Head, J.W., Marchant, D.R., 2010. Kilometer-thick ice accumulation and glaciation in the northern mid-latitudes of Mars: Evidence for crater-filling events in the late Amazonian at the Phlegra Montes. Earth Planet. Sci. Lett. 294, 332–342. Fahnestock, M.A., Scambos, T.A., Bindschadler, R.A., Kvaran, G., 2000. A millennium of variable ice flow recorded by the Ross Ice Shelf. Antarctica. J. Glaciol. 46, 652– 664.

C.J. Souness, B. Hubbard / Icarus 225 (2013) 495–505 Forget, F., Haberle, R.M., Montmessin, F., Levrard, B., Head, J.W., 2006. Formation of glaciers on Mars by atmospheric precipitation at high obliquity. Science 311, 368–371. Glasser, N.F., Gudmundsson, G.H., 2012. Longitudinal surface structures (flowstripes) on Antarctic glaciers. Cryosphere 6, 383–391. Glasser, N.F., Hambrey, M.J., Crawford, K.R., Bennett, M.R., Huddart, D., 1998. The structural glaciology of Kongsvegen, Svalbard, and its role in landform genesis. J. Glaciol. 44, 136–148. Grindrod, F., Fawcett, S.A., 2011. Possible climate-related signals in high-resolution topography of lobate debris aprons in Tempe Terra. Geophys. Res. Lett. 38, L19201. http://dx.doi.org/10.1029/2011GL049295. Hambrey, M.J., 1975. Short notes: The origin of foliation in glaciers: Evidence from some Norwegian examples. J. Glaciol. 14, 181–185. Hättestrand, C., Kolka, V., Johansen, N., 2007. Cirque infills in the Khibiny Mountains, Kola Peninsula, Russia – Palaeoglaciological interpretations and modern analogues in East Antarctica. J. Quatern. Sci.. http://dx.doi.org/10.1002/ jqs.1130. Head, J.W., 2012. Masursky Lecture: Mars Climate History: A Geological Perspective. [#2582] LPSC XXXXII, The Woodlands, Texas. Head, J.W., Marchant, D.R., 2003. Cold-based mountain glaciers on Mars: Western Arsia Mons. Geology 31, 641–644. Head, J.W., Mustard, J.F., Kreslavsky, M.A., Milliken, R.E., Marchant, D.R., 2003. Recent ice ages on Mars. Nature 426, 797–802. Head, J.W. et al., 2005. Tropical to mid-latitude snow and ice accumulation, flow and glaciation on Mars. Nature 434, 346–351. Head, J.W., Marchant, D.R., Dickson, J.L., Kress, A.M., Baker, D.M., 2010. Northern mid-latitude glaciation in the Late Amazonian period of Mars: Criteria for the recognition of debris-covered glacier and valley glacier landsystem deposits. Earth Planet. Sci. Lett. 294, 306–320. Hubbard, B., Glasser, N., Hambrey, M., Etienne, J., 2004. A sedimentological and isotopic study of the origin of supraglacial debris bands: Kongsfjorden, Svalbard. J. Glaciol. 50, 157–170. Hubbard, B., Milliken, R.E., Kargel, J.S., Limaye, A., Souness, C., 2011. Geomorphological characterization and interpretation of a mid-latitude glacier-like form: Hellas Planitia, Mars. Icarus 211, 330–346.

505

Lucchitta, B.K., 1981. Mars and Earth: Comparison of cold-climate features. Icarus 45, 264–303. Madeleine, J.B., Forget, F., Head, J.W., Levrard, B., Montmessin, F., Millour, E., 2009. Amazonian northern mid-latitude glaciations on Mars: A proposed climate scenario. Icarus 203, 390–405. Marchant, D.R., Head, J.W., 2003. Tounge-shaped lobes on Mars: Morphology, nomenclature, and relation to rock glacier deposits. In: Sixth International Conference on Mars. 3091.pdf. Milliken, R.E., Mustard, J.F., Goldsby, D.L., 2003. Viscous flow features on the surface of Mars: Observations from high-resolution Mars Orbiter Camera (MOC) images. J. Geophys. Res. 108 (E6), 5057. Parsons, R.A., Nimmo, F., Miyamoto, H., 2011. Constraints on martian lobate debris apron evolution and rheology from numerical modelling of ice flow. Icarus 214, 246–257. Pearce, G., Osinski, G.R., Soare, R.J., 2011. Intra-crater processes in central Utopia Planitia, Mars. Icarus 212, 86–95. Read, P.L., Lewis, S.R., 2004. The Martian Climate Revisited: Atmosphere and Environment of a Desert Planet. Springer Praxis Publishing. Sharp, R.P., 1973. Mars: Fretted and chaotic terrains. J. Geophys. Res. 78 (20), 4073– 4083. Soare, R.J., Costard, F., Pearce, G.D., Sejourne, A., 2012. A re-interpretation of the recent stratigraphical history of Utopia Planitia, Mars: Implications for lateAmazonian periglacial and ice-rich terrain. Planet. Space Sci. 60, 131–139. Souness, C., Hubbard, B., 2012. Mid-latitude glaciation on Mars. Prog. Phys. Geogr. 36 (2), 238–261. Souness, C.J., Hubbard, B., Milliken, R.E., Quincey, D., 2012. An inventory and population-scale analysis of martian glacier-like forms. Icarus 217, 243–255. 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 and similar flows on Mars. J. Geophys. Res. 84, 8087–8096. Touma, J., Wisdom, J., 1993. The chaotic obliquity of Mars. Science 259, 1294–1297. Ward, W.R., 1992. Long term orbital and spin dynamics of Mars. In: Kieffer, H.H., Jakosky, B.M., Snyder, C.W., Matthews, M.S. (Eds.), Mars. University of Arizona Press.