Mega‐scale glacial lineations formed by ice shelf grounding in the Canadian Beaufort Sea during multiple glaciations

Mega‐scale glacial lineations formed by the raking of ice shelves across the seafloor have been reported from multiple polar regions. Here, we present the first evidence of continental slope situated buried lineations in the southern Canadian Beaufort Sea in present‐day water depths of 220 to 800 m. Three separate surfaces with lineations are defined at sub‐seafloor depths of 40 m to 390 m. All lineations are mostly parallel to the general trend of slope contours. The uppermost surface is recognized over a distance of 56 km. In water depths > 500 m the lineations are parallel to each other at a consistent direction (43°–44°). The second lineated surface is a regionally occurring erosional unconformity. This event has two sub‐sets of lineations: mid‐slope situated lineations oriented at 42°–48°, and lineations closer to the continental shelf break at 55°–59°. The third lineated surface is an unconformable horizon buried up to 390 m below seafloor with lineaments oriented between 30° and 55°. All three sets of lineations are interpreted to have been produced by ice‐ploughing on the paleo‐seafloor through the grounding of an ice shelf. Our observations are similar to those documented along the slope off northern Alaska, Chukchi Rise, and Lomonosov Ridge. Collectively, these observations support the concept of an extensive ice shelf across the Arctic Ocean that grounded locally along its margins during multiple glaciations, including during the penultimate (or an earlier) glaciation. The youngest set of lineations indicates ice movement to the southwest with a suggested source in Amundsen Gulf and/or M'Clure Strait. Tentative age considerations for these youngest lineations indicate the first evidence for an analogous extensive ice shelf configuration for the Last Glacial Maximum.


| INTRODUCTION
The glacial history of the Arctic Ocean is a topic of interest to Quaternary scientists, oceanographers, and environmental scientists who are studying glacial-interglacial cycles, ice shelf dynamics, and associated seabed processes (e.g., Engels et al., 2008;Grosswald & Hughes, 2008;Jakobsson et al., 2014;Niessen et al., 2013;Polyak et al., 2001). This topic also has relevance to those studying the interconnections between the ocean and the atmosphere during glacial times (e.g., Johnson & Andrews, 1986) and the effects of ice shelves influencing glacial flow dynamics at the margins of continental glaciers (Scambos et al., 2004). The concept of ice shelves and glacial ice cover in the Arctic Ocean, including the possibility of a pan-Arctic ice shelf, was originally introduced on conceptual grounds drawing extensively on comparisons with modern ice shelves in Antarctica (Denton & Hughes, 1983;Mercer, 1970). More recently, marine geophysical and geological surveys from various regions of the Arctic Ocean have documented a variety of seabed features, including parallel grooves, that have been interpreted to be caused by the grounding of ice sheets or the grounding of the keels of ice shelves or icebergs (e.g., Engels et al., 2008;Jakobsson et al., 2005Jakobsson et al., , 2010Jakobsson et al., , 2014Niessen et al., 2013;Polyak et al., 2001).
Mega-scale glacial lineations (MSGLs) were first described in the literature by Clark (1993) who reported large-scale streamlined lineations on previously glaciated terrain in northern Canada. He described parallel lineations 8 to 70 km long and 200 to 1300 m wide that were interpreted to form as a result of rapid glacial ice flow. A number of researchers studying active glacial processes in Antarctica have found that MSGLs are associated with ice streams and that they can also be formed at the base of grounded ice shelves in the offshore realm (Dowdeswell et al., 2007, and references cited therein;King et al., 2009). The classic MSGLs (on terrestrial or marine settings) are considered to be a subglacial bedforms associated with grounded ice streams, and are typically regarded as positive relief features (i.e., ridges, Spagnolo et al., 2017). As summarized by Jakobsson et al. (2014), glacial-related lineations have also been identified at greater water depths (800 to > 1000 m) on topographic highs in the Arctic Ocean. It remains a debate if these features should be called MSGLs, since they were probably formed beneath an ice shelf and therefore do not represent the same environment or conditions of a grounded ice stream. Nonetheless, the orientation of these lineations is significant for indicating the flow direction of a grounded ice mass.
The presence of glacial lineations on topographic highs in the Arctic Ocean (e.g., Engels et al., 2008;Dowdeswell et al., 2010) raises the intriguing possibility that at least some parts of the Arctic Ocean were covered by vast and interconnected ice shelves (e.g., Jakobsson et al., 2010). However, as summarized in the review by Jakobsson et al. (2014), there remain many unanswered questions: • Were ice shelves in the Arctic Ocean interconnected?
• Were there multiple ice shelves during past glacial cycles, or did these features exist only during the penultimate glacial maximum shallow high-resolution seismic imaging (e.g., Bellwald et al., 2018Bellwald et al., , 2019Tasianas et al., 2016), we rely primarily on two-dimensional (2D) and three-dimensional (3D) multichannel seismic data to identify sub-seabed features. Two 3D seismic data volumes (informally named Ajurak and Pokak) provided to the Geological Survey of Canada by industry partners provide the basis for the main interpretations. Additional insights were derived from newly acquired 2D seismic surveys and hull mounted 3.5 kHz sub-bottom profiler sonars conducted by the Korean Polar Research Institute (Jin et al., 2015;Jin & Dallimore, 2016). Within the uppermost 400 m of marine sediments beneath the upper slope we have mapped three buried lineated surfaces at different stratigraphic levels. A previous examination of the industry 3D data set, conducted to assess seabed geohazards, identified two of these horizons which were considered to contain glacialrelated lineations (Woodworth-Lynas et al., 2016a, 2016b. These lineations were interpreted to have been formed by seabed scour from a possible melange of icebergs and thick sea ice moving in a southwest direction. They are inferred to have formed during the last glaciation, with an estimated age of 19 to 16 ka based on extrapolated carbon-14 ( 14 C) ages documented in the reports. We seek to examine these features and another deeper lineated horizon in greater detail by documenting their morphology and geophysical characteristics as well as considering the age of the features. Our interpretations broaden the expanse of regional observations of glacially produced seabed lineations observed elsewhere in the Arctic Ocean (e.g., Jakobsson et al., 2010Jakobsson et al., , 2014Jakobsson et al., , 2016, and provide evidence for their source of glacial ice from the western Canadian Arctic Archipelago.

| BACKGROUND
Assuming that the buried lineations we describe are of a glacial origin, a brief review of the present understanding of the glacial history of the Beaufort Shelf and onshore areas of north-western Canada and Alaska is warranted. Regional Quaternary geology mapping efforts have been conducted over the past 50 years by the Canadian and American Geological Surveys and by a variety of university scientists.
These studies suggest multiple Quaternary glaciations affected the Beaufort coast in Canada. Mapping efforts summarized in Rampton (1982Rampton ( , 1988  only reached Herschel Island, and that the coastal areas further west, including northern Alaska, were unglaciated. Rampton's mapping did infer, however, that in order for glacial ice to reach Herschel Island, at least part of the Beaufort continental shelf adjacent to the coast must have been ice covered. The ice on the shelf was thought to have flowed out from the Mackenzie Trough, covering part of the Yukon Shelf and extending locally to areas northwest of Richards Island (Batchelor et al., 2013b). Quaternary mappers working in the western Arctic Islands have documented glacial sediments and landforms in this area that resulted mainly from flow from the north-western extension of the Laurentide Ice Sheet and contributions from the Innuitian Ice Sheet located in the high Arctic islands (Dyke, 2004;Dyke et al., 2002Dyke et al., , 2003England et al., 2006;Vaughan et al., 2014;Vincent, 1983). These researchers recognised westerly ice flow into Amundsen and M'Clure Straits ( Figure 1) and suggested that at least some of the Beaufort continental shelf immediately offshore must have been glaciated. Perhaps the most contentious issue arising from the terrestrial mapping to date has been assigning the age of the glacial sediments and establishing if the ice was cold or warm based. The present consensus is that the late Wisconsin (MIS 2) ice cover was the most extensive in this area and that at least parts of the ice sheet were cold based (England et al., 2006). However, streamlined features (drumlin fields, dispersal trains and lineations) adjacent to many of the inlets on Victoria and Banks Island suggest episodic ice streams and ice shelves during deglaciation (e.g., Dyke, 2008;Hodgson, 1994;Stokes et al., 2005).
The assessment of the geology of offshore areas of the Canadian Beaufort Sea has benefited from industrial activities undertaken to evaluate the hydrocarbon potential of the area. Extensive regional multichannel seismic surveys have been conducted along the entire margin of the eastern Beaufort Sea and many geotechnical coring and site survey studies have been carried out at proposed exploration well sites. Using regional 2D seismic lines, Batchelor et al. (2013aBatchelor et al. ( , 2013bBatchelor et al. ( , 2016 interpret the presence of thick glacial tills in the Mackenzie Trough where Rampton inferred ice cover. They also document extensive trough mouth fans of glacial origin that extend onto the continental slope from M'Clure and Amundsen Troughs. Streamlined seabed features in the M'Clure and Amundsen Gulf also confirm ice movement from these Straits onto the continental shelf during the LGM (Blasco et al., 1990;MacLean et al., 2010MacLean et al., , 2016Niessen et al., 2010;Stokes et al., 2005Stokes et al., , 2006. However, the regional 2D seis-  (Barendregt et al., 1998;Carter et al., 1988;MacCarthy, 1958;Vincent, 1990;Vincent et al., 1984). However, uncertainty arises due to limited exposure of these older sediments and concerns that glacio-tectonism caused by late Wisconsin ice cover could have disrupted the sediment assemblages and assigned dates (Vaughan et al., 2014).

| DATA AND SEISMIC ATTRIBUTE CALCULATIONS
In this study we utilize two 3D seismic data volumes provided by industry as well as newly acquired 2D multichannel seismic (MCS) and sub-bottom profiler data collected using the Korea Polar Research Institute (KOPRI) icebreaker ARAON in 2014. The location of the seismic data sets together with detailed bathymetry is shown in Figure 2.
The industry 3D seismic data were provided as pre-stack timemigrated sections and processing included (in addition to 3D geometry definition), cable depth datum correction, de-bubble phasing designature deconvolution, low-cut filtering, swell-noise attenuation, direct arrival attenuation, channel amplitude correction, water column statics, common-offset noise attenuation, Q-compensation (phase only), high-resolution radon de-multiple processing and anisotropic pre-stack migration. Additional care was taken to suppress the acquisition footprint (in this case predominantly east-west [E-W] oriented).
The 3D seismic data were further processed by the Geological Survey of Canada (GSC) using the similarity attribute (also referred to as coherency attribute) to enhance imaging of structural discontinuities such as faults, channels and the glacial mega-lineations. The similarity attribute is computed as semblance using a range of one neighbouring trace around each input-trace, a 20 ms vertical time window, and within a frequency range of 8 to 80 Hz. Several horizons were derived by manually picking events from the time-migrated 3D seismic data inside a seismic interpretation software package, and exported for further statistical analysis performed inside ArcGIS.
The KOPRI 2D MCS survey used an airgun array of 1200 in 3 volume and a streamer with 120 channels with maximum offset of 1600 m (Riedel et al., 2016). Seismic data were processed (after crooked-line 2D geometry definition) using a sequence including removal of DC offset, muting of direct arrivals and refractions, minimum phase band-pass filtering, surface-consistent deconvolution, stacking-velocity definition using the semblance technique, staking (full-offset range) and post-stack Kirchhoff time migration using a 2D variable velocity field. Details of the processing and interpretation of these 2D MCS data are given in Riedel et al. (2016). The MCS data have been used previously to define interval velocities (Riedel et al., 2016) and we use these velocity-depth functions to convert observations on the 2D and 3D data made in two-way travel time to depth (metres below seafloor, mbsf). The uncertainty in this conversion depends on water-depth and burial depth below seafloor and is practically a measure of robustness of the P-wave velocity-depth functions defined from the MCS data. As described in Riedel et al. (2016), the P-wave interval velocity function is well established in water depths exceeding 300 m but is complicated and laterally fast-changing near the continental shelf edge with rough topography and across most of the continental shelf region due to shallow-water multiples. In the analyses described later, our measurements of lineation depth and width are associated with an additional (individually ascribed) uncertainty based on sample-rate (vertical extent) and 3D bin-size (lateral extent).

| REGIONAL SEISMIC STRATIGRAPHY
Seismic data and stratigraphic information obtained from historical hydrocarbon exploration programmes provide a comprehensive basis to characterize the regional Cenozoic stratigraphy of the Beaufort-Mackenzie Basin (e.g., Dixon, 1996;Dixon & Dietrich, 1990;Dixon et al., 1992;McNeil et al., 2001). The upper-most sedimentary sequence (upper 2-4 km) of the offshore record has been classified into two depositional formations: the Iperk Formation (early Pliocene to mid Pleistocene age) and the Shallow Bay Formation (mid Pleistocene to Holocene age). While the data sources are sparse, especially on the outer continental shelf, a regional angular unconformity associated with the base of the Iperk Sequence is thought to be prominent in the study area (e.g., Dietrich et al., 2010) at 2.0-2.5 s two-way time on regional seismic data (equivalent to a depth of 2.8 to 3 km below seafloor). The base of the Iperk Sequence is deeper than the available 3D seismic data sets we have accessed. The Shallow Bay Formation above the Iperk includes glacial erosion across parts of the continental shelf and was defined by Dietrich et al. (1985) in the Mackenzie Trough using a seismic reflection line; it was later interpreted to occur on the slope tens of kilometres away on a reflection profile (Dietrich et al., 2010) acquired in 1987 as part of the Frontier Geoscience Programme (FGP) some distance west of our study site (e.g., Dietrich et al., 1989). Our interpretations derived from the 3D seismic data F I G U R E 2 (a) Bathymetric map of the study region showing extent of the three identified events with large-scale glacial lineations. Oldest 'event A' (light blue area) is preserved over only a small portion in the eastern study region, 'event B' (light grey area) is seen across most of the study area, and the youngest 'event C' (dark grey area) coincides with much of the area of 'event B'. All seismic lines used in this study are marked as solid black lines and the outline of the two three-dimensional (3D) seismic data volumes Ajurak and Pokak are shown as black rectangles. A seismic section (re-digitized from an analogue paper-record) crossing the study area is shown in the Supporting Information Figure S1 Figure S1). This unconformity represents a significant change in depositional style from an environment associated with rapid deposition and sediment deformation within the slope fan (Rohr et al., 2021), to glacially-dominated erosion and deposition cycles. This slope-fan environment would have been the site of significant meltwater runoff during several Quaternary deglacial events. The slope fans are morphologically distinct from the glacially influenced Amundsen Gulf and M'Clure Strait trough mouth fans described by Batchelor et al. (2013b.

| RESULTS: CHARACTERISTICS OF OBSERVED LINEATIONS
Using the 3D seismic attribute similarity, we scanned the 3D seismic surveys in our study area for events with potential lineations, especially those associated with angular unconformities. Within the uppermost 0.9 s two-way time, or 0.54 s below seafloor, we have identified  Jakobsson et al., 2005Jakobsson et al., , 2010Jakobsson et al., , 2014Jakobsson et al., , 2016Polyak et al., 2001). The character of the lineated surfaces is described in the following sections. Here, we interpret the lineations as predominantly negative relief features (grooves) rather than positive relief ridges.
Representative examples of profiles across each horizon are included in the results shown below and more extensive examples and analyses are given in the Supporting Information. All of the lineated surfaces we observe have significant sediment cover and therefore have limited expression on the present seabed. This is in contrast to other lineations described in the Arctic that have only shallow burial depths and can therefore be identified on multibeam or side scan imagery (e.g., Engels et al., 2008;Jakobsson et al., 2005Jakobsson et al., , 2010Jakobsson et al., , 2016. This reflects our location within the principal Cenozoic depocentre of the Beaufort-Mackenzie Basin with high post-glacial sedimentation rates.

| Oldest (deepest) lineated 'event A'
The deepest reflection event (unconformity) associated with lineations occurs between approximately 0.8 and 0.9 s two-way time (275-390 mbsf, and 600-750 m below present sea level) within the Pokak 3D volume (Figure 4). The unconformity, which occurs within the upper part of a bedded channel-levee complex described by Rohr et al. (2021), has varying dip angles and topography. It has been eroded in many parts of the study area and as a consequence is only approximately 180 km 2 in area. A total of 343 individual lineations have been mapped on this surface (Figure 4)  The width of the grooves varies along the surface of the two 3D data sets, but the majority of the lineations are less than 150 m wide (uncertainty of width is on average 10 m as per 3D seismic data bin size). Using the extracted surfaces, a comparison of groove azimuth and width across the entire region where they are preserved can be made (Figure 5f). Comparing the widths of the grooves over Ajurak and Pokak shows that they may progressively narrow towards the southwest with an increase in the average distance between grooves (i.e., ridge-widening) towards the west (Figure 5f). Sometimes, several grooves occur within a wider depression. In these cases, width was then defined for the individual thinner and the combined wider groove. We also note that multiple grooves commonly merge into one larger feature towards the southwest. However, groove depth is highly irregular along the surface with grooves deepening and shallowing several times along their paths (up to 30 km in length across the Pokak region) and no consistent depth-trend is observed (see also Supporting Information Figure S4). Unlike the sediments which occur above lineated surface 'A', the sediments bedding above surface 'B' do not appear to mimic the groove geometry.

| Lineated 'event C'
Lineated  (Table 1). A progressive change in azimuth following the slope topography is seen in shallower water across the Pokak area (similar to unconformity 'B') with a maximum azimuth of 70 in water depth of 360 m (Figure 7 and S3). The average distance between lineations is apparently less in shallower water depths. The orientation of the lineations is similar in character to those on unconformity 'B'; however, several grooves are much deeper (up to 15 m) than those seen on the unconformity 'B'. The surface area of 'event C' preserved across Ajurak is less than in the Pokak area and thus the statistics derived in this area are less robust. However, the groove widths appear wider at Ajurak than at Pokak (compare to Supporting Information Figures S4 and S5).
T A B L E 1 Width of the grooves derived from surfaces across the two three-dimensional seismic surveys Ajurak and Pokak  While there is general agreement that MSGLs themselves are indicators of glacial flow direction, there is considerable discussion in the literature about their formation mechanism (e.g., Boulton & Clark, 1990;Canals et al., 2000;Clark et al., 2003;Cofaigh et al., 2005;Dowdeswell et al., 2007Dowdeswell et al., , 2010Fowler, 2000Fowler, , 2009Hindmarsh, 1998;King et al., 2009;Shaw et al., 2008;Tulaczyk et al., 2001). Our analyses of the morphology of the lineated surfaces suggest that they are primarily negative relief features, or grooves, and width of the groove and also, but less commonly, building of an intervening ridge ( Figures S4 and S5). Ridge volumes are less than those of the grooves, suggesting that their formation may be a more secondary process, that is, not strictly a function of displacement from the groove. This introduces the possibility that they could be molded by the ice in a dynamic interplay of erosion and re-deposition. We therefore suggest that the lineations seen on all three surfaces mapped are from a grounded ice mass and a grooveploughing-type process. We also assign the term MSGL to our lineations as they are clearly representing large-scale glacial processes along the Beaufort margin and are similar in character to those features described elsewhere in the Arctic as reviewed by Jakobsson et al. (2016).
In addition to the MSGL features themselves, associated sediment characteristics such as evidence for erosion, deformation of bed materials, or sub-ice emplacement of glaciogenic sediments can be useful in the assessment of their origin. Assuming that the lineations we observed were formed by grounding of an ice mass in contact with the sea bottom, we conclude that the most definitive indicator of flow direction would be changes in groove character along the length of the feature. In uniform water depths, such is the case for most of our study area, it is possible that the grooving would become less sharp and well defined in the down flow direction as the resistive feature in the bed of the ice gradually erodes as it is dragged along the sea bottom ( Figure 8). We have examined the grooved surfaces in the 3D data sets to consider if morphology of the lineations suggests the direction of ice-movement (Figures 7, 8 Lineations are less developed across this sloped surface and can be traced only for 800 to 1400 m downslope. With shallowing, (NE to SW), lineations initiate as thin, shallow grooves that progressively deepen and widen. On reaching maximum topographic height, grooves continue but merge so that the overall depth of the groove increases as does their combined widths. Relief from initiation to maximum elevation is about 35 m. We infer ice-flow direction from NE to SW [Color figure can be viewed at wileyonlinelibrary.com] are present along the same profile-distance than to the northeast across Pokak ( Figure 5). Since both horizons are in close proximity to each other, it may be reasonable to assume a similar ice-movement direction for 'event B' as made with higher confidence for 'event C'.
In contrast, the very limited spatial extent of the mapped lineations comprising 'event A' limits any rigorous evaluation and therefore interpretation of flow direction. Event A is further complicated as the orientation of the surface is more steeply dipping.

| Age of lineations
The absolute timing for the events that formed the three observed horizons with MSGL is challenging to assess as to date no sediment cores have been recovered from these horizons. However, extensive shallow coring has been conducted on the upper slope mainly for geohazard studies (Woodworth-Lynas et al., 2016a, 2016b and recently, a 17 m long stratigraphic core has been collected from the upper slope Klotsko et al., 2019) in the eastern part of our study area (location of JPC 15/27 see Figure 2a). The long core is particularly helpful as it has been well dated enabling correlations to be made with high-resolution sub-bottom profiler data (i.e., figure   14 in Klotsko et al., 2019). A detailed correlation is given in Supporting Information Figure S8 between the acoustic data shown in Klotsko et al. (2019) and our database of sub-bottom profiler data (Blasco et al., 2013) and more recently acquired data with the icebreaker ARAON (Jin et al., 2015;Jin & Dallimore, 2016;Riedel et al., 2016).
Correlations of the acoustic sub-bottom profiler data to the stratigraphic framework and 14 C ages were previously made by Woodworth-Lynas et al. (2016a) and King et al. (2017) and we expanded those to the data relevant to our study. Through these assessments we have determined that the main reflector horizons identified can be traced in the sub-bottom profiler data across the Pokak ( Figure 9) and Ajurak ( Figure 10) 3D survey areas. We have focussed on three distinctive seismic horizons as they can be correlated with a high degree of confidence to the Klotsko et al. (2019) assessment (see H2, H3, and H5 on Figures 9 and 10). This approach suggests that the deepest reflector H5 has an estimated age of 18 ka. This horizon is above a layer with low acoustic reflectivity that is situated directly above lineated 'event C'. The character of this layer suggests a different depositional environment than the well lam-

| Regional implications
Our analyses of the buried MSGL features on the continental slope suggest movement of a grounded ice mass across our study area from the northeast to southwest, parallel to the slope contours. This direction is not consistent with a glacial ice source from the Mackenzie Trough to the west or from the mainland immediately to the south (Figure 1). The most likely source is from the Canadian Arctic Archipelago to the east. The question becomes whether our lineated surfaces resulted from a clockwise-rotating extensive ice shelf which was grounded in a narrow band where its Beaufort Sea margin impacted the upper slope (analogous to the process described by Jakobsson et al. [2016]), or whether the ice streams emanating from continental shelf troughs remained fully grounded as they followed the upper slope contours.
In this regard, lineations similar in morphology to our slopesituated examples have been described across the Canadian Arctic Archipelago on land (Dyke, 2008;Hodgson, 1994;Stokes et al., 2005Stokes et al., , 2006Vincent, 1983Vincent, , 1990Vincent et al., 1984) and on the continental shelf in both Amundsen Gulf (Batchelor et al., 2013bMacLean et al., 2015) and M'Clure Strait (Batchelor et al., 2016;. These are attributed to ice streaming from sources in the Arctic Islands westward out of glacially excavated troughs onto the continental shelf (Batchelor et al., 2013b(Batchelor et al., , 2016MacLean et al., 2010MacLean et al., , 2015. Whereas the glacial sediment succession in the inner shelf areas is very thin, further offshore on the shelf,  and Stokes et al. (2006)  to 700 m below the seafloor . In this context, ice emanating from the troughs appears to be thinner than the grounding depth required to form the youngest two lineated horizons (presumably of LGM age). This, in combination with the groove-like nature of the lineations (i.e., an erosional character) suggest that the lineations we observe may not be formed by continuously grounded F I G U R E 1 0 Example of 3.5 kHz sub-bottom profiler data across Ajurak (location see Figure 3) showing uppermost 60-70 m of sediment. The grooved surface of 'event C' is seen overlain by a cover of laminated sediments, conformably filling the grooved surface. The reflection of 'event B' is not imaged here and below the limit of penetration of the sounder system. Approximate ages (extrapolated below H3) were assigned to horizons based on available radiocarbon dates and high-resolution sub-bottom profiler data across the region (compare to Figure 9 ice streams but are rather associated with an ice shelf that locally grounded. The orientation of our lineations suggesting ice movement to the southwest is also puzzling. Irrespective of the grounding depths, ice emanating freely from M'Clure Strait would be expected to trend in a west-northwest direction while ice from Amundsen Gulf would likely trend to the northwest (see Figure 1). This raises a question of what conditions could in theory so strongly deflect movement of a grounded ice mass from the trough axes to a slope parallel orientation.
Full assessment of this topic will require further research; however, a buttressing (confining) force could be a reasonable mechanism to both thicken and deflect the ice stream or its floating (shelf) component.
Given As summarized on Figure 1, several authors have attributed glacial lineations similar to those that we describe to possible western flow along the Beaufort margin (Engels et al., 2008) and the Chukchi Borderlands (Jakobsson et al., 2005(Jakobsson et al., , 2008(Jakobsson et al., , 2016Polyak et al., 2001). Our observations are consistent with this interpretation, providing further evidence that during glacial times an extensive ice shelf occupied the Arctic Ocean and grounded along the upper slope over a distance exceeding 1000 km. On their own, our findings do not dictate complete Arctic ice shelf cover. However, they fill in an impor- The depth difference between events B and C is relatively small and the sediment record between those two horizons is barely imaged with our data set. However, seismic data show that the intervening sediment unit, where present, has an acoustically transparent to weakly laminated character (Figures 3b and 6). The ice shelf that produced the lineations on surface 'B' is therefore suggested to have lifted off and/or retreated to shallower grounding depths during this interval, enabling some postglacial sediments to be deposited, before it re-grounded to form the shallowest set of lineations on surface 'C'.

| CONCLUSIONS
Using 3D  Geological Survey of Canada-Pacific for general support in generating