Ice streams in the Laurentide Ice Sheet: a new mapping inventory

Rapidly flowing ice streams dominate the drainage of continental ice sheets and are a key component of their mass balance. Due to their potential impact on sea level, their activity in the Antarctic and Greenland Ice Sheets has undergone detailed scrutiny in recent decades. However, these observations only cover a fraction of their ‘life-span’ and the subglacial processes that facilitate their rapid flow are very difficult to observe. To circumvent these problems, numerous workers have highlighted the potential of investigating palaeo-ice streams tracks, preserved in the landform and sedimentary record of former ice sheets. As such, it is becoming increasingly important to know where and when palaeo-ice streams operated. In this paper, we present a new map of ice streams in the North American Laurentide Ice Sheet (LIS; including the Innuitian Ice Sheet), which was the largest of the ephemeral Pleistocene ice sheets and where numerous ice streams have been identified. We compile previously published evidence of ice stream activity and complement it with new mapping to generate the most complete and consistent mapping inventory to date. The map depicts close to three times as many ice streams (117 in total) compared to previous inventories, and categorises them according to the evidence they left behind, with some locations more speculative than others. The map considerably refines our understanding of LIS dynamics, but there is a clear requirement for improved dating of ice stream activity.


Introduction
Ice flow in the Antarctic and Greenland ice sheets is organised into a spatial pattern of ice streams draining ice from the interior toward the margin and into the ocean (Bentley, 1987;Rignot, Mouginot, & Scheuchl, 2011). The same spatial organisation of flow toward the ice margin has been suggested for the Pleistocene ice sheets of the northern hemisphere (Denton & Hughes, 1981;Hughes, Denton, & Grosswald, 1977;Kleman & Glasser, 2007; and robust evidence for ice stream activity has been recognised in the glacial landform and sedimentary record Dyke & Morris, 1988;Patterson, 1998; # 2014 The Author(s). Published by Taylor & Francis. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The moral rights of the named author(s) have been asserted. Sharpe, 1988;. The largest ice sheet to grow and disappear was the North American Laurentide Ice Sheet (LIS; Figure 1) for which numerous palaeo-ice streams have been hypothesised (see reviews in Patterson, 1998;Winsborrow, Clark, & Stokes, 2004). Here, we present a new and updated map of palaeo-ice streams in the LIS (including the Innuitian Ice Sheet), building on earlier syntheses (Patterson, 1998;Winsborrow et al., 2004) and updated to include more recent publications and new mapping at the ice sheet scale.

A brief review of previous work
Recognition of the importance of ice streams in the LIS was outlined by Hughes et al. (1977) and Denton and Hughes (1981), who depicted a number of ice streams, based mainly on topographic inferences. Dyke (1984), Dyke and Morris (1988) and Sharpe (1988) were among the first to directly link elements of the glacial landform and sedimentary record with fast ice flow and Prest (1987a, 1987b) included ice streams in their seminal publications describing the Late Wisconsinan and Holocene history of the LIS based on topographic and sedimentary evidence. Subsequently, Patterson (1997Patterson ( , 1998 compiled a map of known ice streams, and she argued that the lobate southern margin of the LIS was produced by ice streams. These were referred to as terrestrial (i.e. land-terminating) ice streams, which do not have any modern analogues. Moreover, the development of geomorphological criteria to identify palaeo-ice streams (Kleman & Borgström, 1996;Stokes & Clark, 1999) has led to a major growth in the number of hypothesised ice streams (De Angelis & Kleman, 2005Ross, Campbell, Parent, & Adams, 2009;Ross, Lajeunesse, & Kosar, 2011;Shaw et al., 2006;Stokes, Clark, & Storrar, 2009). Winsborrow et al. (2004) also provided an inventory and supporting evidence for each of the hypothesised ice streams within the LIS. More recently, new remote sensing products and digital elevation models have permitted the glacial landform mapping at a regional and ice sheet scale (Atkinson, Utting, & Pawley, 2014;Brown, Stokes, & O'Cofaigh, 2011;De Angelis, 2007a;Shaw, Sharpe, & Harris, 2010;Storrar & Stokes, 2007;Trommelen & Ross, 2010), augmenting the Glacial Map of Canada (Prest, Grant, & Rampton, 1968).

Methods
We define ice streams as discrete arteries of enhanced ice flow bordered by slower moving ice (after Swithinbank, 1954, rephrased). Initially, we collated published literature hypothesising ice stream activity. Ice streams reported in literature were checked and mapped and, in some cases, their outlines and tracks were re-mapped from recent satellite and digital elevation imagery that might not have been available to the original authors (see below). This also ensured homogeneity of the mapped ice streams. A small number of ice streams inferred from types of evidence other than that visible in remotely sensed and digital elevation data were adopted without major modification. In addition, areas that had not been previously systematically mapped for ice streams have been surveyed using remotely sensed-data to ensure completeness in the mapping. This has resulted in the identification of several newly hypothesised ice streams (cited as new mapping on the main map). A variety of data were used to distinguish ice stream tracks in the glacial landform record and these are listed in Table 1. The northern portion of the mapped area has been surveyed using a false colour composite Landsat image mosaic (Table 1, Figures 2 -4). Satellite imagery was less useful along the southern and southwestern LIS margin due to the dominance of the agricultural land. Instead, medium to high-resolution digital elevation models (DEMs, Table 1, Figure 5) were used to map this area. The International Bathymetric Chart of the Arctic Ocean DEM   (Table 1, Figure 6) was used for areas north of the 60 th parallel, whereas the continental shelf south of 608 N was mapped from General Bathymetric Chart of the Oceans data (Table 1). High-resolution swath bathymetry has been utilised where available (Table 1, Figure 7).
Ice streams have been mapped based on several types of evidence, which we formally categorise and show on the map. These include the bedform imprint (Figures 2-5 and 7), topographic constraints (Figures 4, 6, and 7) and major sedimentary depo-centres ( Figure 6; Batchelor and Dowdeswell, 2013). Other types of evidence include the sedimentary characteristics of subglacial tills or ice rafted debris traced to its sediment sources. Ice streams inferred from their bedform imprint (after (King, Hindmarsh, & Stokes, 2009;Kleman et al., 2006;Stokes, 2002;Stokes & Clark, 1999, 2001, 2002 have been further categorised into three classes: (i) ice streams with a full bedform imprint (Figures 2, 5, and 7), (ii) ice streams with discontinuous bedform imprint (Figures 3 and 5) and, (iii) ice streams with only an isolated bedform imprint (Figure 4). In addition to these an extra class of ice stream fragment has been introduced where isolated evidence of fast ice flow occurs but no inferences can be made about the outline of the zone of fast ice flow (Figures 4 and 5). Ice stream margins, and in some cases ice streams per se, have been defined by broad-scale topography, i.e. by the limits (and the existence of) glacial troughs (Figures 4, 6, and 7). Where distinct topography was missing, ice stream margins were either recognised as an edge of the bedform imprint (in some cases constituted by a lateral shear margin moraine; see Figure 2 and (Stokes & Clark, 2002) or drawn as undefined (Figure 3). Multiple types of evidence have been found for some ice streams (Figures 4 and 7) whereas other mapped ice streams are only based on one type of evidence (Figures 2 and 3). The strength of the evidence therefore varies and while some ice streams are documented by a robust record, the existence of others is more speculative.  Figure 1 for location). Displaying hypsography separately for each tile enhances topography in areas with low relief. Networks of ice stream tracks, represented by corridors of smooth topography as well as isolated areas of streamlined terrain can be seen in the data. Lateral margins are drawn by a dotted line (where distinguished) and centre-lines of the ice stream tracks are drawn by full lines.
Although our map represents the most complete to date, it is important to acknowledge that some ice streams may have operated, but their evidence has not been preserved. This may be especially true for short-lived ice streams flowing over resistant bedrock (Punkari, 1995;Roberts, Long, Davies, Simpson, & Schnabel, 2010).

Mapping results
The map contains 117 ice streams, which is almost three times more than previous inventories. These vary in size and shape, with some draining large portions of the ice sheet and others being rather minor features, typically draining ice through high-relief coastal areas. Note that we make no inferences on the timing of ice stream operation because very few of the mapped ice streams have any chronological control. However, it is clear that the mapped ice stream tracks represent a time-transgressive imprint of evolving ice stream trajectories, i.e. they cannot have all operated at once. Nevertheless, some broad spatial patterns emerge.
The northernmost part of the LIS covering the Canadian Arctic Archipelago and including the Innuitian Ice Sheet (Figure 1) was drained by a network of ice streams occupying the channels dividing the archipelago's islands and peninsulas. The location and pattern of these ice streams share characteristics with the Siple Coast ice streams of the West Antarctic Ice Sheet (Bennett, 2003;De Angelis, 2007b;Hughes, 1977). The northwestern, western and southwestern margin of the LIS displayed a more complex pattern of sinuous, sometimes cross-cutting, flow trajectories over a relatively flat bed that likely came into operation during different periods of ice sheet growth and decay. Moreover, these were predominantly land-terminating (terrestrial) ice streams. Large terrestrial ice streams also occupied the basins of the Great Lakes at the southern margin of the ice sheet. Further east, the continental shelf fringing the northeastern USA, Atlantic Canada and Labrador hosted a number of ice streams draining into the Atlantic Ocean. Ice streams have also been identified in areas that were close to the centre of a fully grown LIS. These ice streams presumably operated successively during deglaciation when the ice margin gradually retreated into the areas of the main domes in Keewatin, Labrador and Foxe Basin (Figure 1).

Conclusions
A map depicting palaeo-ice streams of the LIS has been produced using previously published hypotheses of their activity, together with new mapping from satellite imagery and digital elevation models, both onshore and offshore. In total, 117 ice streams have been identified, based on a variety of evidence and with some comparatively robust and others more speculative. The map only indicates the spatial distribution of ice streams within the LIS, without providing constraints on their timing of operation. Based on our review of the evidence (see list of sources on the main map), it is likely that the majority relate to the Late Wisconsinan, but we cannot rule out the possibility that some evidence has been preserved from pre-Late Wisconsinan ice sheets. Dating of these palaeo-ice streams is an obvious task for future research. This map, combined with better dating of ice stream operation, will allow for an improved understanding of LIS dynamics and its links to the atmosphere-ocean system. Such reconstructions will also extend the temporal record of ice stream activity, which is currently biased toward short-term observations of modern-day ice sheets.

Software
The mapping was carried out using Esri ArcGIS 10.1. The GEBCO data (Table 1) were exported for use with ArcGIS using the GDA Software Interface provided with the data. The Google Maps platform was used for displaying the high-resolution swath bathymetry data ( Table 1). The final layout of the map was produced in Adobe Illustrator CS4.