The structural glaciology of southwest Antarctic Peninsula Ice Shelves (ca. 2010)

Abstract

The Antarctic Peninsula has recently seen a rapid breakup of its peripheral ice shelves, attributed to atmospheric and oceanic warming in the region. Previous work has illustrated that the final breakup mechanisms are often controlled by the structural glaciology of the ice shelf, and thus understanding the structure of the remaining ‘stable’ ice shelves is of fundamental importance to assess their future response to continued environmental changes. The accompanying map presents the structural glaciology of Bach, George VI and Stange Ice Shelves in the southwest Antarctic Peninsula, from ca. 2010, with a description of each feature presented here.

Keywords:

• Antarctic Peninsula
• Ice Shelves
• structural glaciology
• George VI
• Bach
• Stange

1. Introduction

Over recent decades, the Antarctic Peninsula has seen a rapid recession of its peripheral ice shelves that buttress inland grounded ice (Cook & Vaughan, 2010). Their removal has led to the acceleration and thinning of their former feeder glaciers (e.g. Berthier et al., 2012; De Angelis et al., 2003; Glasser et al., 2011; Hulbe et al., 2008; Rott et al., 2002; Scambos et al., 2004) that have the potential to amplify global sea levels (Vaughan et al., 2003). Attributing the causes of breakup to a single mechanism is difficult due to the complexities of ocean-ice-atmospheric processes. However, the timing of ice-shelf collapse appears to be linked with the southern migration of the −9°C mean annual isotherm (Morris & Vaughan, 2003). Furthermore, (Pritchard et al. 2012), amongst others, have suggested that breakup may also be linked to oceanic warming in the region (Martinson et al., 2008; Robertson et al., 2002).

The structural evolution of an ice shelf is visually recorded in surface features that can be accurately and repeatedly mapped to infer ice dynamics. Certain ice-shelf structures have been shown to play a critical role in determining ice-shelf stability (Braun et al., 2009; Glasser & Scambos, 2008; Holt et al., 2013), so by recording the configuration of surface structures and how they evolve through time, it is often possible to infer their future longevity.

The map described in this paper documents the structural glaciology of Bach, George VI and Stange Ice Shelves on the southwest Antarctic Peninsula that are at, or near, the atmospheric limit of ice-shelf viability (Morris & Vaughan, 2003) and also terminate into a warming Bellingshausen Sea (Holland et al., 2010; Meredith & King, 2005). Here we present an overview of the data and methods used, and a description of the surface structures recorded on the map which can be found as Supplemental Content to this article.

2. Study sites

The study site comprises George VI, Bach and Stange Ice Shelves in the southwest Antarctic Peninsula (Figure 1). George VI Ice Shelf (GVIIS) measured ∼ 23,370 km² (in 2010), Bach ∼ 4540 km² (in 2011) and Stange ∼ 7920 km² (in 2011) (Holt, 2013), and all have been subject to recession and thinning since at least the second half of the twentieth century (see Cook & Vaughan, 2010; Ferrigno et al., 2009; Fricker & Padman, 2012). George VI Ice Shelf is by far the most intensely studied of the ice shelves investigated here, with relatively little work undertaken on Bach or Stange Ice Shelves. The reader is directed to Holt et al. (2013), Smith et al., (2007), Humbert (2007) and Reynolds and Hambrey (1988) for a more thorough discussion of George VI Ice Shelf.

Figure 1. Location of Bach, George VI and Stange Ice Shelves in the Antarctic Peninsula. Wilkins Ice Shelf is not considered in this study, and the reader is directed to Braun et al. (2009) for a thorough structural interpretation. Insert: MODIS Image Mosaic of Antarctic (Haran et al., 2005).

3. Data, methods and software

Structural mapping was carried out in ArcMap 9.3 Geographical Information System (GIS) software using time-separated satellite images following the procedures of Glasser and Scambos (2008) and Holt et al. (2013). Digitised features included the location of the ice front and ice-shelf grounding zone, rifts, fractures, fracture traces, crevasses and crevassed zones (fields), longitudinal structures (elsewhere termed flow stripes, flow bands, foliation), transverse structures, pressure ridges, ice rises, ice rumples and ice dolines.

Structural and surface-feature mapping was performed using Landsat ETM+ (Table 1) Panchromatic Band 8 (0.52–0.90 µm, 15 m) that offers the greatest contrast between surface features and the surrounding ice-shelf surface, and the best spatial resolution. Standard image-enhancement stretches (e.g. Gaussian stretching, equalisation) were applied on an image-by-image basis to improve the appearance of surface features during digitisation.

Table 1. List of Landsat scenes used for structural mapping.

Ice Shelf	Scene ID	Date (dd/mm/yyyy)	Path	Row	Scene Centre
					Lat. (°)	Long. (°)
Bach	L71218111_11120091110	10/11/2009	218	111	−71.63	−72.22
	L71219111_11120110123	23/01/2011	219	111	−73.00	−72.24
George VI	L72218110_11020090211	11/02/2009	218	110	−70.95	−69.38
	L72218109_10920091110	10/11/2009	218	109	−69.64	−67.51
	L72217110_11220100106	06/01/2010	217	112	−67.94	−70.87
	L72217112_11220100106	06/01/2010	217	112	−73.46	−72.42
	L71218111_11120100113	13/01/2010	218	111	−71.56	−72.17
	L71216111_11120100115	15/01/2010	216	111	−72.19	−68.58
	L72216112_11220100115	15/01/2010	216	112	−73.49	−71.06
	L72218109_10920100129	29/01/2010	218	109	−69.64	−67.49
	L72218110_11020100129	29/01/2010	218	110	−70.82	−69.49
	L72218111_11120100129	29/01/2010	218	111	−73.49	−74.16
	L72216112_11220101201	01/12/2010	216	112	−73.52	−71.10
	L72218112_11220120307	07/03/2012	218	112	−73.99	−73.45
Stange	L72221112_11220091115	15/11/2009	212	112	−77.12	−73.30
	L72220112_11220110114	14/11/2011	220	112	−76.97	−73.40
	L72219112_11220110208	08/02/2011	219	112	−77.08	−73.42

Digitising was carried out at three main scales for consistency across all ice shelves: (1) 1:100,000 was used for large surface features such as confluence zones and longitudinal structures; (2) 1:50,000 was used for fractures, rifts, the ice front and the grounding zone; and (3) 1:25,000 was used for crevasses, crevassed zones, transverse structures and pressure ridges. Other features, such as ice dolines and ice rises, were mapped at scales appropriate to their individual characteristics.

Surface meltwater was automatically classified in Definiens Developer 7.0 (eCognition) using a multi-resolution segmentation and image-object classification approach. The Landsat images were segmented using the inherent spectral difference between snow, ice and surface meltwater. A shape factor (that determines the overall geometry of each spectral segment and defines the textural homogeneity of the resulting objects) was set at 0 (on a scale of 0–1) to ensure the spectral reflectance of the chosen image band was the main determinate of the image objects rather than segment shape. A compactness value of 1 (on a scale of 0–1) was also used to ensure that the final segmentation process was based solely upon the distinct spectral properties of the image band. The resulting image objects thus had no pre-determined shape, but reflected the nature of the spectral properties of the meltwater. Resulting image objects were thus shaped according to the contrast between the surface meltwater and surrounding ice surface at a scale comparable to the chosen imagery, with those areas containing surface meltwater classified accordingly based on user-defined statistical thresholds extracted from the segmented image objects. The final map is presented in a Polar Stereographic (metres) projection with a WGS84 datum, and displayed at a 1:1,000,000 scale.

4. Description of glaciological features and structural glaciology

4.1. Ice front

The ice front was identified as the sharp transition from shelf ice to either: (1) open water (typically during austral summer) or; (2) sea ice (austral winter, multi-annual sea ice), observed as a bright-sunlit or dark-shaded sub-linear feature (depending on its orientation with respect to the Sun's azimuth) (Figure 2(A)). The ice front is a clear indicator of the maximum ice-shelf extent for a particular time period.

Figure 2. Feature identification in Landsat ETM+ Panchromatic Band 8 as discussed in the main text. (A) An ice front and adjacent sea ice, (B) the grounding line/ break of slope, (C) longitudinal structures (flowlines), (D) pressure ridges (also termed pressure rollers), (E) crevassing on grounded ice, (F) fractures emanating from the grounding line, (G) rifts filled with ice melange, (H) fracture traces (healed fractures), (I) surface meltwater on George VI Ice Shelf, (J) ice dolines, (K) an ice rise and ice rumple.

4.2. Grounding zone

Identified as a sudden break in slope in the optical imagery, or an area of intense crevassing, that represents the transition from grounded ice of the interior ice sheet and floating shelf ice. The grounding zone has several distinct features including the limit of ice flexure from tidal forcing, the limit of ice flotation, an inflexion point (break of slope) and the point where the floating ice reaches hydrostatic equilibrium (Fricker & Padman, 2006; Vaughan, 1994). Here, the grounding zone has been mapped as the break of slope as it is the most distinct feature in satellite imagery (Figure 2(B)). The true grounding line is likely to be a few hundred metres further upstream. The area between the ice front and the grounding zone represents the total ice-shelf area that fluctuates with a change in ice-shelf mass balance.

4.3. Longitudinal structures

Longitudinal structures (also termed flow stripes (e.g. Casassa et al., 1991)), flow bands (e.g. Swithinbank et al., 1988) and streaklines (e.g. Hulbe et al., 2004) are extended features aligned parallel to the principal ice-flow direction and are observed as dark and light couplets caused by shaded surface relief (Figure 2(C)). They are ubiquitous around Antarctica (Glasser & Gudmundsson, 2012) and typically exist in regions of ‘active’ ice, with their cumulative length governed by the timescale of decay relative to the time taken for ice to travel a certain distance, yet their genesis is unclear. Although some field measurements have been acquired (e.g. Campbell et al., 2008; Hulbe et al., 1997; Ng & Conway, 2004), many of the opposing theories of origin have been developed through numerical modelling (e.g. Gudmundsson et al., 1998), remote sensing-based analysis (Campbell et al., 2008; Casassa & Brecher, 1993; Glasser & Gudmundsson, 2012; Raup, Scambos, & Haran, 2005) and transfer of theory developed on valley-glacier structures (e.g. Reynolds & Hambrey, 1988). Subsequently, three key mechanisms have been proposed to explain their formation:

1.	They form through lateral compression followed by longitudinal extension via processes similar to foliation on temperate glaciers (Hambrey, 1977; Hooke & Hudleston, 1978);
2.	They are indicative of shear margins between flow units (Merry & Whillans, 1993);
3.	They are the surface expression of subglacial bed topography transmitted to the surface in regions of fast-flowing ice (Gudmundsson et al., 1998).

However formed, geophysical surveys (e.g. Ng et al., 2004; Rippin et al., 2006) have shown an internal stratigraphy suggesting that their form is also apparent at depth. They can develop on grounded and floating ice, with the former being capable of crossing the grounding zone without deterioration. It is therefore inferred that vertical flexure does not greatly alter their inherent structure. Furthermore, changes in their linearity can be used to identify historical fluctuations, strain history and deformation rates of ice dynamics (Campbell et al., 2008), otherwise it is assumed that linear features represent a steady flow regime.

4.4. Pressure ridges

Pressure ridges form in various circumstances, but represent areas where the ice shelf has deformed under longitudinal or transverse compression. Unlike longitudinal features, little scientific investigation has been carried out on pressure-ridge structures. In satellite imagery they are identified as singular or successions of elongated shaded ridges above the surface of the surrounding ice (Figure 2(D)).

Pressure ridges on Antarctic Ice Shelves have been recognised as important indicators of stress and strain rates, most strikingly along the Alexander Island grounding zone of George VI Ice Shelf (Reynolds & Hambrey, 1988), although similar features were also found on the McMurdo Ice Shelf (Collins & McCrae, 1985) and in portions of the Ross Ice Shelf (Kehle, 1964). Pressure-ridge dimensions vary, with Wright and Priestley (1922) identifying amplitudes of 12 m and wavelengths of 3.2 km on Beardmore Glacier, Gould (1935) commenting on ridges 9 m high and 1.2 km in wavelength on Shackleton Ice Shelf, and Kehle (1964) describing pressure rollers with amplitudes of 12–15 m and wavelengths of 180–650 m on the Ross Ice Shelf. On George VI Ice Shelf, Reynolds and Hambrey (1988) identified ridges 15 m high where the dominant westward flow caused transverse foliation under increasing longitudinal compression against Alexander Island. More recently, Labarbera and MacAyeal (2011) modelled the formation of en echelon meltwater lakes on George VI Ice Shelf within the troughs between pressure ridges, along the margin of Alexander Island. Pressure ridges are thus indicative of high compressional stresses typically perpendicular to their length.

4.5. Crevasses, fractures and rifts

Fracturing of ice occurs when a temperature-dependant threshold is exceeded, with structures aligned perpendicularly to the direction of maximum tension. Here, crevasses are defined by open fractures that form on grounded ice, and cross the grounding zone with ice motion (Figure 2(E)). The term fracture is given to any surface opening that forms within the ice-shelf itself (Figure 2(F)), with the term rift given to large fractures that visibly penetrate the entire depth of the ice shelf surface (Figure 2(G)). Rifts are often filled with open water, sea ice, or a mélange of calved shelf ice, sea ice and water. All three terms describe structures that form in areas of tensile or shear stresses caused by the motion of ice. Fracture traces are former crevasses, fractures or rifts that leave a distinct imprint on the surface, but do not have a visible opening (Figure 2(H)).

4.6. Surface meltwater

Surface meltwater is easily identifiable (Figure 2(I)) due to the spectral differences between it and the surrounding bare ice or snow of the ice-shelf surface. Visible meltwater indicates areas of surface ablation and can suggest ice-shelf surface-slope orientation where flow direction can be derived. Furthermore, assessing a time-series of surface meltwater extent can indicate changing atmospheric temperatures. Surface meltwater has been identified on Bach and George VI Ice Shelves, but Stange Ice Shelf remains free of meltwater from at least 1973 to present (Holt, 2013).

4.7. Ice dolines

Ice dolines are large, sub-rounded hollows often filled with surface meltwater during the austral summer (Bindschadler et al., 2002; MacAyeal & Sergienko, 2013) that form in areas of extensive surface melt, and may suggest weaker areas of shelf ice (Bindschadler et al., 2002). Ice dolines typically exist in three distinct zones: (1) in previously crevassed ice; (2) where an abundant supply of surface meltwater is available; and (3) in clusters at or near to flow unit boundaries. Indeed, all ice dolines only occur in areas where some meltwater is present. Once formed, the horizontal dimensions of ice dolines alter depending on the stress and strain rate of the surrounding ice, elongating with extensional flow either longitudinally or transversely. Visual inspection of shadowing of individual ice dolines suggests steep-sided cliffs around an internal crater (Figure 2(J)).

4.8. Ice rises and ice rumples

Ice rises and ice rumples are elevated portions of the ice-shelf surface caused by grounding on sub-glacial topography. Ice rises occur where the shelf becomes pinned on the bedrock, whereas ice rumples form where ice-shelf motion is still possible over the bedrock. Both features have similar surface expressions and their identification is dependent on the visibility of other glaciological surface features such as pressure ridges upstream and fracturing downstream (Figure 2(K)).

5. Conclusions

The map described here presents the structural glaciology of southwest Antarctic Peninsula Ice Shelves that forms part of a larger research project into the stability of southwest Antarctic Peninsula ice-shelf systems (Holt, 2013; Holt et al., 2013). We have also provided information on the glaciological structures that appear on Bach, George and Stange Ice Shelves in order for the reader to interpret their glaciological structural regime for ca. 2010.

Software

Esri ArcGIS 9.3 was used for all structural mapping and the production of the final map.

Acknowledgements

The authors would like to thank H. De Angelis, A. Cook, H. Apps and M. Smith for their helpful comments and suggestions that improved the overall quality of this article and Supplementary Map.