Introduction
Polythermal glacier surges in Svalbard are characterised by dynamic and cyclical switches between a decades-long quiescent phase of low velocities and terminus recession and a years-long active phase, during which ice flow velocities can increase to 10–1000 times typical quiescent phase values (Dowdeswell and others, Reference Dowdeswell, Hamilton and Hagen1991; Hagen and others, Reference Hagen, Liestøl, Roland and Jørgensen1993; Murray and others, Reference Murray, Strozzi, Luckman, Jiskoot and Christakos2003). Increased flow velocities in the surge active phase are usually accompanied by the transfer of mass to lower elevations, often resulting in frontal advance (Murray and others, Reference Murray, Strozzi, Luckman, Jiskoot and Christakos2003; Sund and others, Reference Sund, Eiken, Hagen and Kääb2009, Reference Sund, Lauknes and Eiken2014; Kristensen and Benn, Reference Kristensen and Benn2012). Glacier geometry and internal structural properties can change rapidly during surge active phases (Raymond and others, Reference Raymond, Johannesson, Pfeffer and Sharp1987; Sharp and others, Reference Sharp, Lawson and Anderson1988; Lawson and others, Reference Lawson, Sharp and Hambrey1994; Murray and others, Reference Murray2012; Małecki and others, Reference Małecki, Faucherre and Strzelecki2013; King and others, Reference King, Hambrey, Irvine-Fynn and Holt2016). Investigation of glaciological structures such as crevasses, shear planes, foliation and medial moraines have increased understanding of cumulative deformation and variable stress and strain regimes experienced during different phases of the surge cycle (e.g. Sharp and others, Reference Sharp, Lawson and Anderson1988; Hodgkins and Dowdeswell, Reference Hodgkins and Dowdeswell1994; Lawson and others, Reference Lawson, Sharp and Hambrey1994, Reference Lawson, Sharp and Hambrey2000; Lawson, Reference Lawson1996, Reference Lawson1997; Hambrey and Dowdeswell, Reference Hambrey and Dowdeswell1997; Fleming and others, Reference Fleming2013; Hudleston, Reference Hudleston2015; King and others, Reference King, Hambrey, Irvine-Fynn and Holt2016; Sevestre and others, Reference Sevestre2018; Hambrey and Clarke, Reference Hambrey and Clarke2019; Young and others, Reference Young, Flowers, Jiskoot and Gibson2022).
Ice deformation during the quiescent and active phases of surges can lead to a complex array of glaciological structures and detailed structural mapping can help to establish a chronology of structural development (e.g. Lawson and others, Reference Lawson, Sharp and Hambrey1994, Reference Lawson, Sharp and Hambrey2000; Hambrey and Dowdeswell, Reference Hambrey and Dowdeswell1997; King and others, Reference King, Hambrey, Irvine-Fynn and Holt2016; Hambrey and Clarke, Reference Hambrey and Clarke2019). Kinematic structures associated with ductile deformation, such as medial moraines and longitudinal foliation, are largely a passive consequence of ice flow and develop during the quiescent phase. Brittle deformation dominates the active phase, when strain rates can be an order of magnitude higher than during quiescence, forming dynamic/constructive structures such as thrust-faults and crevasses (Sharp and others, Reference Sharp, Lawson and Anderson1988; Lawson and others, Reference Lawson, Sharp and Hambrey1994, Reference Lawson, Sharp and Hambrey2000; Hambrey and others, Reference Hambrey, Dowdeswell, Murray and Porter1996; Lawson, Reference Lawson1996, Reference Lawson1997; Hambrey and Dowdeswell, Reference Hambrey and Dowdeswell1997; King and others, Reference King, Hambrey, Irvine-Fynn and Holt2016; Hambrey and Clarke, Reference Hambrey and Clarke2019). The evolution of active phase structures is typically closely linked to either (1) downglacier propagation of a surge front, where structures such as longitudinal crevasses (where extension is normal to ice flow) and shear planes form in a zone of longitudinal compression at or below the surging ice, and structures such as transverse crevasses (where extension is parallel to ice flow) form above the surge front (e.g. Hambrey and others, Reference Hambrey, Dowdeswell, Murray and Porter1996; Lawson, Reference Lawson1996; Hambrey and Dowdeswell, Reference Hambrey and Dowdeswell1997; Kristensen and Benn, Reference Kristensen and Benn2012; King and others, Reference King, Hambrey, Irvine-Fynn and Holt2016); or (2) up-glacier expansion of crevasse fields where surges initiate at the front of tidewater glaciers (e.g. Hodgkins and Dowdeswell, Reference Hodgkins and Dowdeswell1994; Sund and others, Reference Sund, Lauknes and Eiken2014; Flink and others, Reference Flink2015; Sevestre and others, Reference Sevestre2018).
We aim to investigate the genesis of glaciological structures in the Paulabreen glacier system and constrain their evolution during the most recent surge by (1) mapping 2-D surface structures interpreted from aerial photographs captured during the early stages of surging and after the surge terminated; and (2) mapping and measuring 3-D structures exposed in the tidewater front since surge termination. We present a conceptual model for the structural evolution of the glacier system during the final stages of its recent surge cycle.
Paulabreen glacier system surge history
The glacier system is located in southern Spitsbergen at the head of Rindersbukta, a 12 km-long-tributary fjord of Van Mijenfjorden (Fig. 1). Three main contiguous glaciers comprise the glacier system at present: the main tidewater terminus Paulabreen is flanked by Bakaninbreen (to the northeast) and Skobreen (to the southwest). Bakaninbreen and Paulabreen are trunk glaciers separated by a medial moraine extending from Siggerudfjella. Skobreen is a tributary glacier separated from Paulabreen by a frontal moraine loop (Sund, Reference Sund2006; Kristensen and Benn, Reference Kristensen and Benn2012).
The Paulabreen glacier system has one of the longest and best recorded surge histories of any glacier system in Svalbard (Lovell and others, Reference Lovell2018). The largest surge occurred in ad ~1300, when the combined glacier system advanced to the boundary between inner and outer Van Mijenfjorden (Hald and others, Reference Hald, Dahlgren, Olsen and Lebesbye2001; Ottesen and others, Reference Ottesen2008; Kristensen and others, Reference Kristensen, Hormes and Ottesen2009a; Larsen and others, Reference Larsen2018). The surge moraines of Torellmorenen, Damesmorenen and Crednermorenen were deposited at this time (Kristensen and others, Reference Kristensen, Hormes and Ottesen2009a, Reference Kristensen, Juliussen, Christiansen and Humlum2009b; Larsen and others, Reference Larsen2018) and a short-lived proglacial lake was dammed in the area now occupied by the shallow lagoon Braganzavågen (Lyså and others, Reference Lyså2018). At the mouth of Rindersbukta, three submarine moraines with associated debris-flow lobes on their distal flanks (M1–3 in Fig. 1) record three additional surge maximum positions (Ottesen and others, Reference Ottesen2008; Larsen and others, Reference Larsen2018). The innermost moraine/debris-flow lobe (M1) is coincident with the ad 1898 frontal position of the glacier system as mapped by Kjellström (Reference Kjellström1901), indicating a surge during the Little Ice Age (Ottesen and others, Reference Ottesen2008; Larsen and others, Reference Larsen2018). M2 and M3 are undated but can be very broadly constrained to between ad ~1300 and ad 1898 (Larsen and others, Reference Larsen2018; Lovell and others, Reference Lovell2018).
The glacier system underwent sustained frontal retreat of ~10 km to the head of Rindersbukta throughout the 20th century (Ottesen and others, Reference Ottesen2008; Benn and others, Reference Benn, Kristensen and Gulley2009; Larsen and others, Reference Larsen2018). A renewed phase of quasi-independent surge activity began in 1985 with the development of a surge front in the upper part of Bakaninbreen (Dowdeswell and others, Reference Dowdeswell, Hamilton and Hagen1991; Hambrey and others, Reference Hambrey, Dowdeswell, Murray and Porter1996; Murray and others, Reference Murray, Dowdeswell, Drewry and Frearson1998). The surge front propagated downglacier as a ramp up to 60 m high and at rates of 1.0–1.8 km a−1, before becoming less steep and slowing considerably in the later years of the surge (Murray and others, Reference Murray, Dowdeswell, Drewry and Frearson1998). The surge terminated in 1995 when the surge front was still ~1.8 km from the tidewater terminus (Murray and others, Reference Murray, Dowdeswell, Drewry and Frearson1998, Reference Murray2000; Smith and others, Reference Smith2002). The surge did not propagate laterally across the medial moraine into Paulabreen (Benn and others, Reference Benn, Kristensen and Gulley2009). Independently, the uppermost parts of Skobreen were in the initial stages of a surge active phase in 1990 (Sund, Reference Sund2006). By summer 2003, the entire Skobreen tributary was surging and the frontal moraine loop had begun to advance into Paulabreen (Sund, Reference Sund2006; Benn and others, Reference Benn, Kristensen and Gulley2009; Kristensen and Benn, Reference Kristensen and Benn2012) (Figs 2a, b). A surge front and associated longitudinal crevasses could be identified close behind the moraine loop and transverse crevassing and ice-drawdown was widespread in the upper reaches of the glacier (Sund, Reference Sund2006; Benn and others, Reference Benn, Kristensen and Gulley2009; Kristensen and Benn, Reference Kristensen and Benn2012). The Skobreen moraine loop continued to advance from 2003 to 2005 as the surge front propagated into Paulabreen, recording a total displacement of ~1.4 km (Benn and others, Reference Benn, Kristensen and Gulley2009; Kristensen and Benn, Reference Kristensen and Benn2012) (Figs 2c, d). The surge front reached the Paulabreen tidewater terminus by April 2005 and the terminus advanced at ~4.8 m d−1 between April and August 2005. The terminus had advanced ~1.4 km by surge termination in winter 2005/06 (Benn and others, Reference Benn, Kristensen and Gulley2009; Kristensen and Benn, Reference Kristensen and Benn2012).
The surge activated the lower part of Paulabreen across its entire width. Remote-sensing analysis shows that the medial moraine separating Paulabreen and Bakaninbreen was progressively displaced sideways and, as a result, Bakaninbreen is now land-terminating (Benn and others, Reference Benn, Kristensen and Gulley2009; Kristensen and Benn, Reference Kristensen and Benn2012) (Fig. 2). Both the Bakaninbreen (1985–95) and Skobreen–Paulabreen (2003–05) surges failed to propagate laterally across the medial moraine between Bakaninbreen and Paulabreen, which Benn and others (Reference Benn, Kristensen and Gulley2009) attributed to the presence of a persistent subglacial conduit below the medial moraine. Time-lapse footage of the advancing southwest margin of the Skobreen–Paulabreen surge in 2005 demonstrates plug-like flow, deformation of the lateral proglacial moraine area and the incorporation of large volumes of debris into the ice (Kristensen and Benn, Reference Kristensen and Benn2012). Since 2006, the tidewater terminus has receded ~1 km, exposing the glaciological structures in the ice cliff investigated in this study.
Methods
Two-dimensional glaciological structures exposed on the glacier surface were digitised from georeferenced vertical colour aerial photographs at a maximum resolution of 1:5000. The aerial photographs were captured on 8 August 2003 (as Skobreen began to surge into Paulabreen) and 7 July 2011 (~5 years following surge termination). The 2003 photographs were collected by the UK Natural Environment Research Council (NERC) Airborne Research and Survey Facility (ARSF) and acquired from the NERC Earth Observation Data Centre. The 2011 photographs were collected by the Norwegian Polar Institute and acquired from the TopoSvalbard online archive (toposvalbard.npolar.no). Structures were identified and categorised according to their inferred sequential development using standard structural geological notation and following the examples of previous structural glaciology studies (e.g. Lawson and others, Reference Lawson, Sharp and Hambrey1994; Hambrey and Dowdeswell, Reference Hambrey and Dowdeswell1997; Hambrey and others, Reference Hambrey2005; Hudleston, Reference Hudleston2015; Lovell and others, Reference Lovell2015a; King and others, Reference King, Hambrey, Irvine-Fynn and Holt2016; Hambrey and Clarke, Reference Hambrey and Clarke2019; Jennings and Hambrey, Reference Jennings and Hambrey2021). Fieldwork in April 2013 focused on the identification of planar glaciological structures exposed in the ~1.5 km-long Paulabreen tidewater ice cliff following surge termination. Vertical cross sections were recorded from a photomosaic of the ice cliff. Three-dimensional orientation measurements (strike and dip) of foliation and englacial debris-rich structures (debris layers) were collected using a compass clinometer. Measurements of linear structure (dip and dip direction) were collected from sheared debris laminae, or mineral stretching lineations, within the debris layers (Fleming and others, Reference Fleming2013). Structural data were corrected for magnetic deviation and plotted as equal-area stereographic projections using Stereo32 software (Röller and Trepmann, Reference Röller and Trepmann2008).
Results
Glaciological structures
The five sets of glaciological structures identified on the glacier surface in the 2003 and 2011 aerial photographs are medial moraines, foliation (S1), fracture traces (S2), arcuate fracture traces (S3) and open crevasses (S4) (Fig. 3). Primary stratification (S0) is not visible in the lower reaches of the glacier system covered by Figure 3 but is seen higher up on TopoSvalbard aerial photographs, and we include this within the overall structural sequence for completeness (Table 1). Medial moraines, foliation (S1) and fractures were also identified in the ice cliff in 2013, together with englacial debris-rich structures (debris layers) (Figs 4–6). The structures are first described here, before their distribution, morphology and structural properties as observed on the glacier surface and in the ice cliff are outlined.
The large medial moraines observed on the glacier surface (Fig. 3) can also be identified in the ice cliff (Figs 4, 5a, 6c). The medial moraines are characterised by kilometres-long, tens-of-metres-wide surface debris drapes with a positive topographic expression (Fig. 3) emanating from thin, near-vertical or inclined debris layers (Figs 4, 5a, 6c). Medial moraines typically form (1) through the merger of supraglacial lateral moraines at the confluence of two flow-units or (2) as buried debris layers that have been tightly folded meltout in a glacier's lower reaches (Eyles and Rogerson, Reference Eyles and Rogerson1978a; Hambrey and others, Reference Hambrey, Bennett, Dowdeswell, Glasser and Huddart1999; Hambrey and Glasser, Reference Hambrey and Glasser2003; Hudleston, Reference Hudleston2015). The latter process occurs in conjunction with the formation of strongly developed longitudinal foliation (S1) (Hambrey and others, Reference Hambrey, Bennett, Dowdeswell, Glasser and Huddart1999). Most of the medial moraines observed in the lower reaches of the glacier system separate major flow-units/tributaries and can be traced to supraglacial lateral moraines (Figs 2 and 3). In a few cases, smaller discontinuous medial moraines in close association with longitudinal foliation may be debris-bearing S1 structures that have melted out at the surface (e.g. Hambrey and others, Reference Hambrey, Bennett, Dowdeswell, Glasser and Huddart1999, Reference Hambrey2005).
Foliation (S1) is observed on the glacier surface (Fig. 3) and in the ice cliff (Fig. 5b). The surface expression of foliation forms linear stripes oriented with the direction of ice flow (Fig. 3), termed longitudinal foliation (e.g. Hambrey and Clarke, Reference Hambrey and Clarke2019). In the ice cliff, foliation can be identified as alternating layers of bubble-rich and clear (bubble-poor) ice (Fig. 5b) (Lawson and others, Reference Lawson, Sharp and Hambrey1994). Foliation is considered a secondary structure originating from primary stratification (S0) that has subsequently been deformed and folded. Longitudinal foliation is the result of high degrees of simple shear relating to glacier flow, creating structures aligned parallel to glacier flow with isoclinal folds reflecting differing flow velocities within the ice (Jennings and Hambrey, Reference Jennings and Hambrey2021).
Two generations of fractures can be identified on the glacier surface: fracture traces (S2) and arcuate fracture traces (S3). Fracture traces (S2) are in general short, straight features and are likely to be relict extensional crevasses that have healed (e.g. Hambrey and Dowdeswell, Reference Hambrey and Dowdeswell1997; Goodsell and others, Reference Goodsell, Hambrey, Glasser, Nienow and Mair2005; King and others, Reference King, Hambrey, Irvine-Fynn and Holt2016). Arcuate fracture traces (S3) are typically longer and have a curved morphology (Fig. 3). Such features are commonly interpreted to be shear planes or thrust-faults formed in zones of longitudinal compression (e.g. Hambrey and others, Reference Hambrey, Bennett, Dowdeswell, Glasser and Huddart1999; Appleby and others, Reference Appleby, Brook, Vale and Macdonald-Creevey2010; Roberson and Hubbard, Reference Roberson and Hubbard2010; Lovell and others, Reference Lovell2015a). Fracture traces are also seen in the ice cliff, where they crosscut foliated ice, other fracture traces and debris layers, and in general display one or two dominant orientations (Fig. 4). We are unable to distinguish the fractures observed in section as separate structural elements, but it is likely both healed crevasses (S2) and shear planes (S3) are represented.
Open crevasses (S4) are observed on the glacier surface. Crevasses typically form transverse to ice flow (45° either side of flow perpendicular) in zones of extension and longitudinal to ice flow (45° either side of flow parallel) in zones of compression (Lawson and others, Reference Lawson, Sharp and Hambrey2000; Rea and Evans, Reference Rea and Evans2011; Kristensen and Benn, Reference Kristensen and Benn2012).
Debris layers observed in the ice cliff are thin near-vertical or dipping debris-rich structures extending tens-of-metres from the visible base of the section (Figs 4, 5). Debris layers in PLB1 and PLB2 often display tangential geometries (Fig. 5a) and steepen towards the top. Larger areas of debris are common towards the base of debris layers, which typically taper out towards their upper limits. Debris within the layers varies from sorted sands and gravels (Fig. 5c) to poorly sorted diamicton ranging from silts to pebble and cobble-sized clasts (Fig. 5g). Where debris layers consist of fine material, sheared debris laminae/mineral stretching lineations can be identified (Fig. 5g) (e.g. Fleming and others, Reference Fleming2013). Debris layers exposed in section in surge-type glaciers are typically inferred to have formed at a glacier bed experiencing very large stress gradients by (1) squeezing into basal crevasses, (2) hydrofracturing or (3) thrust-style displacement of pre-existing planar weaknesses (e.g. shear planes, crevasse traces) (e.g. Glasser and others, Reference Glasser, Hambrey, Crawford, Bennett and Huddart1998; Woodward and others, Reference Woodward, Murray and McCaig2002; Rea and Evans, Reference Rea and Evans2011; Lovell and others, Reference Lovell2015b).
Structures on the glacier surface
In 2003, the Bakaninbreen and Paulabreen medial moraines were relatively straight and continuous, tracing the overall flow configuration of the glacier (Fig. 3a). The medial moraine loop separating Skobreen and Paulabreen had begun to advance into Paulabreen as ice surged out of Skobreen (Figs 2b, 3a). By 2011, the Bakaninbreen and Paulabreen medial moraines had been laterally displaced ~500 m to the northeast and the Skobreen moraine loop extended a similar distance laterally into Paulabreen and up to ~1.8 km downglacier (Fig. 3b).
The orientation and morphology of longitudinal foliation (S1) exposed on the glacier surface appears generally unaltered between 2003 and 2011, but both lateral and downglacier displacement is evident. Longitudinal foliation was densely clustered around the Bakaninbreen–Paulabreen and Paulabreen medial moraines in 2003 and extended to the glacier front (Fig. 3a). In 2011 the longitudinal foliation was largely unaltered in Paulabreen above the heavily crevassed zone. Closer to the glacier front, longitudinal foliation was displaced laterally to the northeast together with the medial moraines.
Fracture traces (S2), interpreted as healed crevasses, were found in the central and southwest parts of Paulabreen in 2003 (Fig. 3a). Arcuate fracture traces (S3), interpreted as shear planes, were concentrated up-glacier of the Skobreen moraine loop, closely associated with arcuate topographic ‘waves’ in the ice surface (Kristensen and Benn, Reference Kristensen and Benn2012). By 2011, the arcuate fracture traces (S3) within Skobreen were displaced as the moraine loop advanced into Paulabreen but were otherwise relatively unaltered in morphology and distribution. The lower part of the glacier system was heavily crevassed and only a small area of arcuate fracture traces (S3) concentrated at the southwest margin was visible (Fig 3b). We suggest these are shear planes (S3) rather than healed crevasses (S2) because of their location in a zone of compression at the lateral glacier margin. However, some of the shear planes are likely to have reactivated the healed crevasses that were present in this area in 2003 (e.g. Hambrey and Müller, Reference Hambrey and Müller1978; Goodsell and others, Reference Goodsell, Hambrey, Glasser, Nienow and Mair2005; Appleby and others, Reference Appleby, Brook, Vale and Macdonald-Creevey2010; Rea and Evans, Reference Rea and Evans2011).
Open crevasses (S4) were confined to two zones in 2003: transverse crevasses associated with the tidewater front and short longitudinal crevasses in the centre of Skobreen (Fig. 3a) (Kristensen and Benn, Reference Kristensen and Benn2012). By 2011 most of the lower section of Skobreen and Paulabreen was crevassed. The Skobreen crevasse pattern was dominated by short longitudinal crevasses aligned with the main axis of ice flow to the north. At the moraine loop, the Skobreen crevasse orientation remained similar, transverse to ice flow within the Paulabreen main trunk (Fig. 3b). Crevasses in the central section of Paulabreen were generally aligned transverse to ice flow, indicative of mode 1 fracturing (Benn and others, Reference Benn, Warren and Mottram2007; Rea and Evans, Reference Rea and Evans2011). The densest crevassing in Paulabreen in 2011 was close to the downglacier limit of the Skobreen moraine loop, ~2 km from the tidewater front. Many crevasses in this zone were aligned ~45° from perpendicular to ice flow with two dominant orientations, forming conjugate fracture sets (e.g. Lawson and others, Reference Lawson, Sharp and Hambrey1994; Rea and Evans, Reference Rea and Evans2011). Aside from transverse crevasses at the tidewater front, the lowermost ~2 km of Paulabreen was largely crevasse-free in 2011.
Structures in the ice cliff
A ~1.5 km-long section of the tidewater front of Paulabreen was investigated in 2013, extending from close to where the Bakaninbreen–Paulabreen medial moraine intersects the glacier front (towards the northeast lateral margin) to the point where the southwest margin of Paulabreen meets the lateral proglacial moraine area (Fig. 4a). The ice cliff comprises the central and southwest parts of the glacier system, as Bakaninbreen is currently land terminating. The ice cliff is divided into three sections for ease of reporting: PLB1 (towards the Bakaninbreen–Paulabreen medial moraine), PLB2 (central section) and PLB3 (southwest margin) (Figs 3, 4). The left-hand (northeast) part of PLB1 contains several dipping debris layers with tangential geometries (Fig. 5a) and the easternmost of the large medial moraines within Paulabreen (Figs 4b, 6c). The right-hand (southwest) part of PLB1 contains a series of fractures and areas of clear (bubble-poor) ice.
PLB2 contains several debris layers (e.g. Fig. 5f) and areas of clear ice. The second large Paulabreen medial moraine is exposed here. Its cross section reveals the relationship between foliation, debris layers and the medial moraine (Figs 6c, d). The surrounding foliation is tightly folded but dips consistently to the southwest. Thin debris layers are observed with a similar orientation, which thicken towards the medial moraine. The trace of the medial moraine can clearly be seen and consists of a zone of debris of varying thickness (10–50 cm), also dipping to the southwest. The foliation and debris layers are truncated by the medial moraine.
Debris layers are particularly well exposed in PLB3, often in close association with areas of clear ice (e.g. Figs 5d, e). Here they display evidence for polyphase deformation. For example, Figures 6a and b show flow-parallel debris layers cut by a subparallel debris layer that follows a strike–slip fault.
Structures in the tidewater front typically strike subparallel to parallel to the main axis of ice flow in Paulabreen (Fig. 7). However, significant spatial variation in dip orientation is seen depending on the position along the ice cliff. In PLB1, foliation is typically oriented subparallel to the glacier margins and displays a strong preferred orientation striking northwest/southeast and dipping steeply to the southwest (131/34). A second sub-vertical orientation is observed in the foliation trending north–south, dipping subvertically to the west (359/80) (Fig. 7b-ii). Foliation is tightly folded and a number of isoclinal folds are recognised, with horizontal axes orientated parallel to the strike of the foliation (mean vector 132/03) (Fig. 7b-iii). Debris layers display a similar orientation to the foliation, trending northwest–southeast and dipping steeply to the east-northeast (mean orientation 128/43). As with the foliation, a second sub-vertical orientation is observed trending north–south and dipping subvertically to the west (004/76) (Fig. 7b-i).
Structures in PLB2 are typically also oriented subparallel to the glacier margins, but dip directions vary considerably (Fig. 7c). Two distinct clusters are observed in the foliation data. In the southwest part of the section, the foliation follows a similar pattern to that seen in PLB3, striking north-east-north/south-west-south and dipping to the east-southeast (020/35). In the northeast part of the section foliation strikes west-northwest/east-southeast, dipping to the south-west-south (111/33) (Fig. 7c-ii). Foliation in PLB2 is also tightly folded, with axes orientated parallel to the strike of the foliation (mean vector 302/03) (Fig. 7c-iii). Debris layers in PLB2 show similar bipartite clustering with one cluster from the southwest part of the section striking north–south, dipping moderately to the east (304/40), while those from the northeast part strike northwest–southeast, dipping to the southwest (120/39) (Fig. 7c-i). Stretching lineations are typically sub-horizontal, trending southeast (120/39), although a steeply dipping lineation was recognised (48/224) (Fig. 7c-iv), which could represent thrust-style displacement along a debris layer (e.g. Rea and Evans, Reference Rea and Evans2011; Lovell and others, Reference Lovell2015b).
Structures in PLB3 are generally oriented subparallel to the glacier margins (Fig. 7d). Foliation is oriented subparallel to the glacier margins but is variable. Most of the foliation trends northwest/southeast (338/75), dipping steeply to the northeast. However, a second sub-vertical orientation is observed trending northwest–southeast, dipping moderately to the southwest (135/47) (Fig. 7d-ii). Foliation is tightly folded and a number of folds are recognised. These have horizontal axes orientated parallel to the strike of the foliation (mean vector 135/01) (Fig. 7d-iii). Debris layers are less variable and mostly strike north-northwest–south-southeast, dipping steeply to the east-northeast (mean orientation 349/53) (Fig. 7d-i). Stretching lineations within the debris layers are sub-horizontal (163/08) (Fig. 7d-iv).
These structural data reveal that foliation and debris layers predominantly strike subparallel to glacier flow, which is also indicated by foliation on the glacier surface (Figs 3, 7a). However, measurements at the ice front reveal distinct variations in dip, contrary to what might be expected. At the lateral margin (PLB3), structures predominantly dip to the northeast (Figs 4, 7d). However, the orientation quickly changes along the ice front in a northeast direction and from the medial moraine in PLB2 onwards, structures dip to the southwest (Figs 4, 7b, c). Furthermore, structures display distinct tangential bases (Fig. 5a), suggesting simple shear to the north with increasing intensity towards the base.
Discussion
Glaciological structures exposed on the glacier surface in 2003 and 2011 and in the ice cliff in 2013 are a product of the recent surge history of the glacier system. We combine these data with observations by Kristensen and Benn (Reference Kristensen and Benn2012) to reconstruct the impact of surging on the structure of the glacier system since 1995, capturing the period immediately following the Bakaninbreen surge termination in 1995 through to the late stages of the Skobreen/Paulabreen surge in 2005 (Fig. 8).
Late quiescent phase structural evolution
Figure 8a is a simplification of the configuration of the glacier system following the 1995 Bakaninbreen surge, which terminated when the surge front was still ~2 km up-glacier from the tidewater terminus (Murray and others, Reference Murray, Dowdeswell, Drewry and Frearson1998, Reference Murray2000; Smith and others, Reference Smith2002). The tidewater front was shared evenly between the two main flow-units of Bakaninbreen and Paulabreen. The medial moraine between the two was relatively straight and oriented parallel to the main axis of glacier flow, aside from a slight bulge into Paulabreen, close to the medial moraine's inception point, associated with the passage of the Bakaninbreen surge front in this area from ~1987 onwards (Murray and others, Reference Murray, Dowdeswell, Drewry and Frearson1998). Foliation (S1) in Paulabreen was well-developed and orientated parallel to glacier flow, as were the medial moraines, which we assume were near-vertical (e.g. Eyles and Rogerson, Reference Eyles and Rogerson1978b). Fracture traces and any debris layers were subparallel to flow, reflecting simple shear at the base and margins. The Skobreen and Sokkbreen (and further downglacier, Peisbreen, not shown in Fig. 8a) moraine loops extended into Paulabreen but were not actively impeding flow. Oblique aerial photographs from 1936 show these tributaries had uncrevassed surfaces and moraine loops with a similar morphology to that in 1995, indicating that this was a relatively stable glacier configuration during the quiescent phase (Sund, Reference Sund2006). Marginal crevassing in the uppermost basins of Skobreen (not shown in Fig. 8a) suggests the glacier was in the initial stages of the surge active phase at this time (Sund, Reference Sund2006).
Early surge phase
Figure 8b depicts the arrival of the surge front at the Skobreen moraine loop in summer 2003, which had begun to advance into Paulabreen. Longitudinal crevasses in the centre of Skobreen reflect compressive flow at the surge front, creating topographic waves on the ice surface close to the moraine loop (Kristensen and Benn, Reference Kristensen and Benn2012). Arcuate fracture traces (S3) inside the moraine loop are interpreted as shear planes formed in the compressive zone downglacier of the surge front. The extension of the Skobreen moraine loop into Paulabreen by ~300 m caused southwest–northeast compression and initiated lateral migration of ice to the northeast, evidenced by small displacements of the Paulabreen and Bakaninbreen–Paulabreen medial moraines (Fig. 2b). The displacement began to reorientate the structures in the PLB1 and PLB2 sections of the ice front towards the northeast (dipping to the southwest) from their original orientation subparallel to flow through simple shear (Fig. 7). The absence of crevasses (S4) in the main trunk of Paulabreen indicates that, although the surge had begun to affect the overall structure, it had not yet activated the ice in Paulabreen (Kristensen and Benn, Reference Kristensen and Benn2012).
End of the surge phase
Figure 8c shows the glacier system at the end of the surge, which terminated by spring 2006 (Kristensen and Benn, Reference Kristensen and Benn2012). Paulabreen dominated the entire tidewater front at this time, which had advanced ~1.4 km. Kristensen and Benn (Reference Kristensen and Benn2012) showed that the surge reached the southwest margin of Paulabreen first in April 2005, initiating frontal advance in this area, before extending to the entire front by July 2005. The terminus stopped advancing sometime between December 2005 and February 2006 (Benn and others, Reference Benn, Kristensen and Gulley2009). Transverse crevassing (S4) indicates an extensional flow regime dominated in the lower part of Paulabreen and at the front of Skobreen, where crevasses were oriented perpendicular to the main axis of ice flow in Paulabreen. The frontal zone of Paulabreen was chaotically crevassed, but the area immediately up-glacier remained relatively crevasse-free, suggesting ice moved by plug flow at the surge front (Kristensen and Benn, Reference Kristensen and Benn2012). The Skobreen frontal moraine loop extended a further ~200 m laterally into Paulabreen and ~1.8 km downglacier. The Sokkbreen moraine loop was displaced downglacier a similar distance and was now disconnected from Sokkbreen. The Paulabreen and Bakaninbreen–Paulabreen medial moraines were compressed and displaced laterally by up to 500 m and the tidewater front of Bakaninbreen was squeezed out completely. The lateral compression of Paulabreen rotated structures to the northeast via simple shear in PLB1 and PLB2 (Fig. 7). The reorientation is particularly visible where the two Paulabreen medial moraines are exposed in the ice cliff (Fig. 4). Foliation (S1) in PLB3 at the southwest margin remained subparallel to glacier flow, dipping steeply to the northeast (Fig. 7).
Most englacial debris layers are likely to have formed during the surge as saturated subglacial sediment was elevated into the glacier front via squeezing into crevasses, hydrofracturing and thrust-style displacement (Rea and Evans, Reference Rea and Evans2011; Lovell and others, Reference Lovell2015b). Debris layers are best exposed at the southwest margin, where the glacier partly advanced onto the moraine area. The Kristensen and Benn (Reference Kristensen and Benn2012) time-lapse footage shows large volumes of debris in the surging ice in this area and active proglacial deformation of the moraine surface. Mineral stretching lineations on debris layers in PLB3 provide some evidence for shear subparallel to glacier flow (e.g. Fig. 7d-iv) and large-scale faulting is visible in the ice cliff (e.g. strike–slip fault in Fig. 6a). Together with the presence of shear planes (S3) on the glacier surface in this area (Fig. 3b), these observations are consistent with a zone of intense compression and transpression (strike–slip deformation that deviates from simple shear because of a component of shortening orthogonal to the deformation zone; Dewey and others, Reference Dewey, Holdsworth and Strachan1998) at the southwest margin (Kristensen and Benn, Reference Kristensen and Benn2012).
Near-vertical debris ridges were observed to be melting out at the southwest glacier margin in spring 2013 (Fig. 5h), which will likely form a geometrical ridge network in the lateral proglacial moraine area (e.g. Glasser and others, Reference Glasser, Hambrey, Crawford, Bennett and Huddart1998; Evans and Rea, Reference Evans and Rea1999; Lovell and others, Reference Lovell2015b, Reference Lovell2018; Farnsworth and others, Reference Farnsworth, Ingólfsson, Retelle and Schomacker2016). The geomorphic imprint of the latest surge is also likely to be recorded on the fjord floor (e.g. Ottesen and others, Reference Ottesen2008; Flink and others, Reference Flink2015).
Conclusions
The 2003–05 glacier surge in the Paulabreen glacier system initiated in the Skobreen tributary and propagated downglacier into the lower part of Paulabreen. Glaciological structures exposed on the glacier surface and in the ice cliff allow the structural evolution of the surge to be reconstructed from the late quiescent phase through to the end of the surge phase. The structure of the late quiescent phase is recorded by relatively straight medial moraines and foliation aligned subparallel to glacier flow in Skobreen and Paulabreen. The early surge phase is characterised by the propagation of the surge front from the upper zone of Skobreen to its shared moraine loop with Paulabreen. In summer 2003, the zone of compression associated with the arrival of the surge front at the moraine loop was recorded by longitudinal crevassing immediately up-glacier from arcuate shear planes and topographic ‘waves’ on the glacier surface. The extension of the moraine loop into Paulabreen initiated lateral displacement of medial moraines towards the northeast and the associated rotation of foliation and debris layers. By summer 2005, the surge front had propagated to the tidewater terminus and activated the lower part of Paulabreen, initiating frontal advance and widespread extensional crevassing. At surge termination by spring 2006, medial moraines and frontal moraine loops had been displaced both laterally and downglacier and the terminus had advanced ~1.4 km. The lateral displacement further rotated foliation and englacial debris layers to the northeast via simple shear in the central part of Paulabreen and closed off Bakaninbreen from the tidewater front. Simple shear was greatest at the base of the ice cliff, resulting in tangential geometries of the structures where dip shallows towards the base. The southwest margin experienced intense compression and transpression as surging ice advanced at the lateral margin, evidenced by shearing, strike–slip faulting and the elevation of debris into the ice.
These observations demonstrate that surges leave a clear imprint on the 3-D structure of glaciers, in addition to the more obvious surface changes, which can still be identified in ice cliffs several years after surge termination. Our work provides additional insight into the evolution of structures as surges propagate through glaciers. The study also demonstrates the potential for deciphering complex glacier flow dynamics through detailed structural glaciological investigations. Such work could be particularly useful at glaciers where the surge history is unclear or unknown.
Data
The data presented in this study are available from the corresponding author, HL, upon reasonable request.
Acknowledgements
The fieldwork was undertaken while HL (NE/I528050/1) and EJF (NE/H004963/1) were funded by NERC Algorithm PhD studentships at Queen Mary University of London and the University of Birmingham, respectively, and held External PhD Candidate status at the University Centre in Svalbard (UNIS). Funding support of an Arctic Field Grant from the Research Council of Norway is also gratefully acknowledged. We thank UNIS logistics and Martin Indreiten for fieldwork support, and Alexandra Messerli and Heidi Sevestre for their help during a visit to the site in April 2012. Doug Benn, Sven Lukas, Ian Fairchild and Carl Stevenson provided essential support throughout. Finally, thanks to Neil Glasser (scientific editor) and Mike Hambrey for their help in improving the paper.
Author contributions
HL and EJF designed the study, collected and analysed field data and wrote the paper.