Master’s thesis
Physical Geography and Quaternary Geology, 30 Credits
Department of Physical Geography
Paleoglaciological study of the Ahlmannryggen, Borgmassivet
and Kirwanveggen nunatak ranges, Dronning Maud Land, East Antarctica, using
WorldView imagery
Taisiya Dymova
NKA 227 2018
Preface
This Master’s thesis is Taisiya Dymova’s degree project in Physical Geography and Quaternary Geology at the Department of Physical Geography, Stockholm University. The Master’s thesis comprises 30 credits (one term of full-time studies).
Supervisors have been Arjen Stroeven, Robin Blomdin and Jennifer Newall at the Department of Physical Geography, Stockholm University. Examiner has been Krister Jansson at the Department of Physical Geography, Stockholm University.
The author is responsible for the contents of this thesis.
Stockholm, 2 September 2018
Lars-Ove Westerberg Vice Director of studies
Abstract
Paleoglaciological reconstructions based on glacial geological and geomorphological traces are used to test and constrain numerical models of ice sheet extent and dynamics. MAGIC-DML (“Mapping, Measuring and Modelling Antarctic Geomorphology and Ice Change in Dronning Maud Land”) project is trying to reconstruct the timing and pattern of ice surface elevation changes since the mid-Pliocene across western Dronning Maud Land, East Antarctica. The study area has sparse pre-existing field data and considerable ice sheet model uncertainties.
A remote sensing-based mapping of glacial geomorphology on nunataks and structures on the ice sheet surface is presented for a coastal-inland transect including Ahlmannryggen, Borgmassivet, and Kirwanveggen using high-resolution WorldView imagery. The primary aim of the study is to map traces of a thicker ice sheet on nunatak slopes that were formerly partly or entirely covered during ice surface highstands. Panchromatic and multispectral images were analysed in a multi-step procedure using ArcGIS, including image processing and mosaicking, visual feature recognition, and mapping. The identification of key landforms (such as till veneers and erratic boulders) required the adoption of some assumptions to differentiate, for example, till from regolith.
Where patterned ground was mapped, we infer a presence of till rather than regolith because subglacial erosion is more likely to produce finer material than subaerial weathering. Very large boulders on plateau surfaces are mapped as erratics because they could not have been delivered by slope processes to local highpoints. However, the reliability of derived paleo-ice sheet reconstructions is limited by both the necessary assumptions and the absence of crosscutting relationships between landforms. At face value, the presence of till cover and erratics above the present ice surface on some nunataks indicate thicker ice in the past. According to the geomorphological mapping of the transect, in Kirwanveggen the former ice elevation was at least 100 m higher, in Borgmassivet the ice lowered more than 600 m and in Ahlmannryggen the ice was at least 300 m thicker. Additional mapping of structures on the ice sheet surface is used to yield target field routes for upcoming field season(s) to potential cosmogenic nuclide (CN) sampling locations. The chronology derived from CN dating will permit the delineation of ice sheet surface elevations as targets for ice sheet modeling.
Contents
1 Introduction 5
2 Study area 5
2.1 Physiography . . . . 5
2.1.1 Nunataks . . . . 5
2.1.2 Ice streams . . . . 5
2.1.3 Geological setting . . . . 6
2.2 Morpho-tectonic and glacial history . . . . 8
2.2.1 Palaeozoic: sedimentation on Gondwanaland . . . . 8
2.2.2 Mesozoic (late Jurassic): break-up of the Gondwanaland . . . . 8
2.2.3 Cenozoic (from Eocene until middle Miocene): onset of warm-based glaciation . . . . 8
2.2.4 Cenozoic (from Middle Miocene until Pliocene): onset of cold-based conditions . . . . 9
2.2.5 Quaternary: thinning of the ice sheet . . . . 9
2.2.6 Last Glacial Maximum . . . . 9
2.2.7 Post-LGM warming . . . . 10
3 Background 10 3.1 Geomorphological significance of landforms on nunataks and the ice sheet surface . . . . 10
3.1.1 Ice and snow features . . . . 15
3.1.2 Sedimentary and bedrock features . . . . 16
3.2 Cosmogenic nuclides and their use for ice thinning reconstruction . . . . 17
4 Methods 18 4.1 Data and software . . . . 18
4.1.1 Remote sensing datasets . . . . 18
4.1.2 Dataset processing . . . . 19
4.1.3 Data plotting . . . . 21
4.2 Geomorphological mapping . . . . 21
4.2.1 Ice flow directions . . . . 22
1
4.3 Mapping validation . . . . 23
4.4 Sampling site selections . . . . 24
5 Results 25 5.1 Glacial geomorphology . . . . 25
5.1.1 Landform examples . . . . 26
5.1.2 Distribution of glacial deposits . . . . 26
5.2 Sampling sites . . . . 34
6 Discussion 47 6.1 Paleoglaciological reconstruction . . . . 47
6.2 Exposure ages of the samples . . . . 47
6.3 Evaluation of the WorldView dataset . . . . 49
7 Conclusions 49 8 Appendix 54 8.1 Python scripts written to perform mosaicking in ArcMap . . . . 54
8.2 MatLab scripts used for generating some of the figures . . . . 56
8.2.1 Figure 1 . . . . 56
8.2.2 Figure 2 . . . . 56
8.2.3 Figure 3 . . . . 57
8.2.4 Figure 12 . . . . 58
8.3 The WorldView images used for mapping . . . . 58
List of Figures 1 An overview bedrock topography map of Dronning Maud Land. The study area is shown by the red polygon. The Borgmassivet range and the escarpment are clearly visible as areas of high elevation (brown) inside the red polygon. Jutulstraumen appears as a deep ice stream on the eastern side of the study area. The ice streams and mountain ranges are labelled on Fig. 2. The continental shelf is clearly visible as light blue shallow area compared to darker blue surrounding deep area of the ocean. The bedrock topography is derived from Fretwell et al. (2013). Generated with Antarctic Mapping Toolbox (Greene et al., 2017), see Appendix 8.2.1. . . . 6
2 An overview map of the study area. The study area is shown by the red polygon. The red dotted line indicates the transect shown on Fig. 3. The shaded relief image is derived from MODIS Mosaic of Antarctica (Haran et al., 2014). Ice speed is derived from Rignot et al. (2017) and is displayed in colors explained in the colorbar. The colors on colorbar may slightly differ from the colors on the map since the ice speed map was drawn as a transparent layer above the shaded relief image. Darker colors indicate higher ice speed. Blue arrows indicate ice flow direction and magnitude (Rignot et al., 2017). The blue line indicates the grounding line derived from Fretwell et al. (2013). Generated with Antarctic Mapping Toolbox (Greene et al., 2017), see Appendix 8.2.2. . . . 7
3 The transect of the study area showing the bedrock topography in black, ice cover in light blue and the ocean in dark blue. The figure shows some topographical features of the study area: how Kirwanveggen dams the ice flowing from the inland polar plateau; that the Penck Trough lies below sea level; Borgmassivet reaches the altitudes similar to Kirwanvegen escarpment; Ahlmannryggen is slightly lower and is separated from Borgmassivet by a valley. The data is derived from Fretwell et al. (2013). The transect extent is indicated on Fig. 2. Generated with Antarctic Mapping Toolbox (Greene et al., 2017) . . . . 8
4 a) The EAIS in warmer climate during the Early-Mid Pliocene (earlier than 3 Ma). b) The EAIS in colder climate during the Mid-Late Pleistocene (later than 1 Ma). Dashed line represents the ice sheet level during the warmer climate. SST on the figures means sea surface temperature. Modified from Yamane et al. (2015) . . . . 9
5 Example of the imagery before (a) and after (b) dataset processing. a) Rough multispectral WV images, b) multispectral mosaic. . . . . 20
2
6 Blue ice areas as seen in (a) "natural colour" band combination and in (b) "standard falce colour"
band combination of the WorldView imagery. Differentiation between snow and ice cover is much clearer in the standard false colour as can be seen in these images. . . . 21 7 Images of Flårjuven Buff area. (a) The map showing slope angles of the raster cells. Red colours
represent steeper slopes. Examples of route proposal and values for steepness calculated for the steepest parts of the route (critical points) are shown. Sampling sites are numerated according to the descriptions in Table 4. (b) The WV panchromatic imagery with geomorphological features shown. 22 8 (a-c )WorldView images of till, in the panchromatic band, showing a variety of cases where such
interpretation was based on assumptions (red stars with numbers mark the sampling sites described in Table 4 and shown on Fig. 18 to ??): (a) glacial erratics and patterned ground lying on top of a plateau with approximate elevation of 1200 m a.s.l.; (b) glacial erratics up to 10 m in diameter lying on a plateau that appears flat in the image but actually has 10◦inclination. Elevation approximately 1300 m a.s.l.. Note the different scale used in F igure 8b; (c) patterned ground on top of a plateau interpreted as till; (d) striped bedrock structure can resemble sorted ground. . . . . 23 9 The example of mapping validation. a) Aerial photo showing patterned ground on Grunehogna Peaks
provided by South African research team (J. Newall, pers. comm.). b) WV panchromatic image of the same patterned ground. . . . 24 10 The example of mapping validation. a) Aerial photo showing the eastern part of Grunehogna Peaks
(the location of sampling sites 12 and 13, see Fig. 19) provided by South African research team (J. Newall, pers. comm.). Sediment cover appears reddish brown. Patches of refrozen ice on the supraglacial moraine appear light blue The aerial image also provides an idea of the typical steepness of the slopes in the study area. b) WV multispectral image of the same area. The sediment cover appears dark and homogenous. c) WV panchromatic image of the same area. The texture of bedrock is visible, the sediment cover looks more homogenous, separate large boulders lying above it appear as black dots. Moraine ridges and separate boulders can be distinguished on supraglacial moraines.
d) WV multispectral image showing an example of geomorphological mapping. For the legend see Fig. 15 . . . . 25 11 Google Earth imagery showing a nunatak in Gjelsvikfjella with patterned ground and glacial erratics
on top. They were sampled by Y. Suganuma and his team (J. Newall, pers. comm.) . . . . 26 12 An overview map of BIAs and LSSs. BIAs are indicated as blue polygons, LSSs are indicated by
blue lines. The study area is shown by the black polygon. The shaded relief image is derived from MODIS Mosaic of Antarctica (Haran et al., 2014). Ice flow velocity is derived from Rignot et al.
(2017) and is displayed in colors explained in the colorbar. Contour lines are marked with elevation numers in m a.s.l. Generated with Antarctic Mapping Toolbox (Greene et al., 2017), see Appendix 8.2 . . . . 28 13 Glacial geomorphological map of the area close to the ice shelf. The black polygon is showing the
extent of figure 14 . . . . 29 14 The example of geomorphological mapping of the area close to the ice shelf. . . . 30 15 Glacial geomorphological map of Ahlmannryggen. The legend also applies for the Borgmassivet
geomorphological map (Fig.16). The mapping of Flårjuven area is shown closer on Fig.7b. The mapping of Grunehogna area is shown closer on Fig.21a and 10d . . . . 31 16 Glacial geomorphological map of Borgmassivet. See Fig.15 for legend. The mapping of Borga area
is shown closer on Fig.22a . . . . 32 17 Glacial geomorphological map of Kirwanveggen. . . . . 33 18 The overview map of the sampling sites. Due to a large amount of sampling sites on the Ahlman-
nryggen and Borgmassivet nunataks, these areas are shown separately on Figures 19 and 20. . . . . 35 19 The map of the sampling sites on Ahlmannryggen nunatak range. . . . 36 20 The map of the sampling sites on Borgmassivet nunatak range. . . . 37 21 Grunehogna Peaks shown on a) the glacial geomorphological map (present) and b) the paleoglacio-
logical reconstruction map (past). The transparent layer of paleo ice surface is drawn above the layer showing present bedrock exposure. The paleo ice direction is shown by a large blue arrow in (b).
Grunehogna Peaks appear covered by ice almost everywhere in the reconstruction. . . . 48
3
22 Borga Mountain shown on a) the glacial geomorphological map (present) and b) the paleoglaciological reconstruction map of Borga mountain as inferred from the mapping (past). The transparent layer showing paleo ice surface is drawn over the present topography derived from TanDEM-X. The paleo ice direction is shown by large blue arrows in (b). Borga Mountain appears covered by ice almost everywhere in the reconstruction. . . . 49 23 Kirwanveggen on a) the glacial geomorphological map (present) and b) the paleoglaciological recon-
struction map (past). The paleo ice direction is shown by a large blue arrow in (b). Kirwanveggen appears completely covered by ice in the reconstruction. . . . 50
List of Tables
1 Landform identification criteria on satellite images. . . . 11 2 Table summarizing the properties of cosmogenic nuclides, the atoms from which they are formed,
and in which minerals such atoms occur. Derived from Ivy-Ochs & Kober (2007). . . . 18 3 Datasets used for the geomorphological mapping . . . . 19 4 Identified sampling sites. . . . 38 A1 Panchromatic WorldView imagery selected after dataset processing. Columns refer to the Commercial
Satellite Imagery Naming Conventions of the Polar Geospatial Center. The acquisition time stamp indicates the day and the hour of image acquisition by the acronym yyyymmddhhmmss. The original name indicates the original time stamp, the image type (P: panchromatic, M: multispectral), and the Digital Globe product type (1b: standard, 2a: rectified). . . . 58 A2 The multispectral WorldView images selected after the dataset processing. Columns refer to the
Commercial Satellite Imagery Naming Conventions of the Polar Geospatial Center. The acquisition time stamp indicates the day and the hour of image acquisition by the acronym yyyymmddhhmmss.
The original name indicates the original time stamp, the image type (P: panchromatic, M: multi- spectral), and the Digital Globe product type (1b: standard, 2a: rectified). . . . 62
4
1 Introduction
This project is part of an international collaboration called MAGIC-DML (“Mapping, Measuring and Modeling Antarctic Geomorphology and Ice Change, in Dronning Maud Land”) which aims to build a palaeoglaciological reconstruction of western Dronning Maud Land (DML) in East Antarctica. Glacial reconstructions such as this are based on glacial geological and geomorphological traces (landforms, deposits) and are used to test and constrain numerical models of ice sheet extent and dynamics. MAGIC-DML focuses on the timing and pattern of ice surface elevation changes since the mid-Pliocene across the study area, where pre-existing field data is sparse.
This master thesis project contributes to the paleoglacial research with remote-sensing-based geomorphological mapping of a key area for MAGIC-DML investigations: the coastal-inland transect Ahlmannryggen-Borgmassivet- Kirwanveggen. The new high-resolution WorldView dataset was used to enable the study of this remote region in detail and provide route-planning information for the upcoming MAGIC-DML field season(s).
The primary aim of this study is to map traces of a thicker ice sheet on nunatak slopes that were formerly partly or entirely covered during ice surface highstands and to present plausible paleoglaciological interpretations and reconstructions based on them. Several steps are conducted within the framework of this project:
• acquire, select and mosaic the satellite data;
• map and interpret the geomorphology of the transect including Borgmassivet, Ahlmannryggen, and Kirwan- veggen in DML;
• identify appropriate locations for the collection of rock samples for cosmogenic nuclide (CN) dating;
• reconstruct the paleoglaciology of the study area and estimate maximum past ice elevations;
• evaluate the WorldView imagery for paleoglaciological reconstruction and route planning.
2 Study area
2.1 Physiography
2.1.1 Nunataks
A nunatak is an exposed mountain summit emerging from an ice field (Bharatdwaj, 2006), such as mountain summit towering above the ice sheet surface in Antarctica. The study area covers a transect of the western sector of Dronning Maud Land (DML) that stretches from the ice shelf in the north (4◦22’W 70◦55’S; 1◦29’W 70◦54’S) to the escarpment which forms the edge of the polar plateau to the south. (5◦16’W 74◦6’S; 2◦15’W 74◦3’S) (Fig. 1).
The transect has an approximate N-S orientation and is 350 km long by 130 km wide giving a total area of 40600 km2.
Kirwanveggen lies 350 km inland from the coast (Fig-s 2, 3). It is a part of an escarpment running from SW to NE.
The nunataks there reach elevations from 2100 to 2500 m above sea level (a.s.l.). These mountains serve as a barrier to the ice flow and form a steep step in the ice elevation. The area to the south of Kirwanveggen escarpment is the polar plateau, Amundsenisen (2800 m a.s.l.; Chang et al., 2016; Fig. 1) and the area to the north is a low-lying coastal area, Ritcherflya (2000 m a.s.l.; Chang et al., 2016; Fig. 1).
The Borgmassivet is a nunatak range in the center of the study area covering ca 3000 km2(Fig-s 2, 3).The average altitude of the nunatak summits is 2500 m a.s.l., in places towering ca 800 m above the ice surface.
Ahlmannryggen is a ridge to the north of Borgmassivet (Fig. 2, 3) which is almost completely covered by ice. It is situated 200 km from the margin of the ice sheet and is comprised of nunataks with elevations of up to 1843 m a.s.l.
(Neethling, 1969). The present-day ice surface lies approximately 400 m below the nunataks. Sanae IV, a South African Antarctic research base, is located on the Vesleskarvet nunatak (2◦50’W 71◦40’S), in the north-western part of Ahlmannryggen.
2.1.2 Ice streams
The majority of ice in the study area is drained by the SE branch of Jutulstraumen (Fig-s 1, 2) which is a deep-lying and warm-based ice stream (Høydal, 1996). Jutulstraumen (660 km in length, a drainage area of 120.000 km2, 12.5 km3 a−1 in yearly discharge, and an average ice velocity of 1 km a−1; Herzfeld, 2012) borders Borgmassivet and Ahlmannryggen along its western side. It drains the ice from the Amundsenisen polar plateau into the Fimbul ice
5
Figure 1: An overview bedrock topography map of Dronning Maud Land. The study area is shown by the red polygon. The Borgmassivet range and the escarpment are clearly visible as areas of high elevation (brown) inside the red polygon. Jutulstraumen appears as a deep ice stream on the eastern side of the study area. The ice streams and mountain ranges are labelled on Fig. 2. The continental shelf is clearly visible as light blue shallow area compared to darker blue surrounding deep area of the ocean. The bedrock topography is derived from Fretwell et al. (2013). Generated with Antarctic Mapping Toolbox (Greene et al., 2017), see Appendix 8.2.1.
shelf (Fig. 2). The Penck Trough sits between the Kirwanveggen escarpment and the Borgmassivet (Fig-s 2, 3).
The ice stream that occupies the Penck Trough drains less ice (Rignot et al., 2011) and merges with Jutulstraumen.
Another tributary glacier is Viddalen that drains the ice from Borgmassivet and Ahlmannryggen into Jutulstraumen (Fig. 2). Schyttbreen delineates the study area to the west and it is flowing into the Jelbart ice shelf (Fig. 2).
2.1.3 Geological setting
The geology influences the rate and pattern of erosion in the region (Sugden, 1978) and is the key to the MAGIC- DML sampling strategy. Therefore it is important to describe it in the scope of this study. The Jutulstraumen and Pench trough are lying above the boundary between two different terrains: the Grunehogna and Maudheim provinces (Groenewald et al., 1995). The Grunehogna province comprises the Archean Basement and is lying north-
6
Figure 2: An overview map of the study area. The study area is shown by the red polygon. The red dotted line indicates the transect shown on Fig. 3. The shaded relief image is derived from MODIS Mosaic of Antarctica (Haran et al., 2014). Ice speed is derived from Rignot et al. (2017) and is displayed in colors explained in the colorbar. The colors on colorbar may slightly differ from the colors on the map since the ice speed map was drawn as a transparent layer above the shaded relief image. Darker colors indicate higher ice speed. Blue arrows indicate ice flow direction and magnitude (Rignot et al., 2017). The blue line indicates the grounding line derived from Fretwell et al. (2013). Generated with Antarctic Mapping Toolbox (Greene et al., 2017), see Appendix 8.2.2.
west of these troughs and includes Borgmassivet and Ahlmannryggen. The Borgmassivet and Ahlmannryggen are composed of 3000 myr old granitic rocks (Chang et al., 2016) which are overlain by the Ritscherflya supergroup, a sequence of sedimentary deposits (shallow marine, tidal flat, braided stream and alluvial fan). These are cut by the Borgmassivet intrusion (mafic sills of ca 1107 M yrs old; Grosch et al., 2007), and capped by the Straumsnutane lavas of similar age (Moyes et al., 1995). Kirwanveggen belongs to the Maudheim province which comprises the Proterozoic basement: Precambrian gneisses (ca 1100 M yrs old), the Lower Palaeozoic quartzites known as Urfjell Group, the sandstones of the Permian-Triassic Amelang Plateau Formation, and the Jurassic tholeiitic basalts (Chang et al., 2016; Kleinschmidt et al., 2000; Harris et al., 1989).
7
Figure 3: The transect of the study area showing the bedrock topography in black, ice cover in light blue and the ocean in dark blue. The figure shows some topographical features of the study area: how Kirwanveggen dams the ice flowing from the inland polar plateau; that the Penck Trough lies below sea level; Borgmassivet reaches the altitudes similar to Kirwanvegen escarpment; Ahlmannryggen is slightly lower and is separated from Borgmassivet by a valley. The data is derived from Fretwell et al. (2013). The transect extent is indicated on Fig. 2. Generated with Antarctic Mapping Toolbox (Greene et al., 2017)
2.2 Morpho-tectonic and glacial history
2.2.1 Palaeozoic: sedimentation on Gondwanaland
Before the break-up of the Gondwana supercontinent a large planation surface (sediments that were probably deposited in Early Permian) was distributed across western DML (Näslund, 2001). Now this surface forms some nunatak summits (e.g. in Kirwanveggen).
2.2.2 Mesozoic (late Jurassic): break-up of the Gondwanaland
During the rifting of East Antarctica and South Africa flood basalts covered the Permian sediments (Elliot, 1992).
A new passive continental margin was uplifted, and an escarpment formed at the edge of the highly elevated plateau, separating it from the low-lying seaward land (Naslund, 2001). Kirwanveggen is one of the examples of the current location of this escarpment. The escarpment was retreating because of intensified denudation processes, including predominantly weathering and fluvial erosion (Jacobs et al., 1995). The Ahlmanryggen ridge containing the nunataks north of Borgmassivet (Fig. 2), according to Näslund (2001), is considered a residual erosional feature of the escarpment retreat (an inselberg area). Intense tectonism, related to the continental break-up, produced troughs, which may be a part of rift system (Groenewald et al., 1991). The Jutulstraumen ice stream is currently occupying one of them, the Penck-Jutul Trough (Fig-s 1, 2; Näslund, 2001).
2.2.3 Cenozoic (from Eocene until middle Miocene): onset of warm-based glaciation
After initiation of Cenozoic glacial conditions at the late Eocene pre-existing tectonic and fluvial morphology of western DML was locally eroded by warm-based glaciers (Holmlund & Naslund, 1994). There is an evidence of dendritic fluvial systems existing at that time (Jamieson et al., 2005), which points to cirque and valley glaciers occupying the highlands rather than ice sheet style glaciation.
An alpine landscape (Sugden, 1978) was developed in western DML. The steep cirques mapped in this study area are examples of glacial erosion of this period. The high topography of DML makes it a possible location for ice sheet inception (Jamieson et al., 2010). The Eocene-Oligocene boundary (around 34 Ma) is accepted to be the time of initiation of Antarctic wet-based ice sheet growth, as evidenced from ice-rafted dropstones in marine sediments (Näslund, 2001). The ice sheet growth was caused primarily by declining atmospheric CO2concentrations (DeConto
& Pollard, 2003), opening of the Southern Ocean circulation (Kennett, 1977), cold summers due to a combination of orbital parameters (Coxall et al., 2005) and a 4oC temperature drop (Liu et al., 2009).
Until ca 14 Ma the Antarctic Ice Sheet was fluctuating between warm-based ice sheet and mountain glaciation
8
phases. The former is evidenced by meltwater deposits offshore (Marchant et al., 1993). Since the onset of the Cenozoic, the DML mountains were consistently ice-covered.
2.2.4 Cenozoic (from Middle Miocene until Pliocene): onset of cold-based conditions
Cold-based ice sheet conditions were initiated following a 6-7oC drop in temperature ca 14 Ma (Shevenell et al., 2004), so the glacial erosion of the western DML mountains ceased. Meanwhile the ice streams occupying deep troughs like Penck-Jutul may have been at pressure-melting point at the base similar to the present state (Näslund, 2001). The ice sheet achieved its maximum extent reaching the edge of continental shelf (see the extent of the ice shelf on Fig. 1) by 14 Ma, and retreated to its present state by ca 13.6 Ma (Jamieson et al., 2010).
During the Pliocene Antarctica experienced the Middle Pliocene Climatic Warm Event (3.29-2.97 Ma) during which the ice level above the escarpment mountains was at least 200 m higher than today (Liu et al., 2010). The warmer climate led to an increased precipitation transport from the Southern Ocean, leading to an ice sheet thickening over the polar plateau (Altmaier et al., 2010).
2.2.5 Quaternary: thinning of the ice sheet
In the early Pleistocene (ca 3 Ma) the ice surface level was at least 500 m higher than present in some places of the escarpment area in DML, as evidenced from10Be exposure data and rock weathering analysis of the Sør Rondane Mountains in DML (Suganuma et al., 2014). A decrease in precipitation was inferred as result from steady global cooling since the end of the Pliocene (Fig. 4; Yamane et al., 2015). As a result since the onset of the Quaternary the ice sheet experienced significant thinning (Fig. 4; Suganuma et al., 2014).
Warmer climate
Higher sea surface temperature
Decreased sea ice
More evaporation
Increased snowfall Thicker Pliocene EIAS
Colder climate
Lower sea surface temperature
Increased sea ice
Less evaporation
Less snowfall EIAS thickness decrease in
the interior
Margins of EIAS grow
Figure 4: a) The EAIS in warmer climate during the Early-Mid Pliocene (earlier than 3 Ma). b) The EAIS in colder climate during the Mid-Late Pleistocene (later than 1 Ma). Dashed line represents the ice sheet level during the warmer climate. SST on the figures means sea surface temperature. Modified from Yamane et al. (2015)
2.2.6 Last Glacial Maximum
The Last Glacial Maximum (LGM) is defined as the last period when the ice sheets in both hemispheres reached their integrated maximum (Clark et al., 2009). The East Antarctic Ice Sheet (EAIS) thickened close to the coast because sea level was lowered, which enabled the grounding line to advance further out on the continental shelf (Hättestrand
9
& Johansen, 2005). The ice sheet elevation at the position of the present grounding line was supposedly 600-1200 m higher during the LGM (Lintinen, 1996). The ice sheet reached the outer edge of the shelf around 21 ky BP (Lintinen, 1996; Mackintosh et al., 2013; the extent of the ice shelf can be seen on Fig. 1). The central part of the ice sheet thinned because of the decrease in precipitation (Siegert, 2003) related to cold atmospheric temperatures.
In Ahlmannryggen, glacial deposits are found at elevation up to 1160 m a.s.l. (Neethling, 1969) indicating that it was, at one point, ice-covered at that elevation, but the age of glacial deposition remains unknown.
2.2.7 Post-LGM warming
The EAIS retreat is calculated to have started ca 14 Ka and was completed by ca 7 Ka (Mackintosh et al., 2011). It is thought to be the result of sea level rise and a warming of the ocean waters around the ice sheet (Mackintosh et al., 2011). The culmination of ice sheet recession occurred after Meltwater Pulse 1a (12-6 Ka) (Mackintosh et al., 2014).
This accelerated retreat may be attributed to an instability of the system comprised of ice stream dammed by ice shelf caused by melting close to the grounding line and under the ice shelf. A minimum age of deglaciation of 7652- 8149 Ka in Ahlmannryggen is provided by radiocarbon dating of mumiyo (waxy organic material found in petrel breeding colonies; Steele & Hiller, 1997). It is still unconstrained whether the summits of Ahlmannryggen were overridden by ice during the LGM since this kind of dating provides only the minimal deglaciation age (Mackintosh et al., 2014). Presently the ice sheet in western DML is stable or even growing according to the observations of lichen growing on nunataks close to the ice surface (Lintinen, 1996).
3 Background
3.1 Geomorphological significance of landforms on nunataks and the ice sheet surface
The aim of the geomorphological study is to identify landforms which yield information regarding the history and dynamics of glaciation in the study area. The nunatak landscape is characterized by glacially shaped rocks and sediments surrounded by various ice features. Below the paleoglacial and glacial significance of the mapped landforms is discussed. Additional definitions and morphological descriptions of the landforms are provided in Table 1.
10
Table 1: Landform identification criteria on satellite images.
Landform Morphology Texture/color Optimal
dataset
Scale and features used
for mapping
Possible identification errors
Crevasses
Ice cracks 10’s - 1000’s m long and ranging in width from meters up to 100 m. They cut across the ice flow with different orientations.
Transverse crevasses either occur at the edges of nunatak plateaus, within ice falls or on ice bumps. Splaying crevasses occur close to ice
streams or on prominent ice bumps, intersecting transverse crevasses.
Blue-white colour of the crevass walls; shadows inside the
fracture.
Multispectral WV images in standard false colours.
From 1:2,500 to 1:6,000;
polylines.
When filled by snow, possible confusion with
snowdrift features.
Closed crevasses can be difficult to distinguish
from open crevasses.
Bergschrund
Particular type of transverse crevasse. It occurs at the head of cirques and separates
shallow ice from deep ice. They are a few meters wide and up to several hundred
meters long.
Thin long shadow extending parallel and close to the head of
ice sheet cirques.
Multispectral WV images
in false colours or
panchro- matic WV
images.
From 1:2,500 to 1:3,500;
polyline.
Possible confusion with avalanche crown
fractures.
Ice fall
Irregular broken ice surface which occurs where ice flows over very steep slopes, with a
difference in altitude of hundreds of meters.
The towers of ice forming the ice fall are named séracs. They are separated by
crevasses up to twenty meters wide.
Same characteristics as crevasses, but occurring in a
more irregular pattern. Ice blocks occurring in close proximity help to identify the
ice fall.
Multispectral WV images
in false colours.
From 1:6,000 to 1:8,000;
polygon.
Blue ice area (BIA)
Delimited bare-ice region, up to several square kilometers in some areas, located on
nunatak plateaus or on the lee side of nunataks or on the ice sheet surface in
relation to bumps.
Characteristic homogeneous blue colour.
Multispectral WV images in standard false colours.
From 1:3,000 to 1:8,000;
polygon.
Possible confusion with light blue ice which appears blue because of its hummocky topography
but it is not an ablation centre. Possible confusion
of small blue ice areas with refrozen ice patches.
Longitudinal Surface Structures
(LSSs)
Parallel and unbroken lineations on the ice surface extending continuously up to hundreds of km in length, with an elevation of 1-2 m and a spacing of few km. They occur on ice streams, outlet glaciers and ice shelves.
Subtle ice ridges, close and parallel to each other. Linear,
curved or sinuous.
LIMA images
From 1:100,000 to
1:400,000;
polylines.
Possible confusion with hummocks on ice surface
due to the uneven bed topography.
Snow avalanche
run-out
Conical accumulation of snow, occurring below steep nunatak slopes and ice falls.
Powder snow and ice blocks form conical accumulation zones
which emerge on the ice.
Multispectral or panchro-
matic WV images.
Around 1:4,000;
polygon.
Ice block talus
Talus of broken ice pieces, ranging in size from a few tens of centimeters to around 20
m wide, generally lying under ice falls.
White coloured blocks, outlined by shadows on the surrounding
ice surface.
Multispectral and panchro- matic WV
imagery.
Around 1:3,500;
polygon, points.
Possible confusion with boulders.
Refrozen ice
Sub-circular patches of refrozen ice, extending up to few thousand square meters
in size, often occurring inside supraglacial moraines. Supraglacial lakes were also included in this category for simplicity.
Blue sub-circular patches possibly with lighter small
patches in the middle.
Multispectral WV images
in false colours.
Around 1:1,1000;
polygon.
Possible confusion with BIA.
Supraglacial meltwater
channels
Erosional channel features occurring on the ice surface, in some cases cutting supraglacial
moraines, up to 50 m width, and several km long.
Branching network of channels, light-coloured comparing to the sediments and dark compared to
the ice.
Multispectral or panchro-
matic WV imagery.
Around 1:4,000;
polyline.
Supraglacial moraine limits.
Supraglacial moraine
Supraglacial sediment cover with ridge or apron morphology, up to a few square kilometers in extent. Supraglacial moraine
occurs close to nunatak slopes or on local glaciers surrounding nunataks, often associated with BIAs. Ridges, thermokarst depressions, refrozen ice, and boulders can be
observed .
Area of dark colour, more or less homogeneous depending on the sediment thickness. Possible
presence of internal blue or white spots, due to snow cover
or refrozen ice.
Multispectral or panchro-
matic WV images.
From 1:1,000 to 1:6,000;
polygon.
Possible confusion with bedrock outcrop where the debris cover colour is very dark due to a dense
sediment texture.
Moraine ridges
Linear/curvilinear ridges of sediment, up to 5 km long, that are part of a supraglacial moraine veneer or a till veneer on nunatak
slopes. If they occur in a group, they typically run parallel to each other.
Subtle ridges of dark colour, more or less sharp depending on
the distribution and amount of sediments.
Multispectral or panchro-
matic WV images
From 1:1,000 to 1:6,000;
polylines.
Possible confusion with bedrock ridges where the debris cover colour is very
dark due to a dense sediment texture.
Wind scoop
Wind blown ice/snow depression up to 2 km long that commonly appears close to
obstacles like nunataks.
Well-defined depression that has the same texture and colour as the surrounding snow/ice. Wind
scoops often contain BIAs situated at the base of their lee
side.
Multispectral and panchro- matic WV
imagery.
From 1:3,000 up to 1:6,000;
polylines.
Possible confusion with ice cliffs or with bedrock
ridges covered by snow/ice.
12
Sastrugi
Sastruga is a feature created by katabatic wind erosion and deposition. It has elongated form and can reach up to 2 meters
in height (Mather, 1962).
Sastrugi appear as small parallel features on the ice field. Their
colour is consistent with the colour of ice while their shape is
outlined by shadows. The direction of sastrugi should be
more or less consistent in the area of interest.
Panchromatic WV imagery.
From 1:1,000 up to 1:8,000;
polylines.
Possible confusion with other small-scale wind-blown features
produced by winds blowing in different
direction.
Sediment cover (till or
regolith)
Sediments of various composition covering the bedrock on nunatak plateaus and slopes and sometimes continuing on to the adjacent
ice. Sediment veneer can form small patches between bedrock outcrops or cover almost an entire slope with an extent of up to 0.5 km2.
More or less homogeneous dark colour sediments. Often
’blanket’- like pattern. It can be distinguished from bedrock because it has a darker colour,
can contain boulders and descend down to the ice with an
uneven edge. In some areas where the sediments and the
bedrock appear of the same colour, sediment cover is identified as regolith that was
formed on the same rock.
Panchromatic WV imagery
in order to view with the higher resolution needed to identify
single boulders, or
multispec- tral WV modified band combination,
to distinguish
between bedrock (redder) and
sediments (blacker).
From 1:2,000 to 1:5,000;
polygons.
Possible confusion with bare bedrock when bare
bedrock has a colour similar to the sediments.
Nearly impossible to distinguish between till
and regolith in the satellite imagery. This confusion can be resolved
by using aerial imagery, ground truthing in the field or by looking at the
presence of boulders which are probably erratics and therefore
indicate till. The presence of ridges could
also indicate till, since the ridges are the result
of ice reworking.
13
Patterned ground
Sediment of various compositions forming stripes or polygons roughly hexagonal in shape, of a diameter up to 15 m. They occur
on some nunatak plateaus and erratics can lay on top of them.
Dark area containing stripes or polygonal shapes marked by
thin snow boundaries.
Panchromatic WV images,
for identifi- cation of
single boulders.
From 1:800 to 1:1,000;
polygons.
Possible confusion with frost wedged bedrock
which could have symmetrical structure and stripes similar to the
one of patterned ground (given the geology we could be seeing polygonal
jointing; J. Newall, pers.
comm.).
Cirques
Half-open, semicircular shaped niches located on nunatak slopes, extending up to 5 square kilometers. Cirques are usually ice free at the
head and ice covered at the bottom. They often contain a bergschrund inside.
Bowl-shaped nunatak slopes delineated by clear ridges
circular in shape.
Multispectral or panchro-
matic WV images.
Around 1:6,000;
polyline tracing the cirque ridge.
Concave nunatak slopes.
Erratics
Large boulders up to 10 m in diameter, lying among other sediments on supraglacial moraines, sediment cover, bedrock surfaces,
and patterned ground.
“Three-dimensional” blocks clearly visible since their upper
side looks lighter than the
“two-dimensional” sediment cover underneath. Another side
of the block is outlined by a shadow.
Panchromatic WV images.
Around 1:500;
points.
Regolith derived from the slope processes or bedrock outcrops emerging from the
sediment cover.
14
3.1.1 Ice and snow features
Ice features yield information on the present-day ice sheet dynamics, i.e. its ice flow conditions, subglacial topogra- phy and surface mass balance. Snow features are shaped by wind, and they, therefore, provide information about dominating wind directions in the area.
Crevasses are ice cracks up to hundreds of meters wide and up to thousands of meters long. Crevass orientation and spacing can inform about ice flow characteristics. Transverse crevasses open perpendicular to ice flow direction, under longitudinally extending flow. Splaying crevasses form parallel or obliquely to ice flow and where longitudinal strain rate is compressional or near zero and shear strain dominates (Harper et al., 1998).
Bergschrund is a type of transverse crevasse which separates shallow and deep ice at the head of a cirque or by a steep rock wall. The presence of a bergschrund indicates an increase of ice thickness, deformation and sliding velocity of ice below the crack comparing with ice above it (Mair and Kuhn, 1994).
Ice falls are irregular broken ice surfaces that occur where the ice bed is particularly steep. The presence of an ice fall indicates a pronounced acceleration of the ice flow due to a break in slope at the glacier bed (Smiraglia and Diolaiuti, 2011).
Blue ice areas (BIAs) are defined as being bare-ice regions up to ten square kilometers, characterized by a negative surface mass balance. BIAs are a conspicuous feature of the study area. The spatial coverage of BIAs is considerable in comparison to the rest of the continent (1 percent of Antarctic continent surface; Bintanja, 1999).
The presence of a BIA can give an indication of glacier mass balance, wind direckltion, ice flow characteristics and ice age stratigraphy. They occur where topography blocks snowdrift transport, induces wind scouring due to wind acceleration, or both. Therefore they tend to occur on nunatak plateaus, on the lee side of nunataks relating to a wind direction, or behind undulations in the ice sheet surface. Across these areas, sublimation and wind scouring exceed snow accumulation. The presence of a BIAs influences ice flow; under steady state conditions, ice flows horizontally towards and vertically upward to the BIAs to compensate for surface ablation. A consequence of the upward movement of the ice is a surface ice stratigraphy in the ablation zone, with the oldest ice outcroping closest to the nunatak (Bintanja, 1999; Fogwill et al., 2012).
Longitudinal surface structures (LSSs) are parallel curvy lineations on the ice surface, extending continuously for up to hundreds of kilometres in length, with an elevation of 1-2 m and a spacing of 1-5 kilometres (Glasser et al., 2014). LSSs occur on ice streams, outlet glaciers, and ice shelves and are assumed to form parallel to ice flow direction. They help to understand the conditions at the ice-bed interface and ice flow characteristics. Indeed, they are supposed to be the result of the transmission of uneven basal morphology to the ice surface (Ely and Clark, 2016). Alternatively, they indicate the presence of laterally compressive and longitudinally extensional ice flow, occurring in fast flowing areas or at the convergence of ice tributaries (Ely and Clark, 2016).
Wind scoops are wind-blown snow depressions up to 2 km long that commonly appear close to obstacles. They provide information about dominating wind directions in the area, since they usually form around nunataks due to the channeling of the wind.
Sastruga is a snow ridge up to 2 m in height created by katabatic wind erosion and deposition. Sastrugi are oriented parallel to the wind direction. Therefore they can be used to infer katabatic wind direction (Mather, 1962).
Refrozen ice patches usually occur inside thermokarst depressions of supraglacial moraines. They give indication of above zero surface temperature conditions, probably due to the low albedo of some sediment patches. Supraglacial lakes serve as an accumulation area of meltwater percolating from the adjacent areas (e.g. Lintinen. 1996).
Supraglacial meltwater channels are formed by water moving on the surface of the ice. Running water could be the result of ice melting due to the low albedo of a supraglacial sediment cover.
Snow avalanches indicate instability of the slopes from where they were sourced from, usually due to the steepness and to the presence of ice falls. Snow avalanches often result from cornice collapse events.
15
3.1.2 Sedimentary and bedrock features
Bedrock landforms and sediment cover of the nunataks and on the ice surface often allow an inference of past and present ice flow conditions. These are crucial landforms used for paleoglaciological reconstruction and important for determining the timing and rate of ice surface lowering, using different dating techniques.
Sediment cover or sediment veneers on nunataks slopes and plateaus are blankets of sediment of varying size covering the bedrock and in places, continuing to the adjacent ice. If the sediment cover can be classified as having a glacial origin (e.g. till), the spatial distribution of the the sediment across the nunatak slopes can be used to reconstruct the minimum elevation that the former ice surface reached (Hättestrand and Johansen, 2005). If the till contains stable boulders, that show no signs of having moved or rolled over, CN dating can be used to infer the timing of the last ice sheet thinning (Fogwill et al., 2012).
Patterned ground occurs on some plateaus where the sediment cover is organized into stripes and/or polygons.
The common interpretation for patterned ground in ice-free areas of Antarctica is that it displays the action of periglacial processes on unconsolidated sediments (Sletten et al., 2003). These processes likely occur in till rather than regolith (formed from in situ weathered bedrock) because subglacial erosion is more likely to produce finer material than subaerial weathering. Thermal contraction, cracking of ground surface, filling of the cracks by fine- debris wedges and soil motion due to debris wedge accretion are the main processes producing polygons. Depending on the phase in patterned ground development, the polygons can have different shapes: mature patterned ground polygons are usually more equidimensional and equiangular and with straighter boundaries than the younger ones (Sletten et al., 2003). This can be useful for the paleoglaciological interpretation. Another alternative interpretation of patterned ground is that it forms from till lying on buried ice which contracts and sublimates (Marchant et al., 2002). Using satellite imagery, it is impossible to distinguish between these two kinds of patterned grounds.
Regardless of formation means, patterned ground needs a long time to form, from 103 to 106 years (Sletten et al., 2003). This can provide a minimum exposure age of the patterned ground areas. However, patterned ground could occur on preglacial surfaces much older than 106years, which has been preserved under cold-based non-erosive ice for more than one glacial cycle (Stroeven and Kleman, 1999). CN dating could discriminate this condition. The CN dating could be performed on patterned ground boulders because, even if sediments were moved by frost heave processes, it is unlikely that these big boulders overturned.
Supraglacial moraine is a supraglacial sediment cover with ridge or apron morphology, up to a few square kilometers in size. Supraglacial moraines occur close to nunatak slopes or on local glaciers around nunataks, often associated with BIAs. Depending on the type of supraglacial moraine, they yield information on both present and past ice flow characteristics. Some of them consist of material accumulated from the base of the ice sheet and carried to the surface of the glacier without removal by lateral flow. In this case sediment is carried upward by compressive ice flow caused by a massive obstruction as ice moves toward an ablation centre or by differences in velocity between contiguous ice flows (a zone of fast flow moving into a zone of stagnant ice). The result is the formation of shear moraines and, in some cases, of an adjacent apron structure. These supraglacial moraines indicate the existence of a compressive ice flow and its direction. The time of exposure of the sediments indicates for how long the process of sediment concentration has been continuing (Chinn, 1991; Fogwill et al., 2012). Other supraglacial moraines are made up of talus debris coming from surrounding nunataks slopes and reworked by ice (Hättestrand and Johansen, 2005). When either kind of supraglacial debris covers continue as till veneer on to surrounding nunatak slopes, they give an indication on former ice surface elevations (Hättestrand and Johansen, 2005; Fogwill et al., 2012).
The shape and location of supraglacial moraine fields can also be used to infer local paleoglaciology, since their deformation by local glaciers can be due to an increase in accumulation (Chinn, 1991; Hättestrand and Johansen, 2005).
Moraine ridges are observed as a component of both supraglacial moraines and till veneers on nunatak slopes.
Moraine ridges on the ice are generally shear moraines originated through sediment upward movement due to ice compressive flow. They are usually ice cored since the ablation ceases under a sufficiently thick debris cover. When ice thickens and thins it deposits these moraine ridges on nunatak slopes, giving rise to moraine ridges. CN dating of boulders or clasts located on the surface of moraine ridges, can contribute to the understanding of the timing of ice thickness fluctuations (Fogwill et al., 2012).
16
Glacial erratics provide some of the clearest evidence for past ice level changes because of their position on the tops and slopes of the nunataks above the present ice sheet (Sugden et al., 2005). Erratic boulders have different lithologogies to the underlying bedrock or sediment, and indicate transport and deposition by ice (Fabel et al., 2002).
Through CN dating of erratics, it is possible to date the surface exposure, and so, time since the ice retreated from the nunatak slopes or plateaus (Altmaier et al., 2010). When erratics occur on supraglacial moraines, exposure ages indicate the timing of ice retreat from a specific ice margine position (Altmaier et al., 2010).
Cirques Glacial cirques are half-open, semi-circular shaped niches formed in nunatak slopes. Cirques are the results of erosion by wet-based local alpine glaciers (Fu and Harbor, 2011). Therefore, they were likely not shaped by the present cold-based ice sheet, but were probably forrmed following the Early Cenozoic Antarctic glacial inception (Stroeven and Kleman, 1999), and then preserved under cold-based ice (Näslund, 2001).
3.2 Cosmogenic nuclides and their use for ice thinning reconstruction
One of the goals of this thesis is to contribute to Magic-DML field seasons during which samples for CN dating will be collected in order to determine the timing of past changes in ice surface levels. CNs form in the top few meters of rock exposed at the Earths surface by interactions between cosmic rays and the mineral structure of the rocks or sediment (Table 2; Lal, 1991). As soon as the top layer is eroded away, the cosmogenic "clock" is set to zero, and nuclide accumulation starts from the beginning. CN concentrations in a rock is a function of the time since the rock was exposed to incoming cosmic rays at the Earth’s surface. Therefore, the abundance of CNs within a rock gives information about its exposure history and thus the timing of deglaciation (Phillips et al., 2016).
CN dating is particularly suitable for the Antarctic environment, due to several reasons (Davies et al., 2012):
• shielding by vegetation cover of the bedrock and boulders is unlikely;
• the snowcover burial of the sampling spot is unlikely on nunatak slopes and plateaus due to high winds, especially if the boulder is large;
• there is almost no material for radiocaron dating in the nunatak area.
CN dating is useful in reconstructing the rates of ice thinning, by providing the exposure ages of glacially eroded bedrock and moraines (Gosse and Phillips, 2001). Exposure ages of samples from the top of a nunatak plateau provide the minimum time constraints for when the ice thinned and left the plateau exposed for the first time (Altmaier et al., 2010). In the ice sheet environment it is possile to reconstruct the rate of ice thinning by dating the bedrock and erratics at different elevations. The exposure ages should, in theory, be younger at lower elevations and older at higher elevations above the present day ice surface. This gradient provides the information about the rate of ice thinning (Suganuma et al., 2014). Dating boulders from a supraglacial moraine yields information as to how long sediment concentration has been taking place on the ice (Chinn, 1991).
There are a number of standard assumptions often used while performing CN dating. The most useful glacial landforms for dating are those that are least likely to contain inherited CNs from previous exposure events (leading to age overestimation), or ones that have not been affected by post-depositional processes, such as, burial, exhumation and erosion leading to age underestimation. Quartz-bearing erratics on stable slopes are considered to be good targets for CN dating since they commonly contain less inheritance than bare bedrock (Bentley et al., 2006). The latter has commonly not been sufficiently eroded to reset the cosmogenic clock, because of minimal or absent erosion by cold-based ice. However, cold-based ice may also preserve versatile features like boulder fields and patterned ground (Stroeven and Kleman, 1999).
For the reasons described above a strategy involving of sampling multiple CN analyse is used to determine complex exposure/burial histories and give indications about the thermal regime of the ice sheet in the past. For this purpose the bedrock is dated together with an erratic lying on it. Then if the exposure ages provided by both samples are similar, the ice sheet regime that led to formation of this bedrock-erratic pair is assumed to have been warm based, allowing for the ice sheet to erode enough bedrock to remove pre-existing nuclides. If the erratic provides a significantly younger exposure age than the bedrock, it suggests the thermal regime was cold-based during the last glaciation, and the bedrock was not eroded to a sufficient depth to remove any prior CN inventories.
From the list of the nuclides used by MAGIC-DML (Table 2) it can be seen that the use of various nuclides can provide exposure ages of millions of years before present. Cosmogenic carbon dating can provide the data with tens of thousands of years resolution, and can therefore show more recent ice surface changes while aluminium and berillium isotopes can yield exposure ages of several million years.
17
