Introduction
The mean annual temperature in the Antarctic Peninsula (AP) region has increased by 1.5°C to 3.0°C since the 1950s (e.g. Reference KingKing, 1994; Reference Vaughan, Marshall, Connolley, King and MulvaneyVaughan and others, 2001, Reference Vaughan2003; Reference TurnerTurner and others, 2005). This warming has coincided with the disintegration of ice shelves and the retreat of many land- terminating glaciers in the region (e.g. Reference Skvarca, Rack, Rott and DonangeloSkvarca and others, 1998;Reference Cook, Fox, Vaughan and FerrignoCook and others, 2005;Reference Davies, Carrivick, Glasser, Hambrey and SmellieDavies and others, 2011; Reference Glasser, Scambos, Bohlander, Truffer, Pettit and DaviesGlasser and others, 2011). Land-based glacial mass loss has been most prominent at the northern tip of the AP due to the pronounced glacier acceleration (Reference RignotRignot and others, 2008). Detailed analyses of satellite imagery and some isolated mass-balance measurements together suggest that small land-terminating glaciers are especially sensitive to climate change (Reference RauRau and others, 2004;Reference Skvarca, De Angelis and ErmolinSkvarca and others, 2004). The sensitivity of small land-terminating glaciers to climate is related to a short response time of variations in mass balance (Reference Knap, Oerlemans and CadeeKnap and others, 1996).
Along with direct mass-balance measurements, changes in surface area and volume of glaciers are important indicators of climate changes. Volumetric characteristics are necessary for improved understanding of: (1) the responses of these sensitive glacier systems to climate change;(2) the present contribution of AP glaciers to sea- level rise; and (3) future predictions of AP glacier behaviour. Although changes in glacier mass over long time periods can be analysed by volumetric methods, these do not replace the field-based techniques, which provide detailed measures of glacier change in space and time. Direct methods are timeconsuming and expensive, however, and typically only cover short time periods (Reference Gudmundsson and BauderGudmundsson and Bauder, 1999). Glacier bed topography and ice thickness are important for understanding ice dynamics (e.g. Reference Rippin, Vaughan and CorrRippin and others, 2011a) and for assessing basal pressure and thermal regimes (e.g. Reference Rippin, Vaughan and CorrRippin and others, 2011b). Determination of glacial mass changes may be used to detect shifts in climate and to predict glacial changes associated with future climate conditions (e.g. Reference PatersonPaterson, 1994;Reference Fountain, Lewis and DoranFountain and others, 1999).
In this study, we present the volumetric data for Davies Dome ice cap and Whisky Glacier on James Ross Island (JRI). The name Whisky Glacier is used here for IJR-45 glacier as given in the inventory by Reference Rabassa, Skvarca, Bertani and MazzoniRabassa and others (1982). The name Whisky Glacier was referred to by Reference Chinn and DillonChinn and Dillon (1987) and is given on the Czech Geological Survey (2009) map. It should not be confused with the tidewater Whisky Glacier located in Whisky Bay, as named by the British Antarctic Survey (2010). The changes in elevation, area and geometry between 1979 and 2006 were derived from digital elevation models (DEMs) and aerial photographs. Ice thickness was measured during a field campaign in January and February 2010 using ground- penetrating radar (GPR). Glacier volume and glacier bed elevations were then derived from these data.
Study Area
Approximately 81% of JRI is covered with 138 glaciers, which are concentrated in the central and southern parts of the island (Reference Rabassa, Skvarca, Bertani and MazzoniRabassa and others, 1982). The variety of existing glacier systems and fluctuations of glacier front positions on JRI have been presented by Reference Rabassa, Skvarca, Bertani and MazzoniRabassa and others (1982), Reference Skvarca, Rott and NaglerSkvarca and others (1995) and Reference RauRau and others (2004). mass-balance changes have been studied at Glaciar Bahfa del Diablo on Vega Island (Reference Skvarca, De Angelis and ErmolinSkvarca and others, 2004) and at Davies Dome and Whisky Glacier on JRI (Reference Nývlt, Kopackova, Laska and EngelNývlt and others, 2010; Reference Láska, Nývlt, Engel and BudfkLáska and others, 2012). However, these existing glaciological studies only describe elevation and surface area; data on ice thickness and hence on ice volume are hitherto unknown.
The investigated glaciers are located on the Ulu Peninsula, northern JRI (Fig. 1). This predominantly glacier-free area extends northwards from the central part of the island, which is dominated by the JRI ice field, an area of ice covering 587 km2 (Reference Rabassa, Skvarca, Bertani and MazzoniRabassa and others, 1982). The climate of the Ulu Peninsula is influenced by the regional-scale atmospheric circulation formed by the AP, which generally provides a barrier to westerly winds associated with cyclonic systems centred in the circumpolar trough (Reference King, Turner, Marshall, Connolley, Lachlan-Cope, Domack, Burnett, Leventer, Conley, Kirby and BindschadlerKing and others, 2003). High interdiurnal and seasonal variability of cloudiness, incoming solar radiation and near-surface air temperature can be found along the eastern coast of the AP (Reference MarshallMarshall, 2002; Reference Láska, Budík, Budikova and ProsekLáska and others, 2011) although the long-term trend appears to involve increased frequency of warm airflows advected over the AP as described by Reference Marshall, Orr and Van LipzigMarshall and others (2006).
Location of GPR transects (labelled as D-n and W-n, n being an integer) on Davies Dome and Whisky Glacier (dark grey areas) on the Ulu Peninsula, northern JRI. Other ice bodies are shown in medium grey. Triangles mark the position of meteorological stations.
Contemporary glaciation of JRI is characterized by ice caps, outlet glaciers and different types of alpine glaciers (Reference Rabassa, Skvarca, Bertani and MazzoniRabassa and others, 1982). The JRI ice field covers the southern part of the island where only small masses of rock protrude to the surface. By contrast, most ice has retreated from the northern part of the Ulu Peninsula. Small glaciers have persisted in higher locations with snow accumulation areas above ~460m a.s.l. (Reference Nývlt, Kopackova, Laska and EngelNývlt and others, 2010). The glacial features include an ice cap on Davies Dome (514 m) and on Lachman Crags (641 m), cirque, apron and piedmont glaciers around these plateaux, and valley glaciers carved into Smellie Peak (704 m) and Stickle Ridge (767m). These glaciers are relics of an ice cap that covered the whole island in the last glacial period (Reference Rabassa, Evenson, Schluchter and RabassaRabassa, 1983; Reference Ingolfsson, Hjort, Bjorck and SmithIngólfsson and others, 1992). In the northern part of JRI, this ice cap merged with the northern Prince Gustav Ice Stream, which drained the northeastern AP (Reference Rabassa, Evenson, Schluchter and RabassaRabassa, 1983). During the Holocene, the ice cap was linked to the Rohss Bay Ice Shelf, until it disintegrated in 1995 (Reference Glasser, Scambos, Bohlander, Truffer, Pettit and DaviesGlasser and others, 2011).
Two contrasting types of glacier from the Ulu Peninsula were selected to assess recent volume changes. Davies Dome (DD) is an ice dome (63°52'−63°54' S, 58°01−58°06'W) which originates on the surface of a flat volcanic mesa at >400 m a.s.l. and terminates as a single 700 m wide outlet in Whisky Bay. In 2006, DD had an area of ~6.5 km2 and lay in the altitude range 0-514 m a.s.l. (Table 1). Whisky Glacier (WG) is a cold-based land-terminating valley glacier (63°55'− 63°57'S, 57°56'−57°58'W), which is surrounded by an extensive area of debris-covered ice (Reference Chinn and DillonChinn and Dillon, 1987). In 2006, the glacier covered an area of ~2.4 km2 and ranged from 520 to 215 m a.s.l. Mean annual near-surface air temperatures of−7.8°C and −8.8°C have been recorded since 2009 by automatic weather stations (AWSs) on WG (356 m) and DD(514 m), respectively (Fig. 1). Summer temperatures (November-February) vary from −9°C to +10°C on WG and from −12°C to +7°C on DD. The snowmelt period generally starts at the beginning of November and lasts until the end of February. The meteorological data were collected as part of this study using AWSs on glaciers (Fig. 1).
Morphological characteristics of DD and WG for 2006
Data And Methods
DEMs and aerial photographs of DD and WG were used to extract topographic information for glacier extent and geometry. Two separate DEMs produced by the GEODIS company (Czech Republic) were stereoplotted from aerial photographs obtained by the British Antarctic Survey (BAS) in 1979 and 2006 (Czech Geological Survey, 2009). Ground-control points for the photogrammetric processing were obtained in the deglaciated landscape in 2008 using a dual-frequency differential GPS receiver. The 1979 aerial photos were registered with root-mean-square (rms) errors of 2.0 m in both the horizontal and vertical. Rms errors of the 2006 dataset are 0.7 m and 0.8 m in the horizontal and vertical direction, respectively. Reference MeixnerMeixner (2009) provides a more complete description of the DEM generation method.
The GPR survey was conducted in January and February 2010 in accordance with numerous descriptions of radar investigations of glacier thickness (e.g. Reference Plewes and HubbardPlewes and Hubbard, 2001;Reference Navarro, Eisen, Pellikka and ReeseNavarro and Eisen, 2010). GPR surveys were applied on DD (except over a steep and crevassed outlet glacier) and across the whole surface of WG. GPR data were collected along six radial profiles on DD (Fig. 1). On WG, GPR was carried out along the central longitudinal profile and five transverse profiles (Fig. 1). A longitudinal profiling was carried out from the glacier terminus to the upper reaches of the ablation zone (226-444ma.s.l.). Three transverse profiles were undertaken in the lower part of the snout delimited by lateral moraines and two profiles in the upper zone enclosed by the left lateral moraine and a valley slope facing to the west. GPR profiling was carried out using an unshielded 50 MHz Rough Terrain Antenna and RAMAC CU-II control unit (MALA GeoScience, 2005). The signal acquisition time was set to 2040 ns, and scan spacing ranged from 0.2 to 0.3 m. GPR position was given by GPS: a Garmin GPSMAP 60CSx receiver (Garmin International, 2009) was used to determine coordinates for the corresponding trace numbers with the horizontal rms position accuracy <4m. The GPR profiling equipment was man-hauled at a mean speed of ~0.5 m s-1.
GPR data were processed and interpreted using the REFLEXW software version 5.0 (Reference SandmeierSandmeier, 2008). The raw GPR profiles were filtered using an automatic gain function (AGC-Gain), background removal and static correction. The AGC-Gain was used to amplify deep signals reduced due to pulse dispersion and attenuation. Strong direct wave signal and noise reflections were eliminated using the background removal filter. Finally, the radargrams were corrected for topographic elevation. For this purpose, elevations from the collected GPS data were used. Depth was calculated assuming a wave velocity for cold glacier ice, 0.168 m ns-1 (Reference Narod and ClarkeNarod and Clarke, 1994). The velocity model used to convert two-way travel times to depths affects the accuracy of the ice thickness determination. As signal velocities in glacier ice range from 0.167 to 0.169 m ns-1 (Reference Kovacs, Gow and MoreyKovacs and others, 1995; Reference Pellikka and ReesPellikka and Rees, 2009), the velocity used was accurate to within 0.002 mns-1. Thus, the estimated uncertainty of conversion of travel time to depth is ~1%. Moreover, the accuracy in depth measurement is restricted by the ability to determine the first arrival of the signal (Reference JolJol, 2009). This uncertainty was approximated by a quarter of the signal's wavelength in ice, which corresponds to 0.84 m for the 50 MHz antenna used.
Ice-thickness values converted from two-way travel times along GPR profiles were used to calculate glacier volumes. Because the accuracy of the calculation depends on the algorithm applied we investigated and selected the most appropriate interpolation first. We applied five techniques (inverse distance weighing, kriging, minimum curvature, spatial neighbour and triangulation) to the complete ice- thickness datasets to create grids of glaciers. We then removed the ice-thickness information along one complete section from each of the two datasets. We selected the section D-1, which represents both the upper part of an ice dome on a flat volcanic plateau and the lower termination on mountain slopes. For WG, we chose the section W-4 as it crosses the middle part of the snout. The reduced datasets and the same interpolation techniques were used to generate a new series of grids from which ice-thickness values were assigned for verification points along sections D-1 and W-4. The interpolated values were compared with the measured values and the difference was expressed using standard deviation values. Results of interpolation techniques were also evaluated based on glacier volumes calculated from a complete and reduced dataset. A comparison of the calculated values shows that the minimum curvature technique most accurately represents the ice-thickness data (Table 2). This method was selected for further processing of the data.
Results of glacier volume calculations
The volume of glaciers was calculated using a combination of 2006 surface elevation data and GPR-based ice- thickness data taken in 2010. The elevation of GPS-derived locations along GPR profiles was taken from the 2006 DEM because the accuracy of an altitude measurement was low during the GPR survey. The vertical error of GPS-based coordinates (~8-18 m) is larger than the vertical error of the DEM and elevation changes of glacier surfaces between 2006 and 2010. According to mass-balance measurements on DD and WG the elevation differences ranged from -0.4 to +1.1 m during this period (Reference Láska, Nývlt, Engel and BudfkNývlt and others, 2012). The volume of glaciers for 1979 was calculated based on the 2006 volumes and the difference between 1979 and 2006 surface DEMs. The uncertainty associated with the glacier volume determination was calculated as the product of the glacier surface area and vertical uncertainty. The vertical uncertainty for 2006 is related to the accuracy of ice thickness and to surface elevation changes in 2006-10. The uncertainty for 1979 was calculated as the sum of the glacier surface topography-related uncertainty for 1979 plus the uncertainty of the glacier volume for 2006. The mean ice thickness of DD and WG is 32.4 and 99.6 m, respectively, and average uncertainties of conversion of travel time to ice thickness are 0.3 and 1.0 m, respectively. The total uncertainty of the ice volume per glacier area in 2006 is 3.5 m3 m-2 for DD and 4.1 m3 m-2 for WG. Uncertainties of 5.5 and 6.1 m3 m-2 have been obtained for the 1979 volume of DD and WG, respectively.
Glacier bed topography maps were generated based on the 2006 DEM and the 2010 GpR measurements. The bed elevation was calculated for GPR survey locations by subtracting the ice thickness from the surface elevation. At the margin of DD, the ice thickness was set to zero except for the upper reaches for the outlet where three bounding points were placed. For these points, depth values were approximated from the nearest points of GPR measurements and from the position of a small rock outcrop that protrudes to the surface of the eastern part of the outlet. The bed elevation at the WG limit was determined from marginal measurement points at GPR cross-sections that represent the crest of lateral moraines. The minimum curvature interpolation, which generates a smoothed surface with low standard deviation, was then used to construct an 8 m gridcell DEM of glacier beds. From the DEMs, contour lines at 10 m spacing were constructed for bed topography.
Results
Glacier bed topography
The boundary between the glaciers and underlying bedrock and/or till is recorded as continuous basal reflection in all of our radargrams (Figs 2 and 3). A strong reflection at the glacier bed is clearly visible in DD profiles where the slope of the presumed bedrock is small (Fig. 2a and c). By contrast, the amplitude of reflections is halved in the area below the subglacial eastern flank of the plateau as shown in transect D-3 (Fig. 2b) and in the eastern sections of transects D-1, D-2 and D-6. Beneath WG, the strong and continuous reflection was recorded in the upper half of profile W-1 (Fig. 3a). There are only a few minor diffractions visible along the glacier bed in this section. The signal strength decreases slightly in the middle section of the longitudinal profile ~-1.8km from the glacier terminus. The decreased amplitude prevails in the lower part of the snout as seen along the lower part of the longitudinal profile and in crossprofiles W-4, W-5 and W-6 (Fig. 3c).
Selected GPR transects across DD. (a) Profile D-2 covers the dome glacier in the northwest–southeast direction showing the subglacial surface of the DD mesa and its transition to more undulated bed topography in the eastern part of DD. (b) Profile D-3 shows the lower part of the glacier below the eastern margin of the DD mesa. (c) Profile D-4 shows the upper flat-bottomed part of the glacier. Inverted triangles indicate points of crossover of other profiles.
Longitudinal transect W-1 (a) and cross-sectional transects W-3 (b) and W-4 (c) for WG. The position of the glacier bed along the western margin of WG indicates that glacier ice continues to the west (b, c). The ice thickness on the eastern side of profileW-4 suggests that the right lateral moraine delimiting the snout is also ice-cored (c). Inverted triangles indicate points of crossover of other profiles.
The subglacial topography of DD is relatively complex. The uppermost part of DD is located on a near-horizontal surface. GPR data indicate a slightly undulating glacier bed with an elevation range of <50 m (Fig. 2a and c). Most of the near-flat bed is inclined to the north, and only ~7% of this zone faces southeast or south. An undulating glacier bed extends to half the length of the northwest-southeast axis of DD. Here the reflection boundary descends from 400430 m a.s.l. by 70-90 m to more dissected bed topography (Fig. 2b). The fall line is oriented towards the southeast and represents an eastern flank of the volcanic plateau (Fig. 4a). Steeper slopes and concave wide channel-like depressions are indicated in the radargrams of two sections of the flank. The southernmost of these coincides with the main outlet glacier that drains DD into Whisky Bay. The second slope depression is carved into the central part of the bedrock step ~1 km from DD summit to the east.
(a) Glacier bed topography and (b) ice thickness at DD based on the 2006 DEM and GPR measurements. The investigated dome glacier is marked in dark grey and the outlet glacier is shown in medium grey.
WG bed interface has a simple geometry. The data obtained along the centre flowline show an even longitudinal profile without any subglacial steps or depressions (Fig. 3a). The bed topography is inclined to the north with low average slope angle of ~5°. The mean slope of the basal contact in the lower part of the profile is <2°. The transverse profiles W-2 and W-3 across the central line of WG show an asymmetric bottom of a trough with steeper slopes rising towards the east, where the glacier is constricted by a steep slope of Smellie Peak (Fig. 3b). The western part of the bed topography is characterized by a gentle slope angle, which is within 5° of the glacier surface. As a result, the glacier bed is located up to ~125 m below the western margin of the glacier surface (Fig. 5). The pattern of the bed topography suggests that ice extends as a debris-covered glacier towards the west. Similar conditions are recorded in transverse profiles W-4, W-5 and W-6, which were undertaken in the lower part of WG. These profiles suggest that the glacier ice also extends beyond the eastern margin of WG forming a core of a right lateral moraine (Fig. 3c). The minimum extent of debris-covered ice is ~0.5 km2. However, if the whole area of lateral and terminal moraines is taken into account then the maximum extent of ice-covered moraines is ~3 km2.
(a) Bed topography and (b) ice thickness at WG based on the 2006 DEM and GPR measurements. The investigated valley glacier is marked in dark grey, and the maximum possible extent of debriscored glacier area is shown by dots.
Surface area, elevation and volume change
The area of DD in 1979 was 8.33 ± 0.06 km2 and in 2006 was 6.49 ± 0.01km2. The dome area decreased from 6.23 ± 0.05 to 4.94 ± 0.01km2, losing 20.7% of its area (Table 3). Assuming a constant rate of change, the ice cap therefore lost 0.048±0.002 km2a-1. Almost 70% of measured areal loss occurred in the southeast periphery of the dome at 120-370 m a.s.l.
Changes of surface area, elevation and volume of DD and WG between 1979 and 2006
Over the same time period, WG lost 10.6% of its nondebris-covered area, decreasing from 2.69± 0.02 km2 to 2.40 ± 0.01 km2. The mean glacier loss rate was 0.011 ± 0.001 km2 a-1. The form of WG remained almost the same during this period. In 1979 its terminus descended to 210 m a.s.l. and terminated behind a complex of ice-cored terminal moraines. By 2006 the snout had retreated by ~60m, with a mean retreat rate of 2.2 m a-1 terminating at 215 m a.s.l.
DEM comparisons show surface elevation changes between 1979 and 2006 for each glacier. Surface lowering occurred over 95% of the surface of DD, which lowered on average by 8.5 ±2.8 m. Most of the loss occurred in the southern and western part of the dome (Fig. 6). The site that experienced the greatest reduction in surface elevation (~35 ± 3 m) was occupied on 25 January 2010 by a supraglacial lake (our observation during GPR survey). Surface lowering greater than 25 m occurred on the whole southern part of the dome, where DD glacier descends over rugged relief at lower elevations (300-330 m a.s.l.). The dome glacier surface decreased by ~20 ± 3 m around its western margin, which is located on Davies Mesa at 400460 m a.s.l. The upper parts of the northwestern slope of the dome above 450m a.s.l. experienced lesser changes in surface elevation (<10 m). The surface area located to the north from the highest part of the dome remained either unaltered or increased by <10 m.
Surface topography and elevation change at DD. Surface topography in (a) 1979 and (b) 2006 and (c) surface lowering between 1979 and 2006. The investigated dome glacier is marked in dark grey, the outlet glacier is shown in medium grey and the dotted line represents the 1979 ice outline (b). Isolines within the investigated glacier area in (c) represent surface change.
Surface elevations of WG decreased on average by 10.1 ± 2.8 m over the period 1979-2006. The whole glacier experienced lowering, with the greatest reduction in surface elevations (>25 m) in the upper snow accumulation area (Fig. 7). On average, surface elevations of WG decreased at a rate of 0.37 ± 0.11 m a-1. This value is slightly higher than the mean surface lowering for DD (0.32 ± 0.11 m a-1).
Surface topography and elevation change at WG. Surface topography in (a) 1979 and (b) 2006 and (c) surface lowering between 1979 and 2006. The investigated valley glacier is marked in dark grey, and the dotted line represents the 1979 ice outline (b). Isolines within the investigated glacier area in (c) represent surface change.
GPR data obtained between 25 January and 17 February 2010 reveal the ice thickness of the two glaciers (Table 1; Figs 4b and 5b). At DD, the maximum thickness of 83 ± 2 m was recorded on its highest part near the crossing of the D-2 and D-4 profiles. The second area of greater thickness was discovered in the eastern part of the dome (Fig. 4b). The glacier ice along the subglacial step is ~50−75 m thick.
The mean thickness of WG is greater than on DD, 99.6± 1.8 m compared with 32.4 ± 1.2 m. The greatest thickness was observed in the central part of this valley glacier, where the glacier ice was as much as 158 ± 2 m thick. The ice thickness of WG decreases towards the eastern glacier margin, which abuts a steep valley side. In contrast with the eastern margin, greater thickness values (63-125 m) were recorded along the western margin of WG (Fig. 5b). The basal reflection remains well below the lateral moraine surface, indicating that glacier ice extends beyond the current visible glacier limit.
The volume of DD decreased from 0.23 ± 0.03 km3 to 0.16 ± 0.02 km3 between 1979 and 2006. The glacier lost 30.4% of its volume at a mean annual rate of 0.003 ± 0.002 km3 a-1. For WG, the values of the total volume changed from 0.27 ± 0.02 km3 to 0.24 ± 0.01 km3. The volume loss was 10.6% and the mean annual change was 0.001 ± 0.001 km3 a-1 over the period 1979-2006. The volume change calculated for the dome was almost three times as high as for the valley glacier.
Interpretation and Discussion
Spatial variability in bed topography and surface lowering
The strong reflection observed in the GPR data from the flat- bottomed part of DD is interpreted as the transition from glacier ice to solid bedrock. A possible interpretation of weaker reflectors in the eastern flank of the plateau is that sediments rather than solid bedrock form the glacier bed in this area. A channel-like depression incised in the southeast subglacial margin of the plateau suggests that the bedrock has been deepened and widened by glacial erosion of an outlet glacier (Figs 2a and 4a). The incised terrain in the northeast margin of the plateau (Fig. 4a) indicates thatanother outlet glacier has probably at some time in the past drained the northeastern part of DD towards the Abernethy Flats. The timing of this outlet glacier advance is unresolved by our data, but it might have been synchronous with a mid- Holocene advance of WG (Reference Rabassa, Skvarca, Bertani and MazzoniRabassa and others, 1982) dated by Reference Hjort, Ingolfsson, Moller and LirioHjort and others (1997) at ~460014C BP. The interpolated flatter subglacial relief in the northern part of DD (Fig. 4a) leads us to suggest that DD has not expanded to inundate cirques on the northern margin of the DD mesa. The fact that the steep cirque headwalls have sharp upper edges is also evidence that these rock lips have not been subjected to basal glacier erosion. This contrasts with the southeastern edge of the DD mesa beneath the outlet flowing towards Whisky Bay.
The strong reflection beneath the upper reaches of WG and the straight long profile of the bed suggest that the bed there is solid rock. By contrast, in the lower part of the glacier the reflection from the assumed glacier bed is weak, tentatively indicating a transition from ice to subglacial till. Thick ice at the western periphery of WG and debris- covered ice in adjacent moraine ridges together suggest that WG is more extensive than the present bare-ice surface would otherwise suggest. A large ice thickness along the western margin of the snout supports the hypothesis that a broad zone of the lateral moraine is an internal debris- covered part of WG (Reference Chinn and DillonChinn and Dillon, 1987).
The spatial distribution of surface lowering indicates that the two glaciers are evolving differently. At DD, ice surface reduced mainly at lower altitudes, while the accumulation zone experienced little or slightly positive change. By contrast, the whole surface area of WG experienced lowering, with the greatest magnitude in the upper reaches of the glacier. The local differences in surface lowering probably result from higher air temperatures on WG and from different flow patterns and surface mass balances of the two glaciers. The assumption of the different flow rate is consistent with ice velocities observed at mass-balance measurement sites on the two glaciers. The surface velocities on WG were nearly twice as high as on DD during the period 2009-11 (Reference Láska, Nývlt, Engel and BudfkNývlt and others, 2012). As the ice flow in the flat-bottomed upper part of DD is limited, the increased melting in lower parts of the dome was not compensated by transport of ice mass between the upper and lower parts of the glacier. By contrast, WG is a well-delimited valley glacier with high flow velocities (Reference Láska, Nývlt, Engel and BudfkNývlt and others, 2012) and the spatial distribution of ice volume is controlled by enhanced transport of ice mass. The higher total volume change calculated for DD probably results from lesser thickness of its margins and from the position of the dome, which is more exposed to air masses than is WG.
Comparison to other studies in the AP region
Although the AP ice sheet, island ice caps and valley glaciers cover ~80% of the AP region (Reference Rau and BraunRau and Braun, 2002), volumetric characteristics are known only for a minority of glaciers (e.g. Reference Davies, Carrivick, Glasser, Hambrey and SmellieDavies and others, 2011). Apart from the AP ice sheet, data on ice thickness have been collected only on Alexander Island (Reference WagerWager, 1982), Adelaide Island (Reference RiveraRivera and others, 2005), Anvers Island (Reference Dewart and CraryDewart, 1971), the South Shetland Islands (e.g. Reference MacheretMacheret and others, 2009; Reference NavarroNavarro and others, 2009), JRI (Reference Rabassa, Skvarca, Bertani and MazzoniRabassa and others, 1982) and Dolleman Island (Reference Peel, Mulvaney and DavisonPeel and others, 1988). At JRI, until recently the ice thickness was known only for Mount Haddington ice cap. According to radio-echo sounding, the ice dome is estimated to be ~300m thick (Reference AristarainAristarain, 1980). This is a typical value for island ice caps along the AP as shown in Table 4. On DD the maximum ice thickness of ~83 m is about three times thinner than Mount Haddington, which is not unexpected given the smaller surface area of the dome compared with the extent of the JRI ice cap (587 km2). The ice thickness of WG is similar to the thickness of Spartan Glacier at Alexander Island, which is the only valley glacier in the AP region with known thickness (Reference WagerWager, 1982).
Comparison of the maximum ice thickness of island glaciers along the Antarctic Peninsula
Areal changes observed on DD and WG confirm that the glacier retreat on JRI has increased in recent decades. According to Reference RauRau and others (2004), ice masses on JRI lost 3.9% in area during the period 1975-2002 and, of these, land-terminating glaciers were subject to the smallest retreat rates. By contrast, we find that the DD glacier area decreased by 22.1% and WG by 10.6% over the period 1979-2006. These values are in agreement with retreat rates of DD reported by Reference Davies, Carrivick, Glasser, Hambrey and SmellieDavies and others (2011) for the period 1988-2009. The difference in retreat rate of the two glaciers may be attributed to the faster downwasting of DD. The tidewater front of the DD outlet in Whisky Bay shows more prominent changes than the grounded glacier tongue of WG. During the last 30years the surface area of DD and WG decreased at a mean rate of 0.8% a-1 and 0.4% a-1, respectively. These values are higher than mean annual areal changes of ice caps on Livingston Island (Reference Calvet, Garcla-Selles and CorberaCalvet and others, 1999;Reference Molina, Navarro, Calver, Garcla-Selles and LapazaranMolina and others, 2007) and King George Island (Simoes and others, 1999) over the period 1956-2000 (Table 5). By contrast, the mean annual area decrease of DD and WG is an order of magnitude less than the mean annual area loss of cirque glaciers on King George Island (Simoes and others, 2004). Overall, the increase in glacier retreat observed on JRI is similar to the accelerated area loss reported for glaciers on King George Island in the period 2000-08 (Reference Rückamp, Braun, Suckro and BlindowRuckamp and others, 2011).
Mean annual area changes of island glaciers along the northern part of the Antarctic Peninsula
The mean elevations of DD and WG decreased by 0.3 ± 0.1 and 0.4 ± 0.1m a-1, respectively. These values correspond well with the mean surface lowering of 0.32 m w.e. a-1 observed over the period 1988-97 at Rothera Point on the southwestern AP (Reference Smith, Vaughan, Doake and JohnsonSmith and others, 1998) and with the mean lowering rate of 0.2 ma-1 reported from the summit of Bellingshausen Dome on King George Island (Reference Rückamp, Braun, Suckro and BlindowRuckamp and others, 2011). By contrast, our estimates are lower than values reported from Vega Island, Livingston Island and the eastern part of King George Island. According to recent observations on Vega Island (Reference Skvarca, De Angelis, Domack, Burnett, Leventer, Conley, Kirby and BindschadlerSkvarca and De Angelis, 2003;Reference Skvarca, De Angelis and ErmolinSkvarca and others, 2004), mean annual surface lowering of Glaciar Bahfa del Diablo ranged from to 1.6 m a-1 over the period 1985-2003. Reference Pritchard, Arthern, Vaughan and EdwardsPritchard and others (2009) and Reference Rückamp, Braun, Suckro and BlindowRuckamp and others (2011) reported a similar value of ~1.5 m a-1 on the South Shetland Islands for the periods 2003-07 and 2000-08, respectively. Reference Ximenis, Calvet, Enrique, Corbera, de Gamboa and FurdadaXimenis and others (1999) obtained even higher lowering rates (1-5 m a-1) for glaciers on Livingston Island during the period 1995-97. Comparison of the 1979 and 2006 DEMs revealed a mean annual volume loss of 1.1%a-1 at DD and 0.4% a-1 at WG. The value for WG corresponds with mean annual glacier volume loss (0.2-0.3%a-1) reported for the South Shetland Islands for the period 1956-2000 (Reference Molina, Navarro, Calver, Garcla-Selles and LapazaranMolina and others, 2007). The value obtained at DD is three to four times higher than the mean annual volume loss of Hurd (0.3% a-1) and Johnsons Glaciers (0.2%a-1) according to volume change described by Reference Molina, Navarro, Calver, Garcla-Selles and LapazaranMolina and others (2007).
A comparison of our spatial data with earlier glacio- logical observations suggests that DD and WG were subject to greater retreat, surface lowering and ice volume loss than other island ice caps in the AP region. If the recent rate of volume loss continues, DD and WG could disappear within 62 ± 52 and 227 ± 220years, respectively. This approximated timing corresponds well with the value based on mean rate of areal loss for WG (228±22 years to extinction), but the volumetric change of DD is more rapid than its areal change (104 ± 5 years). However, we assume that area rather than volume has the largest control on the future development of both glaciers. The estimated melting time of WG based on both areal and volumetric changes is consistent with the estimated future evolution of Bellingshausen Dome on King George Island, which is predicted to completely disappear after 285 years if the present warming trend persists (Reference Rückamp, Braun, Suckro and BlindowRuckamp and others, 2011). A total disappearance of cirque glaciers will probably occur much faster, as was shown by Reference Barboza, de Bortoli, Simoes, da Cunha and BraunSimoes and others (2004) for Ferguson and Flagstaff Glaciers on King George Island.
Summary and Concluding Remarks
GPR data acquired in 2010 indicate that the subglacial topography of an ice dome and of a valley glacier in the northern part of JRI differ significantly. A near-flat plateau forms the glacier bed of the upper accumulation area of DD, whereas a steep plateau edge and dissected topography underlie the lower elevated eastern part of the dome. In contrast, almost uniform glacier bed topography exists beneath WG. The maximum ice thicknesses as measured by GPR at DD and WG are 83 ± 2 and 158 ± 2 m, respectively. The average surface elevation of the dome decreased by 8.5 ± 2.8 m between 1979 and 2006, whereas the valley glacier experienced average surface lowering of 10.1 ± 2.8 m. DD lost 22.1% of its area between 1979 and 2006, decreasing from 8.33 ± 0.06 km2 to 6.49 ± 0.01 km2. A dome area of the glacier decreased from 6.23 ± 0.05 km2 to 4.94 ±0.01 km2 (-20.7%) and its volume changed from 0.26 ± 0.03 km3 to 0.16 ± 0.02 km3 (-30.4%). The surface area of WG decreased from 2.69 ±0.02 km2 to 2.40± 0.01 km2 (-10.6%), resulting in the glacier mass loss from 0.27 ± 0.02 km3 to 0.24 ± 0.01 km3 (-10.6%). Between 1979 and 2006, glacial area loss ranged from an average of 0.048 ± 0.002 km2 a-1 (DD) to 0.011 ± 0.001 km2 a-1 (WG). Average volume losses of DD and WG were 0.003 ± 0.001 and 0.001 ± 0.001 km3 a-1, respectively. The observed changes are higher than areal and volumetric losses reported from the AP region between 62° and 68° S. If the recent trend in air temperature continues, DD and WG may disappear in 104 ± 5 and 228±22 years, respectively. However, model projections reported by Reference Carril, Menendez and NavarraCarril and others (2005) show that an increase in temperature trends is expected over the 21st century.
Acknowledgements
This work was funded by the Czech Science Foundation (project No. 205/09/1876) and by the Ministry of Environments of the Czech Republic (VaV SP II 1a9/23/07). Topographic data were provided by the Czech Geological Survey. The study was undertaken in January and February 2009 during a fieldwork stay at the Johann Gregor Mendel Base. We thank P. Váczi, J. Pobo?il and P. Pálka for field and logistic support. Constructive comments and language editing by Jonathan L. Carrivick were highly appreciated. Helpful comments and suggestions from the scientific editor, David M. Rippin, and two anonymous reviewers are also acknowledged.
