Introduction

This paper uses radio-echo sounding (RES) data to provide new, three-dimensional information on internal ice-sheet structure for the Dome Concordia (Dome C) region of central East Antarctica (Fig. 1). An extensive analogue RES dataset of the Antarctic ice sheets, collected by the Scott Polar Research Institute (SPRI)–U.S. National Science Foundation (NSF)–Technical University of Denmark (TUD) consortium between 1968 and 1979, is held in archive at SPRI, University of Cambridge, U.K. (Reference DrewryDrewry, 1983). Approximately 50% of the area of the ice sheets is covered by RES flight tracks, with a focus on the western part of East Antarctica. A pervasive feature of the data is the occurrence of strong reflections from within the ice. Several individual internal reflectors are discrete and continuous over distances on the order of 100 km. Electromagnetic (e/m) reflections of this type occur in ice at boundaries of dielectric contrast. The causes of the reflections are understood to be ice-density changes (at depths of < 1000 m) and acid ice resulting from volcanic aerosol deposition. At depths of > 1000 m, radio-echo layering is caused principally by acid labelling of discrete stratigraphic horizons within the ice sheet (Reference MillarMillar, 1981; Reference Siegert, Hodgkins and DowdeswellSiegert and others, 1998b).

Fig. 1. (a) Location of Dome C. (b) ERS-1 radar altimeter-derived surface elevation in the study area; contours in m a.s.l. (source: Reference Siegert and RidleySiegert and Ridley, 1998).

The Dome C region (Fig. 1) is of particular interest because of its potential significance for palaeoenvironmental studies. The 2700 m ice core taken from Vostok station, central East Antarctica, has provided benchmark evidence of palaeoclimate for the last glacial–interglacial cycle (Reference LoriusLorius and others, 1985; Reference JouzelJouzel and others, 1989, Reference Jouzel1993). A 3300 m ice core from Dome C (about 550 km from Vostok station) is to be obtained through the European Project for Ice Coring in Antarctica (EPICA) (Reference Jouzel, Orombelli and LoriusJouzel and others, 1996). This is likely to yield similarly important palaeoclimate data. Reference Siegert, Hodgkins and DowdeswellSiegert and others (1998a) traced five prominent internal layers, identified from RES data, for >500 km across the East Antarctic ice sheet, between Vostok and the new ice-core site at Dome C. This correlated the stratigraphy and depth–age relationship between the two sites. The present paper enhances understanding of the glaciological context of the future Dome C ice core by mapping the three-dimensional, internal ice-sheet structure in the region.

A 9600 km² area around the Dome C ice-core site has been mapped using 60 MHz radio-echo sounding (Reference Tabacco, Passerini, Corbelli and GormanTabacco and others, 1998). In this survey, the surface and subglacial topography was established. Although internal layering was observed in the RES data, no three-dimensional representation of internal ice-sheet structure was developed. The present paper, therefore, represents the first attempt to map RES layering in three dimensions within central East Antarctica, and to identify internal ice-sheet architecture accordingly.

Acquiring Internal RES Layer Data

Analogue RES data were recorded in two different formats during SPRI–NSF–TUD Antarctic operations: as plots of two-way travel time vs signal strength from single e/m pulses (“A-scope” data; Fig. 2), or as a continuous series of e/m returns where signal strength is identified by brightness rather than an axis variable (“Z-scope” data; Fig. 2). A-scope data were recorded discretely at 15 s intervals; the distance interval varied with the speed of the aircraft, and is referred to as a “CBD unit”. Quantitative information about the reflectivity of individual layers is acquired by determining power reflection coefficients (PRCs), which account for the geometric spreading, refraction and dielectric absorption of radio waves in ice, and the two-way antenna gain of the RES apparatus (Reference Robin, Drewry and MeldrumRobin and others, 1977; Reference ParenParen, 1981; Reference Siegert, Hodgkins and DowdeswellSiegert and others, 1998b). When plotted as two-way e/m wave travel time against real time, continuously recorded Z-scope data yield ice-sheet pseudo-cross-sections.

Fig. 2. Sample analogue RES data: a time-continuous “Z-scope” pseudo-ice-sheet cross-section, and corresponding single-pulse “A-scope” returns; P_r/P_t is the amplitude of the received radio wave relative to the transmitted power. The prominent lowermost P_r/P_t peak represents bedrock, which in this example can be traced from left to right, decreasing in depth.

Three prominent internal layers were traced across four separate flight-lines in the Dome C region: lines 102, 104, 108 and 136 (Fig. 3a). Each individual flight-line has at least two, and up to four, crossing points along it, which have been used for cross-checking to ensure the correct layers are identified on the Z-scope data. The process of identifying and tracing layers is as follows: calculate PRC profiles from A-scope data at 20 CBD unit intervals along each of the four flight-lines; match three PRC maxima at five check locations (11 individual PRC profiles); use PRC profiles to trace the three PRC maxima along the full length of each flight-line at 20 CBD unit resolution (58 individual PRC profiles); use Z-scope data to trace layers corresponding to the PRC maxima over the full length of each flight-line at 1 CBD unit resolution (this step checks the results of the previous two steps); finally, the three RES internal layers (and the bedrock profile) can be marked on analogue Z-scope data consisting of photographic prints, and digitized. Locations and total ice depths at 1 CBD unit resolution are known from the original flight navigation data, so the actual depth of each layer can be calculated from its fractional depth.

Fig. 3. (a) Location of data points within the study area. There are 500 latitude/longitude (x, y) locations, each with the depths of three internal RES layers and the bedrock (z). Flight-line numbers are indicated. (b) 5 km grid raster derived from bedrock depths; lighter shading indicates greater elevation/shallower depth. There are 50 gridcells in the latitudinal direction and 45 cells in the longitudinal direction. A similar raster was generated for each of the three RES internal layers as well as the bedrock, as the basis for contour mapping (see Figs. 4 and 5).

Confidence in the identification of each RES internal layer is high, because of consistency checks from different data formats (A- and Z-scope), and direct tracking from Z-scope prints. Consideration of the digitizing resolution suggests that the absolute accuracy of the determined layer depths is of the order of 100 m; this constitutes only 2.4–4.4% (mean 2.9%) of the ice depth in the study area.

Constructing Internal RES Layer Maps

Some 930 data points, each matched to the depth of the three internal RES layers and of the bedrock, were acquired by the above procedure, in the area 73.0–76.3° S, 118–132° E. However, in order to achieve the densest coverage of data points centred around Dome C, the mapping exercise focuses on an area delimited by the latitude range 73.5–75.75° S (about 250 km), and longitude range 120–127° E (about 225 km); Dome C is located at 74.6667° S, 123.833 3° E, and is therefore very close to the centre of this approximately 56 000 km² area (Fig. 3a). Five hundred vector (x, y, z) data points are located within this area (Fig. 3a), where x corresponds to the latitude of the point, y to its longitude and z to the depth of either internal RES layer 1, 2 or 3 or the bedrock. It was decided to map layer depths rather than elevations because the absolute ice-surface elevations derived from the original flight navigation data are less reliable than the relative depths that may be determined from the RES data (and than recently acquired European Remote-sensing Satellite (ERS-1) surface elevation data). From these vector files, raster files were generated using an iterative minimum-curvature algorithm, in which a two-dimensional cubic spline function is used to fit a surface to the input depth values. Three grid resolutions were examined: 2 km (125 × 111 cells), 5 km (50 × 45 cells) and 10 km (25 × 23 cells). The vector data were not sufficiently densely spaced to justify a 2 km gridcell resolution, while the contours generated from the 10 km gridcell raster appeared unduly generalized. Contour maps were therefore generated from 5 km gridcell rasters (Fig. 3b). The raster was contoured by linear interpolation. Input raster values were pre-smoothed by a weighted average method to remove jagged anomalies and outliers in contouring. Figure 4 presents the final contoured depth maps for each of the three RES internal layers and the bedrock. Note that the necessary interpolation from irregularly spaced vector data suggests the reliability of the contour maps decreases towards the centres of “data-sparse” areas.

Fig. 4. (a) Depth of RES internal layer 1 (age 76 700 BP). (b) Depth of RES internal layer 2 (age 107200 BP). (c) Depth of RES internal layer 3 (age 144 300 BP), (d) Depth of bedrock. Contours in metres. Ages are from radio-echo layer correlation with the Vostok ice core (Reference Siegert, Hodgkins and DowdeswellSiegert and others, 1998a)

Discussion

Ice deformation

Valuable applications of DEMs of ice-sheet internal layers may be as initial boundary conditions in dynamic models of ice-sheet flow, or as final boundary conditions in inverse-modelling procedures. While the flow rate of an ice column decelerates quasi-exponentially towards bedrock, cumulative deformation is greatest near the bed, where ice residence time is greatest as a result of the low flow rates. Beneath a large ice sheet, patterns of deformation will be complicated by large-scale variations in bedrock topography, thermal regime, strain history of the ice column (which reaches a maximum thickness of >4000 m in the mapped area) and residence time: the age of ice at 2200 m depth at Dome C is >145 000 years (Reference Siegert, Hodgkins and DowdeswellSiegert and others, 1998a). Therefore, internal RES layers are an indicator of internal deformation, which can be resolved at a fine scale over large areas. A measure of the extent to which each internal RES layer has been deformed is given by the range in depth of the layer and the standard (rms) deviation of the depth of the layer within the mapped area. The data in Table 1 confirm that both of these measures increase with depth. Two ice-sheet cross-sections across the Dome C area are illustrated in Figure 5. These reveal a trend towards increasing resemblance to the bedrock profile with internal RES layer depth, which is consistent with the data in Table 1.

Fig. 5. Cross-sections showing ice surface, RES internal layers 1–3 and bedrock profiles in the study area along (a) flight-line 102 and (b) flight-line 136. Locations are given in Figure 3a. Note the separate scale for the ice-surface elevation on each profile.

Table 1. Descriptive statistics for internal RES layer and bed depths

Conclusions

Maps have been presented of three ice-sheet internal RES layers in a 56 000 km² region around Dome C, central East Antarctica. These maps can provide boundary conditions for three-dimensional ice-sheet models, taking advantage of information on internal ice-sheet structure and deformation, and allow ice-core data to be extrapolated over large areas (cf. Reference Siegert, Hodgkins and DowdeswellSiegert and others, 1998a). In particular, they enable comparisons to be made between Dome C and other deep ice-core sites. RES layers from beneath Vostok station indicate that ice layering is virtually horizontal at the ice-core site. This is because the ice core is located over an extremely flat subglacial lake. At Dome C, however, where the ice-core site is located over bedrock, the maps of RES layers presented here indicate that ice below 1000 m is dipping by 0.5–1.0°. It is concluded that ice 1000–2000 m below the surface is, or has been, subject to deformation. This finding will be important when ice from the core is examined, and these maps will provide the glaciological context with which to identify the deformation history of the ice core and the Dome C region in general.