2Ottawa-Carleton Geoscience Centre

Department of Earth Sciences

Carleton University

1125 Colonel By Drive,

Ottawa, ON K1S 5B6

CANADA

Key words: permafrost, ground ice, glaciology, Bylot Island, ground-penetrating radar

In the proglacial environment on Bylot Island there are many occurrences of buried ice. Ground-penetrating radar and remote sensing techniques were combined with standard field observations to examine the processes and settings associated with the burial of surface ice, and the potential for preservation of buried ice. Of the different types of surface ice present (i.e. glacier, icing, permanent snow bank), glaciers were found to have the greatest potential for becoming buried, through sediment concentration at the surface during the melt-out of sediment-laden ice. A number of lateral and end moraines were found to have cores of glacial ice over 10 m thick.

Deltaic sedimentation was also found to be effective at preserving buried glacial ice; however, the occurrence of this depositional setting is infrequent. The burial of icing and permanent snow banks was found not to occur to any great extent due to their dynamic and generally erosional settings.

Ground ice on Bylot Island is readily preserved due to the cold ground temperatures. However, in few locations recent fluvial activity had exposed a massive ice body, resulting in the initiation of retrogressive thaw flows. Evidence of past thermokarst activity is widespread, but no evidence of current activity was discovered.

Tabular bodies of massive ice are commonly encountered beneath the ground throughout permafrost regions. While a large body of literature is devoted to the identification and study of ice within the ground, very little is known about subsurface ice within the predominantly coarse-grained surficial materials of the eastern Arctic. Because Bylot Island has spectacular examples of massive subsurface ice contained within electrically resistive terrain, there is an excellent opportunity to use satellite imagery, surface observations, and subsurface geophysical techniques in combination to investigate relationships between the ice and it’s enclosing sediments. Massive subsurface ice bodies can range from less than 1 m to over 50 m in thickness, and from a few square metres to several square kilometres in areal extent. Most occurrences have been discovered within a few metres of the ground surface (e.g. Harry et al., 1988; Moorman et al., 1998; Mackay and Dallimore, 1992; Robinson et al., 1996). However, much deeper bodies have been discovered in deep exploration wells (Mackay, 1971; Dallimore, pers. comm., 1993).

Unlike pingos and ice wedges, the origins of tabular massive subsurface-ice bodies are still a matter of some controversy. The two main processes thought to be responsible for their development are the burial of surficial ice bodies (e.g. glaciers) and the in situ segregation of ice as a result of frost heaving and water injection. From theoretical modelling of the segregation process, it has been suggested that it is possible to develop large tabular bodies of massive ground ice (Taber, 1929; Gorelik and Kolunin, 1993), but laboratory experiments or field observations of the grow processes have yet to verify this.

The purpose of this research project was to investigate occurrence subsurface ice in a periglacial environment and address the potential for its preservation in a high arctic environment such as Bylot Island. In this paper, the processes associated with the burial of glacial, icing, and permanent snow bank ice are discussed, and specific examples are given from three different environments (ice-cored moraines, ice-cored deltas, and icings). The impacts of thermokarst and retrogressive thaw flow erosion on the preservation of ground ice are also addressed.

The complex environment at the base of a glacier makes investigation of basal sediment accretion processes very difficult. However, mechanical and thermal processes have been proposed as the dominant causes of sediment accretion at the base of polar glaciers. Mechanical incorporation of sediment (e.g. shearing), was first suggested as a process capable of introducing appreciable amounts of sediment into the base of glaciers (Goldthwait, 1951). This process has been used to explain several occurrences of thick dirty basal ice sequences in cold conditions (Souchez, 1967; Tison et al., 1993).

Thermal processes such as regelation and basal freezing, originally suggested by Weertman (1961), are now thought to also be a prevalent method of sediment incorporation into glaciers (Goodwin, 1993; Gow et al., 1979; Hubbard and Sharp, 1989; Souchez, 1971; Souchez and Lorrain, 1978). The processes of melting and refreezing, causing the accretion of ice and sediment onto the bottom of a glacier, are most active in situations where the glacier undergoes both warm and cold-based conditions along its travel path. This explains why polythermal glaciers have much thicker basal ice sequences than either temperate glaciers or entirely cold-based glaciers (Boulton, 1970).

Although there is considerable debate regarding the exact nature of both mechanical and thermal basal sediment incorporation processes, in many circumstances (e.g. polythermal glaciers) both sets of processes are likely to act together where basal obstructions create a compressive flow regime (Chinn and Dillon, 1987; Hubbard and Sharp, 1989).

It has been suggested that some of the glaciers on southern Bylot Island are polythermal (Zdanowicz, 1994). As the glaciers flow onto the plains it appears that permafrost grows beneath the margins causing the formation of distinctive structural and sedimentological features within the basal ice sequences. Up ice, the glaciers are thought to have more variable thermal conditions with basal freezing occurring in some locations and sliding occurring in other locations.

Landforms resulting from the burial of glacial ice are readily observable in many proglacial and previously glaciated settings (e.g. kettle lakes). However, the precise processes of ice burial, or the potential for preservation of ground ice has not been thoroughly examined. Research in glacial sedimentology has primarily focused on sedimentary processes related to glacial tectonics (Chinn and Dillon, 1987; Tison et al., 1993) or processes occurring at the sole of glaciers (Goodwin, 1993; Sharp et al., 1994). Several researchers have interpreted the burial of glacial ice in a lacustrine environment (e.g. Ashley et. al., 1985; Shaw, 1977). However, this work has generally been limited t othe examination of sedimentary structures after the ice has melted. Research conducted on the development of supraglacial debris layers resulting from the ablation of ice has focused on the sediment concentration processes, but not on the ice preservation after stagnation (Benn and Evans, 1998; Boulton, 1967; Nakawo, 1979; Sharp, 1949; Shaw, 1988).

Bylot Island offers an ideal setting to examine subsurface ice bodies at various stages of development and preservation, due to its cold climatic conditions.

Bylot Island, located in the eastern Canadian Arctic, (Figure 1) has a central mountainous core of the island consisting of a range of Archean-Proterozoic metamorphic and igneous rock that is covered by a 4500 km² ice field. Glaciers flow from the upland accumulation areas in the centre of the island through deep valleys towards the ocean or onto the plains. In contrast to the mountainous regions, the plains are composed of poorly consolidated sandstone to shale of Cretaceous-Tertiary age. The glaciers become lobate and carve wide valleys as they flow onto the plains. A veneer of unconsolidated glacial drift covers the glacial valleys. Glacial drift provenance evidence indicates a number of periods of local glacial advance have been interspersed with the advance of glaciers originating beyond Bylot Island (Klassen, 1993). The oldest and most extensive native and foreign glaciations, Bylot and Baffin glaciations, respectively, covered most of the island including the study sites. More recent foreign advances, the Eclipse glaciation, and the subsequent native Aktineq glacial response did not affect the study sites as these events were limited to lower elevations. As a result, it is believed that the study sites have not been influenced by foreign glaciations for over 43,000 years BP (Klassen, 1993).

Bylot Island is an arctic desert with a mean annual air temperature of approximately -15 ûC, and an average annual precipitation in the order of 200 mm (AES, 1980). Climate values are approximate due to the distance to the closest weather station in Pond Inlet, and the lack of long term data from that station. The island is well within the zone of continuous permafrost, and the presence of little thermokarst activity indicates that the permafrost is generally stable. The maximum active-layer thickness measured on the island from three years of frost probing and measurement of active-layer detachments was 80 cm. At a monitoring site in the southern lowlands, the temperatures at a depth of 10 cm below the ground surface range from +3ûC to —21ûC with the mean annual ground temperature being roughly —9.6ûC (Figure 2).

Satellite, airborne and surface-based remote sensing techniques were used to image surface and subsurface ice. Data from Landsat Multi-Spectral Sensor (MSS), Thematic Mapper (TM), Radarsat synthetic aperture radar (SAR), airborne SAR, and aerial photography were used to image currently retreating glaciers, seasonal and perennial icings, water flow within a snow pack, and areas potentially underlain by massive subsurface ice. Aerial photographs were also used to create detailed maps of the spatial extent of surface ice bodies such as glaciers and icings and to measure their change through time.

The buried ice study sites were located in areas of unvegetated moraine or where active retrogressive thaw flows were observed. Ground penetrating radar (GPR) was used to measure ice thickness and image the internal structure of surface and subsurface ice and enclosing sediments. The depth to the top of buried ice units and nature of the overlying sediments were also determined with GPR. In addition, GPR was used to map lake bathymetry and the lacustrine deposits within proglacial thermokarst terrain (Moorman and Michel, 1997).

A pulseEKKO IV GPR system was used with 50 MHz, 100 MHz, or 200 MHz antennas employed depending on the site conditions and depth of penetration and resolution required. Surveys were conducted as single transects or in grids with the trace spacing being 2 m or less, enabling interpretation of detailed subsurface features. By acquiring closely spaced gridded GPR data three dimensional estimates can be made of the size and geometry of subsurface features (e.g. ice bodies), using traditional seismic processing techniques as described by Moorman (1998).

As GPR only measures the amplitude and travel time of reflected radar signals, the propagation velocity of the signal must be determined in order to calculate the depth from which reflections originated. This was accomplished by performing velocity surveys to measure the propagation velocities for conversion of travel time to depth values (Moorman, 1998). Interpretation of the GPR profiles was facilitated and verified with the aid of data from boreholes and ice exposures. Boreholes were drilled through the icing with a powered CRREL corer to depths up to 12 m. Boreholes were located at representative sites where the GPR data contained significant features. Where ice was exposed at the vertical sides of the glaciers, in ice caves, stream cut-banks, and active-layer detachments, excavations were undertaken enabling cryostratigraphic logging.

Stratigraphic descriptions of ice exposures and cores were undertaken in the field and ice samples were taken back to the cold laboratory at Carleton University for thin section analysis. Field and lab descriptions included ice contacts and structural features, ice crystal size, shape and orientation, bubble content, size and distribution, and sediment content, structure and distribution.

A wide variety of surface ice types occur on Bylot Island, including permanent snow banks, incipient glaciers, and large valley and piedmont glaciers (Figure 3). Ten of the 18 largest glaciers shown in Figure 3 are currently retreating, and four of these have marginal moraines with surficial characteristics that indicate the moraine is ice-cored.

An example of the variation in rates of glacial retreat on Bylot Island is shown in Figure 4. The photo map was constructed by digitally combining vertical aerial photographs taken in 1958 and 1982. The glacial extents for 1948 and 1994 were then added manually from oblique aerial photographs. As indicated on the map, Glacier B28 (informally named Stagnation Glacier) has retreated about 1.8 km in the last 46 years, while during the same time period, adjacent Glacier B26 (informally named Fountain Glacier) has experienced no appreciable retreat. Lateral and end moraines are almost completely lacking around Fountain Glacier, while the light-coloured unvegetated areas surrounding Stagnation Glacier (Figure 4) indicate retreat of the glacier. The lateral moraines flanking Stagnation Glacier are up to 200 m high with profiles that change daily due to melting of their ice core.

Within the study area eight glaciers have been identified that regularly develop icings at their termini during the winter. Only one icing appears to persist perennially through the melt season when fluvial activity and sedimentation occurs, and is thus a possible candidate for burial. A portion of the icing at the front of Fountain Glacier (Figure 4) has been present in all aerial and satellite images on record since 1948. The up-valley extent of the icing can range from touching the glacier (as observed in the summers of 1991-1999), to 200 m down-valley of the terminus as shown in Figure 4. In 1991 an artesian spring flowed up through the icing in a fountain at the icing blister shown in Figure 5. This fountain also appears on aerial photographs taken in other years when the icing did not cover that area. Since the summer of 1992 an icing blister has been observed at the site of the fountain, but high-pressure water flow has not been observed at the surface of the icing.

Numerous perennial snow banks and incipient glaciers occur within the study area, generally in stream valleys and other depressions on steep north-facing slopes. The seasonal weather patterns determine how many of these features survive the summer. Although these surface features have the potential to become buried, the probability is very low. As the perennial snow banks are located in erosional valley-side environments, sediment that is deposited on the snow or ice surface is normally transported further down slope soon after deposition. As well, the minimal ice flow in incipient glaciers, and thus the lack of basal erosional power, limits the incorporation of sediment into the ice mass from below. As a result there is minimal chance for perennial snow banks or incipient glaciers to develop into buried ice bodies in this high-relief setting.

A good example of the burial of glacial ice, as a result of the incorporation of basal sediments, occurs at Stagnation Glacier. With the recent retreat of this glacier, large ice-cored lateral and end moraines have developed (Figure 4). Although this glacier is entirely constrained in a mountain valley, the contribution of supraglacial debris is relatively small and limited to a narrow margin along the sides of the glacier. Thus, the thickness of the sediment cover and the resulting thermal stability of the buried ice is dependant on the quantity of basal sediments that are brought to the surface.

Figure 6 shows a cross section of Stagnation Glacier and its relationship to its lateral moraines. The GPR profile displays how a core of glacial ice continues beneath the lateral stream, into the lateral moraine. Ice within the moraine is over 10 m thick in some locations. The slope of the lateral moraines range from 20û to 30û, resulting in a surface cover that is susceptible to regular active-layer detachments which expose the ice core. Active-layer detachments appear to result in only a small amount of ice melt and rapidly heal within a few days. Ground-penetrating radar measurements revealed that along the profile over the east lateral moraine shown in Figure 6, the debris covering the ice core did not exceed 2 m in thickness.

The ice core within the west lateral moraine is separated from the main glacial ice body as a result of a more active lateral stream which eroded down through the glacial ice to its base. This has not occurred between the glacier and the east lateral moraine. The connection of the ice core within east lateral moraine to the glacier is at least partially the result of a supraglacial debris cover that extends beyond the lateral stream, protecting the ice beneath.

A series of GPR profiles in front of Stagnation Glacier revealed a buried ice body that extends up to 200 m beyond the current terminus of the glacier (Figure 7). The single tabular ice body ranges from 5 m to 20 m in thickness and is about as wide as the glacier. From large-scale surface morphology it appears that much more of the end moraine was once underlain by ice. In the area between 175 m to 400 m beyond the front of the glacier (starting at 305 m on the distance axis in Figure 7 and extending to the right), a steep slope (29û) results in very unstable surface conditions, thought to be responsible for exposing and melting much of the buried ice. In its current geometry, most of the ice-cored end moraine is restricted to a relatively flat 0.10 km2 area near the front of the glacier. There is little evidence of melt-out induced subsidence; however, the ice is occasionally exposed as a result of erosion of the surface cover by runoff streams. These exposures were observed to be limited in extent and are quickly re-covered with sediment within a few days.

Within 500 m of the terminus of Stagnation Glacier, more than 50 shear planes were observed carrying basal sediments to the glacier surface. It appears that subglacial sediment is either dragged or squeezed into the shear plane as the glacier decelerates near its terminus. The thickest shear zones contain less than 10 cm of sediment and most contain little to no sediment. However, accumulation of the small individual contributions of sediment from individual shear planes results in the snout of the glacier being completely covered.

Ice observed in caves beneath the glacier has a very different appearance than surface ice. The sediment-rich ice sequences are over 5 m thick. The ice has a clear brown appearance and is virtually bubble free. Sand and silt particles were observed throughout the basal ice in thin layers from one to a few grains thick. Coarse-grained sediment, ranging from sand to boulders, was also contained within the basal ice. It was, however, dispersed and in lower concentrations. In general, the ice crystals were an order of magnitude smaller than those of the ice located above. Although basal ice tends to be extremely variable in appearance, this ice is similar to basal freezing or regelation ice in other glaciers (Goodwin, 1993; Herron and Langway, 1979; Hubbard and Sharp, 1989). The upper contact of the basal ice zone was not observed; however, observations of "clean" ice near the surface of the moraine indicate that the maximum thickness of the basal ice zone is less than 10 m. Along the side of the glacier, the basal ice zone was observed to be 0-2 m thick.

It appears that both mechanical (shearing) and thermal (regelation) basal sediment accretion processes are at work in Stagnation Glacier, working together with the local climatic conditions to provide the sediment for burial of the glacier.

Preservation of buried ice in the end moraine of glaciers depends on whether subglacial or proglacial processes bring sufficient debris to the surface to insulate the ice. Where the terrain is relatively flat, little sediment is required to stabilize the thermal conditions of buried ice. Repeated observations of the ground surface, exposed sections, and GPR profiling between 1993 and 1999 revealed that areas covered by as little 1 m of sediment were generally not susceptible to observable rates of thaw settlement. On the steep lateral moraines, the active slope movement is a complicating factor, and the rate of thaw settlement could not measured.

The process of ice burial by deltaic deposition was documented at the site of a lake dammed against the side of Glacier C93 (Figure 8). The lake was created when the northwestward-flowing arm of this glacier expanded and dammed a stream originating in an unglaciated catchment area. Aerial photographs indicate that as the lake grew, it flooded over much of the ice-cored lateral moraine and a portion of the glacier that was damming it Figure 8.

As the unglaciated portion of the drainage basin consists of poorly consolidated sandstone, the main streams feeding the lake were laden with sand. When the stream entered the lake, its sediment load was deposited in the form of a classic Gilbert-style delta. Numerous lake stands are recorded by several raised deltas and many strand lines etched into the valley sides.

Ground-penetrating radar profiles such as in Figure 9 reveal geometric and stratigraphic relationships between till, ice, and covering sediments that indicate a series of fluctuations in the lake water levels. From the size of the deltaic units and the strand lines cut into the valley sides it appears that individual lake stands could have lasted several years. However, water levels in the residual lake have also been observed to drop several metres in a period of two weeks.

The developmental history of the site was interpreted using information gleaned from surficial mapping, stream-cut exposures, and GPR profiling. As Glacier C93 began to shrink, the edge of the glacier retreated back from the ice-cored lateral moraine. When lateral streams flowing into the valley were blocked by the glacier further down ice, an ice-dammed lake formed in the resulting basin (Figure 8a). Over time a terrestrial stream feeding the lake resulted in deltaic deposition on the valley side that progressed over the moraine (Figure 8b) and onto the glacier. A thin lacustrine mud drape was deposited on distal portions of the moraine. Lowering of lake level resulted in the stream down-cutting through the delta and moraine, exposing the ice core. Later, a rise in lake level resulted in deltaic deposition in the previous stream valley, re-covering the ice core (D3 in Figure 8). Over time, the D3 delta stage continued to grow and progressed towards the current edge of the glacier. Recently, the lake has almost completely drained, encouraging down-cutting through the most recent D3 deltaic deposits and into the underlying buried ice. Thermokarst on the moraine has resulted in the development of small ponds and retrogressive thaw flows (Figure 8c).

Deposition of sand and erosion by flowing water has resulted in a complex arrangement of tabular massive ice being covered by both deltaic sands and glacial debris (Figure 8). Although examples of deltaic sands, ablation till, and flow till directly covering glacial ice have been preserved, no examples of the lacustrine mud overlying the glacier surface have been discovered at this site. This is likely because of the unstable thermal and hydrological setting of the lake. The lacustrine mud covering the exposed glacial debris that is situated at higher elevations was generally in the order of several millimetres thick and was never thicker than 3 cm. The lack of lacustrine sediment covering the moraine indicates that either the high water level of this lake was not maintained very long, the runoff entering the lake had a low suspended sediment concentration, or the dark colour of the mud enhanced the solar heating and melting of the underlying ice (Benn and Evans, 1998).

The buried ice at this location varies greatly in appearance and physical properties. Ice encased by the lateral moraine ranged from being heavily laden with very poorly sorted glacial debris, to having a sediment content less than 5% that consisted of only sand and fines. Bubbly, white, sediment-free ice, similar to the exposed ice of the neighboring glacier is prevalent beneath the deltaic sand.

Since lake drainage, the exposed glacial debris has been subject to frequent retrogressive thaw flows and thermokarst. The flow material produced by the retrogressive thaw flows has been redeposited directly on exposed ice, over the lake bottom sediments, and on some of the lower delta levels (Figure 8d). Between 1993 and 1999, the headwalls of the retrogressive thaw flows have retreated at a rate of several metres per year. Changes in the stream course occur on a daily basis as different materials are intersected and the channel becomes blocked by scree fall or mud flows.

Preservation of the ice buried beneath the lateral moraine and deltaic sands at the side of Glacier C93 depends greatly on the erosional potential of the stream and the hydrological base level which is controlled by the glacier dam. In the current configuration, thaw of the ice-rich lateral moraine continues unabated. The areas covered by sand tend to be more stable, and do not move except when undercut by fluvial erosion.

A portion of the icing in front of Fountain Glacier persists perennially, and as such, has the potential to become buried. Repeated GPR surveys revealed that the perennial portion of the icing remains over 7 m thick from year to year. Figure 5 shows the icing thickness in July 1993. As the surface of the icing is roughly planar, the topography shown is that of the bottom of the icing. The majority of the icing mass consists of vertically oriented ice crystals (candle ice) that are 0.1-1 cm in diameter and up to 10 cm long that form from the freezing of quiescent pools or subsurface layers of water (Elver, 1994). In the summer, the majority of ice-melt occurs at the boundaries between the large vertically oriented ice crystals. Melt-water flow through the enlarged intracrystallar spaces, is channeled to larger conduits within or beneath the icing which are partially fed by glacier melt-water (Figure 10). Since the water is effectively carried down by gravity along the intracrystallar boundaries, surficial drainage patterns rarely develop.

The processes with the greatest potential for burying ice are fluvial activity or burial by melt-out of fluvial sediments deposited within the icing. Ground-penetrating radar data revealed that few drainage channels were present in the central perennial portion of the icing (Figure 5). The two active streams in this area carried very little sediment, all of which was carried in suspension and was flushed through the system. There was no indication from the GPR or drilling data of any sediment accumulation within the body of the perennial icing.

It appears that all of the fluvial sedimentation occurring on the icing takes place at the edges where the streams have access to the valley-side sediment source. Several streams near the edge of the icing were observed to deposit a layer of sediment in the channel they carved through the ice. However, the sediment layer left behind after the streams had abandoned their channels was always less than 10 cm thick, which is not great enough to provide thermal stability at the ice surface. By the end of the summer, the sides of the icing completely melt, exposing the edge of the perennial icing as a vertical ice wall with all of the sediment-bearing streams running in the valley bottom along the side. Thus the possibility of ice preservation from the accumulation of sediment from the abandonment of a number of channels was determined to be fairly low.

Due to the U-shape of the valley, burial by colluvial sediment is not likely. Slope movement on the valley sides is generally several hundred metres away from the icing. Under the current conditions deltaic sedimentation is also unlikely. The valley in which the icing resides has a slope in the order of 3-5û and no glaciers flow down to or across the valley floor which could dam the drainage system.

In general, the erosion of massive ice is generally the result of either thermokarst or thermal erosion. Both of these mechanisms were observed on Bylot Island. The terrain beyond Glaciers B7 and C93 (Figure 3), provide good examples thermokarst activity on the island. Ground-penetrating radar surveys across three small lakes (less than 500 m in diameter) showed them to be up to 20 m deep, with steeply dipping sides. Two of the lakes have portions of their bottom that are flat, while the third lake has a symmetrically concave profile (Moorman and Michel, 1997). Lacustrine sediment is only present on the flat or gently sloping bottom portions of the lakes, the steep sides consist of bouldery debris. The morphology of these lakes suggest that they are the result of the in situ melt-out of buried ice. This supports the air photo interpretation of Klassen (1993). It is not known if these lakes represent the melt-out of all of the massive ice in the area, or whether there are still considerable quantities preserved between the lakes.

Continuing large-scale degradation of massive ice in thermokarst terrain is difficult to measure in this region. However, several lines of evidence indicate wide-spread melting of subsurface ice is not currently occurring. The lakes and the surrounding terrain appear stable as there are no signs of surface movement and the vegetation is well established. From the thickness of sediment on the lake bottoms, and the average rate of sedimentation for other arctic lakes, it appears that the lakes are quite old, and thus may have formed during a warmer period sometime in the distant past (Moorman and Michel, 1997).

The current climatic conditions are suggestive of cold stable permafrost. From the ground temperature data (Figure 2), the mean annual ground surface temperature was calculated to be -9.6ûC. Using an estimated geothermal gradient of 42 m/ûC, this surface temperature yields an estimated permafrost thickness of Å400 m.

Although this older thermokarst terrain covers sizable areas (e.g. 30 km2 in the C93 Glacier Valley), evidence of recently activated thermokarst terrain was not observed on the island. Thus, ice that is currently in the process of being buried will likely not melt due to thermokarst under the current climate regime.

A few examples of thermal erosion of subsurface ice were observed on the island, generally consisting of the fluvial undercutting ice-rich valley sides, triggering retrogressive thaw flows. However, these tend to be small and localized.

Glaciers, permanent snowbanks, icings, and buried massive ice are present on Bylot Island. The glaciers are currently undergoing varied rates of retreat, ranging immeasurably small to 1.8 km in 46 years. Some of the rapidly retreating glaciers are becoming buried and preserved, through the melt-out of their entrained sediment.

The use of remote sensing and aerial photography greatly simplified the search for suitable study sites. Historical aerial photographs enabled the determination of previous glacial limits and calculation of the glacial retreat rate. Ground-penetrating radar was shown to be effective for mapping the three-dimensional geometry of surface ice, subsurface ice, and sediments. Internal hydrological features such as drainage tunnels were also imaged with GPR. The depth of GPR penetration depends on site specific factors, however, the base of a 75 m thick section of a glacier was clearly imaged and bottom of a buried ice body, 30 m below the surface, was also imaged. Remote sensing, GPR, and field observation data were found to complement each other, enabling the examination of surface and buried ice at a wide range of scales, and were used to support inferences about burial mechanisms.

From this investigation, it is apparent that melt-out of sediment-laden glacial ice appears to be the most effective and widespread process of ice burial on Bylot Island. With many of the glaciers on Bylot Island currently retreating, this is an optimal environment for not only the burial of glacial ice, but also its preservation beneath the ground. In some situations other processes, such as deltaic sedimentation, are also responsible for the burial of surface ice, but their occurrence was much less frequently observed and at a much smaller scale than proposed by Shaw (1977). Thus, even though the icing in front of Fountain Glacier is a long-term surface feature with the potential for becoming buried, the current sediment depositional processes are not of sufficient magnitude to lead to burial and preservation. Further downstream, the annual portion of the icing is thin (<2 m) and is destroyed annually by stream erosion and solar melting.

The preservation of buried ice depends on climatic conditions and the stability of the terrain. The cold conditions on Bylot Island have resulted in preservation of appreciable amounts of ground ice. However, the prevalence of kettle lakes indicates that a considerable amount of ice has melted since the entire island was last covered by glaciers.

This research was supported by a Natural Sciences and Engineering Research Council grant to F. Michel, and several Northern Science Training Program grants to B. Moorman. Logistical support was provided by the Polar Continental Shelf Project. Special thanks go to Dr. Alan Judge of the Geological Survey of Canada and Dr. Michel Allard for use of GPR equipment. The provision of ground temperature data by Dr. Gilles Gauthier is greatly appreciated. Thanks also go to Ron Pietsch for his assistance with satellite imagery, and M. Elver, D. Kliza and L. Moorman for assistance in the field. Finally the authors would like to thank the people of Pond Inlet for their support of this research, and the anonymous reviewers for their excellent comments that lead to a greatly improved manuscript.

AES, 1980. Canadian Climate Normals (1951-1980). Atmospheric Environment Service, Environment Canada, Downsview.

Ashley, G.M., J. Shaw, N.D. Smith, 1985. Glacial Sedimentary Environments. SEPM Short Course No. 16. Society of Paleontologists and Mineralogists, Tulsa, 246 p.

Benn, D.I. and Evans, D.J., 1998. Glaciers and Glaciation. Arnold, New York, 734 p.

Boulton, G.S. (1967). ‘The development of a complex supraglacial moraine at the margin of Sørbreen, Ny Friesland, Vestspitsbergen.’ Journal of Glaciology, 6, 717-735.

Boulton, G.S. (1970). ‘On the origin and transport of englacial debris in Svalbard glaciers.’ Journal of Glaciology, 9, 213-229.

Chinn, T.J.H. and Dillon, A. (1987). ‘Observations on a debris-covered polar glacier "Whisky Glacier", James Ross Island, Antarctic Peninsula, Antarctica.’ Journal of Glaciology, 33, 300-310.

Elver, M.S. (1994). The stratigraphic and isotopic characteristics of an Arctic icing, Bylot Island, N.W.T. Unpublished M.Sc. thesis, Carleton University, Ottawa, 160 p.

Goldthwait, R.P. (1951). ‘Development of end moraines in east-central Baffin Island.’ Journal of Geology, 59, 567-577.

Goodwin, I.D. (1993). ‘Basal ice accretion and debris entrainment within the coastal ice margin, Law dome, Antarctica.’ Journal of Glaciology, 39, 157-166.

Gow, A.J., S. Epstein and Sheehy, W. (1979). ‘On the origin of stratified debris in ice cores from the bottom of the Antarctic ice sheet.’ Journal of Glaciology, 23, 185-192.

Gorelik, Y. and Kolunin, V. (1993). ‘Mechanisms of layer structures growing during ground freezing.’ Permafrost Sixth International Conference, South China University of Technology Press, Beijing China, pp. 879-884.

Harry, D.G., French, H.M., and Pollard, W.H., (1988). ‘Massive ground ice and ice-cored terrain near Sabine Point, Yukon Coastal Plain.’ Canadian Journal of Earth Sciences, 25, 1846-1856.

Herron, S. and Langway, Jr., C.C. (1979). ‘The debris-laden ice at the bottom of the Greenland ice sheet.’ Journal of Glaciology, 23, 193-207.

Hubbard, B. and Sharp, M. (1989). ‘Basal ice formation and deformation: a review.’ Progress in Physical Geography, 13, 529-558.

Inland Waters Branch (1969). Glacier Atlas of Canada: Bylot Island Glacier Inventory Area 46201. Department of Energy, Mines and Resources, Ottawa.

Klassen, R.A. (1993). Quaternary geology and glacial history of Bylot Island, Northwest Territories. Geological Survey of Canada Memoir, 429, Department of Energy, Mines and Resources, Ottawa. 93 p.

Mackay, J.R. (1971). ‘The origin of massive icy beds in permafrost, western Arctic Coast, Canada.’ Canadian Journal of Earth Sciences, 8, 397-422.

Mackay, J.R. and Dallimore, S.R. (1992). ‘Massive ice of the Tuktoyaktuk area western Arctic coast, Canada.’ Canadian Journal of Earth Sciences, 29, 1235-1249.

Moorman, B.J. (1998). The Development and Preservation of Tabular Massive Ground Ice in Permafrost Regions. Unpublished Ph.D. thesis, Carleton University Ottawa, 308 p.

Moorman, B.J., and Michel, F.A., (1997). Bathymetric mapping and sub-bottom profiling through lake ice with ground-penetrating radar. Journal of Paleolimnology, 18, 61-73.

Moorman, B.J., Michel, F.A., and Wilson, A. (1998). The Development of Tabular massive ground ice at peninsula point, N.W.T., Canada, Seventh International Conference on Permafrost: Yellowknife, N.W.T., Canada. Collection Nordicana, Centre d’études nordiques, Université Laval, St. Foy, Canada. pp. 757-762

Nakawo, M. (1979). ‘Supraglacial debris of G2 Glacier in Hidden Valley, Mukut Himal, Nepal.’ Journal of Glaciology, 22, 273-283.

Robinson, S.D., Moorman, B.J., Judge, A.S., Dallimore, S.R., and Shimeld, J.W. (1992). The application of radar stratigraphic techniques to the investigation of massive ground ice at Yaya Lake, Northwest Territories: Muscox, 39, 39-49.

Sharp, M., J. Jouzel, B. Hubbard and Lawson, W. (1994). ‘The character, structure and origin of the basal ice layer of a surge-type glacier.’ Journal of Glaciology, 40, 327-340.

Sharp, R.P. (1949). ‘Studies of superglacial debris on valley glaciers.’ American Journal of Science, 247, 289-315.

Shaw, J., 1977. Sedimentation in an alpine lake during deglaciation, Okanagan Valley, British Columbia, Canada. Geografiska Annaler, 59A, 221-240.

Shaw, J., 1988. ‘Sublimation till.’ In Goldthwait, R.P. and Matsch, C.L. (eds.), Genetic Classification of Glaciogenic Deposits. Balkema, Rotterdam, pp. 141-142.

Souchez, R.A. (1967). ‘The formation of shear moraines: an example from south Victoria Land, Antarctica.’ Journal of Glaciology, 6, 837-843.

Souchez, R.A. (1971). ‘Ice-cored moraines in south-western Ellesmere Island, N.W.T., Canada.’ Journal of Glaciology, 10, 245-54.

Souchez, R.A. and Lorrain, R.D. (1978). ‘Origin of the basal ice layer from Alpine glaciers indicated by its chemistry.’ Journal of Glaciology, 20, 319-328.

Taber, S. (1929). ‘Frost heaving.’ Geology, 37, 428-461.

Tison, J., J. Petit, J. Barnola and Mahaney, W.C. (1993). ‘Debris entrainment at the ice-bedrock interface in sub-freezing temperature conditions (Terre Adélie, Antarctica).’ Journal of Glaciology, 39, 303-315.

Weertman, J. (1961). ‘Mechanism for the formation of inner moraines found near the edge of cold ice caps and ice sheets.’ Journal of Glaciology, 3, 965-978.

Zdanowicz, C.M. (1994). A sedimentological and glaciological study of basal debris entrainment and transport in polar glaciers or southwest Bylot Island, N.W.T. Unpublished MSc. Thesis, Carleton University, Ottawa, 225 p.

Figure 1 The location of the study area on Bylot Island, Nunavut, Canada. The study area is well within the zone of continuous permafrost.

Figure 2 Minimum and maximum daily ground temperatures from August 1994 — August 1995 at a depth of 10 cm in a low-centred polygon. (Data courtesy of Dr. Gilles Gauthier, Université Laval).

Figure 3. This 1985 satellite image (Landsat band 4) of the study area shows the physiography of southwestern Bylot Island. The faulted contact between the sandstone/shale plains and the metamorphic mountainous regions of the island runs roughly down the centre of the image.

Figure 4 Change detection image created by digitally subtracting a 1958 aerial photograph from a 1982 aerial photograph of the same area. Black represents areas covered by ice in 1958 but not in 1982.

Figure 5 Map of the proglacial icing in front of Fountain Glacier in 1993 showing icing thickness in metres and the location of drainage channels within and below the icing.

Figure 6 A profile of across Stagnation Glacier, constructed from topographic leveling and GPR data. Note the connection between the core of ice within the lateral moraine and the main body of the glacier. The location of this cross section is shown in Figure 4.

Figure 7 Ground-penetrating radar profile and interpreted section through the front of Stagnation Glacier constructed using data from a series of GPR surveys and ice exposures. Note that the there were indications on the GPR profiles that the top of the bedrock was only a few metres below the base of the glacier, however, the data were not continuous enough to indicate on the interpretation. Interpretation of the GPR profiles was aided by the use of stratigraphic data collected from cut-banks exposed by proglacial streams.

Figure 8 The evolution of an ice cored delta on the south side of Glacier C93. Aerial photographs in a-c show the growth and erosion stages of a proglacial delta. d&e show the spatial and stratigraphic relationships of the sediment and ice in July 1995.

Figure 9 Ground-penetrating radar profile across the delta (D1 stage) shown in Figure 8 demonstrates how a cover of deltaic sands helps preserve the ice core within a lateral moraine on the south side of Glacier C93. The stratigraphic data shown on the left side of the profile were used for depth verification of the interpreted contacts.

Figure 10 Ground-penetrating radar profile across a portion of the icing in front of Fountain Glacier depicting icing thickness and the location and depth of drainage tunnels within the ice.