Introduction

Subglacial ground-water flow in karst terrain has been recognized for many years, but information on the processes of drainage at the glacier bed has remained scarce. A large part of Columbia Icefield, Alberta, Canada, appears to be drained through a karst aquifer and Castleguard Cave has allowed access to the glacier bed so that these processes may be studied. The hydrology of this area was described by Reference SmartSmart (1983). Up to 130 km² of glacier bed was shown to be freely drained by karst. In particular, ground-water flow was found at two scales: in large discrete conduits, and as smaller diffuse drips and seeps commonly associated with stalactites. An association between the basal regelation film and the diffuse seepage, and between conduit flow in both the glacier and ground water was suggested. The morphology of recently abandoned pro-glacial surfaces supported this association. A dye test from a moulin on Saskatchewan Glacier demonstrated rapid flow of englacial water to springs over 8 km away in Castleguard Valley.

Continuing work on the subject has prompted this short note which provides further information on the hydrology of the glacier bed and some related observations on subglacial sediment generation.

Further Observations

Aerial photographs and a recent map of Columbia Icefield (National Hydrological Research Institute, 1980) show a pond in a closed depression on the glacier surface. Reference Muir and FordMuir and Ford (1985) illustrated this depression and observed that it is 200 m × 50 m in plan and perhaps 20–30 m deep. There is some evidence of fluctuating water level in the pond, and the icefield is drained centripetally to this point. The depression lies somewhat to the east of an inferred bedrock depression beneath the icefield (Reference SmartSmart, 1983, fig. 6, p. 478) but closely overlies a part of Castleguard Cave exhibiting abundant ground-water flow.

Below the icefield in Castleguard Cave there is usually a wind blowing. This is a chimney-effect wind (Reference Atkinson, Atkinson, Smart and WigleyAtkinson and others, 1983) driven by temperature differences between the cave and the surface. In winter, the wind blows up through the cave and is presumably vented to the surface through solutional shafts in bedrock overlain by about 300 m of glacier ice. There are numerous crevasses in the icefield surface above this point which presumably carry the wind. Attempts to demonstrate air discharge at the surface have so far been frustrated. In summer, the shafts carry both melt water and air down into the cave. Some shafts carry neither water nor air, and are plugged by clean polycrystalline glacier ice. Not only is the intruded ice clean, but ground water draining from the icefield also carries little sediment. In contrast, most of the 18 km of the cave has been occupied by silt–clay laminates presumably of glacial origin. It was originally hypothesized that these were deposited during three glacial epochs (Reference Schroeder and FordSchroeder and Ford, 1983). However, further study of these sediments shows no uniform stratigraphy. Some sections exhibit sequences of erosion-scoured surfaces followed by coarse intraclasts and then a progressive fining upwards. These are micro-turbidites. Other sites show more or less uninterrupted deposition. Sediment appears to have frustrated solutional development of much of the upper reaches of the cave, resulting in a complex series of passages which have been variously formed, infilled and flushed. Anastomosing networks of small tubes have developed in the walls and roof of infilled passages. In places, the sediments are interlain with calcite speleothem possessing a reversed magnetic polarity, suggesting an age of at least 700 ka (Reference Gascoyne, Gascoyne, Latham, Harmon and FordGascoyne and others, 1983). Furthermore, samples of the sediment contain pollen from broad-leafed species not reported from the Holocene record of the area (Reference Gale, Gale, Hunt, Smart and SpencerGale and others, 1985).

Abandoned glacier beds may retain evidence of former basal drainage systems (Reference Walder and HalletWalder and Hallet, 1979), and segments of former channel networks cut into limestone are found throughout the Rocky Mountains. These channels frequently contain abandoned waterfalls (Fig. 1a) and incised meanders (Fig. 1b). In areas where former glacier-bed slopes are oblique to the previous ice surface, most channel remnants are oriented down-dip rather than directly towards the former ice margin (e.g. Reference SmartSmart, 1983, fig. 2, p. 573). Abundant calcite precipitates co-exist with these channel remnants (Fig. 2), although in places striation and precipitate-free patches are found (Fig. 1b). Fine fractures in the lee side of bedrock steps are often marked by annular calcite “crowns” (Fig. 3a and b). These are assumed to mark the outlets of small-scale seeps through bedrock (Reference SmartSmart, 1983).

Fig. 1.

Short Nye-channel segments showing: (a) a retreating waterfall; (b) meander migration with incision. Both these features are about 30 cm in width, and both terminate at the head of karst shafts. A zone free of precipitates and striations is seen in the top right-hand corner of (b).

Fig. 2.

Zones of calcite solution and precipitation on a recently exposed glacier bed represent the action of a regelation water film. The lens cap is 50 mm in diameter.

Fig. 3.

Subglacial “stalactites” show calcite precipitation to have occurred at the outlet of small fractures in lee cavities. The regelation film is inferred to have leaked through bedrock.

Discussion

The waterfalls, incised meanders, and down-dip channels are all characteristic of open-channel rather than closed-conduit flow. These observations plus the perennial cave wind therefore confirm Reference LliboutryLliboutry’s (1983) assertion that water in subglacial conduits is generally under low or atmospheric pressure.

Furthermore, they persist through winter. The channel remnants are by definition “Nye-channels”. They must have been linked by “Röthlisberger-channels” which have left no evidence of their presence, but must have re-formed annually in order to create the Nye-channels. All the channels observed are far larger than those considered by Reference Weertman and BirchfieldWeertman and Birchfield (1983). Exchange of water between the conduit and regelation system is not normally possible (Reference WeertmanWeertman, 1972), but a permeable glacier bed will permit exchange of water. Despite the massive pressure gradient, however, the bulk of the regelation film is not totally drained. If this were the case, calcite precipitates would not be able to form. The precipitate-free areas may be zones of preferred drainage, but the absence of striations suggests cavity formation.

There appears to be little sediment generated at present beneath the accumulation zone of part of Columbia Icefield. The massive carbonate bedrock is certainly resistant to glacial erosion and forms prominent benches. However, this observation is generally consistent with the high elevation of cirques despite their more prolonged occupation by glacier ice. The massive influxes of sediment into the cave in the past appear to mark ancient and profound interglacials, possibly when Columbia Icefield was reduced to a few minor cirque glaciers and the sink points became distal or pro-glacial. The cave appears to have been unable to convey the imposed sediment load and has been repeatedly plugged. Occasionally these plugs have burst, causing scour and subsequent settling which have generated the local turbidite sequences. Cave development has been redirected by sediment fill. The anastomosing passages are hydraulically very inefficient and are well known from karst studies (e.g. Reference Lauritzen and BeckLauritzen, 1981). There is a clear analogy here with the events leading up to glacier surging described by Reference Kamb and KambKamb and others (1985).

Conclusions

Subglacial ground-water flow can be significant in alpine karst areas, but it is unclear whether this is also true for ordinary diffuse ground-water flow. In general, subglacial conduit flow appears to be in open channels. Little sediment is presently being produced beneath the accumulation zone of Columbia Icefield, although abundant material has been injected during past interglacials. The subglacial karst drainage system has been unable to transport all the injected material during these periods. Further work is presently under way on these sites.