Introduction

The tee islands of the Arctic Ocean originate from the break-up of ice shelves, such as the Ward Hunt Ice Shelf, along the northern coast of Ellesmere Island. As calculated from mean freeboard and ice-density relations, the Ward Hum Ice Shelf has an average thickness of 44 m. It is composed of two major stratigraphie units: (1) a lower “basement” ice thought by Reference Marshall and BushneilMarshall (1960, p. 47) to be largely of glacial origin and brine-soaked, but by Reference Nakaya, Nakaya, Muguruma and HiguchiNakaya and others (1962, p. 29) to be of a quite uncertain origin, and (2) an upper unit of inter-stratified iced firn and lake ice lying with angular unconformity (Reference Marshall and BushneilMarshall, 1960; Reference Lyons and LeavittLyons and Leavitt, 1961) upon the basement ice. The unconformity is easily mapped at the surface oils recognized in drill cores because it is marked by a heavy concentration of aeolian dust developed during an ablation cycle which separates the times of formation of the lower and upper shelf stratigraphie units (Fig. 1).

Fig. 1. Map of Ward Hunt Ice Shelf and adjacent ice rises. Ice types identified partly by field and laboratory study, and partly by aerial photography.

This picture of the stratigraphy of the Ward Hunt Ice Shelf is, to some extent, an oversimplification. Radio echo-sounding (Reference Evans, Evans, Gudmandsen, Swithinbank, Hattersley–Smith and Robin.deEvans and others, 1969; Reference Hattersley-Smith, Hattersley–Smith, Fuzesy and EvansHattersley-Smith and others. 1969) suggests that the ice shelf may have a total range in thickness of from 25 to 80 m. Cores from the northern edge of the Ward Hunt ice rise and from the shelf north-east of the ice rise (Reference Ragle, Ragle, Blair and PerssonRagle and others, 1964) show no heavy dust layer nor any basement ice; instead. the lower 20-25 m consist of sea ice. Thus the outer part of the ice shelf, a large part of which calved off in 1961-62 (Reference Hattersley-SmithHattersley-Smith. 1963; Reference NuttNutt, 1966), has a sea-ice basement which differs significantly from the inner basement ice described originally by Reference Marshall and BushneilMarshall (1960) for the Ward Hunt Ice Shelf, and by Reference Nakaya, Nakaya, Muguruma and HiguchiNakaya and others (1962) on ice island T-3. The pétrographie recognition of a difference in the nature of the ice of the lower part of the Ward Hunt Ice Shelf in its inner and outer parts is also consistent with Reference Crowley, Rigsby and BushneilCrowley’s (1961) seismic studies of the ice shelf which showed acoustic differences in the basement ice in these two places. It is the inner basement ice which is the chief subject of the present communication.

Petrography of Basement Ice

Fig. 2. Stratified basement ice, Ward Hunt Ice Shelf.

Most of the inner basement ice, where exposed, shows a remarkably well-developed stratification (Fig. 2) reflecting cyclic variations in the abundance of entrapped gases and average grain-size of the ice. On both T-3, where it constitutes a major part of the ice island. and on the Ward Hunt Ice Shelf, the strata are 20-25 cm in average thickness, and are locally warped into a series of folds with wavelengths of the order of 100 m, and dips on the limbs of as high as 90° (Reference Lyons and LeavittLyons and Leavitt, 1961, p. 14-19; Reference Nakaya, Nakaya, Muguruma and HiguchiNakaya and others, 1962, p. 12). Texturally, much of the stratified ice is characterized by what appear to be very large columnar crystals with indistinct grain diameters of up to 120 cm. The large crystals, however, consist of mosaics of smaller sub-columnar grains showing a sub-parallel orientation of their c-axes normal to the stratification planes, and small-angle grain-boundary relations to one another (Reference Nakaya, Nakaya, Muguruma and HiguchiNakaya and others, 1962, p. 13). Salinity in stratified basement ice is extremely low (Table I; Reference Nakaya, Nakaya, Muguruma and HiguchiNakaya and others 1962, fig. 28). This fact, as well as the optic orientation (c-axis vertical rather than horizontal) and the lack of a characteristic platelet substructure, effectively rule out the possibility that the stratified ice could be sea ice recrystallized in situ. Similar pétrographie arguments may be marshalled against iced firn, glacial ice or lake ice as possible prototypes of the stratified basement ice.

Not all of the basement ice is of the stratified type. Iced firn has been identified in the upper basement ice of both the Ward Hunt Ice Shelf and T-3 by Reference Lyons and LeavittLyons and Leavitt (1961, p. 8) and by Reference Nakaya, Nakaya, Muguruma and HiguchiNakaya and others (1962, p. 20), respectively. The latter investigators also identified 2 m of sea ice at the bottom of T-3. In holes drilled into Ward Hunt basement ice during the 1969 field season (cf. Fig. 1) at locations 1.6 and 2.6 km south-west of the southwestern tip of the island, stratified basement ice and sea ice alternated in the core samples in the manner shown in Figure 3. Identification of ice types was based upon pétrographie criteria in the field and was confirmed in the laboratory by chemical tests (Table I).

TABLE I. δ¹⁸O and chlorinity determination on ice and water samples, northern ellesmere island

18O analyses by S. M. Savin. Cl analyses by L. M. Banos.

Oceanographic Observations

Along the north, east and south sides of Ward Hunt Island, and along the north shore of Ellesmere Island, fresh-water moats open up during the summer melt season. In the moat between Ellesmere Island and the Ward Hunt Ice Shelf (Reference Lyons and LeavittLyons and Leavitt, 1961, fig. 4), fresh water extends 10 a depth of 48±3 m, which is a few meters below the bottom of the ice shelf. In the adjoining Disraeli fiord, Reference Keys, Keys, Johanncssen and LongKeys and others (1969) have shown that the upper part of the fiord is filled with fresh water to a depth of 44 m. Between 44 and 45 m is a brackish water zone, below which is Arctic Ocean sea-water with a salinity of 32%₀. The ice shelf apparently acts as a dam for the surface melt waters draining from northern Ellesmere Island, but interchange with Arctic Ocean water can occur at greater depths (Reference Keys, Keys, Johanncssen and LongKeys and others, 1969, p. 7). Today, as in the past, ice accreting below the Ward Hunt Ice Shelf could conceivably be fresh, brackish or saline ice, depending upon geographic location and climatologie factors i.e. the depths of the fresh, brackish and oceanic zones relative to the ice-shelf bottom). Thinning of the ice shelf during an ablation cycle should increase the likelihood of accreting fresh water or brackish ice on its underside; thickening of the ice shelf during an accretionary cycle would enhance the probability for sea-ice accretion. In any case, complex inter-tonguing relations of ice types are to be expected. Reference SchwarzacherSchwarzacher (1959) has shown that accretion of fresh ice beneath thin sea ice is a common consequence of run-off of melt waters from the pack ice of the Arctic Ocean, and Reference Gow, Gow, Weeks, Hendrickson and RowlandGow and others (1965) have described a similar situation in a part of the McMurdo Ice Shelf of Antarctica.

Fig. 3. Interpretation of stratigraphy of part of Ward Hunt Ice Shelf, based on drill-core and laboratory studies. See Figure 1 for location of holes A and B. Ward Hunt ice rue towards right side of section; trough of syncline between Ward Hunt and Ellesmere Islands to left of section.

Oxygen-Isotope and Chlorinity Data

Table I presents oxygen-isotope and chlorinity data of ten samples collected during the 1969 field season. Three of them (Nos. 6, 7 and 8) were taken by J. Keys and H. Serson from an océanographie station in Disraeli Fiord, and the remainder were collected by the writers. Determinations of δ^l8O were made using techniques described by Reference Epstein and MayedaEpstein and Mayeda (1953). Chlorine was determined using ion-sensitive electrodes. The basement-ice samples come from the holes drilled 1.6 and a.6 km south-west of the south-west edge of Ward Hunt Island (Fig. 1).

The extremely low δ^l8O values of the ice from the Camp Creek and Ward Hunt ice rises (Nos. 12 and 13) are consistent with the pétrographie and stratigraphie identification of iced firn. Because of the number of samples (two), we can draw no conclusions concerning the significance of the differences in δ¹⁸O values with respect to temperature of accumulation, or other factors.

For the fresh-water samples (Nos. 6 and 9) the lower δ^l8O value from Disraeli Fiord is consistent with the fact that it receives melt waters from elevations of up to 1 980 m, and its average δ^l8O value should therefore be more negative than that from Ward Hunt Lake, which receives its drainage from elevations lower than 415 m. Lake ice forming from the melt water in Ward Hunt Lake should have δ¹⁸0 values of about —24 or more positive; that forming on the Ward Hunt Ice Shelf presumably would have values in the range of — 27 to —30.

Because the ice petrographically identified as sea ice (No. 4) has a δ¹⁸O value very close to 0.00, as does the Arctic Ocean water (No. 8), there seems to be little doubt that this ice type is identifiable both chemically and petrographically with considerable certainty.

The three basement ice samples (Nos. 2, 3 and 5) show δ^l8O values more negative than the brackish water in Disraeli Fiord (No. 7) but not as negative as fresh melt water in the same fiord. The δ¹⁸O and chlorinity of samples 3 and 5 imply crystallization of ice from brackish water. Sample 2 seems to have formed from water which was also brackish but its δ¹⁸O value is anomalously low. This sample was taken quite close to the surface in an area from which iced firn has been ablated only recently. It is possible, though not indicated by any other criteria, that a decrease in the δ^l8O value has occurred because of downward percolation by firn melt water. The alternative possibility is that the ice crystallized from water somewhat fresher than that represented by samples 3 and 5.

An obvious conclusion concerning samples 3 and 5, and probably sample 2 as well, is that they represent a newly recognized type of ice, best termed brackish ice, which is a unique but volumetrically significant constituent of Arctic ice shelves and ice islands. Reference CraryCrary (1960, p. 33) had suggested that the freezing of fresh or brackish waters on the underside of T-3 and the Ward Hunt Ice Shelf might have been important in their development. What seems surprising, as gauged by Figure 3, is that approximately half of the basement ice is of brackish origin.

Discussion

Tritium measurements in Disraeli Fiord (Reference Keys, Keys, Johanncssen and LongKeys and others, 1969, p. 4) suggest that each season’s melt water sinks to the interface between the salt and fresh water where it may flow directly out of the fiord, under the ice shelf. However, because the ice shelf thins by ablation, particularly near the edges, there is the possibility that some of the melt water, brackish water, or sea-water may accrete as ice on the underside of the ice shelf. The brackish water is of interest in this regard, because its salinity lies in the range where theory predicts (Reference Weeks, Lofgren and ÖuraWeeks and Lofgren, 1967) that ice growing in the sub-surface should have the c-axis horizontal orientation and platelet substructure characteristic of sea ice. Supercooling and ice-dendrite formation, as observed in the Disraeli Fiord brackish water zone (Reference Keys, Keys, Johanncssen and LongKeys and others, 1969, p. 3), provide a mechanism for explaining the c-axis vertical and small-angle grain-boundary texture so common in the brackish ice. Nucleation followed by floating of dendrites to the underside of the ice shelf, with consequent t-axis vertical preferred orientation, would seem to be an inevitable consequence of undercooling in the brackish water zone. On the other hand, there is no evidence for supercooling of the Arctic Ocean sea-water, and shelf sea ice does, predictably, have its c-axis horizontal.

The concentration of bubbles marking the stratification planes in the basement ice can be understood in terms of the inverse relation between air solubility and salinity in water (Reference DorseyDorsey, 1940, p. 534-49). If brackish ice accretes below the ice shelf, it will expel downward water more saline than the relatively pure ice which is crystallizing, and a condition will eventually be reached where exsolution of gas becomes inevitable. This gas concentrates beneath the year’s annual increment of brackish ice and is partly trapped in the downward encroaching ice front. Flooding of the next summer’s fresh or brackish water under the ice shelf partly dissolves the preceding winter’s gases and allows the cycle to repeat itself.

If the average stratum of basement ice is 20 cm thick, and if this represents an annual increment to the basal edge of the ice shelf because of ablation at the surface, only 220 years as a minimum might be required to bring stratified ice from the bottom of the 44 m thick ice shelf to the surface. This is a surprisingly short time but not inconsistent with a radiocarbon age of 400±150 years which Reference CraryCrary (1960, p. 43) reported for a siliceous sponge which had passed through the basement ice south of Ward Hunt Island.

An extension of the model proposed here for the stratified basement ice is that the individual strata may, on the average, thin or wedge out completely toward the center of the ice shelf. The thickness of the average layer and its nature (brackish ice or sea ice) will depend on hydrologie factors which are, in turn, related to climatology. If the ice shelf maintains a nearly constant thickness, curling up at its edges and, over the past several hundred years, accumulating firn and lake ice in the area between Ward Hunt and Ellesmere Islands, bottom melting is also a necessity under this median part of the ice shelf (Reference Lyons and RagleLyons and Ragle, 1962, p. 93-94). Similarly, accretion of ice under some parts of the McMurdo and Ross Ice Shelves of the Antarctic, and ablation under others, has recently been demonstrated by Reference Gow, Gow, Weeks, Hendrickson and RowlandGow and others (1965) and by Reference SwithinbankSwithinbank (1970).

An interesting feature of the basement ice is its age relative to the upper iced firn and lake ice of the ice shelf. Some of it, immediately below the dust-marked unconformity, must be older than the superjacent iced firn and lake ice, but the lowermost basement ice must be younger than the unconformity. Reference Hattersley-Smith and SersonHattersley-Smith and Serson (1970} have shown that the mass balance of the Ward Hunt Ice Shelf and ice rise has been negative over the past decade, so the present surface is working its way downward into older iced firn and lake ice. In the ice shelf as a whole, therefore, the age increases both from the top downward, and from the bottom upward.

Our interpretation of the stratigraphie-structural relations of the basement ice of the Ward Hunt Ice Shelf is shown in Figure 3. Beneath the heavy-dust ablation unconformity, the basement is gently warped into a broad syncline between Ward Hunt and Ellesmere Islands (cf. Fig. 1). Where the basement ice re-appears at the surface near Ellesmere Island, it is deformed into anticlines and synclines, probably because of stresses exerted on this edge of the ice shelf by glaciers advancing from Ellesmere Island at a time prior to the major ablation cycle during which the heavy-dust layer was concentrated. Reference CraryCrary’s (1960, p. 44-46) estimate of the age of the ablation cycle, based upon somewhat conflicting radiocarbon dates is 1 600 years. However, the oxygen-isotope data of Reference Dansgaard, Dansgaard, Johnsen, Møller and LangwayDansgaard and others (1969) from Camp Century, Greenland, suggest that a major ablation cycle in northern latitudes terminated as recently as 850 years ago.

Finally, the distribution and structure of the ice types comprising the Ward Hunt Ice Shelf (Fig. 1) make it clear that T-3 could not have originated from the area between McClintock Fiord and Markham Bay. McClintock Fiord itself, however, or one of the bays to the west such as Yelverton Bay remain likely sites for its point of origin (cf. Reference Stoiber, Stoiber, Lyons, Elberty, McCrehan and BushnellStoiber and others, 1960).

Acknowledgements

We are indebted to U.S. Army C.R.R.E.L. and the Canadian Defence Research Board for logistic support. Dr G. Hattersley-Smith, J. Keys and H. Serson of the Defence Research Board, and Mr Bruce Petrie were most helpful in the field studies. We are grateful to A. J. Gow for comments and corrections to the manuscript. Financial support for isotopic analyses was provided by N.S.F. Grant GA-1693.

MS. received 8 July 1970