1. Introduction

There is an extensive literature on the application of stable isotopes to glaciological studies, particularly for terrestrial ice, but there have been relatively few applications of stable isotopes to marine ice studies (sea ice, ice shelves). The few published studies include: the identification of the origins of ice in ice shelves (Reference Lyons, Savin and TamburiLyons and others, 1971; Reference Gow and EpsteinGow and Epstein, 1972; Reference MorganMorgan, 1972; Reference Grootes and StuiverGrootes and Stuiver, 1986; Reference Jeffries, Sackinger, Krouse and SersonJeffries and others, 1988b); the oxygen-isotope content of sea ice in the Gulf of St. Lawrence (Reference Tan and FraserTan and Fraser, 1976); the δ ¹⁸O values of frazil ice in Weddell Sea pack ice (Reference Gow, Ackley, Buck and GoldenGow and others, 1987); and, the identification of annual growth layers in pack-ice floes (Reference Friedman, Schoen and HarrisFriedman and others, 1961) and in multi-year land-fast sea ice (Reference Jeffries and KrouseJeffries and Krouse, 1988).

Multi-year land-fast sea ice (MLSI) is a common feature off the north coast of Ellesmere Island (Fig. 1). It is similar perhaps to the “sikussak” of north Greenland fiords (Reference WadhamsWadhams, 1981), and sea ice at the Koettlitz Glacier tongue, Antarctica (Reference Gow, Weeks, Hendrickson and RowlandGow and others, 1965). The MSLI of northern Ellesmere Island grows in place for many years, frequently replacing ice shelf lost by ice-island calvings, and often attaining undeformed thicknesses of as much as IOm (Reference Jeffries and SersonJeffries and Serson, 1986; Reference Jeffries, Sackinger, Shoemaker, Sackinger and JeffriesJeffries and others, 1988a). The MSLI has a characteristic undulating surface topography of long, linear hummocks and depressions, with an average hummock spacing of about 60 m (Reference Jeffries and SersonJeffries and Serson, 1986). On the basis of ice-thickness measurements, it has been suggested that the surface topography is mirrored by an inverted, undulating topography at the bottom side of the ice (Reference Jeffries, Sackinger, Shoemaker, Sackinger and JeffriesJeffries and others, 1988a).

Fig. 1. Location map of ice cores drilled in MLSI, northern Ellesmere Island. Ice core 86–8 (Ward Hunt Ice Shelf) was drilled in an area where brackish ice was found by Reference Lyons, Savin and TamburiLyons and others (1971). Data for 86–8 have been presented elsewhere (Reference Jeffries, Sackinger, Krouse and SersonJeffries and others, 1988b).

Prior to 1984, there had been only a few studies of the properties of this thick, multi-year sea ice, including: some crystallographic investigations of the Markham Bay Re-entrant, formerly attached to the front of east Ward Hunt Ice Shelf (Reference Ragle, Blair and PerssonRagle and others, 1964); an ice-thickness and growth study of Nansen Ice Plug (Fig. 1) (Reference SersonSerson, 1972); and, the structure of the thick sea-ice floe that supported the Soviet drifting station SP-6 (Reference CherepanovCherepanov, 1966). It has been suggested that SP-6 calved from the thick sea ice off the north coast of Ellesmere Island, perhaps even from Nansen Sound (Reference SersonSerson, 1972; Reference Walker and WadhamsWalker and Wadhams, 1979).

Since 1984, we have drilled a number of ice cores in MLSI as part of a program of Arctic ice-shelf and ice-island studies. Ice-core analysis has included salinity, oxygen-isotope (δ ¹⁸O), and tritium measurements. On the basis of the relationship between the salinity and the ¹⁸O content of the coastal waters of northern Ellesmere Island, ice δ ¹⁸0 and tritium values were used to show that: (1) the MLSI has grown from water with salinities varying from as little as 3.2‰ to as much as 31.8‰, a salinity range that includes brackish water as well as sea-water; (2) water salinity varies seasonally; and (3) snow melt is the principal source of fresh melt water being contributed to the water—ice system (Reference Jeffries and KrouseJeffries and Krouse, 1988). MSLI probably includes some brackish ice, but the amount of brackish ice and its properties have not been determined.

Significantly, brackish ice has been identified previously in Ward Hunt Ice Shelf (Fig. 1) on the basis of ¹⁸O, salinity, and crystallographic analysis (Reference Lyons, Savin and TamburiLyons and others, 1971).

There have been few studies of brackish ice and much remains to be learned of its physical and structural properties. This paper emphasizes the physical properties of brackish and sea ice in MLSI, and presents the results of δ ¹⁸O and salinity measurements on 12 ice cores (a total of 70.12 m of ice) drilled at locations shown in Figure 1. All cores were obtained in April or May when air temperatures were low enough to minimize brine drainage. In some cases, the cores were drilled in recognizable hummocks and depressions but, because of the snow-covered, low relative relief, it was not always possible to identify the topographic location. The emphasis of the data analysis and discussion is on the use of S¹⁸O measurements to identify brackish and sea ice in MLSI, to show the amount of each ice type present and the implications for sub-ice water properties and structure, and to examine the differences between brackish and sea-ice salinity. The field and laboratory methods have been described by Reference Jeffries, Sackinger, Shoemaker, Sackinger and JeffriesJeffries and others (1988a).

2 Oxygen-18: A Conservative Tracer for MLSI Studies

The recognition of fresh, brackish, and sea ice in the MLSI cores relies not on structural or salinity analysis of the ice but on the use of stable isotopes as conservative tracers of the water from which the ice grew. In this section the rationale behind this approach to ice-type identification in a fast-ice zone, where the Salinity–δ¹⁸O relationship for the coastal waters is well established, is explained. There are three stages in this approach.

First, it is necessary to define the salinity of fresh, brackish, and sea-water. The U.S. Navy Glossary of oceanographie terms (1966) defines the salinity of brackish water as 0.5–17.0‰. This definition of brackish water salinity is adopted for this study, although from the glaciologist’s point of view it does not take into account the structural changes that occur as ice freezes from progressively less saline water. However, since there is as yet no diagnostic brackish ice structure, as there is for sea ice (cf. Reference Weeks and AckleyWeeks and Ackley, 1982), the identification of brackish ice according to the salinity and oxygen-isotopic composition of the brackish water from which it grew is an appropriate technique. The fresh-water, brackish water, and sea-water salinities are summarized in Table I.

Table. I. Fresh, Brackish, And Sea Ice/Water Classification According To Water Salinity And δ ¹⁸O Values, And Ice δ ¹⁸O Values

*Case 1, where no isotopic fractionation occurs and the ice δ ¹⁸O value is the same as the water δ ¹⁸O value.

†Case 2, where a maximum isotopic fractionation (α = 1.003) occurs.

Secondly, corresponding δ ¹⁸O values must be known for the salinity values of the fresh, brackish, and sea-water categories. For this it is necessary to know the Salinity– δ ¹⁸O relationship in the parent water. In Disraeli Fiord and Milne Fiord (Fig. 1), salinity and δ ¹⁸O are related by (Reference JeffriesJeffries, unpublished)

(1)

Using this relationship, the water salinity values (Table I) can be translated into water δ ¹⁸O values (Table I). The use of Equation (1) for this purpose assumes that the δ −5 relationship applies to the entire study area, and that the primary fresh-water diluent of the sea-water is run-off from the land. This will be discussed further in section 4.

The third stage relates the ice δ ¹⁸O values to water δ ¹⁸O values to identify fresh, brackish, and sea ice. To do this requires that isotopic fractionation on freezing be taken into account. However, as pointed out by Reference Jouzel and SouchezJouzel and Souchez (1982) and Reference Souchez and JouzelSouchez and Jouzel (1984), the amount of isotopic fractionation that occurs during natural processes cannot be known; under natural conditions the problem is complicated by non-equilibrium processes, the inclusion of liquid by the ice, the input and output of water to and from the reservoir, variations in freezing rates, and whether freezing occurs in an open or closed system. In the case of congelation-ice growth, it is reasonable to expect open-system conditions and near-maximum isotopic fractionation. On the other hand, this will be offset by the inclusion of brine in the ice. The inclusion of brine will be particularly critical in a mesh of congealing frazil- or platelet-ice crystals, and in this more closed system the resulting isotopic fractionation might be much less than the maximum. In view of the complex freezing processes that probably occur during MLSI growth, including congelation and frazil/platelet growth, the two extreme cases of isotopic fractionation are considered in this study, and between them they probably account for the majority of the measured δ¹⁸O values in MLSI. Case 1 assumes no isotopic fractionation, so ice δ¹⁸O values represent water values; case 2 assumes a maximum isotopic fractionation (α = 1.003; Reference O'NeilO'Neil, 1968), thus, after subtracting 3‰ from a given ice δ ¹⁸O value, the δ value of the parent water is obtained (Table I).

Assuming that the oxygen-isotope content remains unchanged after ice formation, the isotopic fractionation is small compared to the brine loss from sea ice as it thickens and ages, i.e. the ¹⁸O/¹⁸O ratios of the ice are conservative tracers of the parent water. Some alteration of original δ ¹⁸O values might occur by infiltration and refreezing of melt water at the ice surface. Although Reference Jeffries and KrouseJeffries and Krouse (1988) suggested that melt-water infiltration at the surface of MLSI is minimal, for the purpose of this study, δ ¹⁸O values from the uppermost 0.5 m of ice are excluded to allow for possible infiltration of ¹⁸O-depleted melt water and alteration of the original δ ¹⁸O value, and to exclude refrozen melt water from surface melt pools. On this basis, it is now possible to identify fresh, brackish, and sea ice directly from the ice δ ¹⁸O values. It is stressed that this methodology applies only to the multi-year land-fast sea ice of northern Ellesmere Island, where the δ –salinity relationship of the coastal waters is known. The methodology is “region specific” and cannot be applied to other fast-ice zones, or to multi-year pack-ice floes. Before the results of the analysis are presented, the ¹⁸O content of sea-water and sea ice will be briefly discussed.

On a global basis, sea-water has a mean δ ¹⁸O value of about 0.0‰ (Reference CraigCraig, 1961), with some negative deviations arising from local input of meteoric water (meteoric water is water that has been involved in recent atmospheric circulation and has δ ¹⁸O values less than zero). Unlike sea-water, the ¹⁸O content of meteoric water shows a latitude effect, with precipitation δ ¹⁸O values decreasing (becoming more negative) towards high latitudes (Reference Dansgaard, Johnsen, Clausen and GundestrupDansgaard and others, 1973). In the Arctic Ocean, where meteoric water δ ¹⁸O values are very negative, the surface sea-water has a δ ¹⁸O range of –4.5‰ to +0.3‰ (Reference Vetshteyn, Malyuk and RusanovVetshteyn and others, 1974; Reference Östlund and HutÖstlund and Hut, 1984), due to mixing of meteoric water and sea-water. Assuming maximum isotopic fractionation on freezing, the aforementioned values correspond to ice δ ¹⁸O values of –1.5‰ to +3.3‰. Similar values are found in MLSI, but, as will be shown, a greater proportion of the δ ¹⁸O values in MLSI are much more negative.

3. Results of Oxygen-Isotope and Salinity Measurements

The salinity and isotope data for each core are summarized in Table II. In the combined data set of the 12 MLSI cores, salinity values range from as little as 0.01¾ to as much as 12.06‰. The wide salinity range is matched by a wide range of δ¹⁸O values, from a minimum of –29.9‰ to a maximum of +0.7‰ The wide range of δ¹⁸O values, and the mean values that fall in the brackish ice/water category, indicate a substantial meteoric water input to the water—ice system, and the potential for MLSI growth by the freezing of fresh, brackish, and sea-water.

Table. 2. Data For Ice Thickness, Ice Salinity, And Ice δ¹⁸O Values

* The core length also corresponds to the ice thickness.

† H and D denote hummock and depression, respectively.

‡ Mean values are expressed with ±1 standard deviation. Figures in parentheses are the number of analyses for each parameter for each core.

3.1. Amounts of fresh, brackish, and sea ice in MLSI

Case 1. Assuming no isotopic fractionation, the amount of fresh, brackish, and sea ice in each MLSI core is shown in Figure 2a. MLSI has an overall composition of 29.6‰ brackish ice, 70% sea ice, and 0.4% fresh ice.

Case 2. Assuming maximum isotopic fractionation, the amount of fresh, brackish, and sea ice in each MLSI core is shown in Figure 2b. MLSI has an overall composition of 42.3% brackish ice, 57.3% sea ice, and 0.4% fresh ice.

Whether one considers cases 1 or 2, clearly, brackish ice is quite common, but there is some variation in the amount of each ice type from location to location, and at a given location. Some particular features common to cases 1 and 2 are listed below:

(1) Sea ice is particularly common in the cores drilled in and near Nansen Ice Plug (Figs 1 and 2), and brackish ice is particularly common at Ayles Fiord (Figs 1 and 2), where a single fresh-ice specimen (core 85–7; Fig. 2) was also found. Contrasting salinity and δ¹⁸O profiles for sea-ice core 86–4 (Nansen Ice Plug) and for brackish core 85–5 (Ayles Fiord) are shown in Figure 3.

(2) Where ice cores have been drilled in adjacent hummocks and depressions, the depression cores contain a greater proportion of brackish ice, e.g. cores 85–3 and 85–4 (Fig. 4), 85–5 and 85–6, 85–8 and 85–9 (Fig. 2). Depression ice also has a lower ¹⁸O content and salinity than hummock ice (Table II).

(3) In some cores there are pronounced brackish and sea-ice strata, e.g. cores 85–3 and 85–4 (Fig. 4). The hummock core, 85–4, has two basic strata: a sea-ice layer from the surface to 7.80 m, and a sea-brackish ice layer from 7.80 m to the bottom side. The depression core, 85–3, has three basic strata: a sea-ice layer from the surface to 5.50 m, a sea-brackish ice layer from 5.50 to 7.80 m, and a brackish ice layer from 7.80 m to the bottom side. The sea-brackish ice layers in the central section of core 85–3 and the lower part of core 85–4 consist of thin, interfingering layers of sea and brackish ice.

Fig. 2. Amounts of fresh, brackish, and sea ice in MLSI according to the ¹⁸O content of the ice: case 1 (a), case 2(b). The ice-core identification numbers include H and D for hummock and depression locations, respectively.

Fig. 3. δ¹⁸O(left) and salinity (right) profiles in ice cores 86–4 (Nansen Ice Plug) and 85–5 (Ayles Fiord). Ice core 85–5 is a brackish core and is considerably less saline and has a lower ¹⁸O content than sea-ice core 86–4. Each core shows a salinity increase at the bottom due to winter ice growth.

Fig. 4. δ¹⁸O profiles in adjacent hummock (85–4) and depression (85–3) ice cores, deformed MLSI at the front of Ward Hunt Ice Shelf. The profiles are plotted to allow for the relative surface relief, so that hummock and depression data are at equivalent depths. These profiles assume a case 2 situation, i.e. maximum isotopic fractionation occurred during freezing. Mean δ¹⁸O values for individual layers in each core are shown at the left (85–3) and at the right (85–4). The vertical dashed line separates the sea and brackish ice.

3.2. The ¹⁸O content and salinity of brackish and sea ice

The fresh ice identified on the basis of ice ¹⁸O content (δ¹⁸O, –29.9‰) was found at a depth of 1.78 m in ice core 85–7. This ice was, therefore, some distance below the surface and the ¹⁸O content was unlikely to have been affected by surface processes. Since fresh ice constitutes only a small proportion of the ice cores, it will not be considered further in any detail. The mean δ¹⁸O and mean salinity values of sea and brackish ice, calculated on the basis of no isotopic fractionation (case 1) and maximum isotopic fractionation (case 2), are summarized in Table III. In both cases there is an order-of-magnitude difference between the mean δ ¹⁸O and mean salinity values of each ice type.

Table. 3. Salinity and δ ¹⁸O Data for Brackish and Sea Ice

On the basis of brackish and sea-ice identification according to δ ¹⁸O values, the distribution of salinity values for each ice type has been determined (Fig. 5). For both case 1 (Fig. 5a) and case 2 (Fig. 5b), the distribution of brackish ice salinity values has a pronounced positive skew, with 50% of values occurring in the modal class (0–0.49‰). There is a tail of higher salinity values of as much as 4‰ in brackish ice. Sea-ice salinity values are less skewed and more widely distributed, with a tail of high salinity values exceeding 4.0‰. In both brackish and sea ice, the high salinity values in the tails of the distributions are nearly all associated with the most recent ice growth at the bottom of the cores. The data (Table III; Fig. 5) show that there are two quite different salinity and δ ¹⁸O populations for brackish and sea ice.

Fig. 5. Salinity distributions in brackish ice and sea ice. a. Case 1, no isotopic fractionation, b. Case 2, maximum isotopic fractionation.

4. Discussion

4.2. Brackish water and sea-water under MLSI

The ¹⁸O content of the MLSI cores indicates that low-salinity or brackish water is common below MLSI. Layers of low-salinity water floating upon a sea-water layer and under a sea-ice sheet, or an ice shelf, have been described previously, and it has been common to refer to the low-salinity water layers as “melt pools” (Reference Untersteiner and BadgleyUntersteiner and Badgley, 1958; Reference Gow, Weeks, Hendrickson and RowlandGow and others, 1965; Reference HansonHanson, 1965; Reference Martin and KauffmanMartin and Kauffman, 1974). In the case of MLSI, the term melt pool is, in some instances, somewhat misleading as it suggests an areally limited brackish sea-water stratification confined to the inverted depressions at the bottom side. It is apparent that this is not the case, since brackish ice occurs in hummocks as well as in depressions, i.e. brackish water underlies the entire MLSI sheet at some locations and times.

Brackish sea-water stratification under the entire ice sheet is most likely to occur in the summer, when land run-off is at a maximum. Once run-off ceases or is minimized in the autumn, the stratification will become less pronounced and perhaps even dissipate. This seasonality of water properties and structure gives rise to seasonal or annual layers in MLSI (Reference Jeffries and KrouseJeffries and Krouse, 1988) and pack-ice floes (Reference Friedman, Schoen and HarrisFriedman and others, 1961). On the basis of annual-layer counting in the ice cores from Milne Re-entrant (Fig. 1), it has been estimated that the undeformed MLSI was 20 years old in 1985 (Reference Jeffries and KrouseJeffries and Krouse, 1988). The longevity and immobility of the MLSI provide a stable environment for the growth of annual ice layers associated with the seasonally variable sub-ice oceanography.

The seasonality of water properties is not always a brackish water and sea-water alternation. It is apparent that MLSI can be underlain year-round by brackish water, with varying ¹⁸O content and salinity, but giving rise to 100% brackish ice, e.g. ice core 85–5 (Fig. 3). The year-round persistence of brackish water is further substantiated by the water that entered the ice-core 85–5 bore hole; it had a δ ¹⁸O value of –24.7‰, and was sampled in early May before the onset of melting and run-off. The year-round presence of low-salinity water interposed between ice and sea-water has also been observed near the front of the Koettlitz Glacier tongue, Antarctica (Reference Gow, Weeks, Hendrickson and RowlandGow and others, 1965).

Under-ice, brackish water melt pools are common in inverted depressions and apparently persist year-round in those at some locations; this would account for the greater amount of brackish ice that is found in depressions compared to adjacent hummocks, and the lower mean salinity of depression ice. At other locations, however, the MLSI is apparently underlain by sea-water for most of the time, viz. the Nansen Ice Plug cores (e.g. core 86–4; Fig. 3). Water that entered the 86–4 bore hole had a δ ¹⁸O value of –2.7‰ and a salinity of 29.0‰, i.e. sea-water. The greater proportion of sea ice in Nansen Ice Plug than elsewhere is probably a reflection of the greater mixing of fresh water and sea-water below the ice, due to Nansen Sound being a through channel with a greater sea-water flux.

The stratigraphy of ice cores 85–3 and 85–4 (Fig. 4) is an interesting case and illustrates well the spatial and temporal variations in ice-growth history and water properties below the ice. The stratification can be interpreted in terms of three basic stages in the evolution of the ice cover in relation to the water below the ice (Fig. 6). It is evident from the surface topography of rough, hummocky, weathered ridges of the MLSI at the front of Ward Hunt Ice Shelf that the ice was once deformed (Reference Jeffries, Sackinger, Shoemaker, Sackinger and JeffriesJeffries and others, 1988a). Stage 1 (Fig. 6), represented by the ice from the surface to 3.55 m, was the deformation of a relatively thin sea-ice sheet that grew from sea-water with high (more positive) δ ¹⁸O values. Stage 2 was a period of brackish sea-water stratification below the depression, while the hummock remained underlain by sea-water; hence, the contrast in ice δ ¹⁸O values in the depth interval 5.50–7.80 m (Fig. 4). The brackish sea-water stratification was confined to the inverted depression at the bottom side, but it is apparent from the interfingering of sea and brackish ice in the central section of core 85–3 that there were regular fluctuations in the water salinity, from sea-water to high-salinity brackish water. During stage 3, brackish sea-water stratification was more extensive; low-salinity brackish water only occurred in the inverted depression, while high-salinity brackish water-sea-water stratification occurred at regular intervals below the hummock, similar to stage 2 below the depression. The most likely explanation for the interfingering of thin sea and brackish ice layers in each core is ice growth associated with the seasonal variability of water salinity and structure below the ice, i.e. annual layering in the ice. At the multi-year scale, stage 3 probably represents a 5–6 year period of greater fresh-water penetration below the ice, perhaps due to warmer summers, and/or less mixing of fresh water and sea-water, than occurred during the 5–6 year period of stage 2. The δ ¹⁸O record in ice cores 85–3 and 85–4 indicates that, once deformation has occurred and the ice has consolidated, the deformed MLSI also provides a stable environment for under-ice water stratification and associated ice growth.

Fig. 6. Three stages in the evolution of MLSI at the front of Ward Hunt Ice Shelf interpreted from ice-core (85–3, 85–4) δ ¹⁸O values as a proxy record of under-ice water stratification. Stages 2 and 3 represent summer conditions.

5. Summary And Conclusion

This study has shown that the δ ¹⁸O values in ice cores drilled in multi-year land-fast sea ice can be used to identify fresh, brackish, and sea ice, provided the salinity–¹⁸O relationship in the coastal waters is known. Using this approach, it has been shown that MLSI on the north coast of Ellesmere Island, Canada, contains a significant amount of brackish ice, and that brackish ice and sea ice have quite different salinity and δ ¹⁸O populations. The brackish ice growth probably arises from the extensive brackish sea-water stratification that must exist beneath the ice, frequently in the summer and, in some cases, year-round. The maintenance of the under-ice water stratification and, therefore, the growth of brackish ice can be attributed to the stability of the fast ice which remains in place for many years and causes little disturbance of the water below the ice during that time. A further factor in the growth of brackish ice is the likely presence of an inverted topography at the bottom side which will tend to trap brackish water year-round in depressions.

Measurements and observations of brackish ice in the field and laboratory remain few. Further study is required before a satisfactory explanation for the salinity difference between brackish and sea ice is obtained. The MLSI in this region is a natural laboratory for the study of brackish ice that has grown by congelation and frazil/platelet mechanisms from water of variable salinity. Additional studies are needed and should include structural-crystallographic analysis in order to provide some insight into the questions of crystal structure and sub-structure, ice-growth processes and the inclusion of brine into the ice, and perhaps also how much of the low-salinity ice is contributed by melting and refreezing of the MLSI itself.

Acknowledgements

Since 1982 ice-core drilling on the north coast of Ellesmere Island has been made possible by the generous support of many organizations. M.O. Jeffries gratefully acknowledges the support of the following: Department of Geography, University of Calgary; Polar Continental Shelf Project (G.D. Hobson, Director); Defense Research Establishment Pacific; Arctic Institute of North America; Gulf Canada; Dome Petroleum; Petro-Canada. H.R. Krouse acknowledges NSERC Canada funding of the Stable Isotope Laboratory at the University of Calgary. The support of the Geophysical Institute, University of Alaska (S.-I. Akasofu, Director), and the U.S. Department of Energy (Morgantown Energy Technology Center) is also gratefully acknowledged. Our thanks to O.P. Nordlund for his assistance, and the AES Weather Station in Resolute Bay for allowing us to store our ice cores in their freezer. Dr W.F. Weeks read the manuscript and offered a number of helpful suggestions for changes and improvements.