Accumulation

The core results, which represent the period 1962–73, have been plotted and contoured in Figure 2; additional data have been incorporated from sources referenced in the figure caption. Accumulation-rate values from the weather stations have not been included as they universally give anomalously low values. Reference Walker, Lake, Weller and BowlingWalker and Lake ([^c1975]) have discussed this same problem at length, and they also concluded that the weather-station values seriously underestimate the snow accumulation. A similar map has also been compiled from the surface snow-layer data. The mean value for each traverse has been plotted at the traverse’s geographical mid-point (Fig. 3). In this case, the weather-station snow accumulation values fit into the general pattern well, and this may be due to improved methods of accumulation measurement at the stations introduced prior to 1973.

Fig. 2.

Snow accumulation (in g cm^-2 year^-1) for the period August [962 to August [973 in the eas tern Queen Elizabeth Islands as calculated from measurements made down to the 1962 summer melt-layer in April and May 1974. Bold-faced and Serson (1973) and bold-faced values undefined are from Reference MüllerMüller and others (1977). The isonyets on Greenland are from Mock (1967); the broken lines are interpolations.

Fig. 3.

Snow accumulation (in g cm^-2 year^-1) for the period August 1973 to June 1974 as calculated from snow surface measurements in April and May 1974. Underlined bold-faced values are from the weather stations; bold-faced values are from F. Müller (personal communication).

The regional accumulation-rate distribution of the Queen Elizabeth Islands is very similar between the two maps (Figs 2 and 3). The two dominant features are: a dry area centered over western Axel Heiberg and north-central Ellesmere Islands, and a moist area around northern Baffin Bay. I will now consider the Baffin Bay area. The high accumulation rate on these slopes was first discovered during studies of the Devon Island ice cap by Reference KoernerKoerner (1966) and is now the subject of intense studies by F. Müller and coworkers (personal communication). Reference MockMock (1968) also found very high accumulation rates in Greenland on southerly slopes facing Baffin Bay. The area, which we might consider bounded by the 40 cm isohyet (Fig. 2), extends on the Canadian side of Baffin Bay from the top of the ice fields and ice caps facing Baffin Bay down to sea-level (Figs 2 and 3). The 1973–74 accumulation rate for Carey Øer in Baffin Bay (personal communication from F. Müller) is lower than that of the surrounding ice-cap slopes which, if representative both spatially and temporally, means that Baffin Bay itself is not part of the high-accumulation area. In addition, accumulation measurements on Bylot Island, made in 1975 by the author, showed that in that year the island was outside the high-accumulation area. The general appearance of the glaciers with a generally high ice-edge along the northern coast of Bylot Island, suggests this is typically the case. The high-accumulation area does not, therefore, extend south of Devon Island. The area of highest accumulation on the Greenland side of Baffin Bay appears to be on the south-facing slopes below Camp Century where accumulation rates reach 80 g cm^–2 year^–1 (Reference MockMock, 1968); the detailed pattern between Inglefield Bay and Inglefield Land is not known, but in general the isohyets show no unusual trends between the Queen Elizabeth Islands and Greenland (Fig. 2). The boundary of the high-accumulation area on Ellesmere and Devon Islands is quite sharp in places although the smoothing introduced by the contours in Figures 2 and 3 masks this effect. The steepest part of the gradient is on the uppermost parts of the shadowed west-facing slopes of central Ellesmere Island. For example, measurements of the current balance-year layer in May 1966 showed there was a decrease by 50% in the accumulation-rate on the west side of the central Ellesmere Ice Cap within 2 km of the crest. This emphasizes that the main mechanism in producing the high accumulation-rate is an orographic one, but this will be discussed more fully later. On the southern side of the Thule peninsula in Greenland (Figs 1 and 2) the accumulation gradient was 0.4 g km^–1 for the 3 year period 1962–65 (Reference MockMock, 1968). This is especially steep.

There are substantial variations in the accumulation-rate within the high accumulation region, the most important of which, ignoring the complex topographically induced variations, is a general decrease in accumulation with increasing distance from Baffin Bay. This is not always a linear decrease, however, as repeated profiles from the south-east side of Devon Island ice cap in June 1962, 1963, and 1965 (Reference KoernerKoerner, 1966) and 1967 and 1972 (Fig. 4) have shown. In that area maximum accumulation occurs over a broad zone below 900 m a.s.l.; above that elevation there is an inverse relationship between elevation and accumulation-rate each year which may be a distance-from-source effect. This pattern indicates that most of the precipitation is from low cloud bases although, on Devon Island at least, the ice cap’s own circulation may be sufficiently strong above 900 m a.s.l. to restrict the effects of Baffin Bay- steered cyclones. It is tempting to attribute the high accumulation on the ice caps facing Baffin Bay to the effect of the polynya there; this is usually termed the North Water. The question is whether the presence of a low-pressure cell in northern Baffin Bay draws in moist air masses from further south to deposit moisture on the surrounding ice caps, or whether the moisture is from the polynya itself. The answer to this question has fairly important paleo-climatic implications as it seems highly likely that the polynya did not exist during the Wisconsin period (Reference KoernerKoerner, 1977[b]).

Fig. 4.

Snow accumulation (in g cm^-2 year^-1) for the south-east side of Devon Island for the period August 1971 to June 1972.

Moisture advection into the Baffin Bay area is greatly enhanced by the thermal effect of open water and/or thin ice on air masses. This effect draws and steers cyclones to increase convection and hence precipitation. However, Reference Schriber, Schriber, Stauffer and MüllerSchriber and others (1977) using local δ¹⁸O data calculated that 17% of moisture deposited on Coburg Island was from Baffin Bay itself for the period November 1972 to May 1973. The computations were based on a calculated value for moisture entirely of Baffin Bay origin (δBB) (Reference Schriber, Schriber, Stauffer and MüllerSchriber and others, 1977) and the relationship

where ω is the percentage of moisture from Baffin Bay, δc is the δ value of precipitation with no local contribution (taken from an empirical relationship between measured values of temperature and δ at Coburg Island), and δs is the measured δ value of precipitation on Coburg Island. Koerner and Russell (in press), using the same approach, calculated a 26% contribution from Baffin Bay to sea-level coastal accumulation using their own δ¹⁸O data from the south-east side of Devon Island representing the period August 1971 to late June 1972. They also calculated that at the top of this ice cap the local contribution to moisture was only 8%. Suffice it to say that the fractional oxygen isotope content of the accumulation indicates that a substantial contribution is made to ice-cap accumulation in this area by Baffin Bay.

The accumulation-rate difference between the high-accumulation area and that immediately beyond it (and I would stress here that while there is a general accumulation-rate decrease from Baffin Bay westwards and north-westwards there is a very steep isohyetal gradient along much of the upper and therefore western edge of the high accumulation area) is from an average of 45 to one of <32 g cm^–2 year^–1; i.e. a drop of about 30%. Thus, while the possibility of local, i.e. Baffin Bay moisture being deposited further, afield cannot be discounted, it is possible that the excess accumulation on the slopes facing Baffin Bay is largely from the bay itself.

There is also a slightly higher accumulation-rate along the north-west coast of Ellesmere and Meighen Islands compared to the interior regions further to the south-east. This may be due to the adjoining Arctic Ocean acting as a moisture source or an orographic effect causing precipitation of moisture originating from a much more distant source. The accumulation increase does not of itself prove the Arctic Ocean is a moisture source, but I will return to this when discussing the oxygen-isotope results.

Whatever the cause, however, the accumulation increase is much more restricted than that around Baffin Bay, so that the general accumulation-rate decrease into the dry Ellesmere interior is only in the order of 5 g cm^–2 year^–1 compared to 30 g cm^–2 year^–1 from Baffin Bay. Reference KoernerKoerner (1973) in fact showed there was a significantly higher accumulation-rate around the periphery of the Arctic Ocean ice pack for some 400 km offshore. These increases were of the order of 2.5 to 5.0 g cm^–2 year^–1 and were attributed to greater cyclonic activity there compared to the central Arctic Ocean. The much more rapid increase along the north-west Ellesmere coast must depend on orographic uplift.

The dry area, according to the present data, centers around the western Ellesmere fiords and the Nansen and Eureka Sounds (Figs 1, 2, and 3). The low accumulalion-rates here are due to the shadowing effect of the surrounding mountains with Axel Heiberg Island reaching 1850 m to the west, northern Ellesmere Island 2000 m to the north, and central Ellesmere Island 1800 m to the east. The area is also separated from southern moisture sources by high-lying areas on south-west Ellesmere and Devon Islands, so that air masses generally descend into the area and heat adiabatically to form dry rather than moist air.

Over many areas of the Earth’s surface there is a positive relationship between precipitation and elevation. Such a situation does not appear in the present data from the Queen Elizabeth Islands (Table I). In the few cases where there is a weak but significant relationship between accumulation and elevation it is probably due to wind-drift and removal of snow from the lower elevations. However, the surface-snow samples represent only a 10 month period which may introduce excessive noise into any such statistical analysis. This is not the case with the core results which represent an II year period (August 1962 to August 1973). A multiple regression analysis between these data and latitude in degrees and tenths (z) and elevation in meters a.s.l. (h) gives the relationship

where c is the accumulation in g cm^–2 year^–1. Equation (2) (where the coefficient of correlation r = –0.62 significant at the 1 % level) confirms that there is indeed very little relationship between elevation and accumulation which, overall, is an inverse one. On the other hand, there is a much stronger relationship with latitude which, however, is another expression of what will later be shown to be a prime relationship between accumulation-rate and distance from moisture source and/or distance from the first orographic uplift of moisture-laden air masses.

Table I.Correlation coefficients r and significance levels for ¹⁸O and accumulation Versus Elevation on Various Traverses in the Eastern Queen Elizabeth Islands for the Period August 1973 to May/June 1974.

Ablation

At present the Polar Continental Shelf Project continues to measure the mass balance of the ice caps on Melville, Meighen (Reference ArnoldArnold, 1965; Reference PatersonPaterson, 1969), and Devon Islands (Reference KoernerKoerner, 1970). From these data, and the work of others, accumulation and ablation values for the Queen Elizabeth Islands have been collated (Table II). Because of the diversity of glacier types, aspect, albedos, and periods it is difficult to detect any regional trends within these data. The only significant regionalism in the data is the very low ablation rates on the Ward Hunt ice rise and the Meighen Ice Cap (Table II). They are in fact the only sizeable ice masses in the Canadian Arctic existing entirely below 300 m a.s.l. Reference Hattersley-Smith and SersonHattersley-Smith and Serson (1970) attributed the low ablation on the Ward Hunt ice rise to high prevalence of fog there; this is an Arctic Ocean effect. Alt ([1975]), in a more sophisticated analysis of weather and mass balance, considered the Meighen Ice Cap to owe its existence at such a low elevation to its proximity to the Arctic Ocean which, while it may not act as an important heat sink in summer, does not act as a heat source either. The surrounding land masses, on the other hand, once snow-free, do act as a heat source to air masses moving over them thereby limiting the low-ablation region to a narrow zone along the Arctic Ocean periphery.

Table II.Accumulation and Ablation Rates for Various Periods in the Canadian Arctic

Holmgren (1971) considered that Baffin Bay cyclones by advecting cold air-masses into the high-accumulation region promote lower ablation rates on the lower ice-cap slopes and bring accumulation in summer to the upper parts of the ice cap. Supporting this conclusion is the absence of any increase in ice layering in the firn moving eastwards along the summit ridge of the Devon Island ice cap despite increasing 12 m firn temperatures in that direction (Reference KoernerKoerner, 1966). This combined ice-layer/temperature evidence suggests an overall Baffin Bay warming effect for the whole year but a specific cooling effect for the summer season. On the other hand, Reference KoernerKoerner’s (1970) ablation/elevation profiles from the Devon Island ice cap show no sign of a Baffin Bay effect. However, these profiles do not run directly down to Baffin Bay but south to Lancaster Sound and the south-facing aspect of these glaciers, which increases radiation input, may compensate for any Baffin Bay effect. They are probably not representative of the true Baffin Bay slopes.

To examine the regional variation of melting further, values for the amount of ice-layering in each core for the II year period 1962–73 have been studied. Ice-layer thicknesses were summed for the cores between the 1962 summer surface and the 1973 summer surface. The thicknesses (L _h) were corrected (L _c) to allow for initial density variations from site to site according to

where ρ is the mean density of the surface snow at that site, ρ ₀ = 0.9 Mg m^–3, and L _h is the thickness of the ice layer. This correction to the ice-layer thickness gives the amount of water refrozen to form each ice layer. I should stress here that the ice-layer amount is not a measure of the total melt as this includes melt water formed and refrozen in a more uniform way throughout the firn pack and not in layer form. However, a previous study by Reference KoernerKoerner (1977[b]) has shown that there is a strong relationship between the amount of ice layering formed each summer at the top of the Devon Island ice cap and the annual mass balance for the north-west side so that it is pertinent to include the core ice-layer data in the present discussion of regional variation of ablation.

The ice-layer values are shown in Table III. In this case all the sites are well above the firn line, and are similar in aspect and albedo; the data also refer to the same time-period. The differences between the sites are therefore reduced to those of elevation and geographical position. A multiple-regression analysis between total melt for the II year period in grams water equivalent (m), latitude in degrees and tenths (x ₁), and elevation in meters a.s.l. (x ₂) gives

Table III.Total Melt-Water Refrozen In Ice Layer Form At Each Core Site For The Period August 1962 To May/June 1973 Eastern Queen Elizabeth Islands

This equation (where the coefficient of correlation r = 0.89, significant at the < 1% level) shows that elevation is the dominant variable, with latitude accounting for only 13% of the computed variation and elevation for the rest. From it one can calculate the dry-snow-line elevation, i.e. where m = 0. We calculate 2240 m a.s.l, for lat. 75° N., 1885 m a.s.l. for lat. 80° N., and 1550 m a.s.l. for lat. 85° N. These values are reasonable as Benson (Reference Benson1962, p. 93) found the dry-snow line in Greenland at elevations of 2 360 in and 1720 m at lat. 75° N. and lat. 80° N. respectively.

However, the analysis gives much too low a melt value for sea-level, i.e. less than 2 190 kg m^–2 at lat. 75° N. compared to a measured ablation rate at sea-level on Devon Island of over 10 380 kg m^–2 for the same II year period. To bring the two sets of sea-level values into agreement would mean assuming that the ice layers represent only about 20% of the total melt occurring each summer at each core site; this is unrealistic.

As the core sites are all well above the firn line in each area, values for total melt from the Devon Island studies for the four summers 1962, 1963, 1965, and 1966 have been used to extend the melt/elevation relationship down to sea-level. These values were calculated below the equilibrium-line elevation by adding snow melt to ice melt, between the equilibrium-line elevation and firn line by adding snow melt to subsequent superimposed-ice melt, and above the firn line by calculating a total melt under the assumption that the snow-density change in the active melt layer (identified by its grain shape) is entirely due to melt and not to compaction following grain-shape changes. Ignoring compaction is believed to introduce only a small error.

The values regressed against elevation show a good fit to an exponential equation of the form m = a exp (— bh) where m is inelt rate in mass per unit area per unit time and h is elevation in meters above sea-level. The r coefficients and curves are all shown in Figure 5. Such a relationship explains why the linear fit between the melt values and elevation gives such a low sea-level value; its implications will be discussed later.

Fig. 5.

Melt total for the north-west side of Devon Island for the balance years, 1962, 1963, 1965, and 1966.

Oxygen isotopes

Reference Dansgaard, Dansgaard, Johnsen, Clausen and GundestrupDansgaard and others (1973) have shown a strong relationship between the fractional ¹⁸O content of snow (δ) and elevation (h) on Greenland. Lorius (in press) found a much more complex δ distribution in parts of Antarctica largely due to less uniform atmospheric circulation there and to re-sorting of snow by wind-drift. Our δ picture (see Table I) is more similar to the Antarctic than the Greenland one.

The present data are derived from samples each of which covers only a 10 month accumulation period. D. Fisher (personal communication) found that the standard error of the δ value of any one year’s snow from the Devon ice cores is of the order of 1%₀. As studies by Reference Dansgaard, Dansgaard, Johnsen, Clausen and GundestrupDansgaard and others (1973) in Greenland and also R. M. Koerner and R. D. RussellFootnote ^* on the south-east side of Devon found a δ gradient with elevation of about 0.6%₀ per 100 m, we should be able to see a similar relationship in our data if it exists despite the 1%₀ standard error of each sample. However, of the various profiles only three show a significant relationship between δ and elevation (Fig. 6 and Table I). The most highly significant relationship is for profile number 3 which extends inland from Baffin Bay. Its gradient of 0.64% per 100 m is similar to the 0.65%₀ per 100 m for the Baffin Bay side of the Devon Island ice cap sampled in 1972 (R. M. Koerncr and R. D. RussellFootnote ^* ) and the 0.66%₀ of Reference Dansgaard, Dansgaard, Johnsen, Clausen and GundestrupDansgaard and others (1973) in Greenland. Profile number 4, which has a barely significant relationship between δ and elevation, lies 150 km to the north of number 3 and on the edge of the high Baffin Bay accumulation area. Profile 1 similarly has a barely significant relationship between δ and elevation. All three profiles must owe their relationship to the concurrence of precipitation events with winds moving directly up-slope, i.e. adiabatic cooling. We may consider that the more prevalent the occurrence of orographic precipitation the stronger the relationship. As the strongest relationships occur along the east side of the Devon Island ice cap where R. M. Koerner and R. D. RussellFootnote ^* found an r value for δ/h of 0.97 (significant at the <0.1 % level) and the Baffin Bay side of the south-east Ellesmere Ice Cap, these two areas must get most of their snow accumulation from orographically induced precipitation.

Fig. 6.

¹⁸O profiles for the various traverses in the Queen Elizabeth Islands for the period August 1973 to May/June 1974. For the location of each traverse by number refer to Figure 1.

The question immediately arises as to why profiles 2a and b (Figs 1 and 6) do not show a significant relationship when they also lie in the same high-accumulation area. It seems to be simply a matter of direction; the air masses move westward from Baffin Bay and up-slope and they cross profiles 2a and b without further uplift so that precipitation comes from a single condensation level.

The remaining profiles are similar to 2a. However, while they too show no significant relationship between δ and h, the δ distributions do not at first sight seem to show a random, but rather a uniform, distribution with changing elevation. For example, most of the δ/h gradients are very similar with slightly positive values. So, to see whether the distributions between each profile are significantly different, a variance test was made. This test showed that the profiles divided into three separate groups. Within each group there is no significant difference in mean δ between each profile, but between each group the difference is significant at the <1% level. The distributions, then, are not random. First, the lack of a δ gradient with elevation may be explained in terms of precipitation coming from a single condensation level. This phenomenon has been described before in mountainous regions (Reference Ambach, Ambach, Eisner and PesslAmbach and others, 1972). Secondly, the grouping is a consequence of increasing distance from the moisture source as the air mass is cooled isobarically and the heavier isotope depleted preferentially. Thus profiles 2a and b show the least negative values as they lie closest to the moisture source of the air mass as it penetrates the Queen Elizabeth Islands from the south-east.

The profile most distant from the source, number II (Figs 1 and 6), does not, however, show the most negative δ value, but rather profile number 10 on the “dry side” of Axel Heiberg Island; this warrants discussion. We have already seen that snow accumulation increases slightly with approach to the Arctic Ocean from the south-east. The fact that the west side of Axel Heiberg Island is significantly warmer isotopically indicates the Arctic Ocean as a moisture source, albeit a small one. This is because any air mass that has moved from more southerly latitudes and cooled isobarically en route should show the most negative values along the Arctic Ocean littoral. The Axel Heiberg Island profiles strongly suggest this is not the case; so that some of the west Axel Heiberg Island moisture must be of Arctic Ocean origin. However, the west Axel Heiberg profile has a mean δ value only 3%₀ more positive than the east Axel Heiberg profile, suggesting the Arctic Ocean contribution is quite small and restricted to a fairly narrow zone along its periphery. Alt’s ([1975]) work on the meteorological conditions affecting nearby Meighen Island brought her to a similar conclusion.

We may further depict the regional δ variations in two ways. First, the average δ values have been plotted at the mid-point of each traverse and isolines drawn (Fig. 7). Secondly, the same mean S values have been plotted as a function of distance from source in Figure 8. The source-edge from which the distance has been measured is considered as a smooth coastline along eastern Devon and south and central Ellesmere Islands. This coast is close to the open-water edge during the winter months so that it represents the edge of the last moisture source available to the air mass before it crosses the Queen Elizabeth Islands. Both the map (Fig. 7) and graph (Fig. 8) show the relationship between δ and distance from source (x) which a regression analysis shows to be

where δ is measured in parts per thousand and x in kilometers. The correlation coefficient (r) for this relationship is 0.78 (significant at the <1% level). This means 61% of the 8 distribution over the Queen Elizabeth Islands can be explained in terms of distance from a south-east source.

Fig. 7.

¹⁸O values for the eastern Queen Elizabeth Islands for the period August 1973 to May/June 1974. Bold-faced values are from Müller and others (1977). The contour interval is —2.3 δ units.

Fig. 8.

Relationship between ¹⁸O and the distance from moisture source, eastern Queen Elizabeth Islands for the period August 1973 to May/June 1974.

The overall picture, therefore, from the ¹⁸ O data is one of predominantly orographic precipitation along the slopes facing Baffin Bay but a general isobaric cooling as the air masses move north-westward over the rest of the eastern Queen Elizabeth Islands to bring precipitation from one condensation level and continuously deplete the moisture of ¹⁸O in the same direction. There is also evidence for a secondary but minor moisture source in the Arctic Ocean.