Arctic Ice Cores: Putting Present Climate into Perspective

Roy M. Koerner




Glaciers and ice sheets throughout the world are strongly related to climate. In the past, during periods of intense cold, ice masses have expanded to cover large parts of North America and Europe. Mountain glaciers have also expanded during these times, and it appears from the ocean record that over the past million years the more normal condition is for much colder conditions than today, with ice sheets covering large parts of Canada. However, these are the most dramatic changes, and it must be realized that climate changes through a wide range of wavelengths and amplitudes, from millenial to decadal and from fractions of a degree to several degrees. Ice sheets contain records of all these changes.


Ice-Caps and Glaciers

 An ice or snowfield becomes "permanent" once it is able to survive each summer. To do this, the snowfalls of winter must exceed the ice-melt of summer (Figure 1). Snow survives at high elevations but melts along with some of the underlying ice in the lowermost regions. In the more dynamic ice-caps-those with outlet glaciers reaching the sea-some ice may break off and melt in the ocean.

 Ice formed in the accumulation area moves down toward the underlying rock as it is buried under subsequent snowfalls. It also moves out toward the margin. The ice in the ablation zone also moves toward the margin, but its vertical component is upward, replacing the ice that melts. The thickness of the annual snow layer at the surface varies from about 1.5 metres close to Baffin Bay to as little as 0.15 metres on parts of the more northerly ice caps on Ellesmere and Axel Heiberg islands. In the ablation zone, between 1 metre and 3 metres of ice melts at sea level each summer. The equilibrium line, where snow accumulation equals snow melt, varies in elevation from 900 metres to 1300 metres above sea level in the arctic islands.


Ice Cores

 Because the vertical component is almost entirely downward at the top of an ice-cap, cores for climate studies are best taken there, as they suffer minimal disturbance by flow over irregular bedrock. Such cores contain records of snowfall beginning with the time the ice-cap began its growth and extending to the present day. Everything that falls with the snow, including dust, pollen, and soluble salts is gradually buried. Several surface-to-bedrock cores spanning a time interval of about 100 000 years over a depth increment of 100 metres to 300 metres have been drilled in the arctic islands.

 As it is buried by subsequent snowfalls the surface annual layer thins by compression under increasing thicknesses of overlying snow until it becomes ice at a depth of about 60 metres. At greater depths the layer continues to thin due to ice movement. In an ice-cap 150 metres thick an annual layer at 135 metres is about 10 000 years old and less than 1 millimetre thick. Thus a core cannot be dated by simply dividing the core length by the surface annual layer thickness. Instead, time-scales are determined by detecting seasonal swings in the concentration of micro-particles or ions in the ice, and by picking up high levels of acidity in the cores attributable to well-known, dated volcanic eruptions. These two methods date the ice back to about 7000 years. Beyond, that, the time-scale becomes increasingly inaccurate. We can only get an approximate date for any particular level by comparison with the better-dated Camp Century core in Greenlandl or with the fairly well-dated ocean cores2 .

Perhaps the most important climatic indicator in the ice cores is the oxygen isotope ratio or "delta-value". The more negative the delta-value, the lower the temperature the snow formed at. Figure 2 shows a delta-value profile for a core from the Agassiz Ice Cap on north-eastern Ellesmere Island. The core is from the top of the ice-cap at an elevation of 1702 metres above sea level.

 The profile can be broken into three major sections: an uppermost section (O to 112 metres) with "warm" delta-values; an intermediate section (112 metres to 118 metres) with very "cold" delta-values; and a lowermost section (118 metres to bed) where the delta-values gradually increase toward the bed. The uppermost section was deposited during the Holocene period 10 000 to 70 000 years ago. The temperatures during this period were several degrees colder than today. The lowermost section was deposited during the last interglacial period in warmer conditions than today but also during the "cooling off' transition period which brought the Wisconsin ice-age. When the delta-profile is examined in more detail, the values suggest that, once the ice-cap began its growth, the air temperature above gradually cooled, right up to the main part of the ice-age. It appears that the ice-cap started its growth sometime during the last interglacial period. Any icecap from the previous ice-age must have melted completely during the same interglacial period. There is other evidence in the core (such as pollen concentrations and the texture of the ice itself) to support this interpretation. The most recent 10 000-year section of core shows a period of maximum warmth following close on from the dramatic emergence from ice-age conditions. The climate has been cooling ever since.

 There are a few important points about these profiles relevant to today's debate about the direction in which climate may be heading.

Our view of today's climate depends very much on our terms of reference. Predicting climate is often considered, particularly by the layperson or politician, as the end-product of palaeoclimatic research. So a common question is: "What is the climate going to do in the future?" One has to have some kind of answer ready, and it must be based on a long time-series of high-quality proxy climate data. In our case, delta-value has been used, as it can be reproduced by adding two to four sine waves of different period, amplitude, and phase. Extrapolation can then be made by adding this sine-wave sum or synthesis. The calculations suggest a slight cooling over the next 50 years of about 0.5C to 1.0C annually. Summer cooling in the next 50 years (based on the amount of melting as evidenced by clear ice layers in the cores) may be slightly more than this. Although the standard error of this forecast is of similar magnitude to the change itself, other calculations based on delta-values from Greenland cores give similar "forecasts" to ours. Anthropogenic effects apart, we should not expect the open-sea ice conditions of the 1940s and the 1950s to occur again. Perhaps we should plan for the most pessimistic side of climate in the arctic islands and take the coldest summers from the past 40 years of record as a practical norm.

 Computer models suggest that industrial atmospheric pollutants are going to profoundly affect our climate, if not in our lifetime, at least in that of our children. Just over a decade ago it was suggested that industrial pollutants could cause a premature slip into the next ice-age; this was the "human volcano". Starting with the premise that volcanic activity causes global cooling, it was argued that dust from industrial activity would have a similar but persistently increasing effect. Our cores contain evidence of volcanic activity, mainly from eruptions in Alaska, Yukon and Iceland.

The eruptions appear as high levels of acidity in the cores. We have been unable, up to now, to demonstrate either a cooling or warming effect of volcanic activity by relating the high acid levels to changing oxygen isotope values in the cores. However, Hammer and others3 concluded from work on Greenland cores that high acid background levels in the ice are associated with slightly "colder" delta-values. In other words, climate may change in response to persistent, rather than peak, levels of volcanic activity maintained over many years.

The "human volcano" hypothesis has now been superseded by that of the effect of increasing levels of "greenhouse gases", particularly carbon dioxide. Global circulation models predict dramatic warming over the next 100 years, particularly in the High Arctic. Ice cores have played their part here also. Greenland cores have shown lower levels of carbon dioxide in ice deposited during the last glacial period. However, this is a chicken-and-egg problem. Do the carbon dioxide changes precede or follow the delta-value changes? Are they a cause or an effect?


Climate Monitoring

 Our cores have not yet been analyzed for carbon dioxide concentrations. However, the arctic island ice-caps are being used in another way for climatic research: to determine the way climate is heading. Each summer, melting occurs on the ice-caps. At sea level, 2 to 3 metres of ice may melt and run off onto the surrounding land or into the sea. Melting even occurs at the tops of the ice-caps in most years, although it may be for only a few days. Each spring on the Melville South, Meighen, Devon, and Agassiz ice-caps, measurements are made of the amount of ice and snow that melted the previous summer and the amount of snow remaining from the previous "year" ending with the last melt of summer. This gives a time series of ice-melt and snow accumulation for a "balance-year" beginning one August and ending the next. The Meighen and Devon ice-cap records now cover 27 years. No significant trends of changing snow accumulation or ice-melt rates can be found in the data, although large year-to-year variations may be burying any slight trend that does exist.



 The ice-core record in Figure 2 shows a large range of climatic conditions over the past 100 000 years. Compared to the last 10 years, temperatures show a range from about 10C to 15C colder 18 000 years ago through 3C warmer 7000 to 8000 years ago, to 5C warmer more than 100 000 years ago when the ice-cap began its growth. On a shorter time-scale, during the period of instrumental record, there is evidence from our cores of a 2.5C warming between about 1750 and 1950. The warming trend, which has occasionally been identified as carbon-dioxide induced, can be seen as part of a much longer natural one.

Although global circulation models all predict that greenhouse gases will eventually cause a warming of several C in the High Arctic, the amount of warming varies from model to model. However, the ice-core record suggests that the last interglacial, which may have seen the demise of all Canadian ice-caps, may be used to represent an environment in which the level of carbon dioxide is doubled. Various studies of the palaeo-environment from that period can provide valuable input into estimating the effects of the proposed warming. Data of this nature are already available, much of it within the Quaternary Environments subdivision of the Geological Survey of Canada. However, it must be pointed out that "greenhouse warming" would not persist long enough to completely melt away the ice-caps in the arctic islands. The closest scenario of a carbon-dioxide world might, therefore, be the early part of the last interglacial period when the high arctic ice-caps had still not melted completely.


Roy M. Koerner is a researcher with the Terrain Sciences Division of the Geological Survey of Canada, Department of Energy, Mines and Resources.



1. W. Vansgaard, H.B. Clausen, N. Gundestrup, C.U. Flan~ner, S J. Johnsen, P.M. Kristindottir, and N. Reeh. 1982. "A new Greenland ice core". Science, 218, pp. 1273-1277.

2. C.U. Hammer, H.B. Clausen, and W. Dansgaard. 1980. "Greenland ice sheet evidence of post-glacial volcanism and its climatic impact". Nature, 288, pp. 23~235.

 3. W.F. Ruddiman, A. McIntyre, V. Niebler-Hunt, and J.T. Durazzi. 1980. "Oceanic evidence for the mechanism of rapid Northern Hemisphere glaciation". Quaternary Research, 13, pp. 13-64.

"In This Issue..."