Polar ice sheets: actors in the climate system, archives of the past
Polar ice sheets are an invaluable archive of the past climate if one understands the physical and chemical processes that lead to the deposition, burial, conservation and transport of chemical species, air bubbles and dust in the ice matrix. In this chapter, I describe these processes in order to design an optimum method to track water isotopes in ice. I first explain the role and influence of ice sheets in the climate system, then introduce the ice core records from Greenland and Antarctica used in this study, review the assumptions used for dating the records and summarize the climatic insight they provided. Then I justify the use of water stable isotopes as a thermometer and consider the diffusive properties of firn and ice and their ability to preserve the water isotope record, which motivates our choice of a non-diffusive tracer transport scheme. Finally, I restate the physical laws that rule deformation and storage at depth of annual ice layers, present the principles of ice sheet modelling and close with a discussion of the climate parameterization used for this study.
Ice sheets and climate
Polar ice sheets: big, white and cold
A glimpse through the window of a plane during a transatlantic flight shows vast and intriguing, almost mystical, white fields beyond the reach but not the imagination of the common mortal. Vast, they are indeed. The Greenland Ice Sheet is over 2500 km long from South to North and close to 1000 km at its widest, for a total volume of 2.9 106 km3 (Letreguilly et al., 1991). The volume of the Antarctic Ice Sheet is almost ten times as large and the ice-covered region measures over 4000 km across. The gigantic glaciers of Antarctica and Greenland are called “ice sheets” because, with an ice thickness of 2.5–3.5 km and a spread of several thousands of km, their aspect ratio makes them comparable to the sheets of paper used to print this manuscript. Together, the Greenland and Antarctic ice sheets constitute by far the largest present freshwater reservoir and contain enough water to raise global sea level by 70 m if they were to melt, 61 m coming from the Antarctic Ice Sheet (Huybrechts, 2002) and 7 m from the Greenland Ice Sheet (Letreguilly et al., 1991). The large polar ice masses do not merely store water, they play an active role in the climate system.
Actors in the climate system
Ice looks white. A simple fact, but with crucial implications for the radiative budget of our planet, as most of the external energy brought to the Earth comes from the visible radiation of the Sun. Snow and ice have an albedo of 0.6–1, which indicates that they reflect most of the incoming solar radiation; thus, polar regions tend to cool the planet. The phenomenon is amplified in the winter time when large areas are covered in seasonal snow, leading to a positive feedback. The polar energy
sink also drives the general circulation of the ocean and atmosphere because of the large contrast in temperature between the poles and the tropics, which receive most of the solar energy because of the relative tilt of the Sun’s ray with the normal to the Earth’s surface: energy absorbed in the tropics is exported to higher latitudes through atmospheric and oceanic flow. The positive feedback on the radiative budget is also excited during glacial periods, when large ice sheets were stimulated to grow and cover extensive areas by a decrease in incoming solar radiations above the Arctic Circle, as explained by the Milankovitch theory.
Astronomic theory of climate change
The relative position and orientation of the Earth during its yearly elliptical journey around the Sun determines how much solar radiation is received at different latitudes during a year. The received radiation is controlled by three orbital parameters described in Fig. 2.3: the obliquity or tilt of the Earth polar axis relative to the elliptic plane, the precession of equinox that corresponds to the season when the Earth is closest to the Sun on the ellipse (perihelion), and the eccentricity that describes the degree of elongation of the ellipse. Tilt causes seasons because the hemisphere pointing toward the Sun receives more radiation. Change in tilt between 22 and 25 occurs with a 41 kyr (thousand years) periodicity and controls the intensity of seasonality: the larger the tilt, the larger the insolation contrast between the two hemispheres. Today the Earth reaches the perihelion in January, causing slightly milder winters in the Northern Hemisphere, as opposed to 11 kyr ago when perihelion was approached in July (19 kyr and 23 kyr periodicity). Eccentricity varies in 100 kyr and 400 kyr cycles and controls how the timing of perihelion combines with seasonality (tilt); thus, the relative severity of summer and winter time in each Hemisphere.
Milankovitch (1930) calculated the past position and inclination of the Earth and associated the changes in seasonality with the glacial–interglacial cycles that have characterized the Earth’s climate for the past 2 Myr. Ice ages, cold temperatures and large ice volume, correlate well with the 65 N summer insolation (Emiliani, 1955; Berger, 1978), as shown on the Vostok record Fig. 2.2; thus, small changes in insolation get amplified by feedbacks in the climate system that lead to a global cooling of 5 C and the inception and growth of large ice sheets that drew down sea level by 120– 135 m (e.g., Yokoyama et al., 2000). If insolation is indeed the “pacemaker of ice ages” (Hays et al., 1976), future glaciations can be predicted and the next one should not happen before at least 50 kyr (Berger and Loutre, 2002). However, the astronomical theory of climate does not tell how warm the climate is going to be nor how high sea level will rise over the next decades and centuries; thus, I will especially focus on the climate and behaviour of ice sheets during former periods that most resembled modern conditions and try to improve the understanding of glacial archives by predicting the conditions under which they were generated and preserved.
Figure 2.2: Vostok time series and insolation. Series with respect to time (GT4 timescale for ice on the lower axis, with indication of corresponding depths on the top axis) of: a, CO2; b, isotopic temperature of the atmosphere; c, CH4; d, 18Oatm ; and e, mid-June insolation at 65 N (in W m 2). From Petit et al. (1999).
From http://deschutes.gso.uri.edu/ rutherfo/milankovitch.html.
Ice sheets and fast climate change
Though the Milankovitch theory may well explain the inception and expansion of large ice sheets, isotopic and geomorphologic records display a high frequency variability during glacial periods that is not present in the insolation signal. For instance, deep-sea sediment records in the North Atlantic (Fig. 2.4) contain a series of layers that are rich in sediments also found in eastern Canada and poor in foraminifera, thus reflecting cold sea-surface temperature. These records suggest that intense and short-lived release of large armadas of icebergs released from the Laurentide Ice Sheet exported enough freshwater to shut down the thermohaline circulation of the ocean, effectively cooling down the climate in the North Atlantic region by 5 C. These Heinrich Events (Heinrich, 1988; Bond et al., 1992) are believed to result from surges or oscillations in the ice stream regime of the Laurentide Ice Sheet and have motivated the development of the UBC Ice Sheet Model used for my studies of Greenland (Marshall, 1996). These events reflect the dynamic role of internal ice dynamics on the global climate and stimulate further research to better understand the flow of ice sheets.
Figure 2.4: Ice sheet extent during glacial periods and location of sea-sediment cores containing ice rafted debris (IRD) carried by icebergs released from the Laurentide Ice Sheet. The thick solid line near coast lines maps the maximum limit of ice sheets during the last glaciation. Black patterns map the origin of IRDs. From Bond et al. (1992)
Ice core records
Accumulation of snow on glaciers and ice sheets builds a record of the atmospheric composition by locking water, air and particles in the ice matrix. In central regions of Antarctica and Greenland or in high mountain glaciers, cold summer temperatures limit any melting; thus, yearly snow and ice layers can be preserved, yielding a continuous record of past atmospheric conditions. Therefore modern ice sheets and glaciers constitute a wealthy and abundant archive for glaciologists interested in Earth’s climate history. The following review is inspired from Parrenin (2002).
A brief history of ice core drilling
The potential of the glaciological archive was understood about 60 years ago, but the remoteness and harsh climatic conditions of polar ice sheets and the resulting technical and financial burden have so far limited the use of the “glaciological mine” (Reeh et al., 1987) to a dozen ice cores. Scientific expeditions aimed at understanding the climate in Antarctica started in 1957–1958 during the International Geophysical Year, followed the next year by a 1400-km traverse organized by the Americans and a decade later by deep drilling operations.
The first deep ice core campaign was inaugurated in Northwest Greenland at Camp Century. Bedrock was reached in 1966, with a 1390-m deep core that spans 100 kyr of climatic history (Dansgaard et al., 1969). In Antarctica, 2000 m of ice containing 75 kyr of history were extracted at Byrd near the Ross and Amundsen Sea in 1968 (Epstein et al., 1970). In the early 1970’s, the Russians installed a station at Vostok in central east Antarctica, one of the coldest places on our planet with an average temperature of 55 C and extreme temperature of 89 C. Several ice cores were drilled and successively reached 500 m in 1970, 950 m in 1974 (Barkov and Gordienko, 1976) and much deeper more recently. Also in East Antarctica, the French extracted a 900-m core at Dome Concordia in 1978 (Lorius et al., 1979). Drilling returned to Greenland and in 1981, the Danes, Swiss and Americans pulled a 2000-m ice core at Dye 3 in the southern part of the ice sheet. The record contains over 100 kyr of climatic history (Dansgaard et al., 1982), but the lower part of the record is difficult to interpret because it is strongly affected by ice flow, an inescapable problem for flank flow sites, an issue I will try to address with three-dimensional modelling of ice flow.
Drilling resumed at Vostok in 1982 with a Russian–French association that benefited from American logistic support. High expectations were put on this operation because the low surface ac-cumulation rate, only 2 cm per yr, raising the possibility that the age of the ice might greatly exceed that for the previously drilled sites. Efforts were soon rewarded with a 2000-m ice core spanning 150 kyr of climate history (Lorius et al., 1985), providing the first detailed record reaching beyond the last glacial cycle and into the next one, embracing an interglacial period somewhat similar to modern conditions about 125 kyr ago, an epoch called Eemian or Marine Isotope Stage 5 (MIS 5) in reference to the chronology of marine isotopic records (Martinson et al., 1987). Drilling reached a final depth of 3623 m in 1998, just 80 m above the large subglacial Lake Vostok, with 3310 m of ice spanning 420 kyr of climatic history (Petit et al., 1999, and Fig.2.2), the bottom 300 m originating from basal freeze-on of water from Lake Vostok. The record thus contains four glacial–interglacial cycles and reaches MIS 11 (395–420 kyr BP), a long interglacial period of special interest because its astronomical parameters resemble modern conditions (Drowler et al., 2002).
In the early 1990’s, Europeans and Americans drilled two deep ice cores 28 km away from each other near the 3300-m-high Summit of the Greenland Ice Sheet. The European GRIP project reached the 3029-m-deep bedrock in July 1992 (GRIP Members, 1993), the American GISP2 project 3087 m the following summer (Grootes et al., 1993). Their deep ice from the last inter-glacial period led to much speculation about rapid climate variability. However, comparison of greenhouse gases with the Vostok record clearly proved the Greenland cores had been corrupted by strong flow disturbances that had mixed ice deposited during the Eemian (Chappellaz et al., 1997). A new drilling program aiming at better sampling Eemian ice in Greenland was started in 1996 at NorthGRIP, 300 km North of the Summit sites. Bedrock was finally reached in 2003 and appears to contain ice up to 123 kyr BP, that is ice reaching into a stable part of the Eemian (NGRIP Members, 2004).
Besides the Vostok program, other deep ice cores have recently been drilled in Antarctica. The Japanese went to Dome Fuji, 1500 km away from Vostok and closer to the Atlantic Ocean, and extracted 2503 m of ice before the drill got trapped into the ice. The ice core record showed three glacial cycles strikingly similar to Vostok, providing evidence of an homogeneous climate over most of Antarctica (Watanabe et al., 2003). Europeans started a new deep drilling project (EPICA) at the French–Italian base of Dome C (DC) in 1998. In 2003, they recovered 3190 m of ice spanning the longest-to-date ice-core record with up to eight glacial cycles of climatic history (EPICA community members, 2004).
Methods for dating ice
Although annual ice layers can preserve their original chemical content, the climatic history can be obscured by the thinning and deformation processes resulting from ice flow; thus, determining the age of ice can be a real challenge. Annual layers tend to merge at greater depth, especially in Antarctica where initial layers are already thin because of low surface accumulation. As previously seen, flow disturbances have also particularly affected sites like GRIP and GISP2. Each ice core record has its own flow and accumulation rate characteristics, therefore many strategies have been applied to recover the essential chronology of events.
I distinguish three main classes of methods for extracting the age of ice along a core: (1) Count-ing of annual layers, which is accurate (Alley et al., 1997) if layers are thick enough. It has been applied for dating the youngest 37.8 kyr of the GISP2 record (Meese et al., 1997) and from 8.2 to 11.5 kyr BP for GRIP (Johnsen et al., 1997). Accumulation rate is too low in central Antarctica for this method. (2) Use some level of comparison to a chronology or to dated events, either by referring to time markers (e.g., 10Be peak, 14C, or U-Th Genty et al., 2003) and other dated time series, or by applying an orbital tuning by linking the measured changes in the core to the well known insolation signal (Berger, 1978). Time markers are useful but limited to 10’s kyr. Orbital tuning assumes a linear response of the climate system to change in insolation, introducing a 6 kyr uncertainty on age (Parrenin, 2002). The method has been widely used both on ice and marine core records (e.g., Meese et al., 1997; Martinson et al., 1987; Bassinot et al., 1994). Chronologies can also be syn-chronized by correlating the isotopes of atmospheric oxygen, a method also compatible with marine records because the composition of the ocean and the atmosphere change simultaneously (Bender et al., 1994), or by methane concentration to compare Antarctic and Greenland records (e.g., Blunier and Brook, 2001), because methane is well mixed in the atmosphere. (3) Ice flow modelling, which requires assumptions concerning the simplicity of ice flow, on past surface elevation and origin of ice, and an accumulation model. This method is satisfying because it includes the physics of ice, but is highly dependent on the accumulation scenario, which is generally tied to the isotopic variations in the record. Such relations have been shown to fail at particular periods (Cuffey and Clow, 1997), explaining for instance the continuing debate on the chronology of Summit ice cores (Grootes et al., 1993; Meese et al., 1997; Johnsen et al., 1995, 2001; Shackleton et al., 2004). Vostok experiences all the difficulties of the modelling approach because of a distant origin of deep ice (Ritz, 1992), rough underlying topography (Siegert and Kwok, 2000) and possible hiatus in surface accumulation (Parrenin et al., 2001). The methods can be combined (e.g., GRIP-SS09 chronology, Johnsen et al., 2001) and Parrenin (2002) has recently developed a highly promising “federative approach” that picks up the best features of all these dating techniques to obtain the age–depth relationship of the Vostok, DC and Dome Fuji records.
What do we learn from ice cores?
Polar archives contain a rich and diverse record of atmospheric history. Water stable isotopic ratios in the ice reflect the temperature where the snow formed and second order information on the tem-perature of the oceanic site where the water first evaporated. Gas bubbles sample the composition of the atmosphere, for instance carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O), so that past greenhouse gases concentrations can be compared to temperature. The isotopic composition of gases also relates to past sea levels and biologic productivity through their 18O signature, to sources and sinks of methane through 13C and to the rhythm of climate change, especially through argon and nitrogen isotopes (40Ar,38Ar, 36Ar, 15N, 14N). Aerosols in snow and ice (Na+, Na2+, Cl , SO24 , NO23 , 10Be, dust) reflect the chemistry of the atmosphere, mark volcanic eruptions and measure cosmogenic production of radioactive isotopes.
Ice records have played a key role in understanding global climate change. In the early 1980’s, Vostok helped validate the orbital theory of climate that had already been observed in marine sedi-ments. Measurement of CO2 and CH4 in bubbles provided compelling evidence of the link between greenhouse gases and climate change (e.g., Barnola et al., 1987) and proved that greenhouse-gas levels in the atmosphere over the past hundreds of kyr had never been as high as those of the in-dustrial era, raising awareness of the impact of humankind on the climate. Ice cores from GRIP and GISP2 attested of the extreme sensitivity of the climate system, showing that rapid warmings of 10 C can occur in less than a human lifetime (Dansgaard et al., 1993; Severinghaus and Brook, 1999).
Past and future climate
Ice core records have opened a window on our past that helps scientists realize the fragility of the equilibrium between the different actors in the climate system. Though mankind is a recent player, looking at past periods can help scientists distinguish the natural variability of climate from changes induced by human activity. Interglacial periods with climate similar to modern conditions are particularly interesting. Given the difficulty of obtaining useful information on the distant past, great effort and hopes have been placed in the most recent interglacial. This was the motivation for drilling in Greenland, where the high snow accumulation rate offers high resolution data. A variety of records for that period indicate that the climate was warmer than modern conditions, sea level was 5 m higher and greenhouse-gas concentration was close to the pre-industrial period, 30% less than today. Therefore the current increase in temperature and greenhouse-gas emissions raises concern about the possible disintegration of the Greenland Ice Sheet or the West Antarctic Ice Sheet in the near future (Oppenheimer, 1998). I will assess the contribution of the Greenland Ice Sheet to past sea level rise by using an array of deep Greenland ice cores to constrain predictions of the past extent of the ice sheet. Because the current 10-kyr-long warm period, the Holocene, has already lasted nearly as long as the Eemian, the possibility of an imminent glaciation has been suggested, but the astronomical theory predicts that the Holocene could last for another 50 kyr (Berger and Loutre, 2002) if human intervention is taken out of the equation. Orbital calculations also indicate that the best analogue to the modern state is MIS 11, 400 kyr ago, an exceptional 30-kyr-long interglacial period when sea level was probably 13 to 20 m above current levels. Many questions arise from that discovery: was climate warmer than today? Why and how did sea level rise? Why did it last so long? These questions are very important, especially for those living in coastal regions. The recently drilled EPICA-Dome C record with its 800 kyr of history might help answer some of these questions. I believe that my method can also provide useful insight by combining ice core records to constrain the reconstruction of polar ice sheets during interglacial periods, test the sensitivity of ice sheets to climate warming and estimate the resulting effect of their full or partial desintegration on sea level and ocean composition.
Inferring past temperature from water isotopes
Stable isotopes of oxygen and hydrogen in water are widely used to reconstruct past temperature from ice and marine sediments because origin effects and temperature-dependent processes influ-ence the final composition of water isotopes found in polar ice, sea water, atmospheric vapour and meteoric water. This section explains the different processes and outlines the principles of the iso-topic paleo-thermometer.
Stable water isotopes
Oxygen has three stable isotopes, 16O (99.76% of all O), 17O (0.04%) and 18O (0.20%); hydrogen has two stable isotopes, 1H (H hereafter, 99.985% of all hydrogen) and 2H (or D, deuterium, 0.015% of all H), and one radioactive isotope, 3H (tritium). Together, oxygen and hydrogen form water, but they are also found in many other combinations in the Earth’s hydrosphere, biosphere, and geosphere. For instance, hydrogen constitues 75% of normal matter in the universe by mass whereas oxygen is the most abundant element in the Earth’s crust. Among the nine possible configurations of water, the most common are H162O, HD16O and H182O. The proprieties of these types of water are presented in Table 2.1.
Table of contents :
CHAPTER 1: Introduction
1.1 Context and objectives
1.2 Thesis outline
CHAPTER 2: Polar ice sheets: actors in the climate system, archives of the past
2.1 Ice sheets and climate
2.1.1 Polar ice sheets: big, white and cold
2.1.2 Actors in the climate system
2.1.3 Astronomic theory of climate change
2.1.4 Ice sheets and fast climate change
2.2 Ice core records
2.2.1 A brief history of ice core drilling
2.2.2 Methods for dating ice
2.2.3 What do we learn from ice cores?
2.2.4 Past and future climate
2.3 Inferring past temperature from water isotopes
2.3.1 Stable water isotopes
2.3.2 Isotopic fractionation
2.3.3 Fractionation during precipitation
2.3.4 Fractionation during evaporation
2.3.5 Rayleigh distillation
2.3.6 Principle of isotopic thermometry
2.4 Conservation of the water-isotope signal in snow and ice
2.4.1 Diffusion in the firn
2.4.2 Diffusion in ice
2.5 Modelling ice-sheet evolution
2.5.1 Dynamics of ice flow
2.5.2 Principles of ice sheet modelling
2.5.3 Parameterization of the climate over ice sheets
CHAPTER 3: Three-dimensional tracer modelling
3.2 Transport of provenance variables
3.2.1 Technical context
3.2.2 Tracer principles
3.2.3 Eulerian, Lagrangian, or semi-Lagrangian scheme?
3.3 Specificity of the age–depth relationship
3.3.1 Sources of error
3.3.2 A balance-based interpolation method
3.3.3 Age boundary conditions
3.3.4 Three-dimensional implementation
3.4 Depositional model
3.4.1 An archive like a “birth registry”
3.4.2 Present isotopic distribution
3.4.3 Past isotopic distribution
3.5 Practical application
3.5.1 Tracer transport flow chart
3.5.2 Construction of tracer stratigraphy
3.5.3 Global properties
3.5.4 Average isotopic concentration
3.5.5 Application of tracer modelling
CHAPTER 4: Global stratigraphy of the Greenland Ice Sheet
4.2 Model and experiment
4.2.1 Ice dynamics
4.2.2 Climate forcing
4.2.3 Practical details
4.3 Depositional provenance stratigraphy
4.4 Stratigraphy at ice core sites
4.4.1 Isotopic stratigraphy
4.4.2 Borehole catchment
4.5 Age of Greenland ice
4.5.1 Age of deep ice
4.5.2 Average age of the ice sheet
CHAPTER 5: Constraints on Greenland ice cores and glacial history
5.2 Validation process and effect of model parameters
5.2.1 Methodology and validation process
5.2.2 Best parameters
5.2.3 Sensitivity of parameters
5.3 Modelling the GRIP records
5.3.1 Age–depth profile
5.3.2 Borehole temperature
5.3.5 Ice origin and ice-divide migration
5.3.6 Constraints from GRIP
5.4 Constraints from other Greenland cores
5.4.2 Dye 3
5.4.3 Camp Century
5.5 Minimal configuration during the Eemian
CHAPTER 6: Global stratigraphy of the Antarctic Ice Sheet
6.1 Description of the model
6.1.1 Ice dynamics
6.1.2 Climate forcing
6.2 Experimental design
6.2.1 Adjustments near ice core sites
6.2.2 Model spin-up
6.2.3 Validation process
6.3 Depositional provenance stratigraphy
6.3.1 “East–west” profiles
6.3.2 “North–south” profiles
6.4 Deposition age
6.4.1 Age stratigraphy
6.4.2 Age of deep ice
6.4.3 Average age of the ice sheet
6.5 Isotopic stratigraphy at ice core sites
6.5.1 Dome C
6.5.3 Dome Fuji
CHAPTER 7: Simple vs. complex ice flow models for deep Antarctic records
7.2 Prediction of age–depth in an ice core
7.2.1 Ice flow dating model
7.2.2 Age–depth for the ice sheet model
7.3 Depositional conditions for the deep East Antarctic records
7.3.1 Surface mass balance
7.3.2 Ice origin
7.3.3 Paleo-elevation of depositional ice
7.4 Internal ice flow properties at deep ice core sites
7.4.2 Velocity profiles at domes
7.4.3 Velocity profiles at Vostok
7.4.4 Simple vs. 3D calculation of velocity
7.5 Discussion and conclusion
CHAPTER 8: Global budget of water isotopes inferred from polar ice sheets
8.2 Conditions for an accurate diagnosis
8.2.1 Ice sheet reconstruction
8.2.2 Deposition rate of water isotopes
8.2.3 Stability of the Antarctic Ice Sheets
8.2.4 Presentation mode
8.3 Volume and composition of the East Antarctic Ice Sheet
8.3.1 Ice volume and sea level
8.3.2 Isotopic composition of the EAIS
8.3.3 Effect of the EAIS on ocean composition
8.4 Volume and composition of the West Antarctic Ice Sheet
8.4.1 Ice volume and sea level
8.4.2 Isotopic composition of the WAIS
8.4.3 Effect of the WAIS on sea water composition
8.5 Volume and composition of the Greenland Ice Sheet
8.5.1 Ice volume and sea level
8.5.2 Isotopic composition of the GIS
8.5.3 Effect of the GIS on sea water composition
8.6 Discussion: ice sheets and sea water composition
8.6.1 Sea level vs. sea water composition
8.6.2 Last Glacial Maximum
8.6.3 In an ice-free world
CHAPTER 9: Conclusion
9.2 Glaciological results
9.3 Suggestions for future work