Comments on the origin of dust in East Antarctica for present and ice age conditions

Journal of Atmospheric Chem&try 14: 129-142, 1992. © 1992 Kluwer Academic Publishers. Printed in the Netherlands.

C O M M E N T S ON T H E O R I G I N O F DUST IN EAST A N T A R C T I C A F O R P R E S E N T

AND I C E A G E C O N D I T I O N S

A. Gaudichet*, M. De Angclis**, S. Joussaume***, J.R. Petit**, Y.S.

Korotkevitch**** and V.N. Petrov****

* Laboratoire de Microscopie Analytique Appliqu6e aux Sciences de la Terre, Univ. Paris XII, Av. Gdn~ral de Gaulle, F-94010 Creteil, URA-CNRS 1404

** Laboratoire de Glaciologie el G~ophysique de l'Environnement, CNRS, rue Molidre, BP

96, F-38402 Saint Marlin d'lleres Cedex *** Laboratoire de Meteorologie Dynamique, CNRS, 24 rue Lhomond, F-75231 Paris **** Arctic and Antarctic Research Institute, 38 Beringa St., Leningrad 199226, USSR

ABSTRACT. We have studied the distribution of 327 clay mineral particles retrieved from four Antarctic ice samples corresponding to present and Last Glacial Maximum (LGM) climate conditions. Illite, chlorite, smectite and kaolinite were identified in all samples. Focusing on kaolinite, because of its use as a possible tracer of low latitude soils, we find a significantly smaller amount for LGM samples while the dust concentration in snow during the LGM was about 30 times higher than for present climate conditions. This can be interpreted as change in the contribution of the Australian source with climate.

A second approach was based on the modeling of the desert dust cycle using an Atmospheric General Circulation Model (AGCM) under both present-day and ice age conditions. Unlike mineralogical results, the model suggests the prevalence of the Australian dust source in tile deposits over East Antarctica under both present-day and LGM climate conditions. However the model fails to reproduce the strong increase in dust deposits during the LGM. This discrepancy could be partly due to the lack of a higher latitude dust source in the model.

The stronger dust input recorded in ice cores for the LGM could be related to an additional active high latitude source (possibly close to South America) overlapping the atmospheric background coming from low latitude areas.

Key words : continental dust, Antarctica, paleo-environment, glacial age.

1. Introduction

Due to its geographical location and climatic conditions, Antarctica is an ideal site for studying

the background of atmospheric composit ion and the long-range transport processes of aerosols on

130 A. GAUDICHET ET AL.

a global scale. The study of ice cores provides information on the atmospheric environment for

time periods going back many thousands of years and including the last glacial period. The

atmospheric aerosol can be reconstructed assuming a direct, although not well understood, link

between atmospheric and snow composition (Pourchct ct al., 1983, Legrand et al., 1988).

This work deals with the dust emitted from continents and incorporated in the Antarctic

precipitation. The bulk input can be quantified relatively easily by direct measurement of, for

example, aluminium. It has been shown that ice lormed under the Last Glacial Maximum

conditions (LGM, c.a. 18 kyr BP) contained up to 30 times more dust than the ice formed during

the Holocene period. This increase has been observed in West Antarctica (at Byrd station,

Thompson and Mosley-Thompson, 1981) and in East AnU~rctica (at Dome C station by Petit et

al., 1981, and at Vostok station by De Angclis ct al., 1987, and Petit et al., 1990) (Fig. 1) and

interpreted as the consequence of the glacial climatic conditions characterized by greater aridity

over the continents and enhanced atmospheric circulation (Cragin et al., 1977, Petit et al., 1981,

Briat et al., 1982).

Since the potential dust sources are most Iikcly the three continents of the Southern

Hemisphere, the geographical area of the source, the transport path and the source strength

variations during the glacial climate are open questions. To address this problem, we followed two

different approaches, i.e. mineralogical investigations and modeling :

1 - The mineralogical nature of clay species, mainly depending on weathering processes, reflects

their parental soil type and indirectly their geographical origin. For example, in Antarctic and sub-

Antarctic areas, we find illite, which is ubiquitously distributed, chlorite, which is typically present

in high latitude areas, and kaolinite, which is produced by more intense weathering processes

occurring at low latitude. Kaolinite can then be considered as an indicator of Australian soils

(Griffin et al., 1968, Moriarty, 1977, Gaudichct ct al., 1989).

2 - Simulations of dust transport using an Atmospheric General Circulation Model (AGCM) make

it possible to take into account the full complexity of the atmospheric continental dust cycle, i.e.

mobilization, transport, diffusion and removal processes, and may therefore be used to investigate

the transport of mineral particles towards Antarctica. This approach can be applied to both present

and glacial climate conditions.

Considering clays as helpful indicators of the location of dust sources, the aim of the present

study is to investigate the mineral clay content of recent snow levels at two sites (Vostok and

South Pole stations), and secondly, to compare the distribution of different clay species for recent,

mid-Holocene and LGM samples from Vostok station. Mineralogical results are then compared

with results from a numerical simulation of the dust transport for present day and glacial climate

conditions.

COMMENTS ON THE ORIGIN OF DUST

South America

131

bathymotric 3ntour

- Polar circle

/ / " \ \

===================================================

S o u t h A t l a n t i c O c e a n "-.

Indian Ocean

South Pacific Ocean \

4 0 ° S . N . . _ . / / ~ / / . _ . .

Figure 1. Map of the Southern Hemisphere including 150 m bathymetric contours and deep ice

core drilling locations in Antarctica (By: Byrd, DC: Dome C, VK: Vostok).

132

2. Mineralogical study

A. GAUDICHET ET AL.

Four samples were selected from Vostok and South Pole stations in Antarctica. Three (N°I, N°2,

N°3) of them are from the 2083 m deep ice core recovered at Vostok station (78 ° 28'S, 106 °

48'E, 3488 m a.s.l.). The climatic record deduced from stable isotope measurements, and the

dating have been published by Jouzel et al. (1987). The ages are 50 yrs for sample N°I, 5 kyrs

for sample N°2 (mid-Holocene period) and 24 kyrs for sample N°3. As shown along with stable

isotope content in Fig. 2, the aluminium concentration (De Angelis et al., 1987) of sample N°3

is about 30 times higher than that of the two other samples. Sample N°4 is from a shallow core

recovered at South Pole station (90°S, 2800m a.s.1.) and corresponds to snow deposited about 35

yrs ago. Each sample covers 2 to 5 years of accumulation.

Age (kvrs)

-440

~'~ - 4 6 0

-480 --Deuter ium ~ l ~ t w ~ ' ' 3 " - -

- 5 0 0

50 -- ~

5OO

DEPTH (m)

150

~,luminium

Figure 2. Deuterium record and A1

concentrations along the 2083 m deep Vostok ice

core (redraw from Jouzel et al., 1987 and from

De Angelis et al., 1987). Sample numbers are

indicated.


2.1. ANALYTICAL PROCEDURE

133

First, each sample was decontaminated by subcoring for snow samples and by rinsing for ice

samples (Batifol, 1987, De Angelis et al., 1987). Aliquots (20 to 60 ml) of the final melted sample

were filtered through a Nuclepore membrane (diameter 25 mm, porosity 0.2 pm) and then

transferred onto transmission electron microscope cupper grids. Particles were individually

analysed under a transmission electron microscope (TEM, JEOL 100 CX IIS) fitted with a

chemical microanalysis system (Energy Dispersive X-Ray Spectrometer; TRACOR 5200).

In randomly chosen areas, clay particles were identified by both their lamellar morphological

features and their hexagonal electron diffraction pattern. Chemical analysis was used to identify

the species of each preselected clay (Fig. 3). For this, we made a semi-quantitative assessment of

the elemental composition using the peak ratio method, taking Si as a reference element. With this

method, the chemical composition is accurate to within better than 3%. All four samples were

processed in the same way to prevent any bias in the analytical procedure (Gaudichet et al., 1986).

About 80 clay particles were studied per sample (327 particles in all).

2.2 RESULTS

Most of the particles are in the sub-micron range. The number and the relative percentage of each

identified clay species are reported in Table 1. Illite remains the dominant species (62 to 71%)

followed by smectite (9 to 20%), chlorite (4 to 19%) and kaolinite (1 to 12%). A statistical

comparison was made between the distribution of clay species in the four samples using the

hypergeometric law. In the two recent snow samples (from Vostok and South Pole Stations) the

relative amounts of kaolinite (10 to 12%) are not significantly different, whereas the South Pole

sample is characterized by a higher amount of chlorite (probability 0.99). The comparison between

the three samples from Vostok station suggests that the amount of kaolinite is significantly lower

(probability 0.985) in sample N°3, which corresponds to the LGM climate conditions. By contrast

the chlorite content is not significantly different from one sample to the other (probability about

0.80). The most noticeable result is the presence of kaolinite (approximately 10%) in Holocene

samples, while it is scarce in the LGM sample even though the AI concentrations were about 30

times higher.

134 A. GAUDICHET ET AI

r.u; d

C • , j

~ . ~

-2

I ! i . . . . . ., . . . . . . . . . . . . . . • I

Figure 3. Transmission Electron Microscope microphotographs of a Vostok recent snow sample.

Part a: field of a filter showing few particles and filter pores. Part b: morphology of a kaolinite

particle slightly damaged by the electron beam after the microanalysis.Part c: typical pseudo-

hexagonal electron diffraction pattern of phyllosilicate (here, kaolinite). Part d: kaolinite

microanalysis spectrum characterized by high Si and A1 content. The Cu signal originates from

the electron microscope grid.

COMMENTS ON THE ORIGIN OF DUST 135

TABLE 1. Number and relative abundances of the different clay species in Vostok and South

Pole snow and ice samples

1 2 3 4

VOSTOK VOSTOK VOSTOK SOUTH POLE

snow 1927 ice 5000 yrs ice 24000 yrs snow 1955

CLAYS n % n % n % n %

Illite 55 65 53 66 58 71 50 62

Smectite 14 16 16 20 13 16 7 9

Chlorite 5 6 3 4 10 12 15 19

Kaolinite 10 12 8 10 1" 1 8 10

Vermiculite 1 1 - -

TOTAL 85 100 80 100 82 100 80 100

*The Kaolinite amount is significantly lower in sample N°3 (probability .985 with the

hypergeometric law).

3. Modeling approach

The transport of desert dust has been modeled using the AGCM of the Laboratoire de

M6t6orologie Dynamique as fully described in Joussaume (1990). Dust is raised by the simulated

surface winds over all the desert areas, the extents of which are predicted by the model itself

through the soil moisture parameter. Dust is then diffused vertically by turbulence within the

planetary boundary layer and by convective motion, transported by simulated winds and removed

by gravitational settling, interception at the surface due to turbulence and by ralnout processes.

For this first study, only one size range of particles was considered (around 1 gin) and no

interaction with radiation or clouds was included.

Simulatic~ns were performed for both the present-day and LGM (18 kyr BP) climates under

February (Fig. 4-a) and August (Fig. 4-b) conditions. The simulated results for the present-day

climate have been presented and compared to the available observations of dust transport by

Joussaume (1989). The simulated impact of the climate change from present day to full glacial

136

Australia February

A. GAUDICHET ET AL.

South America

modem

II 8O

@ lO

1.25

0.16

LGM

SAHARA F E B M O D [ ~ / ~

[ ] S. AFRICA [ ] S. AMERICA [ ] AUSTRALIA

F E B L G M

I I

0 lO0 200

Dust concentration in snow arbitrary units

Figure 4. Simulated dust plumes over Antarctica for dust originating from Australia and South

America. The plumes are obtained for present day and LGM climate conditions and are displayed

in kg AU, where AU is an arbitrary unit as the dust mass is defined up to an arbitrary constant

in the model. Contributions of the various dust sources to snow concentrations (ratio of dust mass

deposits to snow accumulation rate), averaged over East Antarctica (90°E - 150°W and 67°S-90°S),

are also shown. Part a: simulation for February.


Austral ia A u g u s t South A m e r i c a

137

m o d e m

8o

@ 10

1.25

. . . . . . ° °

0.16

L G M

• SAHARA

AUG MOD [ ] S. AFRICA [ ] S. AMERICA [ ] AUSTRALIA

A U G L G M

I , , 0 100 2OO

Dust concentrat ion in s n o w arbitrary units

Figure 4. Part b : simulation for August


conditions on the atmospheric dust cycle is described by Joussaume (1989 and submitted). The

average global atmospheric dust content is only weakly affected by the climate change, but on

a regional scale, an increase of at least a factor of two is shown by the model over Europe, the

westem North Atlantic ocean and southeastem Australia. However, the dust concentration in snow

(ratio of dust removal flux to snow accumulation) remains practically unchanged over Antarctica.

This discrepancy with ice core data is discussed by Joussaume (1989 and submitted). It could

result from deficiencies in the simulated change of source areas. To address this problem,

indicators of the origin of the dust found in ice core samples are particularly useful to test the

model results.

To study the origin of dust arriving in East Antarctica, six types of particles have been

considered according to their geographical origin. They correspond to the three southem

hemisphere desert areas located in South America, South Africa and Australia, and to the three

desert areas of the northern hemisphere located in North America, the Sahara and Arabia-Asia.

In the present paper, we will discuss the origin of dust deposits in East Antarctica simulated for

present day and LGM period with regards to microparticle analyses.

Under present-day climate conditions, Australia is the prevailing source of dust in East

Antarctica, mainly during the winter season which drives the annual mean budgets. The stronger

winter season input results from both a stronger transient activity in the southern mid-latitudes and

from stronger surface winds over Australia. South America is a secondary dust source for

Antarctica and is limited to westem Antarctica (Fig. 4).

During the LGM, Australia was already the prevailing source of dust for East Antarctica. The

simulated dust concentration in snow slightly increases when averaged over East Antarctica (from

90°E to 150°W and 70°S to 90°S) but remains virtually unchanged in the Vostok area (reduced

by a factor of 0.9, Joussaume, 1989 and submitted).

4. Discussion and conclusions

Previous mineralogical studies on ice cores have shown, both for Dome C (over the last 30 kyrs)

and Vostok (over the last 150 kyrs) stations, the scarcity of kaolinite in all the samples (Gaudichet

et al., 1986, Gaudichet et al., 1988). But in these previous studies, the number of clay particles

analysed was insufficient (only 30 clay particles in each level), and the uncertainty of the results

is therefore too high to demonstrate any significant difference of the clay species distribution

between the different levels analysed. In this study, wc find about 10% kaolinite in recent snow

samples as well as in mid-Holocene samples, instead of the 0% previously found in Dome C and

Vostok samples. This difference is due to the small number of clay particles analysed in the


previous study and/or to the slightly different methodology used. In the work of Gaudichet et al.

(1986 and 1988), all types of particles were counted in a given randomly chosen field while in

this study we counted only clay particles, selected according to their morphology and electron

diffraction pattern. Our protocole should therefore yield more reliable and significant results for

clay particles.

Regarding the representativeness of our samples with respect to the environmental conditions

of a given climatic period, the interannual variability could affect the transport of dust from source

areas on a very short time-scale, but unlikely on the 2 to 5 year periods covered by each sample

selected inside the strong and weak inputs corresponding to LGM and Holocene climatic

conditions respectively.

Very few quantitative mineralogical analyses are available for Antarctica. Kumai (1976) studied

93 snow crystal nuclei at South Pole station and obtained 7 kaolinite among 55 clay particles.

This represents about 13% kaolinite, which is in agreement with our data for recent snow at South

Pole (about 10%) and Votsok (about 12%) stations. Considering kaolinite as an indicator of the

dust origin, our results suggest that, under present time and mid-Holocene climatic conditions,

dust was transported from Australia towards East Antarctica. However, under glacial climatic

conditions, the low kaolinite content suggests a minor contribution of the Australian source to the

strong dust input occurring at this time.

During the glacial period, there is evidence that most of Australia was cold and dry, with

active surface winds over the continent moving sand dunes (Bowler, 1976). Therefore it is

unlikely that the Australian dust source was much less efficient than today. Another source may

have overlapped the Australian contribution. South America has already been proposed as a good

candidate (Gaudichet et al., 1986, 1988, Burckle ct al., 1988). Its more poleward location

compared to the other continents could favor a strong dust transport towards Antarctica. Moreover

the 120 m lowering of the sea level during the ice age exposed a large continental shelf of

erodible soils around South America which may have been a potential dust source at this time.

The contribution of this source is also suggested by De Angelis et al. (this issue) from chemical

analyses.

The AGCM model does not simulate any great increase of the dust concentration in snow

during the LGM and indicates the prevalence of the Australian source for East Antarctica under

present-day and LGM climatic conditions (Fig. 4). Both simulated results disagree with ice core

data. Several reasons could explain why the model fails to reproduce a stronger input of dust

during the LGM. First of all, the desert dust model includes many oversimplified parameters

(Joussaume, 1990). Secondly simulated results may also be biased by the systematic errors of

AGCMs. Note that AGCMs are not very good at reproducing the present-day climate of


Antarctica (Schlesinger, 1984). However, the discrepancy between model results and dust studies

may also reflect the occurrence of another source underestimated by the model for the LGM

climate. The model generates its own dust source regions lbr both climates according to the

simulated ground moisture parameter. The model can therclbre overlook certain dust source areas

because of erroneous ground moisture values. For example, the model overestimates precipitation

in southern South America as a result of the smoothing of the Andes topography.

More mineralogical analyses in deep ice cores and also on aerosols collected in sub-Antarctic

and Antarctic areas are required to confirm these preliminary data. The modeling approach could

also be improved by performing several experiments to test climate sensitivity with respect to

various assumptions on dust source location during the last glacial age (e.g. South America).

5. Acknowledgements

We thank the following organisations for their support: EPF (Exp6ditions Polaires Franqaises,

France), PNEDC (Programme National d'Etudc de la Dynamique du Climat, France), NSF-DPP

(National Science Foundation-Division of Polar Programs, USA), TAAF (Terres Australes et

Antarctiques Fran~aises, France) and Soviet Antarctic Expeditions (USSR).

6. References

Batifol, F. (1987) 'Cd, Cu, Pb and Zn in Antarctic snow and ice: a contribution to their

geochemical cycles' (in French), Thesis, USMG of Grenoble, France.

Bowler, J.M. (1976) 'Aridity in Australia: age, origin and expression in eolian landforms and

sediments', Earth Sc. Rev. 12, 279-310.

Briat, M., Royer, A., Petit, J.R. and Lorius, C. (1982) 'Late glacial input of eolian continental

dust in the Dome C ice core : additional evidence from individual microparticles analysis', Arm.

of Glaciol. 3, 27-31.

Burckle, R.H., Gayley, R.I., Ram, M. and Petit, J.R. (1988) 'Diatoms in Antarctic ice cores :

Some implications for the glacial history of Antarctica', Geology 16, 326-329.

Cragin, J.M., Herron, M.M., Langway, C.C. and Klouda, G. (1977) 'Intcrhemispheric comparison

of changes in the composition of atmospheric precipitation during the late cenozoic era', in

Dunbar Maxwell (ed.), Polmar Oceans, Proceed. of tile Polar Oceans Conference (1974),

Montreal, Canada, 617-631.

De Angelis, M., Barkov, N.I. and Petrov V.N. (1987) 'Aerosol concentrations over the last


climatic cycle (160 kyr) from Antarctic ice core', Nature 325, 318-321.

De Angelis, M., Barkov, N.I. and Petrov, V.N. 'Change in continental aerosol sources over

Antarctica during the last climatic cycle (this issue).

Gaudichet, A., Petit, J.R., Lefevre, R. and Lorius, C. (1986) 'An investigation by analytical

transmission microscopy of individual insoluble microparticles from Antarctic (Dome C) ice

core samples', Tcllus 38B, 250-261.

Gaudichet, A., De Angelis, M., Lcfevre, R., Petit, J.R., Korotkevitch, Y.S. and Petrov, V.N.

(1988) 'Mineralogy of insoluble particles in the Vostok antarctic ice core over the last climatic

cycle (150 kyr)', Gcophys. Rcs. Letters 15, 1471-1474.

Gaudichet, A., Lefevre, R., Gaudry, A., Ardouin, B., Lambert, G. and Miller, J.M. (1989)

'Mineralogical composition at Amsterdam island', Tcllus 41B, 344-352.

Griffin, J.J., Windom, H. and Golbcrg, E.D. (1968) 'The distribution of clay minerals in the world

ocean', Deep Sea Res. 15,433-459.

Joussaume, S. (1989) 'Climate simulations lbr the last glacial maximum using a general

circulation model including isotopes and desert dust modeling' (in French), Thesis, University

of Paris VI, France.

Joussaume, S. (1990) 'Three-dimensional simulation of desert dust particles using a general

circulation model', J. Geophys. Rcs. 95, 1909-1941.

Joussaume, S. 'Palcoclimatic tracers: 'an investigation using an atmospheric general circulation

model under ice age conditions, part 1: desert dust', submitted to

Jouzcl, J., Lorius, C., Petit, J.R., Genthon, C., Barkov, N.I., Kotlyakov, V.M. and Petrov, V.M.

(1987) 'Vostok ice core: a continuous isotope temperature record over the last climatic cycle

(160,000 years), Nature 329, 403-408.

Kumai, M. (1976) 'Identification of nuclci and concentrations of chemical species in snow crystals

sampled at South Pole', J. Atm. Sc. 33, 833-841.

Legrand, M.R., Lorius, C., Barkov, N.I. and Pctrov, V.N. (1988) 'Vostok (Antarctica) ice core:

atmospheric chemistry changes ovcr the last climatic cycle (160,000 years)', Atm. Env. 22, 317-

331.

Moriarty, K.C. (1977) 'Clay minerals in the southeast Indian Ocean sediments, transport

mechanisms and depositional environments', Marine Geolog. 25, 149-174.

Petit, J.R., Briar, M. and Royer, A. (1981) 'Ice Age aerosol content from East-Antarctica ice core

samples and past wind strength', Nature 293, 391-393.

Petit, J.R., Mounicr, L., Jouzel, J., Korotkevitch, Y.S., Kotlyakov, V.I. and Lorius, C. (1990)

'Paleoclimatological and chronological implications of the Vostok dust record', Nature 343, 56-

58.

142 A. G A U D I C H E T ET AL.

Pourchet, M., Pinglot, F. and Lorius, C. (1983) 'Some meteorological applications of the

radioactive fallout measurements in Antarctic snows', J. Geophys. Res. 88, 6013-6020.

Schlesinger, M.E. (1984) 'Atmospheric general circulation model simulations of the modem

Antarctic climate', in 'Environment of West Antarctica: potential COz-induced climate',

Committee on Glaciology, Polar Research Board, National Research Council, 151-196, National

Academy Press.

Thompson, L. and Mosley-Thompson, E. (1981) 'Microparticle concentration variations linked

with climatic change. Evidence from polar ice cores', Science 212, 912-815.

Documents

Comments on the origin of dust in East Antarctica for present and ice age conditions