1. Introduction
During recent decades a steadily increasing number of studies on the chemical composition of snow and ice on polar ice sheets have been performed in order to understand environmental changes. The variations of the impurity concentrations recorded in the snow are commonly interpreted as changes in the atmospheric environment, despite the fact that the link between the concentrations in the air and those found in the snow and ice is not well understood (e.g. Reference Wolff and BalesWolff and Bales, 1996). Many of the ice-core studies have focused largely on a few deep-drilling sites, but we need to improve our understanding of the spatial distribution of chemical species before we can resolve some of the questions involving origin, transport and deposition of the impurities in the Antarctic snow.
The interior of Antarctica is of particular interest because of its unique combination of remoteness from other continents, which excludes to a large degree any local anthropogenic influences, and its extreme atmospheric and meteorological conditions. In recent years, Dronning Maud Land has been the focus of increased attention because it has been selected as a deep-drilling site for the European Project for Ice Coring in Antarctica (EPICA). Several studies from this project concerning accumulation and oxygen isotopes in the area have recently been published (Reference Isaksson, vanden Broeke, Winther, Karlöf, Pinglot and GundestrupIsaksson and others, 1999; Reference Oerter, Graf, Wilhelms, Minikin and MillerOerter and others, 1999, Reference Oerter2000; Reference Van den BroekeVan den Broeke and others, 1999; Reference HolmlundHolmlund and others, 2000; Reference KarlöfKarlöf and others, 2000). Previous studies on the polar plateau include several Swedish Antarctic Research Programme (SWEDARP) expeditions (Reference Isaksson and KarlénIsaksson and Karlen, 1994a, Reference Isaksson and Karlénb; Reference Isaksson, Karlén, Gundestrup, Mayewski, Whitlow and TwicklerIsaksson and others, 1996) and one International Trans-Antarctic Scientific Expedition (ITASE) traverse (Reference Mayewski and GoodwinMayewski and Goodwin, 1999) under SWEDARP (Reference Richardson, Aarholt, Hamran, Holmlund and IsakssonRichardson and others, 1997; Reference StenbergStenberg and others, 1998).
Presently only a few glacio-chemical studies have been published from western Dronning Maud Land. The first was performed on the Riiser-Larsen Ice Shelf during the Norwegian Antarctic Research Expedition 1978/79 (Reference GjessingGjessing, 1984), while the more recent studies concern the spatial variation of major ions in snow pits along transects from the coast and inland (Reference StenbergStenberg and others, 1998; M. Stenberg and M. Hansson, unpublished information). These studies report high concentrations of marine-derived ions (Na+, Mg2+, Ca2+, Cl−, K+, MSA) in coastal areas, with a rapid decrease further inland, while the non-sea-salt ions (NO3 −, nssSO4 2−) were not a function of the distance to the coast. Recently, there have also been results from high-resolution continuous flow analysis (CFA) covering the past 2000 years from the polar plateau presented (Reference Sommer, Wagenbach, Mulvaney and FischerSommer and others, 2000b).
In this paper we present chemical data from a 20 m firn core obtained on Amundsenisen in the area around 75° S, 2° E at 2900 m a.s.l. during SWEDARP 1991/92 (Fig. 1). At this site, Reference Isaksson, Karlén, Gundestrup, Mayewski, Whitlow and TwicklerIsaksson and others (1996) estimated the mean annual accumulation to be 7.7 cm w.e., and the mean annual temperature, as indicated by the 10 m borehole temperature, to be about −43.8°C. This paper focuses on the glacio-chemical record from this core, which is estimated to cover the period 1865–1991. The accumulation and oxygen isotope record are discussed in more detail in Reference Isaksson, Karlén, Gundestrup, Mayewski, Whitlow and TwicklerIsaksson and others (1996). To summarize, the δ 18O record shows an increasing trend equivalent to a temperature increase of 0.8°C since 1865, of the same order of magnitude as the calculated overall warming of the Southern Hemisphere (Reference Briffa and JonesBriffa and Jones, 1993). The accumulation of 7.7 cm w.e. a−1 appears to be rather stable throughout the record and does not reveal any clear trend over the 1865–1991 time period (Reference Isaksson, Karlén, Gundestrup, Mayewski, Whitlow and TwicklerIsaksson and others, 1996). More recent studies in the same area have also found very stable accumulation rates through time although the annual mean is somewhat higher at this particular site (Reference Oerter, Graf, Wilhelms, Minikin and MillerOerter and others, 1999, Reference Oerter2000; Reference KarlöfKarlöf and others, 2000; Reference SommerSommer and others, 2000a).
Map of the area on Amundsenisen discussed in this paper. Indicated on the map are the core site discussed in this paper, one snow-pit location from ITASE 1993/94 used for comparison and the traverse line with the coring locations from the Norwegian Antarctic Research Expedition (NARE) 1996/97.
2. Sampling and Analytical Procedures
The core was drilled using a Polar Ice Coring Office (PICO) lightweight coring system with a 3 in (7.6 cm) diameter barrel. The snow/firn density was determined immediately after retrieval by measuring and weighing each core section on an electronic scale. The core was packed in plastic bags, sealed and stored in insulated wooden boxes before further transportation to a freezer on board the ship to Sweden and later to facilities at Stockholm University.
The core was transported frozen from Stockholm to the University of New Hampshire where it was subsampled at 3 cm resolution for major anions, cations and methane-sulphonic acid (MSA). About 2 cm of the outer core section was mechanically cut away to clean the core. It was processed in a cold room (−10°C), using special procedures to minimize contamination (Reference Buck, Mayewski, Spencer, Whitlow, Twickler and BarrettBuck and others, 1992). The sampled core sections were put in pre-cleaned and dried containers. The samples were melted in these containers and aliquots were taken for major-ion, MSA and oxygen isotope analysis. The major ions were analyzed within 2 hours of melting; the MSA and δ 18O samples were refrozen. The cations were analyzed on a CS12 column, and the anions on an AS4A-SC 2 mm column. In both cases, suppressed chromatography was used. MSA was later analyzed on an AS4 column using suppressed conductivity. The methods, including the accuracy of the measurements, are described in Reference Whitlow, Mayewski and DibbWhitlow and others (1992).
3. Dating
The total number of samples per year varied between 3 and 10. An annual stratigraphy was obtained by examining seasonal variations in the sodium and nitrate records (Fig. 2). Comparing these records to the seasonal variation of oxygen isotopes suggests nitrate peaks during summer and sodium during winter, in agreement with other sites on the polar plateau (Reference Legrand and DelmasLegrand and Delmas, 1984). The seasonal cycles in the oxygen isotope record can be observed in the uppermost 4 m, where the densification processes cause diffusion of the signal (Reference JohnsenJohnsen, 1977). In order to avoid using occasional double peaks as dividing lines for years, we used two ions, which have different seasonal input, to determine the different years. Part of a year’s accumulation is likely to be missing due to uneven distribution of snow and wind erosion through the year. The volcanic record interpreted from the non-sea-salt sulphate record provided additional dating reference horizons (Fig. 2) (see section 4.2). We estimate the dating of the core to be ±3 years; however, in the oldest part of the core, before the Krakatau eruption, age control is lacking, so the dating errors there could be about ±5 years. The core was found to cover the period 1865–1991.
The upper 5 m, comprising 21 years, of the sodium and nitrate profiles, which were used to obtain an annual time-scale for the core together with the complete profile of the sulphate record from the Dronning Maud Land snow core (from Reference Isaksson, Karlén, Gundestrup, Mayewski, Whitlow and TwicklerIsaksson and others, 1996). The proposed volcanic peaks are listed in Table 2.
4. Ion Records
4.1. Sea salts
In our core, 77% of the magnesium, 59% of the chloride, 43% of the calcium and 6% of the sulphate are of sea-salt origin, in agreement with snow-pit data from the same area (Reference StenbergStenberg and others, 1998). The calculation of the minimum sea-salt contribution follows the method described in Reference StenbergStenberg and others (1998), using Na+ as the sea-salt indicator following Reference Legrand and DelmasLegrand and Delmas (1988).
Mean concentrations of Na+, Cl− and Mg2+ are similar to those reported from other high-altitude areas in East Antarctica (Table 1). These major sea-salt components have seasonal variations that are the result of different amounts of input during an annual cycle. If we use the oxygen isotope record as an indicator of the seasons during an annual cycle, then all three ions peak during the time when δ 18O values are lowest, i.e. during winter. This has also been shown in a pit study in the same area (Reference StenbergStenberg and others, 1998) and by CFA on several cores from this part of the polar plateau (Reference Sommer, Wagenbach, Mulvaney and FischerSommer and others, 2000b). Aerosol measurements at the South Pole (Reference Cunningham and ZollerCunningham and Zoller, 1981; Reference Bodhaine, Deluisi, Harris, Houmere and BaumanBodhaine and others, 1986; Reference Tuncel, Aras and ZollerTuncel and others, 1989) clearly show that high input of sea-salt elements is most frequent during the winter storms. The timing of sea-salt input contrasts with that at coastal sites such as Neumayer, 500 km north of the Amundsenisen coring site, where sea salt peaks in the late summer to autumn, at the time of minimum sea-ice extent (Reference Wagenbach, Görlach, Moser and MünnichWagenbach and others, 1988).
Concentrations of all the analyzed ions for the Dronning Maud Land core site, estimated to cover the time 1865–1991
Temporal variations in sea salts are thought to be associated with changing intensity in surface wind speeds near the ocean surface (e.g. Reference Petit, Briat and RoyerPetit and others, 1981). The temporal variability of the sea-salt species through this record is not large, except for several high concentration spikes through the 1880s and 1890s, (Fig. 3). The most pronounced peaks appear during the winters of 1894 and 1895, and coincide with the Thompson Island eruption which is believed by Reference LambLamb (1967) to have occurred in 1895–96 at high southern latitude (54° S, 5° E). This eruption is thought to have destroyed an island and most likely injected large amounts of sea salts into the atmosphere. Reference Legrand and DelmasLegrand and Delmas (1988) suggested that this eruption explained sea-salt peaks around 1894 in a core from Dome C, but there have been no reports from other sites in Antarctica.
The ion-concentration records of chloride, magnesium, sodium, MSA, non-sea-salt sulphate, calcium, nitrate and the chloride/sodium ratio, in ng g−1. The oxygen isotope record and accumulation records are included for comparison. For each record the raw data are plotted together with smoothed data using Gaussian weighting coefficients (approximately equivalent to a 7 year moving average).
The excess Cl− component is 41%, and a reaction involving sea salts and acid aerosols producing gaseous HCl has been proposed as the major source for excess Cl− in Antarctica (Reference Legrand and DelmasLegrand and Delmas, 1988). The reaction is more efficient when weather conditions are calm, usually during summer, and the higher the Cl/Na ratio (the indicator for excess Cl−), the stronger the fractionation. Both the Cl/Na ratio of 2.8 in our core (Table 1) and a summer peak agree with the current knowledge of excess Cl.
Throughout most of our record there are minor fluctuations of the Cl/Na ratio around the mean value. However, during the late 1950s a gradual increase began, culminating in a significant rise around 1980 (Fig. 3), suggesting that the content of HCl in the atmosphere has changed. We have no explanation for this and have not seen it reported elsewhere, although one possibility might be changes in the fractionation of sea salts due to long-range transport.
4.2. Sulphate
Sulphate is generally the dominant ion in high-altitude Antarctic snow, and comes in the form of atmospheric H2SO4 mainly from marine biogenic emission, with sporadic input from volcanic activity. In this core the sea-salt part of sulphate is < 6% of the total sulphur (Table 1), and the concentrations show summer maxima, all in agreement with previous studies from the polar plateau (Reference Delmas and BoutronDelmas and Boutron, 1978; Reference DelmasDelmas, 1982; Reference Legrand and DelmasLegrand and Delmas, 1984). A well-developed seasonal cycle with summer maxima is often present both in aerosol measurements and in the snow on the polar plateau. The explanation for this could be summer intrusion of stratospheric air masses in combination with enhanced vertical mixing of the troposphere during the summer (Reference Cunningham and ZollerCunningham and Zoller, 1981), or could involve photochemical processes (Reference Legrand and DelmasLegrand and Delmas, 1984).
The most pronounced sulphur peak occurs in summer 1964/65 (Fig. 2; Table 2), corresponding to the eruption of Mount Agung, Indonesia (1963), in good agreement with what is known about the intensity of this eruption (Reference Delmas, Legrand, Aristarain and ZanoliniDelmas and others, 1985; Reference Legrand and DelmasLegrand and Delmas, 1987). Other peaks are interpreted as being due to the eruptions of Mount Ngauruhoe, New Zealand (1974), and Azul, Argentina (1932), and one broad peak coincides with the eruptions of Tarawera, New Zealand (1886), Falcon Island, South Shetlands, Southern Ocean (1885), and Krakatau, Indonesia (1883) (Fig. 2; Table 2). The latter two eruptions appear as a clear double peak in areas with higher accumulation (Reference Cole-Dai, Mosley-Thompson and ThompsonCole-Dai and others, 1997). Subsequent studies in the same area confirm some of the eruptions (Reference KarlöfKarlöf and others, 2000; Reference OerterOerter and others 2000).
Volcanic eruptions as identified in the non-sea-salt sulphate record
4.3. MSA
MSA (CH3SO3H) and sulphur dioxide are two products of marine biogenic dimethyl sulphide (DMS; (CH^S) emission. Sulphur dioxide is converted to non-sea-salt SO4 2− in the atmosphere, but MSA does not appear to have any other source than DMS (Reference Lovelock, Maggs and RasmussenLovelock and others, 1972). Therefore, MSA is frequently used as an indicator of the strength of the marine biogenic source (e.g. Reference Ivey, Davies and AyersIvey and others, 1986; Reference Saigne and LegrandSaigne and Legrand, 1987).
The mean MSA content in this record (Table 1) is comparable to other data collected from the snow on the Antarctic plateau (Reference Legrand, Feniet-Saigne, Saltzman and GermainLegrand and others, 1992; Reference StenbergStenberg and others, 1998). The MSA record does not show any pronounced seasonal variations. The lack of seasonality in MSA has been reported previously from other inland sites in Antarctica (Reference Ivey, Davies and AyersIvey and others 1986; Reference Legrand, Feniet-Saigne, Saltzman and GermainLegrand and others, 1992), in contrast to coastal areas where seasonal cycles are clearly present in both snow (Reference Mulvaney, Pasteur, Peel, Saltzman and WhungMulvaney and others, 1992; Reference Welch, Mayewski and WhitlowWelch and others, 1993; Reference StenbergStenberg and others, 1998) and air samples (Reference Wagenbach, Wolff and BalesWagenbach, 1996).
The MSA record shows a number of pronounced peaks throughout the record (Fig. 3). Reference Legrand and Feniet-SaigneLegrand and Feniet-Saigne (1991) also find such peaks in a South Pole core, which they correlate with El Niño–Southern Oscillation (ENSO) events. This has been confirmed by more recent studies from South Pole by Reference Dibb and WhitlowDibb and Whitlow (1996) and Meyerson and others (in press). In our record we have 17 pronounced MSA peaks, which seem to match with ENSO events as listed by Reference Quinn, Diaz and MarkgrafQuinn (1992) (Fig. 4; Table 3). Reference Quinn, Diaz and MarkgrafQuinn (1992) reports 34 large-scale ENSO events between 1864 and 1987. Some of these occur close together, and are therefore difficult to separate, and at least two of them comprise more than one event (these peaks are marked 1864, 1866 and 1929–31, 1932 in Fig. 4). The largest peaks are assigned to the ENSO events of 1991–95 and 1929–31, 1932, neither of which was unusually strong, which suggests that the implied link to ENSO is not simple. The MSA record has high concentrations in the uppermost snow layers, corresponding to 1991–92, which could be due to the beginning of the extended ENSO event of 1991–95 (Reference Trenberth and HoarTrenberth and Hoar, 1996). In several cases, the MSA peaks correspond well with the timing of ENSO events, but with a dating error of as much as ±3 years, there is a risk of false correlation since the average frequency of ENSO during this time period is 3.6 years (Reference Quinn, Diaz and MarkgrafQuinn, 1992). When more records become available through EPICA, a stacked MSA record produced by several snow cores may provide different results.
The concentration record of MSA from the Amundsenisen ice core. The MSA peaks are correlated with ENSO events, which are indicated by years, after Reference Quinn, Diaz and MarkgrafQuinn (1992). The corresponding ENSO events are listed in Table 3 together with other relevant information.
The ENSO chronology together with the corresponding MSA peak in the 20 m snow core
Several shallow-pit studies in Dronning Maud Land have also clearly indicated increased MSA concentrations during recent ENSO events (Reference StenbergStenberg and others, 1998; M. Stenberg and M. Hansson, unpublished information). These results suggest that the pronounced MSA peaks are both spatially reproduced and preserved down the snowpack.
Furthermore, although the amplitudes of the corresponding ENSO events in our record from Amundsenisen and the South Pole core (Reference Legrand and Feniet-SaigneLegrand and Feniet-Saigne, 1991) do not correspond very well, both records show low concentrations and a lack of major peaks between the mid-1940s and late 1950s, a period with little ENSO activity according to Reference Quinn, Diaz and MarkgrafQuinn (1992).
The high MSA concentrations during some of the ENSO years suggest that this is more than a regional phenomenon and that the large-scale circulation pattern is affected. Reference White and PetersonWhite and Petersen (1996) presented intriguing evidence of changes associated with the Antarctic continent on interannual time-scales. Using data concerning atmospheric pressure at sea level, wind stress, sea surface temperature and sea-ice extent over the Southern Ocean, they track a system of coupled anomalies. During a period of 4–5 years this system of anomalies, which they call the Antarctic Circumpolar Wave, moves eastwards around the continent. They suggest that its initiation is associated with El Niño in the equatorial Pacific. If the atmosphere and ocean are coupled in such a stable way, it may be reasonable to infer the MSA–ENSO link presented here.
The implications for the source area of MSA are discussed by Reference Legrand, Feniet-Saigne, Saltzman and GermainLegrand and others (1992) by using the difference in timing between coastal and high-altitude sites to infer a low- or mid-latitude source for the high-altitude areas, and a local source for the coastal zones of Antarctica. They base their arguments on the difference in the MSA/nssSO4 2− molar ratio (R), which is latitude-dependent, high-latitude areas having a higher value (Reference BerresheimBerresheim, 1987; Reference Bates, Calhoun and QuinnBates and others, 1992). The climatic implications of changing MSA content in snow/ice cores are focused particularly on the value of R, which is supposed to have a direct correlation to the air temperature (Reference Legrand, Feniet-Saigne, Saltzman and GermainLegrand and others, 1992). The average value of the MSA/nssSO4 2− molar ratio for this core is 0.125, corresponding to a source at low or mid-latitudes.
There are significant increases in R during ENSO events, also found by Reference Legrand and Feniet-SaigneLegrand and Feniet-Saigne (1991) in the South Pole record. They suggested an enhanced contribution from high-latitude sources during ENSO events. Some of the ENSO events have a corresponding nssSO4 2− peak, but not all. This implies either that the non-sea-salt sulphate does not arrive simultaneously with MSA or that it has additional, as yet unknown, sources. It thus appears that MSA consists of a mixture of local and distant material, the former attracted by the relative proximity to the coast (about 500 km) and the latter by the high altitude.
Reference StenbergStenberg and others (1998) found that MSA and sulphate have different spatial distributions in this area, implying that their sources are different and that the molar ratio R should be used with great caution. Other studies of the spatial variability in MSA suggest that Antarctic high-altitude sites have an additional source of MSA (Reference Kreutz and MayewskiKreutz and Mayewski, 1999), perhaps through the upper troposphere (Reference Legrand and Feniet-SaigneLegrand and Feniet-Saigne, 1991).
4.4. Non-sea-salt calcium: a crustal indicator
The non-sea-salt calcium is considered to be primarily of continental origin and is therefore used as an indicator of changes in wind speed over continental areas, and of changes in transportation strength (Reference Mayewski and LyonsMayewski and Lyons, 1982; Reference De Angelis, Barkov and PetrovDe Angelis and others, 1987).
The non-sea-salt calcium fraction is 57% of the total calcium content in this core (Table 1), which is somewhat higher than reported from the same area over the past 2000 years (Reference Sommer, Wagenbach, Mulvaney and FischerSommer and others, 2000b). The record shows seasonal variations, with maxima during the austral summer, in agreement with results from CFA by Reference Sommer, Wagenbach, Mulvaney and FischerSommer and others (2000b). Aerosol measurements at the South Pole (Reference Tuncel, Aras and ZollerTuncel and others, 1989) show high concentrations of crustal material in the summer, which gradually decrease to a minimum during the austral winter. Reference HoganHogan (1984) observed that terrestrial aerosols at South Pole are associated with the upper-troposphere or the lower-stratosphere source layer. The seasonal distribution is largely determined by the strength of the cyclonic wind barrier around the continent and the temperature inversion layer, both of which are stronger during winter. Thus changes in input of dust at high altitudes in Antarctica have been interpreted as changes in the summer upper-tropospheric wind.
In general, the calcium record indicates an overall increase between 1865 and 1940 and a decrease thereafter, but the record also consists of cycles, which seem to have a periodicity of 10–20 years. Particularly strong peaks or increases occur during the periods 1880–85, 1905–12, 1920–40, 1960 and 1975–77, with calcium increases of > 15 times the mean value. This is in general agreement with the peaks in the sea-salt record. These peaks correspond to periods with lower δ 18O content, i.e. colder periods (Fig. 3). The most pronounced calcium increase takes place between 1920 and 1940 and is associated with a cold period.
4.5. Nitrate
The nitrate record from Amundsenisen has a total concentration of 54.5 ng g−1 (Table 1), lower than the South Pole concentration but much higher than for most other sites on the polar plateau (Reference Legrand and DelmasLegrand and Delmas, 1986, Reference Legrand and Delmas1988). The nitrate record shows clear seasonal variations (Fig. 3), with summer peaks, as at other high-altitude sites (e.g. Reference Wolff and DelmasWolff, 1995). The highest concentrations are found in the uppermost 0.6 m, most likely because of gaseous interaction with the atmosphere through the porous snow (density around 0.35 g cm−3), as at many low-accumulation sites in Antarctica (e.g. Reference Mayewski and LegrandMayewski and Legrand, 1990; Reference Dibb and WhitlowDibb and Whitlow, 1996; M. Stenberg and M. Hansson, unpublished information). The source for nitrate in Antarctica is widely debated (see Reference Wolff and DelmasWolff (1995) for a review of this topic).
5. Conclusions
None of the ion records from this high-altitude firn core from western Dronning Maud Land show any trend over the time period 1865–1991. The MSA concentrations show a substantial increase in connection with ENSO years. This correspondence is not perfect throughout the core; whether this is due to dating problems or mechanisms within the teleconnection is difficult to say. Results from Reference White and PetersonWhite and Petersen (1996) strongly suggest an ENSO influence in both the oceans and atmosphere around Antarctica. In addition, a recent meteorological study focused on western Dronning Maud Land using model precipitation data suggests that the variability in accumulation is directly linked to ENSO, so one can expect this signal to be seen in ice cores from this area (Reference Noone, Turner and MulvaneyNoone and others, 1999).
Shallow-snow-pit studies of the spatial variations of major ions, including MSA (Reference StenbergStenberg and others, 1998; Stenberg and Hansson, in press), suggest that this core site is representative also for a larger part of this area of the polar plateau. Due to the extensive EPICA pre-site investigations, many more shallow and medium-deep ice cores have recently been drilled in the same area, and the results so far available suggest that the spatial variability is large and redistribution by wind may cause problems with correlation of short time periods (Reference SommerSommer and others, 2000a, Reference Sommer, Wagenbach, Mulvaney and Fischerb). At this point there are no other published MSA records from this area that cover the time period discussed in this paper. However, more chemical data will soon become available from this area on the polar plateau, which will provide a more complete picture of glacio-chemical conditions.
Acknowledgements
This work was done as a part of the first author’s Ph.D. at Stockholm University and was financed by a Swedish Natural Science Research Council grant to W. Karlén. Excellent field support came from SWEDARP under the leadership of O. Melander. A. Bodin helped with the drilling, and A. Moberg with the filtering of the data. The map was made by A. Estoppey. The paper benefited from helpful comments by D. A. Peel and two anonymous referees.
