Introduction
Situated 1300 km inland at an elevation of 2850 m a.s.l., the South Pole is neither on the highest plateau nor at the most central location on the Antarctic continent. However, in addition to being the most important station for Antarctic atmospheric-chemistry research, the Amundsen–Scott base lies in a geographical area particularly well suited to glaciochemical studies (Reference Legrand, De Angelis and DelmasLegrand and Delmas 1984), due to a relatively stable snow-accumulation rate (Jouzel and others 1983, Mosley-Thompson and others 1985), sufficiently high to limit the smoothing of the seasonal variations in the various parameters recorded in the snow layers. On the East Antarctic plateau (such as Dome C or Vostok stations) the weakness of the accumulation rates is an advantage for long-term glaciochemical studies but a disadvantage for detailed studies covering a few centuries. At the South Pole, the 1000 year level can be reached relatively easily with the aid of drilling equipment. The detailed study of such a glaciochemical record, although time-consuming (analysis of several hundred samples is required for good time resolution), is not an impossible task.
In the context of a deep ice-core drilling program at the South Pole, the results which can be obtained from such an ice core are most valuable. They allow “calibration” of glaciochemical methods for a relatively well-known period, even if knowledge of the past environment is far less well documented for the Southern than for the Northern Hemisphere.
For the period under consideration, the only South Pole data available hitherto relate to microparticles, obtained by Reference Mosley-Thompson and ThompsonMosley-Thompson and Thompson (1982) from a 101 m firn core drilled in 1974, and to nitrate concentrations, obtained by Parker and others (1982). Shorter sequences have been studied by Reference HerronHerron (1982), Reference BoutronBoutron (1982) and Reference Legrand, De Angelis and DelmasLegrand and Delmas (1984) for other parameters of interest to atmospheric chemistry, whereas Reference Giovinetto and SchwerdtfegerGiovinetto and Schwerdtfeger (1966), Jouzel and others (1983) and Reference Stauffer and SchwanderStauffer and Schwander (1983) have investigated the accumulation rate, stable isotopes, and (δ18O) and CO2 concentrations respectively.
During the summer of 1983–84, several ice cores were drilled in the immediate vicinity of the Amundsen–Scott base. One of them (PS1, 130 m long) has been used for glaciochemical studies. This paper reports on the measurements made along this ice core, assumed to cover at least the last 1000 years. The purpose of this investigation was to assess the temporal variations in background atmospheric chemistry at the South Pole, assuming that snow chemistry is directly linked to the chemical composition of the aerosol. This assumption is a matter of serious discussion, particularly among atmospheric chemists. Since the developments exposed by Reference JungeJunge (1977), the debate has been reactivated time and time again, particularly by Davidson and others (1981) and by Reference BarrieBarrie (1985). In fact, the chemistry of snow is not necessarily the same as that of the aerosol, but, for a given element or compound and at a given location, concentrations in the snow and in the air are most likely to vary in parallel. This assumption appears to be quite reasonable in terms of the study presented below and will not be discussed further in this paper.
What are the general features of the present-day chemistry of snow at the South Pole? Delmas and others (1982) and, later, Reference Legrand, De Angelis and DelmasLegrand and Delmas (1984) have demonstrated that the chemical composition of soluble impurities determined in South Pole precipitation is dominated by three gas-derived mineral acids (H2SO4, HNO3 and HCl). The other constituents are essentially the different components of sea salt (in particular NaCl). The ion budget is small, generally less than 10 μEql−1, except during the 2 years following explosive volcanic eruptions of global significance (in which case the ion content can be as much as 20 μEq l−1). The proton accounts for from one-third to nearly one-half of the total ion budget.
Neutralization of the acidity is very slight, as indicated by very low ammonium concentrations. Finally, insoluble impurities are almost negligible (less than 5% by weight of the total deposition).
Well-marked seasonal patterns have been detected for several of the ions (Reference HerronHerron 1982, Reference Legrand, De Angelis and DelmasLegrand and Delmas (1984). For instance, mineral-acid concentrations are at a maximum in summer, as are sea-salt compounds in autumn or winter.
Regarding the origin of these impurities (apart from sea salt), H2SO4 appears to be primarily marine-biogenic, HNO3 is derived from atmospheric nitrogen fixation by lightning, and HC1 results from the interaction of H2SO4 with sea-salt particles. This very schematic description of the chemical composition of the snow at the South Pole will be developed in a more detailed manner in the discussion of our results.
Dating the PS1 Firn Core
The accurate dating of ice cores in areas of low accumulation rates (<10 g cm−2 a−1) is a real problem, except when well-defined reference levels are available (as in Greenland). For the last 30 years, total beta radioactivity reference levels exist, allowing very accurate dating (±1 year) of the snow layers (Jouzel and others 1983). Investigations of longer time series have demonstrated that the mean accumulation rate over several decades, as far as it can be evaluated, is relatively stable at the South Pole (Reference Mosley-Thompson and ThompsonMosley-Thompson and Thompson 1982, Jouzel and others 1983, personal communication from A.J. Gow). Mean values are in the range 6.6–12.5 cm−2 a−1 (Jouzel and others 1983).
More recently a double-conductivity spike has been clearly identified by electrical conductivity measurements (ECM) in five of seven South Pole firn cores (Legrand and others 1987). It has been attributed to two major volcanic eruptions in the Southern Hemisphere at the beginning of the ninteenth century (Tambora in 1815 and another eruption around the same time ±8 years). With the aid of these new reference levels, a mean accumulation of 6.5–9.3 g cm−2 a−1 (depending on the firn core under consideration) has been calculated for the last 170 years. For the PSl firn core, the signal of the Tambora eruption has been detected at 27.1 m depth, leading to a mean accumulation of 7.97 g cm−2 a−1 from the surface up to the present time.
Microparticle measurements on a 101 m firn core (Reference Mosley-Thompson and ThompsonMosley-Thompson and Thompson 1982) led to an estimated annual net accumulation equal to 7 g cm−2 a−1 over nine centuries, whereas the stratigraphical study of a recent 202 m ice core (drilled 50 m from PSl) gives a mean accumulation of 8 g cm−2 a−1 for the last 2000 ± 60 years (personal communication from A.J. Gow).
Taking into account the conductivity measurements on PSl (Legrand and others, in press) and Gow’s estimate for the 202 m ice core nearby, we have chosen the mean value of 8 g cm−2 a−1 for dating the entire PSl firn core. Using this value the age of the bottom of the firn core is A.D. 1083
years, to allow for uncertainty about the accumulation rate in the range 6–10 g cm−2 a−1).
Sample Preparation and Analysis
Chemical measurements were made on 42 sections (40 cm long) selected every 2 m along the 130 m PSl firn core. Each sample represents about 1.9–3.7 years of precipitation (depending on depth) and the mean time interval between consecutive samples is about 24 years. In order to obtain chemical data representative of background conditions, samples at depth levels where ECM detected acid spikes (in particular the volcanic signals) were rejected.
Stringent precautions were taken in order to minimize contamination problems. All tools and containers which came into contact with the samples were first washed several times in double-deiontzed water and the sub-coring was done with mini-drilling equipment placed in a cold clean-room according to a procedure described by Legrand and others (1984). After preparation, the samples were stored in a solid state at −15° C in double-sealed polyethylene bags and were melted at room temperature just before analysis. The bags were pierced by the tip of the syringe used to inject the melt-water sample into the ion chromatograph. These additional precautions were taken, with respect to the methodology described by Legrand and others (1984), to improve the quality of the ammonium determinations.
All major soluble ions were determined by ion chromatography (on Dionex 2010i equipment), using two 3 ml aliquots (the first for the cations, the second for the anions). Pre-concentration columns were used for both measurements.
Uncertainties and detection limits were calculated from the linear regression coefficients of the calibration curves, obtained with the aid of standard solutions.
Results and Discussion
The results of the chemical measurements are summarized in Table I, In addition to the mean values (42 samples), we also give the variation range, We shall now a (the natural variability) and the precision of the measurements.
The composition reported in Table I, although incomplete (since Ca2+, Mg2+ and H+ were not determined), confirms that at the South Pole the sea-salt components (indexed by Na) are not dominant for the 1000 year time interval which was covered by this study. Gas-derived compounds (in particular, nitrates and sulfates) are the major constituents of the snow impurities. Non-sea-salt sulfate (excess sulfate, SO4 *) has been calculated from total sulfate according to the relationship
SO4 *=SO4 – 0.25Na
Our results are generally in good agreement with the chemical data already available for South Pole snow over various periods, except for the ammonium values, which are nearly one order of magnitude lower than the values of Reference Legrand, De Angelis and DelmasLegrand and Delmas (1984) (3 ng g−1), which, in turn, are significantly lower than the values (>10ng g−1) of Reference Parker and ZellerParker and Zeller (1979). As already emphasized in the section on the experiments, we have considerably improved the analytical protocol in order to minimize the influence of gaseous contaminants (in this case NH3) during laboratory work. Earlier NH4 measurements were erroneously high. The present values are most probably correct (Saigne and others, in press), allowing for the relatively high analytical uncertainties (±100%) at sub-ppb levels (Table I).
For NH4 + and K+ concentrations (very close to the detection limit of the method), the uncertainties are significantly higher than for other ions. S04 * is excess sulfate (non-marine sulfate). The results of four data sets published by others are also given: (1) Reference BoutronBoutron (1982), (2) Reference HerronHerron (1982), Parker and others (1982), (4) Reference Legrand, De Angelis and DelmasLegrand and Delmas (1984).
Sodium, chloride, excess sulfate and nitrate ion concentrations along the PS1 firn core. Solid lines were drawn by using a spline function where ρ = 0.1.
An obvious conclusion stemming from our observations is that the mean neutralization of the atmospheric acidity in this geographical area is of the order of 1%, i.e. negligible.
We shall now focus the discussion on the time variations in the major ions Na, Cl, SO4 and NO3. The results obtained for the 42 depth levels have been plotted as a function of depth (Fig. 1).
Sodium and chloride
Crustal sodium has been found to be very low in the aerosol collected at the Amundsen–Scott Station and therefore marine sodium is largely dominant (∼90%; Reference Cunningham and ZollerCunningham and Zoller 1981). According to the same authors, Cl has a purely marine source. However, Na is better than Cl for indexing sea-salt deposition, due to the possible presence in the Antarctic atmosphere of Cl partly in gaseous form (HCl), as shown indirectly by Maenhaut and others (1979) and Reference Cunningham and ZollerCunningham and Zoller (1981) from air measurements, and by Reference Legrand, De Angelis and DelmasLegrand and Delmas (1984 and in press) in snow. Therefore Na (rather than Cl)
concentration changes may be used to reconstruct the past variations in sea-salt deposition at the South Pole. Such variations could be linked to changes in the production, transport and deposition of marine aerosol particles. The strong seasonal pattern (with maxima in winter) found for Na at the South Pole (Reference Cunningham and ZollerCunningham and Zoller 1981, Reference Legrand, De Angelis and DelmasLegrand and Delmas 1984) is related partly to the frequency of storms in the sub-Antarctic ocean and partly to their degree of penetration over the Antarctic continent. For long-term variations, changes in the snow-deposition rates may be relevant.
The Na (as well as Cl) depth profiles obtained from our measurements exhibit no definite trend over time, the values varying in the range 4.4-1.15 ng g−1 (and 20-48 ng g−1 for Cl). This suggests no major long-term changes in the three causes of variability listed above.
Additional information of environmental concern could be drawn from the variations in the Cl/Na ratio (R), which at several Antarctic locations has been found to be significantly different from its reference value (1.8) in bulk sea-water (Reference Legrand, De Angelis and DelmasLegrand and Delmas 1984, Reference Palais and LegrandPalais and Legrand 1985, Legrand and Delmas, in press). For the South Pole itself, the measurements indicate that on average the aerosol sea-salt particles are Cl-depleted (Reference Cunningham and ZollerCunningham and Zoller 1981), whereas the sea salt deposited with the snow is Cl-enriched (Reference Legrand, De Angelis and DelmasLegrand and Delmas 1984). This shows that the snow measurements include a certain amount of HCl not taken into account by aerosol sampling. Consequently the Cl/Na ratios determined in the snow may be either higher (the general case at the South Pole in recent decades) or lower than 1.8, according to the amount of HCl deposited together with the fractionated sea-salt aerosol.
The chemical cause of this fractionation is most probably the reaction of H2SO4 aerosol droplets with the sea-salt particles. This reaction has been proposed for temperate latitudes (Reference ErikssonEriksson 1960, Yue and others 1976, Reference Kritz and RancherKritz and Rancher 1980, Reference Clegg and BrimblecombeClegg and Brimblecombe 1985) as well as for Antarctic latitudes (Legrand and Delmas, in press).
How does R vary over the last millennium? Has it always been higher than 1.8 or did it reach values lower than 1.8 in the past? The variation range reported in Figure 1 is 2-6 and shows no definite trend over time. This relative stability of R is an indication that no radical change in the behaviour of the regional atmosphere around the South Pole occurred over the last millennium, at least with respect to this sea-salt fractionation. This is not the case, for instance, in the Dome C area, where a 220 year profile (Legrand and Delmas, in press) has shown that R changed radically between the nineteenth and the twentieth centuries, passing from values lower than 1.8 to values higher than 1.8. This could be an indication that what occurs in West Antarctica (the source of air masses crossing over the South Pole) may not necessarily be similar to what occurs over the East Antarctic plateau (where Dome C is located). It is worth noting that no fractionation at all has been observed for the sea salt in snow at Byrd Station, a West Antarctic site located at a lower elevation and with a more maritime weather regime than the South Pole.
Sulfate
In South Pole precipitation, excess sulfate generally represents more than 90% of the total sulfate. Its origin is probably essentially marine-biogenic, as proposed and discussed elsewhere (Reference Delmas, Briat and LegrandDelmas 1982, Legrand and Delmas, in preparation), except after large volcanic events of global significance, which may affect the Antarctic troposphere by the transport of volcanic sulfur (as H2SO4) through the stratosphere. As we have selected here firn sections in which the volcanic influence is assumed to be minor, the sulfate depth profile reported in Figure 1 is assumed to represent the input of marine-biogenic sulfate to the Antarctic continent. The variations observed are linked either to transport phenomena, or to changes in the marine-biogenic activity of the sulfur-production areas (assumed to be essentially the sub-Antarctic ocean). Variations in snow-accumulation rates may also be responsible for changes in concentration (Reference LegrandLegrand 1985). The variability of excess sulfate is not linked to that of sodium, probably because these componds are not carried by the same kind of aerosol (even though they originate in the same geographical areas).
The general trend of the sulfate profile reported in Figure 1 demonstrates that no major long-term change has occurred in the deposition of this compound at the South Pole. A similar conclusion can be drawn for Dome C over the last 220 years (Reference LegrandLegrand 1985), despite a significant change in the neutralization of H2SO4 by sea salt in the snow at this site.
Nitrate
The origin of nitrate found in Antarctic snow is unclear. From the examination of numerous new data sets, Reference Legrand and DelmasLegrand and Delmas (1986) have proposed N-fixation by lightning at mid-latitudes and subsequent long-range transport to high latitudes as the most likely source of Antarctic nitrate. Snow measurements have also demonstrated that NO3 ions are present over the ice cap as gaseous HNO3. The South Pole is the Antarctic location which exhibits the highest nitrate concentrations in snow: 1.67 μEq l−1, whereas 0.50-0.7 μEq l−1, 0.2-0.4 μEq l−1 and 0.6-0.8 μEq l−1 have been found for similar periods at D57 (Terre Adélie), Dome C (East Antarctic high plateau) and Byrd Station (West Antarctica) respectively. We have no satisfactory explanation, but this spatial variability, if not linked to pure deposition phenomena, could be an indication of a relatively short life span for HNO3 in the Antarctic atmosphere (of the order of a few days – a rather controversial assertion) and the injection of HNO3 from the upper to the lower tropospheric layers over the South Pole, i.e. at the center of the Antarctic vortex.
Finally, the sampling was unsuitable for detecting sudden NO3− increases (spikes lasting 1 or 2 years) like those found at D57 (Zanolini and others 1985).
The Little Ice Age and the chemical record – conclusion
Climatic changes can modify atmospheric circulation patterns and the strength of the aerosol sources. The consequences of such environmental changes are clearly recorded in deep ice cores encompassing the last glacial age. The Little Ice Age is a minor climatic change which affected mainly the Northern Hemisphere from the sixteenth to the nineteenth centuries. Thompson and others (1986) found evidence of this event in the δ18O, microparticle and conductivity profiles of the tropica! Quelccaya ice cap in Peru between about A.D. 1600 and about A.D. 1900. The δ18O results at the South Pole (.Jouzel unpublished) indicate no major changes over the last millennium. Our chemical depth profiles are also negative with respect to a possible record of the Little Ice Age (this period corresponds to depths of 72–14 m). In Antarctic ice, ice ages are marked by considerable increases in the sea-salt and crustal-dust inputs (see, for instance, Petit and others 1981 or Reference LegrandLegrand 1985). This observation suggests that an increase in the sea-salt deposition could also be expected at the South Pole for the period of the Little Ice Age. No such definite fluctuation in the natural variability of the Na and CI values, except possibly a soft “bump” from 1600 to 1700, is visible in Figure 1. With stronger winds, the Cl/Na ratio would be nearer to its reference level of 1.8 (Legrand and Delmas, in press), but our Cl/Na profile does not exhibit such a trend for these years. Finally, the concentrations of the acid anions (NO3, excess SO4 and excess Cl) are neither exceptionally high nor exceptionally low over this period.
From our measurements we therefore conclude that the environmental impact of the Little Ice Age, if any, was negligible at the South Pole.
On the other hand, this new data provides clear evidence of the stability of atmospheric environmental conditions in this area of the Antarctic over the last millennium, a conclusion which is not necessarily applicable to the entire continent.
These data also provide an assessment of the natural variability of the major chemical parameters measured along deep ice cores, a factor which should be taken into account in the interpretation of long-term glaciochemical profiles.
Acknowledgements
We thank J R Petit and M Legrand Tor their participation in the field work and for helpful discussions, and J Jouzel and A J Gow for making available their unpublished data. Financial support was provided by the CNRS (Programme Interdisciplinaire sur l’Environnement), T.A.A.F. (Terres Australes et Antarctiques Françaises) and the U.S. National Science Foundation (Division of Polar Programs).
