Ice-core location and recovery

Dyer Plateau is located in the middle of the Antarctic Peninsula (Fig. 1). The crest of the plateau (70°4O′ S, 64°52′ W, elevation 2002 m a.s.l.) is approximately 150km west of Dolleman Island which lies along the west coast of the Weddell Sea (Fig. 1). In the 1988/89 and 1989/90 austral-summer field seasons, several cores of various lengths (50-235 m) were drilled at three sites along a 10 km east-west transect across the ice divide. All cores were drilled and handled in the field with stringent contamination-control procedures (Reference Mulvaney and Peel.Mulvaney and Peel, 1988); and, after being transported frozen to ice-core processing facilities, the cores were stored at −30° C prior to the analyses.

Scope of study and analysis

To study snow chemical composition and its spatial variability, one core from each of the three sites was selected. Samples from one continuous section (4-5m in length) in each core were analyzed for ionic concentrations, and the oxygen-isotope ratios (δ¹⁸O) were measured on all of these cores to assist in dating. Table 1 provides information on the core sections analyzed, the sampling scheme and the dating of the sections.

Table 1.

The three drill sites on Dyer Plateau are located along a 10 km east-west transect across the ice divide. One section of each of the three cores was selected to correspond to the decade 1950-60. based on preliminary depth-age estimates. Dating was accomplished by counting δ¹⁸O and SO₄ ²⁻ cycles from the top of each core. Non-sea-salt sulfate calculation is according to Equation (1) and the Na⁺/SO₄ ²⁻ ratio in sea water (see text)

Continuous measurements of anion concentrations have been conducted on the top 112 m (dated AD 1761 1989) of a 235 m core (Cl) and on part (107 181 m, AD 1505-1784) of another 235 m core (C2) drilled 1 m away from core Cl. The relatively high annual snowaccumulation rate of 450 mm w.c. allowed analyses of 6-15 samples per year. A total of 3850 samples was analyzed to produce the 485 year record.

Sample preparation and analytical procedures

All snow and ice samples were decontaminated prior to chemical analysis. A pre-cleaned band-saw was used to decontaminate firn cores by removing a 20mm thick layer from all exposed surfaces. Below the firn/ice transition which occurs between 55 and 60m, ice samples were washed with ultra-pure deionized water. A procedural blank (frozen ultra-pure deionized water) was included with each batch of samples prepared for analysis. Before washing and melting, all sample preparation was conducted on a clean-air bench in a cold (−10° C) work room. The decontaminated samples were melted at room temperature in sealed plastic (polyethylene containers and were analyzed in a class-100 clean-room laboratory immediately after melting.

All cation and anion measurements were made by ion chromatography. The analytical procedures for the measurements of unions (Cl⁻, NO₃ and SO₄ ²⁻) have been described elsewhere (Reference Mosley-Thomoson, Dai., Thompson, Grootes, Arbogasl and Paskievkh.Mosley-Thompson and others, 1991). for the cations, a combination of Dionex FAS SEP CATION I and II columns was used with an eluant of 3 m M DAP (diaminoproponic acid) in 10 m M HCI. A cation micromembrane suppressor column (Dionex) was regenerated continuously with a solution of 10 m M KOH. The column combination and a column-switching technique allow simultaneous determination of sodium (Na⁺), ammonium (NH₄ ⁺), potassium (K⁺). magnesium (Mg²⁺) and calcium (Ca²⁺ ions. However, the Mg²⁺ and Ca²⁺ concentrations of most samples were below the detection limits of this analytical method (≈20 μg l¹). As discussed by other authors (Reference Saigne, Kirchner and Legrand.Saigne and others, 1987; Reference Mulvaney and Peel.Mulvaney and Peel, 1988), the laboratory measurement of NH₄ ⁺ in Antarctic snow is subject to serious contamination from the dissolution of NH₃ in ambient air. Although an effort was made to control the exposure of melting samples to the Laboratory atmosphere, the NH₄ ⁺ data are not considered contamination-free and are therefore not reported here.

For typical concentrations (0.2-2.5 μequl⁻) of Na⁺, Cl⁻, NO₃ ⁻ and SO₄ ², the analytical error of measurements is 10% or less, with negligible procedural blank values. Due to extremely low concentrations (≈0.1 μequl⁻¹), K⁺ measurements carry an error of about 100%.

Dating

Annual cycles of SO₄ ²⁻ concentrations and those of δ¹⁸O are very well preserved along the entire depth of all cores (Fig. 2). Therefore, all cores were dated by counting annual cycles in continuous δ¹⁸O and/or S0₄ ²⁻ concentration profiles. A “year” is defined as the firn or ice layer between two adjacent minima in δ ₁₈O or SO₄ ²⁻ concentration. The thickness of that layer is converted to water equivalent using density measurements along the core and represents net accumulation in that year.

Fig. 2.

An example of the well-preserved seasonal variations of δ¹⁸O and SO₄ ²⁻ in a section of a Dyer core (core W1; see Table 1 for detais). The core is dated by counting δ¹⁸O cycles from the top of the core, assuming that δ¹⁸O and SO₄ ²⁻ maxima occur in the austral summer.

For the most recent 200 years, the dating errors are negligible, as indicated by the accurate dates for several known volcanic horizons (e.g. Tambora in AD 1816).

No analyses were made on either core Cl or core C2 below 181 m due to very poor core quality. Core C2 suffers from poor core quality (probable loss of ice during drilling) between 112 and 130 m (AD 1706-63). In core C2 the dating below 112 m was refined by comparing prominent SO₄ ²⁻ horizons, believed to be volcanic, with those in a well-dated core drilled in 1985 at Siple Station (Fig. 1) at the base of the Antarctic Peninsula (Reference Mosley-Thomoson, Dai., Thompson, Grootes, Arbogasl and Paskievkh.Mosley-Thompson and others, 1991). Prominent volcanic events in the Dyer cores match those in the Siple Station core very well (Reference ThompsonThompson and others, 1994). The accumulated dating uncertainty at 485 years is estimated to be ±3 a.

Layer-Thinning Correction

As with all ice cores, annual snow layers at Dyer thin as they are buried under successive new snowfalls and are stretched due to the outflow of ice from the ice divide. A complementary geophysical study at the Dyer ice-core sites provides information about the characteristics of the ice flow (personal communication from C. Raymond and others, 1994). This allows the annual layer thicknesses or net annual accumulation to be corrected for thinning at depth (Reference ThompsonThompson and others, 1994), as shown in Figure 3. In the 112-130 m section of core C2 where significant core loss was encountered, it was difficult to determine accurately annual layer thicknesses. Data for this section are not shown in Figure 3; instead, a linear extrapolation of the thinning curve for this section (1706-63) is presented and subsequently used in the annual-flux calculation (see Figure 3 and caption).

Fig. 3.

Ice flow thins the thicknesses of old annual layers. This graph shows the ratios of annual layer thicknesses corrected far thinning over the measured layer thicknesses in cores C1 and C2. The shaded area indicates the poor-quality section of the core. The dashed line represents lineal regression fit to the data. which is used in the flux calculation fur the poor-quality section.

Calculation of nss SO₄ ²⁻

On Dyer Plateau, the extremely low concentrations of insoluble dust (Reference ThompsonThompson and others, 1994) indicate the lack of continental-dust aerosols in the atmosphere and hence in the snow. This leads to the assumption that the only important source of Na⁺ in snow is sea-salt aerosol; therefore, the non-sea-salt part of SO₄ ²⁻ in total S0₄ ²⁻ may be estimated this way:

(1)

where R ^Na = 0.12 is the ratio of SO₄ ²⁻ to Na⁺ in bulb sea water. For the spatial-variability study, where Na⁺ measurements are available. nss-SO₄ ²⁻ concentrations are calculated according to Equation (1), and are presented in Table 1.

As will be discussed later, the ratios of Na⁺ to Cl concentrations in snow adhere closely to that of bulk sea water, so it is probable that nearly all Cl⁻ in the snow is derived from sea salt. Consequently, nss SO₄ ²⁻ may he calculated using the SO₄ ²⁻/Cl ratio in sea water (R _Cl = 0.103):

(2)

Since only anion concentrations were measured in the 181 m section of cores C1 and C2. Equation (2) is used for nss-SO₄ ²⁻ flux calculations which are in turn employed in the annual time series of nss SO₄ ²⁻ for the last 485 years.

Flux calculation

The air-to-ground flux of a chemical species in an ice-core sample is calculated according to:

(3)

in which sample sizes represent accumulation in water equivalent. The sum of the fluxes of all samples in an ice-core year is the annual flux for that year, or:

(4)

where i denotes individual samples in one year.

Annual fluxes of Cl⁻, NO₃ and nss SO₄ ²⁻ for the 485 years; (1505-1989) contained in the 181 m core section have been calculated using Equations (3) and (4). Layer thinning affects the annual layer thickness as well as the sample sizes and consequently the calculated fluxes. To compensate for this effect, the annual flux calculated according to Equations (3) and (4) is multiplied by the ratio of the corrected (for thinning) annual layer thickness to the measured thickness, as shown in Figure 3. The thinning-corrected flux data, along with annual accumulation rates, are shown in Figure 4.

Fig. 4.

Cubic-spline smoothed data are shown for (a) annual accumulation rates, and fluxes of (b) Cl⁻, (c) NO₃ ⁻ and (d) nss SO₄ ²⁻ for the past 485years. The shaded area in (a) indicates the poor-quality section of the core, where accumulation rates may be underestimated. The identification of prominent volcanic events in (d) is based on dating and comparison with a similar record from Siple Station (Reference Mosley-Thomoson, Dai., Thompson, Grootes, Arbogasl and Paskievkh.Mosley-Thompson and others, 1991).

Snow Chemical Composition

In the first part of this study, core sections from three separate cores/sites are examined for snow chemical composition. The time period under discussion here is the decade 1950-60. No major volcanic eruptions or other unusual events affecting the atmospheric chemistry of the high southern latitudes were reported in this period. Therefore, the chemical composition of these samples is considered to be representative of the prevalent back-ground. snow chemistry in this area.

Average concentrations of the ions measured and other relevant data are presented in Table 1. It can be seen that K⁺ concentrations are significantly lower than the others and carry a large (100%) analytical uncertainty; hence K⁺ is a minor ion in snow. Since no other ions are present in concentrations above the detection limits of the ion chromatogntphy methods (with the exception of H⁺ and HCO₃ ⁻). Table 1 demonstrates that Na⁺, Cl⁻, NO₃ ⁻ and S0₄ ²⁻ are the major ionic species in snow. Selected meltwater samples were measured for pH for a qualitative indication of sample acidity (we are aware of the limitations of pH measurements of meltwater samples). A mean pH value of 5.3, with a range of 5.0-5.6, was obtained for a large number (≈800) of samples. These pH values suggest that H⁺ probably exists in the snow at concentrations comparable to those of the other major ions.

The ratios of Na⁺ to Cl⁻ concentrations in all snow samples (Table 1) are close to that of bulk sea water (0.86). This suggests that sea-salt aerosols over the Dyer Plateau region have not experienced significant fractionation which would result in lower Na⁺/Cl⁻ ratios (Reference Legrand and Delmas.Legrand and Delmas, 1988). Together with data from other Antarctic Peninsula locations (Reference Mulvaney and Peel.Mulvaney and Peel, 1988), these results suggest that formation of HCl from sea salt and atmospheric acids (most probably sulfuric and nitric acids) may be a phenomenon limited to the dry atmosphere of the interior Antarctic continent (Reference Legrand and Delmas.Legrand and Delmas, 1988).

Among the cations measured here, only H⁺ and Na⁺ have concentrations comparable to those of the major anions (Cl⁻, SO₄ ²⁻ and NO₃ ⁻). As discussed previously, the only major source of Na⁺ and Cl⁻ is probably sea salt. Since all cations and anions in sea salt must balance ionically, this leaves the non-sea-salt major anions (nss SO₄ ²⁻ and NO₃ ⁻) to balance the only other cation, H⁺. This leads to the conclusion that the principal chemical compounds from non-sea-salt sources are HNO₃ and H₂SO₄, consistent with results of snow chemistry studies in other parts of Antarctica (e.g. Reference Legrand and Delmas.Legrand and Delmas, 1984).

Spatial Variation of Chemistry on Dyer Plateau

Local spatial variations in snow chemistry are important when long-term changes in atmospheric composition over a large area are to be inferred from a single ice-core record. Knowledge of local variability helps in the selection of ice-core drilling sites that are most representative of the location. As part of a survey program to identify a deep ice-coring site, several shallow (50 m) and intermediate (100-235 m) cores were drilled in the Dyer Plateau area. The ionic concentrations in three cores across the ice divide are used to assess the spatial variability of the snow chemistry.

When comparing cores from closely located sites, it makes little sense to try to match individual samples since it is impossible to obtain snow samples deposited simultaneously, except perhaps when snow samples are collected in situ during a precipitation event. Even specific years in separate ice cores are difficult to compare, for the definition of the length (or the number of samples) of an ice-core “year” is somewhat arbitrary and the comparison may not be between the same lengths of the “year”, given the finite number of samples in a year, even if the same “year” is being evaluated. Thus, sections spanning several years are necessary for appropriate comparison. In the three columns of Table 1, the average ionic concentrations of each core section are listed for the three sites. Six to seven complete sulfate and δ ₁₈O cycles (years) are contained in each section with assigned dates. Since the core sections were selected assuming similar accumulation rates at all three sites, which are quite different (see Table 1), they correspond to different time intervals (Table 1, row 3). Nonetheless, the large number of samples in each section ensures that comparison between the sections or the sites can be made using average concentrations and variances.

The average concentrations of all ions and Na⁺/Cl⁻ ratios show rather small differences among the sites (Table 1), even with the somewhat different 6 year periods covered by the three core sections. These small variances indicate that there is no directional preference for the deposition of the ions, including the sea-salt aerosols. The exceptionally small variance of nss SO₄ ²⁻, coupled with the fact that its seasonal cycles are very regular, strongly suggests that deposition of this species is a large-scale atmospheric process and that there is little local influence on or post-depositional alteration of its signal in snow.

Accumulation and Flux in the Last 485 Years

As discussed elsewhere (Reference ThompsonThompson and others, 1994), Figure 4a shows that the net snow-accumulation rate on Dyer Plateau is relatively stable over the last five centuries. Nonetheless, several trends are discernible in the 485 year record. First, the accumulation rates between AD 1700 and 1750 (Fig. 4a; shaded area) fall well below the 485 year average. As discussed above, the core quality in this section is rather poor, which casts doubt on the validity of the reduced accumulation rate. Results from other Dyer cores currently under analysis may confirm or refute this period of low accumulation. However, strong positive cross-correlation with the Siple Station accumulation record (Reference ThompsonThompson and others, 1994) suggests that the low net accumulation at this time period may not be a spurious event, but may be more spatially extensive (Reference ThompsonThompson and others, 1994). Secondly, a small, gradual increase in net annual accumulation rate begins in the late 19th century and lasts to the present. Over this interval the annual accumulation increases from an average of 43 to about 52 cm w.e.a⁻¹. Similar increases in other ice-core records from the Antarctic Peninsula indicate that the trend may be widespread in the region (Reference ThompsonThompson and others, 1994).

This small increase in annual accumulation since the late 19th century may account for the concurrent slight increase in the NO₃ ⁻ and nss-S0₄ ²⁻ fluxes in the same time period (Fig. 4). No trend is apparent in Cl. From 1700 to 1760, all three anion species exhibit below-average fluxes; however, this probably reflects the loss of ice in this section of the core (zero concentration and flux for lost ice).

Among the three anions, Cl⁻ shows the largest temporal variability, which suggests that sea-salt deposition reflects the sporadic intrusion into the Dyer Plateau area of synoptic air masses containing sea-salt aerosols. The nitrate flux is relatively constant for the entire time period, as the largest annual NO₃ fluxes are only twice the average. This implies that the sources of N0₃ in the atmosphere are probably quite stable From year to year and on the time-scale of centuries. It is also clear that the deposition of NO₃ ⁻ is not significantly affected by fluctuations in local meteorological conditions. The nss-SO₄ ² flux profile is characterized by large peaks 2-5 times above the background level. These peaks represent several known and suspected vulcanic eruptions in the last 485 years (Reference ThompsonThompson and others, 1994), as indicated in Figure 4d. A detailed discussion of the nss-SO₄ ²⁻ volcanic chronology will be presented elsewhere.

Dry vs Wet Deposition of Anions

An examination of the relationship between the net accumulation rate and the ionic flux may reveal characteristics of the processes by which specific atmospheric chemical species are deposited on to the snow surface (Reference Davidson, Oeschger and LangwayDavidson, 1989). An atmospheric constituent, as either paniculate or gaseous aerosols, may be incorporated into or scavenged by water droplets or snow grains and be deposited as a direct result of precipitation (wet deposition); or the atmospheric species may fall without accompanying precipitation dry deposition). In the case of wet deposition, the flux can be expected to be proportionally related to the amount of precipitation or accumulation. Conversely, the flux by dry deposition should be independent of the accumulation rate. Thus, the relationship between flux and accumulation may be described simply as

(5)

where F _d and F _w represent dry- and wet-deposition flux, respectively. A _n is the snow-accumulation rate. The parameter b is related to the scavenging efficiency of an atmospheric species by precipitation and should be constant under given atmospheric conditions. If the flux plotted as a function of accumulation is a straight line, the intercept of the line should lie the dry-deposition part of the flux. A positive slope of the straight line would indicate that the deposition is dependent on precipitation (wet deposition).

The annual fluxes of Cl⁻, NO₃ ⁻ and nss SO₄ ²⁻ are plotted in Figure 5 as functions of the annual net accumulation rate reconstructed from the layer thick-nesses. Because flux is calculated as the product of both concentration and accumulation, there is a tendency for the slope to be positive. The dashed lines in Figure 5 represent the least-square linear fits to the data.

Fig. 5.

Annual fluxes of (a) Cl⁻, (b) NO₃ ⁻ and (c) nss SO₄ ²⁻ are shown as functions of annual net accumulation. Flux (in μg cm⁻² a⁻¹) and accumulation data for the period 1706-63 were removed from the data set due to poor our quality. Circles in (c) represent known and suspected volcanic years and the dashed lines are linear-regression fits to ”*” data.

All three linear fits exhibit positive slopes, suggesting that the deposition of all three species is somewhat dependent On precipitation. However, there are significant differences among the plots. For example, the linear regression line for NOδ3⁻ shows the best fit to the data (the correlation coefficient is (0.74; see Table 2). This is a strong indication that HNO₃ aerosols serve as efficienl condensation nuclei or that they are efficiently scavenged by precipitation. The linear fit to NO₃ flux has the smallest intercept (0.103 μg cm⁻² a⁻¹), suggesting that dry deposition contributes less than 10% of the average NO₃ flux (1.06 μg cm⁻²) deposited annually.

Table 2.

Statistical summary of the annual net accumulation and flux data, as illustrated in Figure 5. Data for 1706-63 are excluded, due to the probable loss of core during drilling. Values in parentheses are obtained after conspicuous volcanic years are removed from the data set

Sea-salt aerosols are generated by bursting bubbles of ocean spray. In Antarctica, must airborne sea-salt aerosol quickly settles along the coastal areas due to gravity (Reference Legrand and Delmas.Legrand and Delmas, 1985). Dried sea-salt aerosol particles in the dry Antarctic atmosphere are not considered good condensation nuclei; therefore, the amount of Cl⁻ in snow is not expected to be strongly dependent on accumulation rate. As expected, the Cl⁻ data show a large scatter in Figure 5a and the correlation between Cl⁻ flux and accumulation is rather poor (r = 0.42). In fact, even the weak correlation may reflect the bias toward a positive correlation resulting from the flux calculation, as discussed above.

The case of nss SO₄ ²⁻ is more complicated than the other two species. Although the nss-SO₄ ²⁻ flux, similar to Cl⁻. appears to be poorly correlated with accumulation (r = 0.45; Table 2), it is probable that large inputs from explosive volcanic sources distort the normal nss-SO₄ ²⁻ deposition accumulation relationship. When years with high nss-SO₄ ²⁻ fluxes from known and suspected volcanic eruptions are removed from the data set. the correlation between the remaining nss-SO₄ ²⁻ flux and accumulation is significantly improved, with r increasing from 0.45 to 0.56 (Table 2). By Contrast, removal of the volcanic years does not result in a significant change in either the No₃ or Cl⁻ regression results (Table 2). The nss-SO₄ ²⁻ intercept also increases from 0.20 to 0.38 μg cm⁻² a⁻¹ with the removal of volcanic years. This would mean that in non-volcanic years, only 25% or 0.38 μg cm⁻² a⁻¹ of the nss-SO₄ ²⁻ flux (average: 1.88 μg cm⁻² a^−l) may be attributed to dry deposition.