Antarctic mass balance has received increased attention because of the ongoing discussion of an enhanced greenhouse effect, where one of the important questions is how the warming would affect the Antarctic ice sheet. Accumulation rate is a fundamental parameter for evaluating the Antarctic mass balance. Spatial and temporal accumulation-rate variability can be obtained from seasonal records of stable isotopes and chemical species in firn and ice core. Firn- and ice-core analysis is also one of the few tools available for studying the environmental and climatic conditions of the Antarctic continent, where meteorological data are scarce.
The aim of this work is to improve knowledge of the snow accumulation-rate distribution in the northern Victoria Land region, for a better understanding of the ice-cap mass balance and the phenomena related to global change in Antarctica. Northern Victoria Land presents varied characteristics: it is a typical coastal region (with sudden and intense transport phenomena on a small scale with a strongly seasonal character), yet it also has plateau-like areas, resulting from the presence of the Transantarctic Mountains at a short distance from the sea, where transport phenomena on a global scale can sometimes be observed (Reference Udisti, Becagli, Traversi, Vermigli and PiccardiUdisti and others, 1998a). The chemical and physical analyses of firn and ice cores in areas with large altitude variations and very different annual snow accumulations permit the observation of environmental variations over several decades, with detailed analyses of the seasonal behaviour of the main and secondary sources of atmospheric aerosol. Moreover, the evaluation of the snow-accumulation rate can give experimental calibration to satellite measurements of the glacial area mass balance.
For these purposes, shallow firn cores were collected at three stations situated at different altitudes and distances from the sea, to obtain stratigraphic, isotopic (δ l8O) and chemical profiles of seasonal markers (H2O2, methanesulphonic acid (MSA) and non-sea-salt sulphate (nssSO4 2−)). The spatial and temporal distribution of these parameters allowed an evaluation of the seasonality and altitude effects on the snow composition. A reliable stratigraphic dating was obtained by critical comparison between the isotopic profile and a new chemical multiparametric profile, obtained using a linear combination of the three above-mentioned seasonal markers.
Stable-isotope stratigraphy has long been successfully used to determine palaeoclimatic conditions in deep ice cores recovered from both Greenland and Antarctica (Reference DansgaardDansgaard and others, 1993; Reference JouzelJouzel and others, 1993; Reference PetitPetit and others, 1999), but shorter time periods have also been investigated (Reference Peel, Mulvaney and DavisonPeel and others, 1988; Reference GoodwinGoodwin, 1991; Reference Mosley-Thompson, Dai, Thompson, Grootes, Arbogast and PaskievitchMosley-Thompson and others, 1991; Reference Isaksson and KarlénIsaksson and Karlén, 1994). The oxygen and hydrogen isotopic composition of polar snow is mainly related to the condensation temperature (Reference DansgaardDansgaard, 1964), but conditions prevailing in the source regions (Reference Merlivat and JouzelMerlivat and Jouzel, 1979), variation of sea-ice extension (Reference Bromwich and WeaverBromwich and Weaver, 1983) and seasonal variation in the timing of precipitation events (Reference Steig, Grootes and StuiverSteig and others, 1994) can also affect its value.
The temperature dependence of the δ values implies that the seasonal variations of this parameter are preserved in a δ 18O (or δD) record. However snowdrifts and homogenization processes in the firn can affect the original δ profile. The seasonal δ amplitude may decrease with depth, due to diffusion processes, which act more effectively in the firn layers when accumulation rates are low (Reference JohnsenJohnsen, 1977). This is particularly true on moving inland towards the centre of the Antarctic Plateau.
H2O2 is principally produced in the atmosphere by radical reactions originating from the photolysis of O3. Therefore, the maximum concentration of H2O2 is found in snow precipitation occurring in the period of maximum solar radiation, from late spring to late summer. The H2O2 which is contained in the snow can remain stable for long periods of time (Reference Neftel, Davies, Tranter and JonesNeftel, 1991). In areas with high accumulation rates, where neither wind-redistribution effects nor diffusion by summer melting significantly perturbs the structure of the snowpack, the seasonal variation of H2O2 content is one of the most reliable chemical methods for sequentially dating successive snow layers (Reference Sigg, Staffelbach and NeftelSigg and others, 1992; Reference Neftel, Fuhrer, Niki and BeckerNeftel and Fuhrer, 1993; Reference Piccardi, Udisti and CasellaPiccardi and others, 1994a; Reference UdistiUdisti, 1996).
The nssSO4 2− found in Antarctic snow is mainly produced by the oxidation of biogenic dimethylsulphide (DMS), volcanic emissions and crustal sources (Reference Delmas and JaeschkeDelmas, 1986; Reference Legrand, Delmas and CharlsonLegrand and others, 1988; Reference ShawShaw, 1988). In summer, the main contribution to nssSO4 2− comes from the biogenic DMS (Reference Savoie and ProsperoSavoie and Prospero, 1989; Reference Legrand, Feniet-Saigne, Saltzman, Germain, Barkov and PetrovLegrand and others, 1991; Reference DelmasDelmas, 1994; Reference Saltzman and DelmasSaltzman, 1995; Reference Wagenbach, Wolff and BalesWagenbach, 1996), so the nssSO4 2− aerosol-concentration maxima are found in the period of phytoplanktonic bloom and in the period immediately after that (January–February).
MSA derives solely from the oxidation of phytoplanktonic DMS, and is therefore an unequivocal marker of biological marine activity (Reference Saltzman, Savoie, Zika and ProsperoSaltzman and others, 1983, Reference Saltzman, Savoie, Prospero and Zika1986; Reference Legrand and SaigneLegrand and Saigne, 1988; Reference Savoie and ProsperoSavoie and Prospero, 1989; Reference Legrand, Feniet-Saigne, Saltzman, Germain, Barkov and PetrovLegrand and others, 1991; Reference Udisti, Casella, Piccardi, Restelli and AngelettiUdisti and others, 1993; Reference Piccardi, Udisti and CasellaPiccardi and others, 1994a; Reference Legrand and DelmasLegrand, 1995; Reference Saltzman and DelmasSaltzman, 1995; Reference UdistiUdisti, 1996; Reference Wagenbach, Wolff and BalesWagenbach, 1996; Reference De Mora, Wylie and Dickde Mora and others, 1997). The seasonal trend of MSA is the same as that already seen for nssSO4 2− (same DMS source and similar atmospheric transport processes). MSA summer maxima are often observed in snow (Reference Mulvaney, Coulson and CorrMulvaney and others, 1993; Reference Udisti, Casella, Piccardi, Restelli and AngelettiUdisti and others, 1993; Reference Langway, Osada, Clausen, Hammer, Shoji and MitaniLangway and others, 1994; Reference Piccardi, Casella and UdistiPiccardi and others, 1996b; Reference UdistiUdisti, 1996) and in aerosol (Reference Ayers and GrasAyers and Gras, 1991; Reference Ayers, Ivey and GilletAyers and others, 1991; Reference Saltzman and DelmasSaltzman, 1995; Reference Wagenbach, Wolff and BalesWagenbach, 1996; Reference De Mora, Wylie and Dickde Mora and others, 1997) measurements. However, some authors (e.g. Reference Mulvaney, Pasteur, Peel, Saltzman and WhungMulvaney and others, 1992; Reference Minikin, Wagenbach, Graf and KipfstuhlMinikin and others, 1994) show a dephasing between nssSO4 2− and MSA maxima, with MSA maxima occurring also in the winter period, probably due to post-depositional phenomena. In our measurements of snow-pit and shallow firn-core samples in northern Victoria Land, such dephasing was never found; thus, MSA was used, in combination with nssSO4 2−, for snow-pit and firn-core dating (Reference Piccardi, Udisti and CasellaPiccardi and others, 1994a; Reference UdistiUdisti, 1996).
2. Sampling Area
Terra Nova Bay is located in a border area between regions with different physiographic and glaciological characteristics (Fig. 1). The mountains of the Transantarctic Chain of northern Victoria Land, with their alpine glaciers and névé reaching heights of >3000 m a.s.l., are located to the north. To the south lie glaciers that drain local nunatak areas and in part the most peripheral area of the ice sheet. The principal outlet glaciers of Victoria Land, David, Reeves and Priestley Glaciers, drain into Terra Nova Bay. These constitute the Nansen Ice Sheet (Priestley and Reeves Glaciers) and the Drygalski Ice Tongue (David Glacier). Terra Nova Bay is one of the regions of Victoria Land where the East Antarctic ice sheet is nearest to the sea (about 75 km). The proximity of the ice sheet to the sea and the presence of the outlet glacier valleys are responsible for the confluence of katabatic winds in Terra Nova Bay (Reference Bromwich and KurtzBromwich and Kurtz, 1984). The intensity and persistence of the predominantly wintry katabatic winds create a permanent polynya in the bay.
Information about snow accumulation and mean annual temperatures in the Ross Sea area comes from meteorological, stratigraphic and chemical analyses of snow pits and firn cores and from 10 m borehole temperatures (Table 1). In the Terra Nova Bay area, mean net accumulation values of about 150 kg m−2 a−1 were measured for small local glaciers fed by wind-blown snow, and average ablation values of about 200 kg m−2 a−1 were measured for the blue-ice areas (Reference FrezzottiFrezzotti, 1993). The Terra Nova Bay area is located between the surface balance isopleths of 200 and 150 kg m−2 a−1 on the map of surface mass-balance rates (D″–D′ drainage system) drawn by Reference Giovinetto and BentleyGiovinetto and Bentley (1985).
2.1. Site selection
Several authors (Reference WatanabeWatanabe, 1978; Reference Pettré, Pinglot, Pourchet and ReynaudPettre and others, 1986; Reference GoodwinGoodwin, 1990) have pointed out the influence of katabatic winds on the net annual accumulation. These winds are the main element acting on the glacier surface that causes high surface ablation rates and snow transport/redistribution and makes the interpretation of chemical, isotopic and physical profiles difficult. Surface air does not blow radially and uniformly away from the highest part of the ice sheet, but rather is highly irregular, with regions of pronounced confluence and divergence. To identify the sites as little disturbed as possible by the action of katabatic winds, a study of the surface wind field of the the Terra Nova Bay area (about 100 000 km2) was carried out (Reference FrezzottiFrezzotti, 1998). Remote sensing can provide information about the surface wind field through the location, direction and aerial extension of epiglacial aeolian morphology and ablation-area survey. The survey was carried out using aerial photographs and satellite images at scales between 1: 50 000 and 1: 100 000. Field investigation on sastrugi, snowdrifts and other aeolian morphologies was performed in order to check the remote-sensing survey. A final map was drawn at 1: 500 000 scale (Reference FrezzottiFrezzotti, 1998). Comparison between different documents, taken several years apart, and field surveys shows the time persistence of the shape and size of the aeolian morphologies, at the analysis scale. These analyses of the surface wind field of Terra Nova Bay indicated that McCarthy Ridge (163°01′ E, 74°33′ S), Styx Glacier (163°45′ E, 73°55′ S) and Hercules Névé (164°58′ E, 73°07′ S) were least influenced by the katabatic winds (Fig. 1).
McCarthy Ridge is a plateau area of approximately 100 km2, located at about 700 m a.s.l. between the Eisenhower Range to the east and the Nansen Ice Sheet to the west, with the Priestley Glacier valley to the north and the Reeves Glacier valley to the south. This sampling site is in the path of the katabatic winds that flow down along the Priestley and Reeves valleys but is protected from these winds by the Eisenhower Range. The site is 40 km from the permanent polynya of the Terra Nova Bay.
Styx Glacier plateau, approximately 150 km2 in area, is located between the Campbell and Tinker Glacier valleys, about 50 km from the sea (Wood Bay) and at about 1700 m a.s.l. The analyses of the wind field showed that the katabatic winds coming from the west and southwest pass over the Deep Freeze Range and affect the leeward area of this chain as seen from the Campbell Glacier right hydrographic. On the eastern part of the Campbell Glacier valley, where the Styx Glacier plateau lies, the wind effect is strongly reduced, causing some snowdrifts in the west–east direction.
Hercules Névé is a plateau of approximately 1100 km2, at about 3000 ma.s.l., and about 75 km from the sea (Lady Newnes Bay); together with the adjacent Evans Névé, it makes up the largest ice cap of northern Victoria Land. The sampling site is located on the ice divide between the glaciers that flow into the Ross Sea and those that flow into the Pacific Ocean. The distribution of the wind morphology exhibits a radial distribution, showing that the area is not affected by katabatic winds. Temperature values of −34.5° and −33.1°C at 7.5 and 10 m depth, respectively, were measured in a borehole.
Three firn cores were drilled during the Italian Antarctic Campaigns 1991/92 and 1992/93 at McCarthy Ridge, Styx Glacier and Hercules Névé stations (core lengths 6.5, 9 and 7.5 m, respectively).
The firn-core sections (about 50 cm long; d = 10 cm) were kept frozen and brought to the cold laboratory of the University of Milan. After decontamination by removing a thin external layer, subsamples were obtained by cutting the firn cores every 3 cm (Hercules Névé) and 5 cm (McCarthy Ridge and Styx Glacier). Chemical, stratigraphic and stable-isotope analyses were carried out on these samples.
3.2. Firn-core analysis
Before samples were taken for the chemical and isotopic analyses, the firn cores were studied to provide a detailed logging of the internal characteristics of the firn. The size and shape of the crystals and pores and the presence and thickness of each ice crust were measured to obtain a continuous stratigraphy of these characteristics.
3.2.2. Isotopic composition
The measurement of oxygen isotopic composition was carried out according to the Reference Epstein and MayedaEpstein and Mayeda (1953) technique of isotopic equilibration of CO2 with water. The McCarthy and Styx samples were prepared manually and measured by a Finnigan Delta S mass spectrometer for their 18O/16O ratios, while the Hercules Névé samples were prepared using an automatic equilibration device (Isoprep 18) in line with an Optima VG mass spectrometer.
The results are reported as δ units per mil vs Vienna standard mean ocean water (V-SMOW) isotopic standard (Reference GonfiantiniGonfiantini, 1978).
The mean standard deviation of oxygen measurements was ±0.07 per mil (1σ) for the manual preparations and ±0.10 per mil for the automatic preparation device.
3.2.3. Chemical analysis
The determination of H2O2 was carried out using a flow analysis method with a spectrofluorimetric detector. This method is based on the formation of a fluorescent dimer from the reaction of p-hydroxyphenylacetic acid with H2O2 in the presence of peroxidase (Reference Hwang and DasguptaHwang and Dasgupta, 1985). The measurements were performed with an RF-551 Shimadzu spectrofluorimetric detector connected to a peristaltic system.
The determinations of MSA and of the ionic species necessary to calculate nssSO4 2− (total SO4 2−, Cl−) were carried out by ion chromatography (IC) using a Dionex IC 4000i apparatus, Dionex AS4A and AS11 anion separator columns and a conductometric detector with a membrane conductivity suppressor.
The analytical determinations and method performances for these substances are discussed in detail in Reference Piccardi, Udisti and CasellaPiccardi and others (1994a), Reference Udisti, Bellandi and PiccardiUdisti and others (1994) and Reference UdistiUdisti (1996).
4. Results and Discussion
4.1.1. McCarthy Ridge
The firn core has a fairly undifferentiated stratigraphy (Fig. 2a) with a very weak variation in the pore size. According to Reference KuhnKuhn (1970), these crystals are characteristic of normal snowfalls with formation temperatures below −20°C. No ice crusts were observed along the core.
4.1.2. Styx Glacier
In the upper part, from the surface to 4.2 m depth (Fig. 2b), there is a large, irregular porosity with an average pore dimension of 2–3 mm. Inside this area there are levels composed of denser facies (average porosity 1 mm) with thickness ranging from a few cm (probably due to the very cold snowfall) to 20 cm (probably indicative of an entire season).
The intermediate part, at 4.20–8.30 m depth, is less porous, with an average pore size of about 1 mm and more regular pores. This area is much more homogeneous than the upper area and presents less dense levels (maximum thickness 5 cm, porosity 2–3 mm).
The lower part, from 8.20 m depth to the bottom of the hole, has a pore size of <1 mm. Ice crusts 0.5–2 mm thick were observed.
4.1.3. Hercules Névé
Here, the transition from the less dense to the denser facies occurs at 3 m depth, in a gradual but faster way (Fig. 2c). Ice crusts are present here too, especially in the superficial part, with a thickness of 1–2 mm. These are also summer crusts, but are probably due to vapour condensation formed in sunny periods since strong katabatic winds are not present here. In this station as well, the ice crusts cannot be used for annual dating.
4.2. Dating and accumulation rate by δ 18O and chemical profiles
Table 2 reports the basic statistical parameters for nssSO4 2−, MSA, H2O2 and δ 18O values measured at the three stations.
The nssSO4 2− concentration refers to the contribution of sulphate not coming from sea spray and is calculated using the following formula:
where [SO4 2−] and [Cl−]are the concentrations measured in the sample and 0.139 is their concentration ratio (w/w) in sea water. As only inorganic anions were determined by IC in all the firn-core samples (sample volume was often insufficient for cation analysis), we used Cl− concentration as the sea-spray marker. Actually, in northern Victoria Land the more reliable sea-spray marker is Na+ because Cl− may come from other sources (e.g. volcanic HCl) or may undergo fractionating effects able to change the original sea-spray composition during transport (Reference Piccardi, Becagli, Traversi, Udisti, Colacino, Giovannelli and StefanuttiPiccardi and others, 1996a; Reference Udisti, Traversi, Becagli and PiccardiUdisti and others, 1998b). Névértheless, previous nssSO4 − measurements on snow-pit samples collected at the same stations indicated that the use of Cl− or Na+ as a sea-spray marker gives very similar nssSO4 2− values when the nssSO4 2− concentration levels are not too low (Reference Piccardi, Udisti and CasellaPiccardi, 1994a; Reference UdistiUdisti, 1996). A satisfactory correlation (R = 0.97, slope = 0.91) was obtained between nssSO4 2− values calculated by sodium concentration and by chloride concentration, for 184 of the Hercules Névé samples for which the volume was sufficient to determine the Na+ content. Similar results were obtained by Reference Ayers, Ivey and GilletAyers and others (1991) in aerosol measurements.
Negative nssSO4 2− values were sometimes found in the samples. These were usually found in snow and ice samples with high sea-spray content, collected in winter (low biogenic activity) at low-altitude coastal stations (Reference Maupetit and DelmasMaupetit and Delmas, 1992; Reference Piccardi, Udisti and CasellaPiccardi and others, 1994a, Reference Piccardi, Becagli, Traversi, Udisti, Colacino, Giovannelli and Stefanutti1996a, Reference Piccardi, Casella and Udistib; Reference UdistiUdisti, 1996; Reference Wagenbach, Wolff and BalesWagenbach, 1996).
Figure 3 shows the smoothed (order 5) nssSO4 2− and original Cl− profiles at the three stations. The “background” chloride values decrease with increasing altitude, showing the highest mean value at McCarthy station (1570 ± 1037 μg L−1) and the lowest at Hercules Névé (279 ± 366 μg L−1). The depth/concentration profiles are modulated by sudden spikes, occurring mainly in the winter period and probably due to salt storms. Comparison between the nssSO4 2− and Cl− profiles shows that the nssSO4 2− negative values, especially evident for the lowest station, are generally coincident with the Cl− winter peaks.
The negative values may result from an incorrect interpretation of the SO4 2−/Cl− or Na+ ratio in sea water, from analytical errors in anion determination or from fractionating effects of the sea-spray components during the transport of the marine aerosol (Reference Keene, Pszenny, Galloway and HawleyKeene and others, 1986; Reference Hawley, Galloway and KeeneHawley and others, 1988; Reference Piccardi, Becagli, Traversi, Udisti, Colacino, Giovannelli and StefanuttiPiccardi and others, 1996a; Reference WagenbachWagenbach and others, 1998). An unequivocal opinion about the correction of the nssSO4 2− calculation to avoid negative values is lacking, so in Table 2 we report the statistical parameters related both to all the values (positive and negative) and to the positive ones only.
For dating purposes, we applied the multiparametric method for chemical dating proposed by Reference UdistiUdisti (1996). An essential abstract of the method is reported here.
The chemical concentration/depth profiles (Figs 4a–c, 5a–c and 6a–c) were normalized by dividing each analytical value by the maximum value found in the same temporal series in the interval of amplitude 2q + 1 (the chosen value, q data before and q data after) around the considered analytical value. q is an arbitrary value that permits better identification of summer peaks even when a low summer peak for a year is near to very high summer peaks belonging to adjacent years. The ideal aim of the normalization procedure is to attribute a value of 1 to summer peaks, independently of their analytical value. For all the firn cores considered, an interval value of 21 was chosen (q = 10). The smoothed sum (Figs 4d, 5d and 6d) of the three normalized profiles gives a new depth profile with maximum values of 3 for depths at which all the parameters show a summer peak (normalized height = 1), and minimum values when the temporal series do not present high concentration values.
For dating purposes, the negative nssSO4 2− values are not a problem, because the normalized sum is built to spot the summer maxima, while the negative values are mainly found in the winter period.
For each firn core, the definitive dating decision comes from visual comparison of the relative normalized sum (chemical seasonal marker) and the isotopic profile.
The possible temporal dephasing between the different parameters used for chemical and isotopic profiles can sometimes result in a small shift between the relative summer maxima. However, the agreement between the two temporal series is generally good.
4.2.2. McCarthy Ridge
Figure 4 shows the depth profiles of the analytical parameters used for the McCarthy Ridge firn-core dating. Figure 4a–c show the concentration/depth profiles of H2O2, nssSO4 2− and MSA, respectively. A sharp seasonal behaviour is evident for all the concentration profiles, with summer peaks relatively low in some years and very high in others. For example, the H2O2 profile shows a summer concentration maximum of 70 μg L−1 at around 220 cm depth and of 150 μg L−1 at 600 cm depth. This summer peak is also very broad (550–650 cm), which could indicate a large summer snow accumulation for that year or a very low winter snow accumulation for two or more successive years (see below).
The nssSO4 2− profile (Fig. 4b) includes quite a large number of negative values (about 25%) and shows more homogeneous summer peaks. By contrast, the MSA concentration profile (Fig. 4c) shows higher values for the first 2 years, with a maximum value of 143 μg L−1 at 210 cm depth.
The observed variations of the δ 18O values obtained along the profile (Fig. 4e) are related to seasonal temperature variations. The δ 18O seasonal trend seems to be quite regular, except for two small intervals at depths of 300–450 and 580–670 cm. At about 240 cm, a large negative δ 18O value suggests a low precipitation temperature.
For this station, the best seasonal tracer seems to be H2O2. However, a comparative examination of all three concentration profiles can give a more reliable summer peak identification. Figure 4d shows the relative normalized smoothed (order 3) sum. In the normalized sum, it is possible to recognize nine well-defined summer peaks, indicated by the numbers 1–5, 6–8 and 10, and two small peaks (5′ and 8′). Peak 9 will be discussed later. The main problem with the dating interpretation is the large peak at 550–650 cm depth that could either be single or include three summer peaks (8, 8′ and 9).
The weak changes in firn facies evidenced by visual stratigraphy are not relevant for dating purposes. It is only possible to observe, close to the bottom section, a fairly good correlation between summer peaks 8–10 and the denser firn facies. However, the 2 mm ice crusts at 560 cm depth correspond to H2O2 peak 8, the widest peak of all the stratigraphy.
The chemical profile may be compared to the δ 18O profile (Fig. 4e). Peaks 1–5, 6–8 and 10 are fully confirmed. The peak indicated as 5′ is not visible in the δ 18O profile and is discarded. The same is true for peak 8′, which is very small in the two series. The isotopic profile contains, instead, peak 9, which is visible with difficulty in the normalized sum. We consider this as a summer peak, because of its well-defined shape in the isotopic profile. In the end, we found 10±1 summer peaks. An uncertainty of 10% is acceptable for estimating a reliable mean snow-accumulation rate. The ten peaks identify nine accumulation years in the depth range 160–680 cm, so an accumulation rate of 57.8 cm snow a−1 can be calculated. By considering the annual mean density of each year (calculated on about 30 firn-core sections), a mean accumulation rate of 260 kg m−2 a−1 is estimated. This result is in full agreement with the value of 270 kg m−2 a−1 found for a 2 m deep snow pit excavated at the same station in the Italian Antarctic Campaign 1990/91 (Reference Piccardi, Udisti and CasellaPiccardi and others, 1994a). Referring to the mean accumulation rate, the first peak may be ascribed to summer 1988/89 and the last peak to summer 1979/80.
4.2.3. Styx Glacier
The concentration/depth profiles for the Styx Glacier station of H2O2, nssSO4 2− and MSA are shown in Figure 5a–c, respectively. Here too, a clear seasonal behaviour, with summer maxima and winter minima, can be observed for all the components. The highest H2O2 summer peaks are located at the surface (about 220 μg L−1 and at 440 cm (about 110 μg L−1). The nssSO4 2− concentration/depth profile (Fig. 5b) shows very few negative values (2.3%, as opposed to 25% at McCarthy Ridge). For this station, higher than McCarthy Ridge, the sea-spray contribution is much lower (Fig. 3) and the contribution of secondary marine aerosol (nssSO4 2− and MSA) becomes relatively more important. The highest nssSO4 2− concentration was measured at the second summer peak (about 270 μg L−1), and corresponds to the highest concentration for the MSA profile (Fig. 5c; about 58 μg L−1).
The seasonal variations of δ 18O observed along the firn core recovered at Styx Glacier (Fig. 5e) show a regular pattern, with a slight negative trend of the mean value in the first 350 cm followed by an extremely high and wide peak (at about 425 cm) probably corresponding to relatively high depositional temperatures. This peak is almost coincident with the large H2O2 peak mentioned above.
The normalized sum profile is plotted in Figure 5d. By comparing the normalized sum and the δ 18O profiles, 18 sharp summer peaks (Nos 1–10, 12, 13–14, 16–21) can be observed. Peaks 8–10 have different heights but exactly the same shape in the chemical and isotopic profiles. Peaks 11, 12′, 15 and 19 are absent or very low in the normalized sum profile but well defined in the δ 18O profile; by contrast, peak 17′ is evident in the chemical profile but not in the δ 18O profile. By considering peaks 11, 17′ and 19 as more reliable, it is possible to identify 21±2 summer peaks (uncertainty of 10%), corresponding to 20 years of accumulation in the depth range 65–925 cm. A mean snow accumulation rate of 43.0 cm a−1 can be calculated. By considering the annual mean density for each year (calculated on about 36 firn-core sections), a mean snow-accumulation rate of 203 kg m−2 a−1 is estimated. Using this mean accumulation rate, the first peak can be related to summer 1990/91, and the last peak to summer 1970/71.
The value of 203 kg m−2 a−1 is considerably higher than that of 160 kg m−2 a−1 calculated for a 2 m deep snow pit excavated during the 1990/91 Italian expedition at the same station (Reference Piccardi, Barbolani, Bellandi, Casella and UdistiPiccardi and others, 1994b; Reference UdistiUdisti, 1996). The higher accumulation rate can be explained by very high snow accumulation for some years: 1982 (peaks 9–10: 320 kg m−2 a−1), 1979 (peaks 12–13: 316 kg m−2 a−1) and 1977 (peaks 14–16: 334 kg m−2 a−1). The last two of these high-accumulation periods include the most uncertain dating (the disregarded 12′ and 15 peaks). If just one of these is an annual peak, we obtain a mean annual accumulation rate of 184 kg m−2 a−1, which is close to the previously reported one.
Visual stratigraphy cannot be used as a tool for dating purposes. However, the comparison with chemical and stable-isotope records permits a correlation of the thin layers of facies 1 and 2 with summer peaks. The denser layers at depths of 2, 2.7, 6.8 and 7.15 m correspond to peaks 4, 6, 16 and 17, respectively. The other layers and the ice crusts are not related with summer peaks.
4.2.4. Hercules Névé
At this station, located at 3000 m a.s.l., the mean H2O2 concentration value is slightly higher than the mean values calculated for the two lower stations (Table 2), and the H2O2 depth profile (Fig. 6a) shows relatively high values even in winter. Anyway, the seasonal pattern is clear in the whole firn core except for the three oldest years. By contrast, this depth range (from 650 cm to the bottom) is well resolved by the temporal trends of the other two parameters, particularly nssSO4 2− which have unclear profiles at other depths (e.g. at 200 and 350 cm depth; Fig. 6b).
The sea-spray contribution at this station is very low (Fig. 3), and the secondary marine aerosol is dominant with respect to the sea-spray contribution, at least during summer (Reference Piccardi, Becagli, Traversi, Udisti, Colacino, Giovannelli and StefanuttiPiccardi and others, 1996a). As a result, the mean nssSO4 2− concentration is >60% of the total sulphate, and only a few negative values (about 5%) were observed, all during the winter periods.
The MSA concentration profile (Fig. 6c) shows a sharp decrease of the summer peaks in the first 3 years. The summer maxima decrease from about 55 μg L−1 for the first year to around 10–20 μg L−1 in the depth range 200–760 cm. The MSA values are lower than those measured at the lower stations, and for some samples they are only 10 times higher than the detection limit (0.2 μg L−1). Not withstanding this, the seasonal pattern of both the original and normalized profiles is well defined, and this parameter has an important role in dating, even at this high station. The MSA determination is a direct analytical measurement and is not subject to incorrect interpretation, unlike nssSO4 2− determination.
The δ l8O record obtained at Hercules Névé (Fig. 6e) shows strong seasonal variations in the uppermost 4 m of the firn core (up to a value of 9–10‰), followed by a smoother trend in the part below, suggesting the presence of homo-genization effects, probably caused by water-vapour diffusion in the porous firn. The scarce precipitation at this high-altitude site may explain the partial obliteration of the seasonal isotopic signal. The chemical species are not affected by this phenomenon. Therefore, a comparison of isotopic and chemical profiles is possible only for about the first half of the firn core.
By assuming the chemical profile (normalized sum; Fig. 6d) as a discussion base, 20 well-defined summer peaks (Nos 1–20, Fig. 6d) and 4 not completely resolved, probable summer peaks (peaks 4′, 5′, 7′ and 9′) can be distinguished. Fortunately, all the ambiguous peaks are located in the first part of the firn core, so it is possible to confirm their correspondence with summer peaks using the isotopic profile. By comparing Figure 6d and e, we can observe that peak 9′ has the same shape and relative height in the chemical and isotopic profiles. Peak 7′ is missing in the isotopic profile. The 4–4′ and 5–5′ pairs are apparent in both profiles, even if the relative heights are reversed. We assumed peak 9′ to be a summer peak and we discarded the 7′ peak. The 4′ and 5′ peaks remain ambiguous.
In this firn core, the homogeneity of the firn facies does not allow one to follow the seasonality at all. Although a few ice crusts could be related to certain peaks (1 and 3 m depth), the whole stratigraphy does not seem to be related to the chemical and stable-isotope records.
In conclusion, we can identify 22 ± 2 summer peaks (uncertainty of 10%). The resulting 21 annual snow layers correspond to the 20–740 cm depth range, leading to a mean snow-accumulation rate of 34.3 cm a−1. From the mean annual density values (measured every 5 cm on the entire firn core) we obtained a mean accumulation rate of 159 kg m−2 a−1 in the period 1972–92. This mean accumulation value is in good agreement with those reported by Reference MaggiMaggi and others (1998) and Reference StenniStenni and others (1999) for the same station, and by Reference Allen, Mayewski, Lyons and SpencerAllen and others (1985) for a nearby area (182 kg m−2 a−1; core M1, Rennick Glacier area).
4.3. Effects of altitude and seasonality on chemical composition
Figure 7 shows the altitude effect and the seasonal pattern of Cl−, totSO4 2−, nssSO4 2−, MSA and H2O2 for the three firn cores. Each plot reports the mean concentrations (calculated on total, winter and summer samples) as a function of the station altitude. The summer and winter samples were selected on the basis of the sum normalized profiles. Each sample was considered to be a “summer sample” when its normalized sum value was higher than one-half of the maximum peak value, and a “winter sample” otherwise.
The Cl− concentration (Fig. 7a) is highest at the lowest station, with a maximum mean value in winter (2180 ± 1870 μg L−1), but it decreases quickly in the first altitude step (700–1700 m). In the second step, the concentration decrease with altitude is less marked. The seasonal pattern shows a similar trend: the mean concentrations vary a lot with seasonality (winter maximum) at McCarthy, but such differences gradually decrease when the altitude increases. At Hercules Névé station (3000 m a.s.l), a seasonal trend of the mean values is lacking, although the Cl− spikes occur mainly in winter (Fig. 3). By contrast, nssSO4 2− concentrations (Fig. 7c) are higher in summer at all stations, and the decrease with increasing altitude is more gradual. For the two higher stations, the mean concentrations (total, summer and winter) are very similar (around 50 μg L−1), showing similar seasonal patterns. From the comparison between the Cl− and nssSO4 2− patterns, an altitude-induced fractionation effect is evident. This is due to the different decreasing trends: as the altitude increases, the secondary marine aerosol (biogenic source) becomes progressively more important than the sea-spray contribution. totSO4 2− (Fig. 7b) shows a trend which combines the Cl− and nssSO4 2− patterns. In fact, both sources (sea spray and biogenic marine) are important for this component, and they have contrasting seasonal patterns. So the mean totSO4 2− concentration values show a decrease with increasing altitude similar to that of Cl−, and their seasonal variation is very low, because the winter sea-spray maxima are counterbalanced by summer maxima of the biogenic contribution. It is interesting to note that at the lowest station the seasonality of sea-spray is prevalent (winter maxima), whereas at the highest station we observe higher mean concentration values in the summer (the nssSO4 2− contribution becomes dominant).
The decreasing trend of nssSO4 2− with increasing altitude and its seasonal pattern are fully confirmed by the MSA concentration trend (Fig. 7d), showing that the biogenic contribution is dominant with respect to the other sources (volcanic, crustal) for nssSO4 2− at coastal sites. Even at the highest station, MSA shows marked seasonal variations, with summer maxima. A sharp seasonal pattern is also shown by the H2O2 mean concentration profiles (Fig. 7e). Unlike the previous patterns, this component does not show a concentration decrease with altitude. This is because H2O2 is produced directly in the atmosphere over the whole Antarctic ice sheet, and the aerosol transport processes and their fractionating effects do not affect its atmospheric concentration levels. The H2O2 concentration in snow should therefore be more closely related to the accumulation rate, as is shown by its progressive increase with altitude.
The different contributions of primary and secondary marine aerosols as a function of altitude and seasonality heavily affect the snow composition. In winter, at the lowest station, the sea-spray content is >90% of the total marine contribution. In summer, at the highest station, the secondary marine aerosol (nssSO4 2− and MSA) constitutes about 50% of the marine input. These results fully confirm those previously obtained for snow-pit samples collected at the same stations (and at another intermediate-altitude site) over a shorter time interval (Reference Udisti, Becagli, Castellano, Traversi, Vermigli and PiccardiUdisti and others, 1999).
4.4. Comparison of annual accumulation rates
Figure 8 shows the annual accumulation rates for the three sampled stations. The McCarthy firn core shows the largest annual variability, with a maximum value of 426 kg m−2 a−1 and a minimum of 130 kg m−2 a−1 annual accumulation rates at Styx Glacier plateau are 111–335 kg m−2 a−1, and the Hercules Névé station shows the lowest annual changes, 83–264 kg m−2 a−1. Although the dating of each firn core is subject to an uncertainty of 10%, some observations can be made from the comparison of the annual accumulation trends. A very good correlation is shown among all the firn cores in the years before 1983. We can recognize years characterized by high snow accumulation (1982, 1981, 1979, 1977, 1973) and drier years (1980, 1978, 1974). For the upper part of the firn cores, the correlation is less evident, but we must also consider that in this part we also find the main uncertainties in the firn-core dating. From visual observation of the annual trends at the Styx Glacier and Hercules Névé stations, 1991 and 1990 are also characterized by high snow-accumulation rates. This result may be confirmed by the field observation of the high summer snow precipitations during the Italian Antarctic Campaigns 1990/91 and 1991/92 at Terra Nova Bay.
The accumulation-rate distribution of all the available data for northern Victoria Land shows an overall negative correlation with altitude and distance from the coast (Table 1). The maximum values are measured at McCarthy Ridge and Aviator Glacier station, the sites closest to the coast and at the lowest elevations; low accumulation rates are measured at Hercules Névé and Priestley Névé stations, located at high altitudes and far from the sea. The geographical distribution of the accumulation rates indicates that the depositional regime may depend on the orography and the direction of the storm. The E10 site, studied by Reference Allen, Mayewski, Lyons and SpencerAllen and others (1985), shows anomalously low accumulation values, probably because of local effects. For the northern Victoria Land area (D″–D′ drainage system), Reference Giovinetto and BentleyGiovinetto and Bentley (1985) and Reference Vaughan, Bamber, Giovinetto, Russell and CooperVaughan and others (1999) reported net surface accumulation rates on grounded ice of 260 and 188 kg m−2 a−1, respectively. The improved estimation of 170 kg m−2 a−1, calculated as the mean value of the data shown in Table 1, represents a decrease of 35% and 10% with respect to previous estimations. Reference FrezzottiFrezzotti (1997) estimated a 24.8 ± 9.5 Gt a−1 ice discharge for the glaciers of northern Victoria Land (D′′–D′ drainage system). By comparing these data to Reference Giovinetto and BentleyGiovinetto and Bentley’s (1985) accumulation value of 40.9 ± 20% Gt a−1, Reference FrezzottiFrezzotti (1997) suggested a combined positive mass balance of these glaciers due to the 40% surplus of snow accumulation. Considering the estimated ice-discharge rate of 24.8 ± 9.5 Gt a−1 for the D′′–D′ drainage system (148 000 km2) and assuming a steady state, a surface snow-accumulation balance of 167 kg m−2 a−1 is calculated. This is very close to the net surface accumulation rate of 170 kg m−2 a−1 calculated in this paper. The difference between our results and previous ones may result from overestimation of snow accumulations due to the use of historical data from snow-pit stratigraphy, as suggested by Reference Frezzotti, Tabacco and ZirizzottiFrezzotti and others (2000) for eastern Dome C.
4.5. Stable-isotope/temperature relationship
Linear relationships between δ 18O and the mean annual surface temperature at the sampling site (δ/T) have been observed for different regions of Antarctica (Reference Lorius and MerlivatLorius and Merlivat, 1977; Reference Peel and ClausenPeel and Clausen, 1982; Reference Peel, Mulvaney and DavisonPeel and others, 1988; Reference Dahe, Petit, Jouzel and StievenardQin and others, 1994). In a general way, the geographical dependence of this relationship relies mainly on the difference in the local climatological situations and in the moisture of the source regions supplying the precipitations over the different Antarctic areas. Recently, the use of the δ/T relationship to correlate the ice-core isotopic profiles to the past temperature variations has been partly questioned, mainly regarding the Greenland ice sheet, because of the different spatial and temporal patterns of such a relationship (see Reference JouzelJouzel and others, 1997, for a comprehensive discussion). However, Reference JouzelJouzel and others (1997) suggested that >70% of the isotopic changes observed in the ice-core records are explained by temperature changes alone. Therefore, better knowledge of δ 18O distribution in the Antarctic ice sheet is needed in order to obtain a database that can be used to validate isotopic models (both Rayleigh-type distillation and general circulation models). The spatial δ/T relationship obtained in this paper for northern Victoria Land is shown in Figure 9. We used all the mean δ 18O values obtained from the shallow firn cores drilled up to now in this area in the framework of the Italian Antarctic Project (Reference StenniStenni, 1999), and the mean surface temperatures reported in Table 1. A fairly good correlation (R = 0.90) can be observed, with a δ/T gradient of 0.81‰ °C−1. This value is slightly higher than the Reference Lorius and MerlivatLorius and Merlivat (1977) gradient of 0.75‰ °C−1, found in Terre Adelie along the Dumont d’Urville and Dome C traverse, and generally used in East Antarctica. The Reference Lorius and MerlivatLorius and Merlivat (1977) regression line is reported in Figure 9, for comparison. The slight difference between the two lines could be due to the different climatic conditions of the oceanic source regions delivering moisture to the Antarctic continent, as suggested by Reference Dahe, Petit, Jouzel and StievenardQin and others (1994) and Reference StenbergStenberg and others (1998). We must also bear in mind that the northern Victoria Land sites are located at higher latitudes than those described by the Lorius and Merlivat equation. Moreover, the northern Victoria Land area examined here is more mountainous, with sub-adiabatic lapse rate (0.5°C (100 m)−1), whereas the plateau area studied by Reference Lorius and MerlivatLorius and Merlivat (1977) has a near-dry-adiabatic lapse rate (1.1°C (100 m)−1).
We suggest that the northern Victoria Land region could be more influenced by the Pacific Ocean sector and, on a regional scale, by the nearby Ross Sea. The annual cycle of the sea ice in the Ross Sea may play a role in the isotopic composition changes of snow precipitations. If part of the air-mass cooling occurs over the sea ice, a more isotopically depleted water vapour will reach the coast. In fact, the new regression line (Fig. 9) is shifted toward lower δ 18O values than the Lorius and Merlivat one. Indeed, isobaric cooling of air masses over sea ice would lead to precipitations with higher δ/T gradients than adiabatically cooled air masses (Reference DansgaardDansgaard, 1964).
This study underlines the importance of a multiparametric approach in order to date ice cores from such a meteorologically and topographically variable area as the coastal region of northern Victoria Land. The use of several seasonal markers reduces the likelihood of mistakes due to singular events that can affect an individual parameter. The comparison between the δ 18O profile and a composite chemical profile (normalized sum of H2O2, nssSO4 2− and MSA) has given good results for summer peak identification. The seasonal signal of the parameters used for dating was very sharp, even for the highest station (Hercules Névé; 3000 m a.s.l.), because of its relatively high mean accumulation rate.
As expected, the mean accumulation rates show a sharp decrease from the coast (260 kg m−2 a−1) to the margin of the Antarctic Plateau (159 kg m−2 a−1). This difference can be principally attributed to the different altitudes. Altitude is also a fundamental parameter for controlling the chemical composition of snow precipitations, especially for the sea-spray and marine biogenic components. In fact, altitude-induced aerosol-fractionating processes change the absolute and relative contributions of the two marine sources.
The accumulation values are in good agreement with those obtained from snow pits and firn cores sampled in the same area. The mean accumulation value for all the available data (170 kg m−2 a−1) leads to an improvement of the previous estimations for the northern Victoria Land region. The new value is very close to that required for a zero net surface mass balance according to ice discharge.
A linear relationship between the mean δ 18O values and the mean annual surface temperatures gave a δ/T gradient of 0.81‰ °C−1, slightly higher than the value of 0.75‰ °C−1 previously calculated for the Terre Adelie inner regions, indicating possible different sources and transport processes of the air masses reaching the coastal area of northern Victoria Land. A better knowledge of the regional correlation between δ 18O and surface temperature, also including more continental sites, will be possible in the future through the isotopic analysis of surface samples collected during an Italian traverse (in the framework of the International Trans-Antarctic Scientific Expedition) from Terra Nova Bay to Dome C.
Research was carried out in the framework of a Project on Glaciology and Palaeoclimatology of the Programma Nazionale di Ricerche in Antartide, and financially supported by Ente per le Nuove Tecnologie, Energia e l’Ambiente (ENEA) through a cooperation agreement with Università degli Studi di Milano.