3.1. Calculation of sea-salt and non-sea-salt fractions

Na⁺, Mg²⁺ and Cl^– have possible alternative sources with respect to sea spray. Na⁺ and Mg²⁺ also originate from crustal primary aerosol, and Cl^– is also supplied by adsorption processes of gaseous HCl, in turn mainly originating from acid–base chemical exchange between NaCl and HNO₃ or H₂SO₄, into aerosol particles or directly on the snow surface (Reference Aristarain and DelmasAristarain and others, 1982; Reference Legrand and DelmasLegrand and Delmas, 1988; Reference FrezzottiGjessing, 1989; Reference Mouri, Nagao, Okada, Koga and TanakaMulvaney and others, 1993; Reference UdistiWagenbach and others, 1998; Reference Aristarain, Delmas and BriatAristarain and Delmas, 2002).

The Na⁺ sea-salt fraction (ssNa⁺) was evaluated by subtracting from the measured total Na⁺ concentration the crustal contribution, calculated on the basis of the Ca²⁺ content. Since Ca²⁺ also has marine and crustal sources, its non-sea-salt fraction (nssCa²⁺) must also be evaluated. We used a simple two-equation system to evaluate sea-spray and crustal contribution of Na⁺ and Ca²⁺:

where (Na⁺/Ca²⁺)_crust is the mean ratio in the Earth’s crust and (Ca²⁺/Na⁺)_seawater the mean ratio in bulk sea water. Ratios are expressed as weight on weight (w/w).

Though ssSO₄ ^2– was also considered as a sea-spray component, we do not discuss it for two main reasons: (1) ssSO₄ ^2– is a minor fraction of the sulfate budget at the great majority of the sampled sites (the major contribution comes from oceanic biogenic source of H₂SO₄; Reference BecagliBecagli and others, 2005); and (2) a large uncertainty affects the ssSO₄ ^2– calculation because the bulk sea-water ratio usually used (SO₄ ^2–/Na⁺ = 0.253 w/w) was changed in winter sea-spray aerosol by precipitation of mirabilite (Na₂SO₄), temperature-tuned (Reference UdistiWagenbach and others, 1998). In addition, Reference Rankin, Auld and WolffRankin and others (2002) supposed that the ice flowers formed on the sea-ice surface during its seasonal growth by volatilization– condensation processes, similarly weakened in SO₄ ^2– with respect to sea-water bulk composition, could constitute a major source of sea-spray aerosol, at least for coastal sites and in the winter period. The contribution of mirabilite precipitation and ice-flower growth to the global budget of annual ssSO₄ ^2– deposition in areas located at different distances from the sea and altitudes is difficult to quantify, making the uncertainties related to ssSO₄ ^2– calculation too large.

Possible enrichments or depletions of Mg²⁺ and Cl^–, with respect to the bulk sea-water composition, are discussed in section 3.3, based on their linear regression with ssNa⁺ and on the trends of the Mg²⁺/ssNa⁺ and Cl^–/ssNa⁺ ratios.

3.2. Geographical variability

Figure 2 shows the ssNa⁺ percentage fraction as a function of distance from the sea for all the sampling sites. The sea-salt contribution to the total Na⁺ budget is always dominant (>80% for most sites). It is lower only at a few stations in NVL characterized by crustal contribution from ice-free areas of the Transantarctic Chain and at some sites in Wilkes Land affected by a particular atmospheric circulation pattern (Reference Legrand and DelmasMagand and others, 2004). Indeed, as a consequence of increasing remoteness, the percentage of ssNa⁺ shows a slight and noisy decreasing trend as distance from the sea increases.

Fig. 2. Percentage contribution of ssNa⁺ with respect to total Na⁺ reported as a function of distance from the sea.

In order to obtain a higher significance between the sea-spray snow content and the site location (altitude and distance from the sea), total depositional flux was here preferred to concentration, as a parameter less affected by the large changes in accumulation rate observed in the studied area. It is important to clarify that flux represents the total net deposition of an atmospheric component by both wet and dry removal processes. The flux is closely related to the atmospheric aerosol load only for components measured at sites where dry deposition is dominant (Mayewsky and Legrand, 1990; Reference UdistiUdisti and others, 2004b). Even so, the relative contribution of wet and dry deposition to the snow content of atmospheric components is valuable by a graphic elaboration, if a number of sites are sampled in the same area (Reference Kreutz, Mayewski, Pittalwala, Meeker, Twickler and WhitlowLegrand and Delmas, 1987; Reference FattoriFischer and Wagenbach, 1996). This approach applied to coastal and inner areas is described in section 3.5.

Total (wet and dry) depositional fluxes of Na⁺, Mg²⁺ and Cl^– are calculated by multiplying concentration (mgL^–1) by the mean annual accumulation rate (kgm^–2 a^–1) at the sampling site:

Snow composition along the coast–inland transect reflects aerosol fractionating effects occurring during the transport of humid air masses from the oceanic source areas to the deposition site. In this way, the parameters mainly affecting the reconstruction of the load and chemical composition of the atmospheric aerosol are accumulation rate, altitude and temperature. In the NVL–DC–DdU area, these factors are interrelated: a positive relationship links distance from the sea and altitude, while these two parameters are inversely correlated to the temperature (Reference Frezzotti and FloraFrezzotti and others, 2004; Reference Legrand and DelmasMagand and others, 2004). We have chosen distance from the sea as the reference parameter in evaluating the sea-spray spatial trend in this Antarctic region.

The inverse correlation between sea-spray atmospheric load as revealed by the snow composition, and distance from the sea is clear in Figure 3. The sharpest decrease in depositional fluxes of Na⁺, Mg²⁺ and Cl^– occurs in the first 200km from the sea (see also smaller plots at lower y scale). From 500 km inland, the fluxes of the three components reach stable values about two orders of magnitude lower than those measured at the coastal sites. When plotted as a function of altitude (not shown), depositional fluxes show the most rapid decrease in the first 2000ma.s.l. This pattern is caused by the fast decrease in atmospheric load of aerosol particles mainly belonging to the super-micrometric classes, which undergo more efficient atmospheric scavenging processes by wet and dry deposition (short atmospheric residence time) (e.g. Reference Migliori, Becagli, Benassai, Fattori, Traversi, Udisti and ColacinoMinikin and others 1994; Reference Udisti, Bellandi and PiccardiUdisti and others, 1998; Reference BecagliBecagli and others, 2003; Reference Traversi, Becagli, Castellano, Largiuni, Udisti, Colacino and GiovannelliTraversi and others, 2004). Aerosol measurements at Terra Nova Bay (Reference Innocenti and ColacinoKerminen and others, 2000; Reference McInnes, Covert, Quinn and GermaniMigliori and others, 2002; Fattori and others, 2004, Reference Fattori, Bellandi, Benassai, Innocenti, Mannini, Udisti and Colacino2005) and at Dome C (Reference Udisti, Becagli, Castellano, Traversi, Vermigli and PiccardiUdisti and others, 2004a) show the sea-spray components are mainly distributed in the ‘coarse’ aerosol fraction, with a prevalent mode around 2.1–1.1 μ m (Reference Gragnani, Smiraglia, Stenni and TorciniInnocenti and others, 2004). The trends of sea-salt components shown in Figure 3 show that the first 200km from the sea, corresponding to a ~2000ma.s.l. altitude step, could be considered a critical threshold for sea-spray penetration in the central Antarctic area.

Fig. 3. Na⁺ (a), Mg²⁺ (b) and Cl^– (c) depositional fluxes as a function of distance from the sea. Smaller plots inside the panels show the same plot at lower y scale.

The decreasing pattern can be described by a power function. Figure 3 shows the curve-fits (correlation coefficient R = 0.68–0.71, n = 621). The constant flux values from 500km inland for Na⁺ and Mg²⁺ could support the hypothesis that background sea-spray aerosol reaches the central areas of Antarctica due to the long residence time in the atmosphere of sea-spray particles in the accumulation mode (~0.3–1 μm) (Reference RöthlisbergerShaw, 1989; Reference Mulvaney, Coulson and CorrMurphy and others, 1998).

The different pattern of chloride flux (Fig. 3c), which shows a continuous decreasing trend inland (although slightly less so with respect to the first 200 km step), can be explained by the increasing effects of post-depositional re-emission of HCl into the atmosphere as accumulation rate decreases, i.e. the distance from the sea increases (see section 3.4).

3.3. Correlations between Mg2+ or Cl– and ssNa+

To identify possible other sources of Mg²⁺ and Cl^– or fractionating effects of sea spray during the atmospheric transport, we plotted the Mg²⁺ and Cl^– concentrations vs the ssNa⁺ concentration and the Mg²⁺/ssNa⁺ and Cl^–/ssNa⁺ ratios as functions of ssNa⁺ flux and distance from the sea.

Figures 4a and 5a show good correlations between Mg²⁺ or Cl^– and ssNa⁺ at all sites (R = 0.966 and 0.984, respectively). The correlation is also seen at sites where sea-spray input is low. The linear regression slopes (0.120 for Mg²⁺, 1.77 for Cl^–) are very close to the bulk sea-water ratio (0.129 and 1.81 w/w, respectively), confirming that sea spray mainly drives the relationships.

Fig. 4. (a) Correlation plot between Mg²⁺ and ssNa⁺. (b, c) Mg²⁺/ Na⁺ ratio reported as a function of ssNa⁺ depositional flux (b) and distance from the sea (c). Lines in (b) and (c) represent the Mg²⁺/ Na⁺ ratio in sea water.

Fig. 5. Same as Figure 4, but for Cl^– instead of Mg²⁺.

While the Mg²⁺ vs ssNa⁺ regression line shows an insignificant y intercept (about 1 ppb), the Cl^– vs ssNa⁺ intercept value (about 23 ppb) shows that Cl^– in snow precipitation is supplied by other sources than sea spray, which becomes more and more evident as the sea-spray load decreases (low ssNa⁺ flux, i.e. when the distance from the sea increases; Fig 5b). Figure 4b and c show the pattern of the Mg²⁺/ssNa⁺ ratio as a function of the ssNa⁺ flux and distance from the sea. When ssNa⁺ decreases, i.e. at low sea-spray input, a large increase in data scattering is visible, but no particular trend is shown. This scattering is mainly attributed to the natural variability of the atmospheric transport of sea spray because the analytical reproducibility at ppb levels is better than 3% for both components (total reproducibility better than 10%). By plotting the Mg²⁺/ssNa⁺ ratio vs distance from the sea (Fig. 4c), the values stay near the bulk sea-water ratio (shown by the line at 0.129 w/w) without sharp shifts. We conclude that sea spray is the main source of Mg²⁺ at all stations and no significant fractionating effects occur during the transport or deposition processes. By contrast, Cl^– exhibits a clear, large enrichment, with respect to the bulk sea-water composition, as the sea-spray load decreases (Fig. 5b). The Cl^–/ssNa⁺ ratio reaches values up to five times higher than sea-water composition when the ssNa⁺ flux is lower than 2×10⁴ μ g m^–2 a^–1. Figure 5c shows this enrichment is not abrupt, but progressively increases with distance from the sea. Therefore, other sources than sea spray for Cl^– deposition at inland Antarctic sites must be considered. As discussed in section 3.1, the major non-sea-salt source for Cl^– is the deposition of HCl, mainly formed by acid–base exchange between NaCl and atmospheric acidic species (H₂SO₄, HNO₃). The contribution of this additional input becomes more evident when the main source (sea spray) is low.

3.4. Post-depositional re-emission of HCl

The increase of the Cl^–/ssNa⁺ ratio by the extra contribution of HCl deposition is evident only because we considered mean values calculated on the uppermost snow layers, related to the first 0.5–1.0m at sites with low accumulation rate. Actually, post-depositional processes leading to reemission into the atmosphere of HCl directly deposited on the snow surface, absorbed into cloud droplets or adsorbed on the snowflake and ice-crystal surface heavily affect chloride depth profiles. Snow acidity is the key parameter for preservation of chloride in the snow layers. Glacial snow, enriched in crustal particles, showed much higher chloride concentrations by transformation of volatile HCl in nonvolatile salts (Reference Rankin, Wolff and MartinRöthlisberger and others, 2002; Reference UdistiUdisti and others, 2004b). In the present condition, when Antarctic snow is acidic, especially at inland sites where H₂SO₄ and HNO₃ dominate the ionic budget (Reference McInnes, Covert, Quinn and GermaniMigliori and others, 2002; Reference Udisti, Becagli, Castellano, Traversi, Vermigli and PiccardiUdisti and others, 2004a; Reference Fattori, Bellandi, Benassai, Innocenti, Mannini, Udisti and ColacinoFattori and others 2005), chloride is preserved only when the site accumulation rate is relatively high (Reference StenniTraversi and others, 2000). Here, we aim to evaluate the snow accumulation rate threshold able to preserve the original chloride deposition.

Figure 6 shows the depth profiles of the Cl^–/Na⁺ ratio in sites where high-resolution (2.5–5.0 cm) chemical analysis was carried out on snow-pit and firn-core samples. Accumulation rates range from 213 kgm^–2 a^–1 at D66 to 15 kgm^–2 a^–1 at M2. At the first three stations (D66, Evans Névé and Talos Dome), the Cl^–/Na⁺ ratio does not show a decreasing trend with depth. Their heavily smoothed trends (line in the plots) show constant values (1.5–2) close to the bulk sea-water composition (1.81 w/w). By contrast, the stations located far from the sea and/or characterized by low accumulation rates show a sharp decrease in the Cl^–/Na⁺ ratio from high superficial values (higher than sea water) to values equal to or lower than sea-water composition on the snow-pit bottom. In such stations, where sea-spray contribution is low, the HCl extra source to chloride is clearly seen in the uppermost layers (Cl^–/Na⁺ higher than sea water), but accumulation rate is not sufficient to preserve the HCl original deposition, leading to progressive loss of chloride. In addition, the high snow acidity, as revealed by ionic balances (Reference Udisti, Becagli, Castellano, Traversi, Vermigli and PiccardiUdisti and others, 2004a), causes chemical exchange between NaCl particles and acidic species (mainly H₂SO₄), leading to further loss of chloride (Cl^–/Na⁺ values lower than sea water).

Fig. 6. Cl^–/Na⁺ depth profiles from snow pits and firn core related to selected sites with different accumulation rates.

To evaluate the accumulation rate threshold for the chloride preservation, we plotted a Cl^–/Na⁺ decreasing index (D1) as a function of accumulation rate (Fig. 7). The Cl^–/Na⁺ is the decrease in Cl^–/Na⁺ ratio for each depth meter evaluated in the uppermost 2.0m snow layer. Cl^–/Na⁺ ratios were measured on the smoothed curves shown in Figure 6, in order to minimize the high background noise. Two simple linear regressions were drawn for high accumulation rates (where no re-emission occurs: Cl^–/Na⁺ DI = 0) and for low accumulation rates. The intercept between the regression lines roughly indicates the accumulation rate from which chloride is preserved in the snow in the present acidic conditions. This threshold is about 80 kgm^–2 a^–1 for the NVL–DC–DdU sector.

Fig. 7. Cl^–/Na⁺ decreasing index as a function of accumulation rate. See text for calculation of decreasing index.

3.5. Wet and dry contributions to net depositional fluxes

The net depositional flux of sea spray is the result of three contributions (Reference FattoriFischer and Wagenbach, 1996):

where F _d and F _w indicate the dry and wet deposition fluxes and DF accounts for post-depositional processes (such as reemission into the atmosphere of chloride). The term ΔF can be disregarded for Na⁺ and Mg²⁺ because they are irreversibly deposited in the snow. It also plays a minor role for chloride if only very superficial layers are considered. Therefore, for the uppermost layers we have:

where C _p is the concentration associated with the wet deposition and A is the accumulation rate. By assuming C _p constant for precipitation occurring along coast–inland pathways (e.g. Reference FattoriFischer and Wagenbach, 1996), a linear regression should be obtained when plotting F _tot as a function of accumulation rate; the y intercept (flux axis) of such a regression line represents the contribution of the dry deposition.By plotting Na⁺, Mg²⁺ and Cl^– total fluxes as a function of accumulation rate (Fig. 8a–c), an exponential regression seems better able to represent the trends of NVL–DC and sectors (see Fig. 8a–c).

Fig. 8. Na⁺ (a), Mg²⁺ (b) and Cl^– (c) depositional fluxes as a function of accumulation rate: grey dots represent Wilkes Land sites and dark dots represent NVL sites. Smaller plots inside show the same plot at lower y scale.

where F is the values of F _tot when A = 0 and can be considered as the dry deposition flux and K is a term related to the wet deposition which is different as a function of airmass transport pathway.

The highest fluxes measured at the NVL coastal sites indicate a larger sea-spray input than that occurring in WL at similar accumulation rate. This is due to the different atmospheric circulation pattern in the two sectors: NVL is affected by air masses coming directly from the Ross Sea, whereas Wilkes Land presents more continental conditions and low intrusion of atmospheric moisture correlated with the persistence of higher atmospheric pressure in the central areas of the Wilkes Land transect (Reference Legrand and DelmasMagand and others, 2004). The exponential regressions drawn in Figure 8a–c improve the estimation of the y-intercept values, allowing evaluation of the dry contribution (when accumulation rate tends to zero). Unlike that observed for H₂SO₄ and MSA (Reference BecagliBecagli and others, 2005), dry deposition seems to play a minor role in sea-spray scavenging processes. This means that, in the Ross Sea and Pacific Ocean sectors, wet removal accounts for the majority of sea-spray deposition both in coastal and in inland areas.