Spatial and temporal variations of methanesulfonic acid and non sea salt sulfate in Antarctic ice

A simultaneous glaciochemical study of methanesulfonic acid (MSA) and non-sea-salt sulfate (nss-SO4-) has been conducted on the Antarctic plateau (South Pole, Vostok) and in more coastal regions. The objective was to investigate marine sulfur emissions in very remote areas. Firstly, our data suggest that MSA and nss-SO4 present in antarctic ice are mainly marine in origin and that DMS emissions have been significantly modulated by short term (eg. El Nino Southern Oscillation events) as well as long term climatic changes in the past. Secondly, our study of spatial variations of these two sulfur species seems to indicate that the atmosphere of coastal antarctic regions are mainly supplied by local DMS emissions whereas the atmosphere of the high plateau is also influenced by DMS emissions from more temperate marine latitudes. Thirdly, our study of the partitioning between MSA and nss-SO4 suggest that the temperature could have been an important parameter controlling the final composition of the high southern latitude atmosphere over the last climatic cycle; colder temperature favoring the formation of MSA. However, our data also support a possible role played by changes in the transport pattern of marine air to the high antarctic plateau.


Introduction
DMS emissions from the oceans may affect Earth's climate through the production of submicrometer sulfate particles which may act as cloud condensation nuclei and modify the marine cloud albedo (Charlson et al., 1987).Although it is now clear that marine biogenic DMS emissions represent the major source of such aerosol in marine remote areas, some important gaps in our knowledge remain in the sulfur cycle at high latitudes.
First, there have been few DMS studies in the marine boundary layer and in surface water of these high latitude regions where relatively high primary marine productivity is expected (EI-Sayed and Weber, 1982).Until now, only a few measurements have been carried out on DMS oxidation products in the marine boundary layer south of about 40 ° S (Berresheim, 1987 ;Pszenny et al., 1989) and such studies generally have covered very short time periods.
A second knowledge gap relates to the fate of the DMS within the atmosphere at high latitudes.Indeed methanesulfonic acid (CH3SO3H, abbreviated as MSA) and sulfur dioxide (converted in the atmosphere into non sea-salt sulfate, nss-SO4) are two major end products from the atmospheric oxidation of DMS provided by the marine biota.Molar MSA/nss-SO 4 ratios (R) ranging from 3 to 10 % have been generally observed in marine remote area at mid and low latitudes (Saltzman et al., 1983 ;1986) but this ratio seems to increase at higher latitudes (Berresheim, 1987 ;Pszenny et al., 1989 ;Savoie et al., 1989 ;Savoie et al., 1990 ;Minildn and Wagenbach, 1990).It is therefore important to increase our understanding of the sulfur cycle in these very remote areas.
Polar ice cores can provide a unique record of environmental changes and be therefore useful to obtain information on the variability of marine DMS emissions in the past and on the fate of DMS in the high latitude atmosphere.Only a few studies have investigated MSA and nss-SO 4 content of polar ice.Legrand et al. (1991) have suggested that DMS emissions of southern hemisphere oceans have been modulated by the climate in the past.In a recent study, Legrand and Feniet-Saigne (1991) suggest that DMS emissions from the antarctic oceans were enhanced during ENSO developments.In this paper we investigate MSA and nss-SO4 content of snow and ice deposited in various antarctic stations (coastal as well as central antarctic), over different time periods.

Experimental
On each sample, we have determined MSA, sulfate, sodium and calcium contents using an ion chromatograph (Dionex 4000 i).Using a gradient pump system, an AS5 separator column and a sodium hydroxide eluant (carbonation prevented by a continuous helium flow), it is possible (as shown in Figure 1) to measure MSA and SO 4 on the same sample.At the beginning of the analysis, a weak NaOH concentration (5.10 -4 N) is used to obtain a good separation between formate (HCOO) and methanesulfonate (CH3SO3) ions (Figure 1).Later on, we increase the NaOH concentration (up to 3.10 .2N) to elute the SO4 ion within less than fifteen minutes.As far as we prevent our sample from formate contamination (such a contamination occurs when liquid samples are stored in polyethylene bags for instance, see Legrand and Saigne, 1988 ;Saigne et al., 1987), the accuracy is 5 % (at 95 % confidence level) in the range 1 to 20 ng.g "1 of MSA with a 5 ml sample volume.Cation measurements were performed using working conditions reported elsewhere (Salgne et al., 1987) with an accuracy of 10 %.Ice coring, transport and storage of ice cores create high contamination levels on the external part of the cores.To remove surface contamination, two procedures can be used: 1) washing cores with ultrapure water at the room temperature (-20°C) or, 2) mechanically removing the exterior of cores in cold room (-15°C).In their first MSA study along the deep Dome C ice core, Salgne and Legrand (1987) washed ice core sections with ultrapure water.This method successfully employed for the analysis of mineral species (successive fractions from the outer to the inner part of the core exhibit high values on the exterior of the core and then a constant value further inside, see Figure 2a) appeared to be adequate for MSA study along the Dome C (Figure 2b 0 core.When using this procedure on the Vostok core, we observed a rather slight but persistent decrease of MSA concentrations to the center of the core (Figure 2b2).Note that, in contrast to the Dome C core, the Vostok core sections were stored during 24 hours at -3°C to prevent breakage of this core which appeared to be extremely fragile.
When using the washing procedure on an ice core section close to the firn ice transition (Figure 2b3), the exterior-interior profile exhibits a more dramatic picture.Because corresponding profiles obtained for mineral species (not reported here) do not exhibit such an exterior-interior trend, it is difficult to explain by a weak contamination dragged by the washing procedure to the interior of these cores.
Subsequently we have designed an ice core lathe equipped with a stainless steel knife to mechanically remove the outer part of the core until 95 % of the ice volume was discarded.
Using this procedure which does not require any previous storage of ice at temperature close to 0°C, the exterior-interior profiles for mineral species as well as for MSA exhibit a plateau a) For mineral species (Na, SO4, K) on a 250-m depth Dome C core section bl) For CH3SO~ on the same core section b2) For CH3SO 3" on a Vostok core section b3) For CH3SO 3" on a D10 core section Concentration (in ng.g 1) are reported as a function of the melted fraction of ice volume (%V).
(Figure 3).It also becomes clear that in contrast to the mineral species, no dramatic MSA contamination is observed on the outer part of the Vostok core.
The importance of the choice of the decontamination procedure is clearly demonstrated in Figure 4 on which we report SO4 and MSA values observed along the Vostok core at similar depths by the washing procedure and by using the lathe.SO4 data obtained using the two procedures agree rather well whereas we can observe systematically lower MSA values by washing Vostok core sections.Although it remains unclear why MSA and mineral species seem to have very different behaviours when ice cores are washed with ultrapure water, our data strongly recommend using the mechanical procedure in future MSA studies along deep ice cores.In the marine boundary layer, low R values are generally observed at low latitudes (Figure 6) but recent studies (Berresheim, 1987 ;Pszenny et al., 1989) reported a strong increase of this ratio (60 to 80 %) at 60 ° S. In this way, ratios observed in coastal antarctic precipitation are in rather good agreement with atmospheric observations at high southern latitudes.More questionable are the lowered ratios observed in high antarctic plateau precipitation and several possible explanations can be suggested.First, we can invoke the input of a non-marine SO4 source acting more efficiently over the antarctic plateau than in coastal areas.The stratospheric input has historically been presented as a possible contribution to the SO4 budget of central Antarctica.There is now a general agreement that this contribution is weak (see Legrand and Delmas, 1990) except after large volcanic eruptions (Legrand and Delmas, 1987).Second, if we assume that end products (MSA and nss-SO4) present in central as well as in coastal antarctic regions both originate from the same high latitude marine DMS emissions, our data could suggest a different branching ratio (DMS ---> SO2, DMS ---> MSA) related to different photochemical conditions occurring over the central plateau and over the surrounding ocean.For instance, the relative content of NO 3 and OH radicals of the atmosphere over the ocean and over the continent could modify the DMS oxidation scheme, but due to a present lack of knowledge of possible DMS oxidation schemes occurring under polar conditions, it remains difficult to assess such a explanation.More recently, a size distribution study (Pszenny, 1990) has shown that MSA particles have a larger size range than have that of nss-S04.It may be therefore possible that the lowering of R over the antarctic plateau reflects a more efficient uptake of MSA by sea salt particles near the coast.On the other hand, the study from Minikin and Wagenbach (1990) reveals an underestimation in evaluating the nss-SO4 at high latitudes in connection with a possible sea salt fractionation phenomena which can lead to some important errors in the calculation of R in these regions.
Finally, to explain lowering R values towards the high antarctic plateau, Salgne and Legrand (1987) suggested that these regions are mainly supplied by end products of DMS emitted at more temperate latitudes thereafter long-range transported in the middle troposphere.
3.2.Temporal variations of MSA and R 3.2.1.Short term variations.Seasonal variations of MSA and nss-SO 4 in snow may help to unravel the origin of these species.As previously discussed (Legrand and Delmas, 1988) the South Pole is a very suitable location for such studies.In this section, we examine the years 1976-1983 at the South Pole.This was a period during which seasonal variations were particularly well recorded, as evidenced by the sodium profile (Figure 7).
The Na profile exhibits a mean value of 10 ng.g x, with maxima (16 ng.g ~) in winter and minima (3 ng.g -~) in summer.Atmospheric aerosol measurements at the South Pole (Cunhingham and Zoller, 1981 ;Tuncel et al., 1989) show similar seasonal sodium variations with marked maxima in winter.These variations arise from the more frequent advection of marine air masses over the high antarctic plateau in winter.Nss-SO 4 exhibits seasonal cycles which are smaller in amplitude than those of sodium, with weak maxima in summer and minima in winter.These variations differ markedly from that in coastal antarctica, where the amplitude of seasonal variations in nss-SO4 is very large, in fact much larger that of sodium (Savoie et al., (C1/Na maxima) seasons (Legrand and Delmas, 1988).
Wintertime concentrations of nss-SO4 at South Pole are comparable or 6 times greater than in coastal areas (Table 1).
In contrast to Na, MSA exhibits very weak seasonality in South Pole snow, with slight maxima occurring during the early spring (Figure 8).Again, this weak seasonal cycle differs from that observed in coastal Antarctica, where winter to summer variations in MSA are as large as a factor of 30 (Table 1).As a consequence of the different seasonal cycles of MSA and nss-SO4 at South Pole, the ratio R (MSA/nss-SO 4) also exhibits seasonality, with a fall-winter maximum of 15-21% and a spring-summer minimum of 10-15%.
The marked differences between the seasonality observed at South Pole and that in coastal aerosols suggest that the high latitude coastal source may not dominate the sulfur budget of the high antarctic plateau.One might expect that lransport between the coast and plateau would be most efficient during the winter (as evidenced by the sodium seasonality) but during this time, there is little atmospheric nss-SO4 and MSA available in the coastal air masses for transport.The R ratios observed at South Pole throughout the year are surprisingly low, well below the 30% expected for the coastal antarctic source region.This suggests that the the source regions are located at low to mid latitudes and that the long-range transport plays a major role in the sulfur budget at the South Pole.Of course, the possibility must also be considered that the ratio of MSA to nss-SO4 in South Pole precipitation has been altered from that at the source region due to selective deposition of MSA relative to nss-SO4 as a result of particle size differences.The particle size distributions of MSA and nss-SO4 over Antarctica are unknown at present and further investigation of this point is needed.
As depicted in Figure 7, a large increase of MSA occurred in 1982 and 1983.This MSA increase is accompanied by a small nss-SO4 change, resulting in a large increase of the R ratio  6), we can expect a mean nss-SO4 content of about 55 ng.g 4 and a mean R of 15% over the high antarctic plateau (Figure 9).Such a picture is in rather good agreement with our  data (Figure 8).For the 1982-1983 event, the large increase of MSA is mainly Inked with local emission increase as reflected by the small change of nss-SO4 (Figure 9).This enhanced contribution of the local DMS source suggests a more efficient transport between the central area and surrounding oceans or enhanced high latitude DMS emissions during these years.Because the 1982 perturbation is not accompanied by higher Na content (Figure 7), we can therefore conclude (as Legrand and Feniet-Saigne, 1991) that the perturbation reflects enhanced DMS emissions at high latitudes.Such perturbations appear rather often further down in south polar snow layers and it was recently suggested that they coincide with years of strong ENSO development (Legrand and Feuiet-Saigne, 1991).The physical link between ENSO event and high latitude DMS emissions is not still elucidated and further studies are required.
3.2.2.Long term variations.MSA and nss-SO4 depth profiles along the Vostok ice core have been recently examined by Legrand et al. (1991).Under present climatic conditions, with mean MSA and nss-SO4 levels of 6 and 100 ng.g -1 respectively, the Vostok site seems to be mainly influenced by DMS emissions from mid latitudes (see Figure 9).This absence of input from high Stades A to H are defined in Legrand et al. (1991).One of the notable features of these MSA and nss-SO4 records over the last climatic cycle is that both species increase during glacial ages.These data suggest that during very cold time periods over the last climatic cycle, DMS emissions of the southern hemisphere oceans have been significantly enhanced.Moreover, the R ratio itself seems to be modulated by the climatic conditions as reflected (Figure 10 a) by the good relationship (r=0.78) with the temperatures (Ts) at Vostok in the past.Such a good correlation between R and the climate (colder temperature being favouring MSA formation) may be indirect evidence for temperature dependence of the branching ratio of the DMS oxidation process.Indeed, as suggested by laboratory experiments (Hynes et al;1986), the temperature dependance of OH addition (MSA formation) versus H abstraction (SO 2 formation) during the DMS oxidation by OH should increase the MSA production at colder temperatures.However our Vostok data suggest that other explanations are possible as well.Indeed, as shown in Figure 10 b, R values are also well correlated with the Na content (r=0.73).Such a correlation could be the result of the good correlation between Na content and the climate (cold climate leading enhanced sea salt production, Legrand et al., 1987) or the reflect of a more efficient transport of MSA on sea salt particles to the central antarctic.We need therefore further studies to understand this link between the climate and the MSA/nss-SO 4 ratios during the past.

Conclusions
This study of spatial and temporal variations of MSA and nss-SO4 contents of antarctic snow and ice provides some useful information on the sulfur cycle at high latitudes.Our data show that the atmosphere of the East antarctic plateau contains sulfur species (MSA and sulfate) in similar proportion than in the mid latitude marine atmosphere and that its sulfur content is strongly modulated by climatic conditions in the past.At the South Pole, the MSA budget seems to be additionally influenced by more sub-antarctic DMS emissions, especially during years corresponding to ENSO developments.Finally, coastal regions are more exposed to local DMS emissions which seems to be characterised by relatively rich MSA content.
Further works are needed to evaluate the relative contributions of local and global DMS emissions to the sulfur budget of snow deposited in various antarctic regions.

Figure 2 :
Figure 2 : Decontamination of ice sections by washing with ultrapure water :

Figure 3 :Figure 4 :
Figure 3 : Demonstration of the external contamination of an ice core section (Vostok at 1775 m depth) for mineral species (Na ÷ and SO4).At the opposite, no contamination is observed for MSA.

Figure 5 :Figure 6 :
Figure 5 : MSA (in ng.g -1) and MSA/nss-SO 4 molar ratio R (in %) in snow deposited at several antarctic sites under present climatic conditions.VK abbreviates Vostok, DC Dome C, PS South Pole, LD Law Dome, JRI James Ross Island, and M-VK the traverse Mirny-Vostok.JRI data are from A. Aristarain (personal communication), LD data from Ivey et al. (1986).

Figure 7 :Figure 8 :
Figure 7 : Seasonal variations of MSA, nss-SO4, R, and Na content of south polar snow deposited between 1976 and 1983.W denotes winters seasons.

Figure 9 :
Figure 9 : Relative contribution of local and global DMS emissions to the MSA and sulfate content of snow deposited at Vostok and South Pole (see text).

Figure 10 :
Figure 10 : Temporal variations of R along the Vostok ice core (the last climatic a) as a function of surface Vostok temperatures (Ts).b) as a function of the Na content of the ice (Nam).
R=45%.Referring to high R values observed at high latitudes, our data suggest an enhanced contribution of high latitude DMS emissions towards the antarctic plateau during this event.We tentatively have tried to reconstitute MSA and nss-SO4 origins at the South Pole in terms of local and more temperate DMS emissions.Concerning background levels, we assume that almost a half of MSA originates from local DMS emissions, an other half from north of 40°S.Using R values of 6% and 100% for temperate and high latitude emissions respectively (as suggested by the figure