2.1 Sampling area

Aerosol and DMS sampling are performed in two Antarctic summer campaigns (AC): 2018–2019 and 2019–2020. The campaigns lasted from the beginning of November until the end of January in the area surrounding the Antarctic Italian base, Mario Zucchelli Station (MZS – $\mathrm{74}{}^{\circ }{\mathrm{42}}^{\prime }$ S, $\mathrm{164}{}^{\circ }{\mathrm{07}}^{\prime }$ E), located at Terra Nova Bay, along the coast of the Northern Foothills to north-east of Gerlache Inlet (Fig. 1). There is a persistent polynya in the sea facing the base, the extent of which is shown in the average ice maps for the two campaigns (Fig. 1). Figure 1 also includes average DMS concentrations (nM) in surface waters for the period December–January, according to the climatology of Hulswar et al. (2021).

Figure 1Average sea ice cover during campaign 2018–2019 (a) and 2019–2020 (b) with sea ice data from NSIDC (https://nsidc.org, last access: 4 November 2021) and (c) average DMS (nM) in surface waters during December–January according to the DMS climatology of Hulswar et al. (2021). Figure on the bottom (d) reports the enlarged map of the Terra Nova Bay with Mario Zucchelli Station (MZS – red star) and the aerosol sampling site (Campo Icaro – light blue star). The geographical location of the volcano Erebus (Er.V) and the Antarctic stations reported in Table 1 are reported in (a), (b) and (c).

2.2 Aerosol sampling and analysis

In order to avoid possible contamination from the base, the site chosen for aerosol sampling is Icaro Camp ($\mathrm{74}{}^{\circ }{\mathrm{42}}^{\prime }{\mathrm{43}}^{\prime \prime }$ S, $\mathrm{164}{}^{\circ }{\mathrm{07}}^{\prime }{\mathrm{00}}^{\prime \prime }$ E) located about 2 km south of MZS. The aerosol sampler was installed on the hill facing the sea at about 30 m above sea level.

Aerosol sampling was performed at 12 h resolution by a low volume sequential aerosol sampler (Giano – Dado lab Srl Milano) equipped with PM₁₀ sampling head operating at a constant air flow of 2.3 m³ h⁻¹, in accord with the European rule EN12341. Aerosol samples were collected on Teflon filters (PALL, Germany), 47 mm in diameter, with 2.0 µm nominal porosity and 99.5 % sampling efficiency for 0.3 µm aerodynamic particle diameter. The filters were shipped to Italy, being kept at −20 ^∘C and stored in Petri plastic dishes until they were cut, extracted and analyzed.

In the laboratory, the PM₁₀ mass was determined by weighting filters before and after the sampling on a five-digit microanalytical balance (Sartorius) equipped with an ionic cannon to avoid mass fluctuations due to the electrostatic charge of filters. Before weighting, filters were stored in a dryer for 48 h with 50±5 % relative humidity. A fourth of each filter was devoted to the ion determination by ion chromatography (IC). The 1/4 filter was extracted in ultrapure water (resistivity > 18 MΩ) in an ultrasonic bath at room temperature, then the ionic content was determined by three ion chromatographs (two ICS-1000 and one DX500 Thermo Fisher Scientific Inc., USA) equipped with Gilson 222 XL autosampler. This system makes it possible to simultaneously determine both anions (inorganic and selected low molecular weight organic anions) and cations within 10 min. Details on IC measurements are reported in Becagli et al. (2022).

In order to exclude the sea salt contribution to the total ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ budget, the non-sea salt (nss) SO${}_{\mathrm{4}}^{\mathrm{2}-}$ was calculated as follows:

where ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ and Na⁺ are the measured concentrations in the aerosol samples (as ng m⁻³) and (${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}/{\mathrm{Na}}^{+}$)_sw is the ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ to Na⁺ ratio in sea water 0.25 w/w (Henderson and Henderson, 2009).

2.3 Gaseous DMS sampling and analysis

DMS sampling is performed near the main MZS building at about 5 m distance from the sea at 2 m above sea level. The sampling was done at sub daily resolution (typically four samples in 24 h) by filling electropolished stainless steel canisters by compressing the air at 4 bars within several minutes with a membrane pump (Millipore XX5522050). Before sampling, the canister was filled and emptied two times with ambient air in order to wash the canister, avoiding memory effects from the previous samples.

DMS measurements were made in the MZS laboratories by using a gas chromatograph equipped with a flame photometric detector (HP6890, 393 nm). DMS is trapped in an ethanol bath at −70 ^∘C on a porous polymer resin based on 2.6-diphenilene oxide (Tenax^®) contained in a sample loop. DMS is injected in the GC by thermal desorption in boiling water. Working conditions are reported in detail by Legrand et al. (2001). Daily calibrations were achieved by using a permeation tube (VICI Metronics, Santa Clara California) thermostated at 30 ^∘C. The permeation tube was calibrated and its stability was checked, resulting in less than 5 % changes within one year (Preunkert et al., 2007). The limit of quantification (LOQ) is 0.2 ng, leading to an atmospheric detection limit of 12 pptV for the 6 L volume of air usually trapped into the Tenax^®. In order to reduce the LOQ, a high amount of air (up to 25 L) is trapped into the Tenax^® at the beginning of the field campaign when DMS concentration was low.

2.4 DMSP in sea water sampling and analysis

During both ACs, sea water samples were collected at the sea surface in two sites about once a week when the piers were free from sea ice and the meteorological condition allowed the use of a boat. The two sampling sites were chosen in ice-free water about 2 miles from the coast and from the sea ice margin and 2 miles from each other. Few samples in November 2018 were collected in a hole in the pack ice, about 1 km from MZS in the Gerlache Inlet. The samples were collected in Schott^® bottles, acidified to pH < 2 by adding a small amount of distilled HNO₃, then hermetically sealed and maintained at 4 ^∘C until the analysis. The analysis was accomplished at the Korea Polar Research Institute (KOPRI) laboratories. The preserved DMSP sample was hydrolyzed to gaseous DMS using 10 M NaOH (addition of 0.25 mL per mL sample) and was allowed to react overnight in the dark. Then, DMS was measured by using a gas chromatography equipped with a pulsed flame photometric detector (GC-PFPD) as described in Park et al. (2014). The DMSP sample was measured in duplicate and the analytical precision was generally better than 5 %.

2.5 UV-A and SW irradiance measurements

Measurements of downwelling photosynthetically active radiation and shortwave irradiance were made at Icaro Camp throughout 2018–2019 and 2019–2020. Measurements of UV-A and UV-B irradiances were added during the 2019–2020. The shortwave irradiance was measured with a compact Kipp and Zonen SP Lite sensor, while the photosynthetically active radiation (PAR) with a LI-COR 190R. Both instruments were calibrated by the manufacturer before deployment. In addition, the radiometers were installed at the Lampedusa Climate Observatory before deployments in Antarctica in 2018 and 2019, where they were compared with instruments continuously running at the site. The calibration of the shortwave irradiance radiometers at Lampedusa is traceable to the World Meteorological Organization World Radiation Reference scale (e.g., di Sarra et al., 2019). The PAR calibration scale has been maintained at Lampedusa, relying on the initial manufacturer calibration and on the local calibration of a multi-band radiometer through the Langley plot method (Trisolino et al., 2017).

A UV-A and a UV-B broadband radiometer, a Delta-T UV2/ap and UV2/bp, respectively, were added for the 2019–2020 AC. The radiometers were calibrated at Lampedusa before the campaign by comparison with measurements of spectral irradiance performed with a double monochromator Brewer spectrometer. Nominal spectral response functions (peaked respectively at 373 ad 313 nm for the UV-A and UV-B, corresponding bandwidths of about 31 and 26 nm) were used in the determination of the broadband calibration. The Brewer spectrophotometer is regularly calibrated on site with 1000 W FEL lamps (di Sarra et al., 2008).

2.6 Wind speed and direction data

Wind speed and direction data were measured at Eneide automatic weather station (AWS) ($\mathrm{74}{}^{\circ }{\mathrm{41}}^{\prime }{\mathrm{45}}^{\prime \prime }$ S, $\mathrm{164}{}^{\circ }{\mathrm{5}}^{\prime }{\mathrm{32}}^{\prime \prime }$ E) that takes part of the Meteo-Climatological Observatory at MZS and Victoria Land maintained by the Italian National Programme of Antarctic Research (http://www.climantartide.it, last access: 15 September 2021). This AWS is nearest to both the DMS and aerosol sampling sites and it has been used in this work.

2.7 Satellite data of sea ice and chlorophyll

Daily maps of Southern Hemisphere sea ice cover were obtained from the National Snow & Ice Data Center. Information on sea ice extent is derived from the analysis of satellite passive microwave brightness temperature data from the Nimbus-7 Scanning Multichannel Microwave Radiometer (SMMR) and from a series of Special Sensor Microwave Imager (SSM/I) and Special Sensor Microwave Imager/Sounder (SSMIS) instruments (Fetterer et al., 2017). The nominal spatial resolution is 25×25 km². Data were downloaded from https://nsidc.org/data (last access: 22 December 2021).

Satellite-derived daily L3 datasets of surface chlorophyll-a concentration with a 4 km spatial resolution from the European Space Agency's GlobColour Project (http://hermes.acri.fr, last access: 15 September 2021) were obtained from the Copernicus Marine Environment Monitoring Service (CMEMS, https://marine.copernicus.eu/, last access: 15 September 2021). The Chl-a product is derived by reprocessing the merged observations from five satellite radiometers (MODIS on Aqua, VIIRS from Suomi-NPP and JPSS-1, and OLCI from Sentinel 3a and 3b). The GlobColour dataset is a common and appropriate choice for phytoplankton dynamics studies, even in the Southern Ocean (Ardyna et al., 2017; Cole et al., 2015).

2.8 Backward trajectories calculation

Table 1Mean and standard deviation of DMS, MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations in coastal Antarctic sites.

^a summer mean calculated over the years 1999–2003; ^b summer mean calculated over the years 1997/98, 2001/02, 2002/03; ^c summer mean calculated over the years 1996/97, 1999/00, 2001/02.

Download Print Version | Download XLSX

Ten-day back trajectories are calculated with the HYSPLIT-model (Stein et al., 2015). We used the ensemble method of the model that has been incorporated directly into the code, so that trajectories are automatically computed about a three-dimensional cube, about the starting point at 300 m above MZS. The initial positions are not offset, just the meteorological data point associated with each particular trajectory, so that all trajectories start from the same point (https://www.ready.noaa.gov/documents/Tutorial/html/traj_ensem.html, last access: 22 December 2021). The trajectories are based on the meteorological fields of the Global Data Assimilation System (GDAS1) provided by the US National Weather Service's National Centers for Environmental Prediction (NCEP) at 1 degree resolution (https://www.ready.noaa.gov/gdas1.php, last access: 22 December 2021).

3.1 Factors controlling atmospheric DMS: wind speed and direction, Chl and sea ice extent

The difference between the DMS evolution in ACs 2018–2019 and 2019–2020 shown in Fig. 2 may be due to different causes: biological (DMS production from phytoplankton), physical (DMS flux from sea water to the atmosphere) or chemical (different rates of DMS oxidation in the atmosphere).

The role of wind speed has been firstly investigated. The DMS sea air flux depends on oceanic DMS concentrations and on the transfer velocity coefficient (k_w), which is a function of wind speed (Nightingale et al., 2000; Vlahos and Monahan, 2009; Bell et al., 2017; Zavarsky et al., 2018). Moreover, the measured DMS atmospheric concentrations at coastal sites are expected to be heavily influenced by the wind direction as DMS sea-to-air transfer occurs only at the ice-free ocean surface and air masses coming from the ice sheet do not contain DMS.

In the AWS hourly dataset, wind coming from the continental ice sheet, i.e., wind direction between 200 and 350^∘, were identified in the period 6 December–6 January when maximum DMS release in the atmosphere is measured in both campaigns. Surprisingly, the fraction of time with winds blowing from the ice sheet is higher in 6 December 2018–6 January 2019 (when more samplings with the DMS concentration >75th percentile are observed) than 6 December 2019–6 January 2020 (68 % and 48 % of time, respectively, see Table 2), suggesting that the different evolution of the DMS concentration is not produced by a different pattern of wind direction.

Data display, conversely, a dependency on wind speed. The distribution of wind speeds for cases with high DMS concentrations (>75th percentile) displays that the 82 %–83 % of wind speeds are in the range 1–10 m s⁻¹ in the two campaigns.

Table 2Number of data and percentage of DMS concentrations higher than the 75th percentile, wind direction from the ice sheet and wind speed lower than 10 m s⁻¹.

Download Print Version | Download XLSX

This is consistent with a modelized DMS transfer rate as function of wind speed that increases as wind speed increases up to 10 m s⁻¹, but decreases as wind speed increases (Vlahos and Monahan, 2009).

If we consider the wind speed velocity in the range 1–10 m s⁻¹ as the best conditions for DMS transfer rate by looking at the wind distribution in the abovementioned time period for the two years, we can see an opposite pattern than expected with a lower percentage of favorable wind speed in the year with the higher number of DMS concentration higher than the 75th percentile (Table 2).

Therefore, large differences in the DMS sea-to-air transfer velocity do not seem to occur and are not expected to be the cause of the different behavior of DMS evolution in the two ACs. Unfortunately, we do not have measurements of DMS in sea water that could have confirmed this hypothesis, but it is consistent with previous modeling and experimental evidence assessing that the large year-to-year variability of atmospheric DMS concentrations can not be explained by changes of meteorological processes controlling the k_w factor or by changes of atmospheric oxidants, but most likely by changes in oceanic DMS concentrations (Sciare et al., 2000; Kettle and Andreae, 2000; Marandino et al., 2007; Bock et al., 2021).

Therefore, the DMS concentration in the atmosphere does not seem to be controlled by physical processes at the ocean/atmosphere interface, but is likely related to biological processes in the ocean.

Previous studies found that DMS concentrations in sea water are related to phytoplanktonic biomass (expressed by Chl-a concentration) or to primary productivity that are, in turn, controlling the biogenic aerosol (Minikin et al., 1998; Preunkert et al., 2007).

Some of the highest DMS concentrations in sea water worldwide (>300 nMol L⁻¹) have been reported from the Ross Sea, Antarctica, associated with seasonal blooms of the phytoplankton Phaeocystis antarctica. (Ditullio et al., 2003; Gambaro et al., 2004), a high-DMSP producer (Liss et al., 1994).

Figure 3 shows the time series of measured atmospheric DMS and marine DMSP concentrations and of satellite-derived Chl-a in the two ACs. The satellite determinations of Chl-a are averaged over the region 162.5977^∘ E, 77.666^∘ S; 171.2109^∘ E, 72.2168^∘ S, which corresponds to the area of a rectangle 600×300 km covering the polynya area facing MZS. Evidently, the first maxima in Chl-a are followed by increased DMS values in the atmosphere. Also, the first Chl-a double-peak in late November, early December 2019 is higher (1.2–1.8 mg m⁻³) than the first peak in mid-December of 2018 (0.9 mg m⁻³). A similar pattern is visible in DMS concentration, with a higher peak in early December 2019 (921 pptV = 37.7 nMol m⁻³) than in late December 2018 (620 pptV = 25.4 nMol m⁻³). This seasonal DMS accumulation may be caused by a combination of factors, including high DMS production rates, limitation of bacterial DMS consumption at low temperatures and saturation of biological DMS consumption rates (Ditullio et al., 2003; del Valle et al., 2009).

Figure 3Time series of measured atmospheric DMS, DMSP in sea water, and satellite-derived Chl-a (daily, area-averaged concentration in the region 162.5977^∘ E, 77.666^∘ S; 171.2109^∘ E, 72.2168^∘ S, corresponding to a rectangle 600×300 km) in the two ACs.

Download

Particularly relevant is the time lag of about 15 d in both Antarctic campaigns between the Chl-a and the DMS peaks. This delay is expected because DMS emissions depend on the physiological state of the phytoplankton. In particular, large emissions are connected with the phytoplanktonic senescent phase associated with stress factors, such as increasing solar radiation due to the shallowing of the depth of the mixing layer (Simó et al., 1999; Vallina and Simó, 2007), consumption of nutrients (Sunda et al., 2007; Zindler et al., 2014), grazing (Savoca and Nevitt, 2014) and bacterial decomposition (Kiene and Bates, 1990; Lomans et al., 2002). The senescent phase, and the DMS emission, generally follows the maximum of phytoplankton biomass coincident with the peak of Chl-a concentration.

It is interesting to note the presence in both the ACs of a second Chl-a peak in January. The Chl-a concentration peak in January is higher than the one in December and it is not associated with an increase in DMS concentrations into the atmosphere (cf. Fig. 3), despite the high concentration of DMSP in surface sea water. To understand this mechanism, the DMS concentration in sea water would have been useful, but as we do not have these measurements, we can speculate that the DMS concentration in the surface ocean at any given time reflects a complex balance between its biological source, bacterial consumption, photochemical oxidation and ventilation to the atmosphere. Regarding the biological source, the difference in community composition, which is dominated by Phaeocystis antarctica in early summer (November–December) when sea ice melts and later by diatoms (Bolinesi et al., 2020; Innamorati et al., 2000), can affect the DMS concentration in sea water in early and late summer. Indeed, it has been shown that in the Ross Sea, the DMS:chl-a ratio (58–78 nMol µg⁻¹) was significantly higher in waters dominated by Phaeocystis antarctica compared to diatom-dominated waters (2–12 nMol µg⁻¹) (DiTullio and Smith, 1995). Regarding the effect of biological DMS consumption, Del Valle et al. (2009) found that while it remained relatively low and constant throughout the spring (0.05–0.21 d⁻¹), it becomes higher in summer (0.22–0.98 d⁻¹; i.e., faster biological turnover). The spring slow biological turnover probably contributed to the DMS buildup during the early bloom, while the fast biological turnover (leading to the formation of DMSO) helped in producing low DMS concentrations in summer (3.2–16.8 nMol L⁻¹). The higher biological DMS consumption in January than in December can explain the apparent anomaly of the higher concentration of DMSP in near-surface sea water and low DMS in the atmosphere. Unfortunately, we could not collect sea water samples during the period of maximum atmospheric DMS concentration due to lack of safe conditions which prevented reaching the open sea.In addition to the different levels of Chl-a and DMS during the two Antarctic campaigns, the Chl-a peaks occurred with a different timing. In particular, in summer 2018–2019, the Chl-a peak occurs later than in the 2019-2020 summer.

It is well known that phytoplanktonic blooms of Phaeocystis antarctica occur at the beginning of sea ice melting (Arrigo et al., 1998; Stefels et al., 2018). Figure 4 shows the sea ice coverage, as determined by satellite observations, in two areas of the Ross Sea, for 2018–2019 and 2019–2020. When the whole Ross Sea area is considered (an area about 800×800 km²), it appears that the ice cover is higher in 2019–2020 than in 2018–2019, which seems in contrast to the timing of Chl-a increases. However, analyzing in detail the area of the polynya facing the sampling site (about 100×100 km² wide), we may observe that sea ice starts melting earlier and decreases faster in 2019–2020 than the previous year. This evolution of sea ice in the region surrounding MZS could produce a shorter but more intense phytoplanktonic bloom in the polynya area in 2019–2020, and seems to be consistent with the observed evolution of DMS and related parameters. This suggests that the polynya areas close to MZS play a dominant role for the phytoplanktonic cycle and production of biogenic aerosol precursors.

Figure 4Percentage of the area covered by sea ice for the (a) Ross Sea (800×800 km²) and (b) the restricted area of polynya in the Ross Sea (100×100 km²) for the two ACs.

Download

3.2 The relationship between DMS and its oxidation products (MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$): a schematic representation of mechanisms

The simultaneous measurements of atmospheric DMS and its oxidation products allow us to study the dynamic processes occurring in the atmosphere leading to the formation of biogenic particulate matter. These processes can be summarized by the simple model reported in Fig. 5, where the box represent the atmosphere over the sampling site (and relative concentration of DMS, MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$). F_DMS, F_MSA and F_nssSO4 are the flux of DMS, MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ incoming (F-in), outcoming (F-out) or formation (Fox). DMS_sw and DMS_LR represent the concentration of DMS in sea water and from sea water far from the sampling site (long-range), respectively. MSA_LR and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}{}_{\mathrm{LR}}$ represent the concentration of long-range transported species and nssSO${}_{\mathrm{4}}^{\mathrm{2}-}{}_{\mathrm{volc}.}$ represent the volcanic nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$.

Figure 5Schematic representation of the processes related to the measured concentration of DMS, MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ at MZS. See text for the abbreviations' meaning.

Download

In this section, we discuss the overall evolution of DMS, MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ and their relationships, with the aim of identifying periods for which occurring processes with respect to the DMS oxidation pathways can be determined. Specific periods in which near/far sources of the different compounds and oxidation processes for different DMS emission conditions are identified and discussed as explanatory cases.

Figure 2 shows that in both ACs, MSA displays a time evolution similar to nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ with simultaneous peaks. Conversely, the time evolution of MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ differs from that of DMS in the two ACs: (i) during the 2019–2020 AC, maxima of biogenic aerosol compounds occur with a short time difference (24 h) with respect to the DMS peaks and (ii) during the 2018–2019 AC, the largest MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ peak occurs one month later than DMS.

In the period 15–18 December 2019 (Fig. 6), MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ maxima are associated with DMS. This case offers an exceptional example to understand the dynamics of biogenic aerosol formation. In this period, air masses arrived from the area of the Ross Sea surrounding the sampling site (Fig. 7a) passing at low height on the near-sea areas not covered by sea ice, therefore, in correspondence with strong DMS emissions from sea water. The wind speed and direction are almost constant at about 10 m s⁻¹ from the marine sector (180–220^∘ N). The measured relative humidity is 100 % and UV radiation is attenuated by clouds until 17 December at 00:00 (Fig. 6). Looking at the scheme in Fig. 5 from 16 to 17 December 2019, we assumed that F_DMS-in is constant and quite high; at the beginning of DMS emission (16 December 2019), F_ox-MSA and $F_{\mathrm{ox}\text{-}\mathrm{nss}{\mathrm{SO}}_{\mathrm{4}}}$ just started; therefore, the concentration of DMS in the box depends on the equilibrium between F_DMS-in and F_DMS-out. At this wind speed, F_DMS-out is probably low with respect to F_DMS-in, as the concentration DMS_sw (driving the sea-air flux) is high. In these conditions, the DMS concentration reached a maximum of 32.8 nMol m⁻³ on 16 December (the average over the period 09:00–21:00 LT). Due to the constant wind speed and direction (Fig. 6), we can assume that DMS emission from the ocean remains constant also in the following days (DMS_emitted), when UV radiation increase in the following 24 h stimulated the DMS oxidation processes, F_ox-MSA and $F_{\mathrm{ox}\text{-}\mathrm{nss}{\mathrm{SO}}_{\mathrm{4}}}$ become relevant and at constant F_DMS-in and F_DMS-out, the concentration of MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ increases and DMS decreases. MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ reached the maximum concentration on 17 December (8.3 and 9.9 nMol m⁻³ for MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, respectively, for the sampling time 09:00–21:00 LT) when the DMS concentration was 12.6 nMol m⁻³.

Figure 6DMS, MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, wind speed and direction, relative humidity (RH) and UVB for the time 15–19 December 2019 and 9–14 December 2019. DMS and UV-B data are averaged over the time interval of the corresponding aerosol sampling (12 h).

Download

Therefore, the 17 December in the box of Fig. 5 should have

If DMS_emitted=32.8 nMol m⁻³ and DMS in the box is 12.6 nMol m⁻³,

But the DMS_lost is due to the formation of MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, therefore,

DMS_lost= MSA + nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ $=\mathrm{8.3}+\mathrm{9.9}=\mathrm{18.2}$ nMol m⁻³ that is in agreement with the value of 20.2 previously calculated with the approximation of the constant DMS emission for these days.

Therefore, in this situation of constant wind speed and direction for the 17 December, we can suppose that F_MSA and F_nssSO42− in and out are negligible and the concentration of MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ in the box are mainly due to the Fox-MSA and Fox-nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, for this reason, reflecting the $\mathrm{MSA}/{\mathrm{nssSO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratio of freshly formed biogenic aerosol.

On the following day (18 December), the abrupt change of wind direction (Fig. 5) transport on the sampling site different air masses, therefore, progressively increasing F_DMS-out, F_MSA-out $F_{\mathrm{nss}{\mathrm{SO}}_{\mathrm{4}}\text{-}\mathrm{out}}$, leading to MSA, nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ and DMS concentration decreases in the box of Fig. 5.

On the days 16–17 December, the MSA/nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratio ranges from 0.68 to 0.94 mol mol⁻¹. As reported by Fung et al. (2022), the BrO reaction with DMS in the gas phase and O₃ reaction in the aqueous phase are the two main processes for DMS loss in the southern high-latitude ocean, accounting for 50 %–60 % and 20 %–30 %, respectively. Both these processes lead to the formation of MSA in the aerosol phase (Fung et al., 2022). In particular, the reaction with O₃ in the aqueous phase could be particularly efficient at this time when high relative humidity is measured (100 %). This high $\mathrm{MSA}/\mathrm{nss}$${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratio can be measured only in the freshly formed secondary biogenic aerosol, as MSA in the aerosol phase can be transformed in nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ by reaction with OH radicals (Fung et al., 2022), leading to a decrease of the $\mathrm{MSA}/\mathrm{nss}$${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratio in the aged aerosol.

A similar situation occurred also in the period 9–15 December 2019 and 3–7 January 2020 (the latter reported as an example in Fig. 6 and the corresponding backward trajectories in Fig. 7b), even though with a different intensity of DMS emissions.

Figure 7Ensemble of 10-day backward trajectories at 300 m a.g.l. (above ground level) arrival height for the days 16 December 2019 (a), 4 January 2020 (b), 10 December 2018 (c), 26 December 2018 (d), 9 January 2019 (e) and 15 January 2019, (f) together with the ice cover for these days. Trajectory height along the route and sea ice percentage are presented as color or gray scales, respectively.

In the 2018–2019 AC, conversely, MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ maxima do not strictly coincide with high DMS concentrations. In particular, the first peaks of MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ occur on 10–11 December 2018, about 10 d before the main DMS peak. Sea ice is still present near the sampling site and DMS can not escape locally to the atmosphere, therefore, F_DMS-in (cifr. Fig. 5) is low and, therefore, the concentration of DMS in the atmosphere (represented by the box in Fig. 5) is also low. At this time, MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ were very likely transported from areas far from the sampling site (i.e., MSA_LR and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}{}_{\mathrm{LR}}$ – Fig. 5), where an early phytoplankton bloom was taking place likely due to the sea ice melting in the external boundary of the sea ice belt around Antarctica (Gabric et al., 2005, 2018). Backward trajectories show air masses coming from the ice sheet (cf. Fig. 7c); therefore, we can suppose that MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ came from oceanic sectors far from the sampling site. In this period, the MSA/nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratio was quite low (about 0.2), indicating generally aged air masses.

In the period 19–30 December 2018, several DMS peaks were measured, associated with low or moderate MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations (Fig. 8). In this period, DMS emissions from the ocean are expected to be high and to be captured by the air masses traveling over the sea. This period is characterized by high wind speeds; DMS concentration spikes are higher when wind speed drops (Fig. 8). In this case, due to the high variability of wind speed with values up to 25–30 m s⁻¹ F_DMS−in (Fig. 5) can be high, but in this condition, F_DMS-out (Fig. 5)is also high. Therefore, the DMS-laden air masses are transported away before a relevant amount of product (MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$) are formed. Therefore, even if a small amount of measured MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ in the atmosphere (i.e. box in Fig. 5) can come from F_ox-MSA and $F_{\mathrm{ox}\text{-}{\mathrm{nssSO}}_{\mathrm{4}}}$, the main part comes from F_MSA-LR and $F_{{\mathrm{nssSO}}_{\mathrm{4}}\text{-}\mathrm{LR}}$. As nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}{}_{\mathrm{LR}}$ can arise from the further oxidation of MSA in the water phase along the transport (F_MSA-ox, Fung et al., 2022) and from volcanic sources (nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}{}_{\mathrm{volc}.}$), the measured $\mathrm{MSA}/{\mathrm{nssSO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratio is lower and more variable with respect to those measured in freshly formed biogenic aerosol. The backward trajectories show that air masses came from the Weddell Sea, far from the sampling site (Fig. 7d), crossing the ice sheet and passing at low elevation over the Terra Nova Bay polynya just before arriving at the sampling site. The presence of generally aged air masses in this period is confirmed by the low $\mathrm{MSA}/{\mathrm{nssSO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratio (0.35, on average). Therefore, in these periods, air masses containing MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ came from an oceanic sector far from the sampling site, whereas the DMS enters the air mass over the polynya just before the sampling site.

Figure 8DMS, MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, wind speed and direction, relative humidity (RH) and short-wave irradiance for the time 19–30 December 2018 and 9–20 January 2019. DMS and SW irradiance data are averaged over the same time interval of the aerosol sampling (12 h).

Download

In the two periods 9–13 and 15–18 January 2019, high MSA and nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations (>5 nMol m⁻³) are measured, while DMS concentrations are very low (<3 nMol m⁻³). In these periods, the wind speed was very low (<5 m s⁻¹) (Fig. 8). Backward trajectories show air masses coming from the ice sheet for the period 9–13 January (Fig. 7e) and routes extremely variable sometimes coming from the northern part of the Ross Sea uncovered by ice at this time (Fig. 7f). Indeed, the $\mathrm{MSA}/{\mathrm{nssSO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratio is quite low on 9–13 January (0.40 Mol Mol⁻¹) when air masses come from the ice sheet and, therefore, far from the sampling site, while it reaches the highest measured value (up to 0.96 Mol Mol⁻¹) during 15–18 January. This suggests the presence of freshly formed biogenic aerosol from DMS formed in the northernly Ross Sea.

3.3 Quantification of the biogenic aerosol contribution to PM₁₀

In the previous section, we highlighted the variability of the MSA/nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratio as a function of the air masses' aging. In order to find a characteristic branch ratio for the DMS oxidation, the MSA concentrations are reported versus nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations in Fig. 9. A somewhat different pattern of MSA concentration with respect to nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations appears. We separated data into two classes respectively for nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ values lower and higher than the somewhat arbitrary value of 3 nMol m⁻³. The slope of the MSA-nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ relationship appears to change approximately around this value. Small changes in the results that will be discussed occur when a different threshold value in the range 2.5–4 nMol m⁻³ is chosen.

Figure 9Scatterplot of MSA concentrations versus nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$. The two regression lines are calculated for the nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ concentration lower (red crosses, black line) and higher (blue dots and line) than 3 nMol m⁻³.

Download

The value of the threshold allows us to split the dataset in a low and high biogenic aerosol load. The presence or high or low biogenic aerosol load is related both to the air masses' direction and timing. About 60 % of data with a nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ concentration lower than 3 nMol m⁻³ come from the ice sheet (direction from 200–350^∘ N, as reported above) and the remaining 40 % are related to a time before the beginning of sea-ice melt and, therefore, before the phytoplanktonic bloom.

MSA shows a relatively high correlation with nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ for both datasets (even though the correlation is worse for low nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ concentrations), suggesting that both species have a common source, as expected for DMS oxidation. The slope of the regression line is higher for the class with nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ > 3 nMol m⁻³. The two different slopes can be associated with different situations: for nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$-concentrations below 3 nMol m⁻³, the MSA-nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ relationship is expected to be produced by aged biogenic aerosol with possible additional nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ contributions coming from the oxidation of SO₂ emitted by the near volcano Erebus (Boichu et al., 2010) (Fig. 1) or from long-range transport from northern latitudes (Minikin et al., 1998).

Conversely, nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ values higher than 3 nMol m⁻³ appear to be generally associated with the presence of freshly formed biogenic aerosol. The abscissa-intercept of the regression line is 2.1 nMol m⁻³, corresponding to 202 ng m⁻³. Once this background contribution is subtracted, the MSA/nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ derived from the regression line is 0.84±0.06 Mol Mol (that is the same value expressed as w/w as MSA and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ have almost the same molar mass). We assume that this value can be considered as the mean branch ratio between the two species in the newly formed biogenic aerosol in summer at this high southern latitude. Several other studies report quite different MSA/nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratios, both in aerosol and snow layers (Becagli et al., 2005; Legrand and Pasteur, 1998; Minikin et al., 1998; Mulvaney and Wolff, 1994; Preunkert et al., 2008; Zhang et al., 2015). However, some of these determinations are affected by fractionation effects in the aerosol during the transport from source regions to the sampling site and by different deposition processes. In this study, thanks to the closeness and strength of the DMS source during periods of ice-free polynya in the short-range of the sampling site and to the opportunity to find air masses containing DMS and both its freshly formed oxidation products, we believe that it is possible to obtain a characteristic MSA/nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ branch ratio for the freshly formed biogenic sulfur-oxidized aerosol.

Considering the samples when nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ is higher than 3 nMol m⁻³ as representative of the presence of freshly formed biogenic aerosol, is it thus possible to quantify the role played by newly formed biogenic aerosol on the total PM₁₀ mass. On average, the sum of biogenic nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ (i.e., by subtracting the nss${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ background) and MSA accounts for 17 % of the PM₁₀ mass, with maxima in single samples as high as 56 %. This contribution is relevant and its quantification is important also from a climatic point of view, as freshly formed biogenic aerosol can constitute an important source of cloud condensation nuclei over the Southern Ocean.