1 Introduction
The marine atmosphere over the Southern Ocean and Antarctica, being far from the few adjacent populated continental southern hemispheric land masses, is our closest present-day analogue to the pre-industrial state due to its pristine character (Hamilton et al., 2014). Concerning the regional aerosol these environmental conditions imply lower concentrations of any anthropogenic aerosol components than in the marine atmosphere of the northern hemisphere. Knowledge of the atmospheric composition under such pristine conditions provides a baseline for understanding climate changes in the global warming discussion of the industrial era.
Natural aerosol sources, however, can be very important in this region. Due to the strong and persistent westerlies over the Southern Ocean the mechanical source of sea salt particles dominates particulate mass in the regional marine boundary layer (Gong et al., 1997; Jiang et al., 2021). Furthermore, in the presence of sea ice (Wagenbach et al., 1998), emissions from blowing snow were recently demonstrated to be a significant (possibly dominant in winter) sea-salt source compared to the common sea-salt emissions from the open ocean in the Antarctic regions (Yang et al., 2019; Frey et al., 2020). The increased understanding of the ocean surface microlayer (Riley, 1963; Garrett, 1967; Aller et al., 2005) added a host of biogenic and some enriched inorganic ions to the source of sea salt particles (Leck and Bigg, 1999; Leck and Bigg, 2005).
The high marine primary productivity in surface waters of the Southern Ocean (Westberry et al., 2008) causes very high concentrations of the waste product dimethyl sulfide (DMS) compared to other oceanic regions, (Hulswar et al., 2021), that is vented to the atmosphere. After photooxidation and multiphase oxidation this DMS yields the particulate sulfur components, mainly sulfate and methane sulfonic acid (MSA), (Hoffmann et al., 2016), and is a major regional natural aerosol source. This source exhibits large spatio-temporal and seasonal variability because of the strong dependence of the DMS production in the surface sea waters and related gas-to-particle conversion processes on both oceanic and atmospheric conditions and plays a vital role in the climate system and its variability (Charlson et al., 1987; Andreae and Crutzen, 1997).
Beginning with the discussion of global background or baseline atmospheric parameters in the 1970s, (e.g., Junge, 1972), a number of atmospheric monitoring stations with aerosol programs were established in pristine environments of the Southern Hemisphere. At the same time, geopolitical developments led to a rapidly increasing number of research stations in coastal Antarctica that also established aerosol monitoring. Supply voyages to these Antarctic stations, more recently complemented by an increasing number of research cruises, have greatly increased the geographical coverage of aerosol data over the Southern Ocean. Despite the many technical, logistic and meteorological challenges aerosol measurements have been made in this region for over 40 years. They started from simple electrical and optical methods in earlier years to the present extremely detailed physical and chemical measurements directed at specific aerosol sources and processes (e.g., Alroe et al., 2020; Brean et al., 2021; Sanchez et al., 2021; Simmons et al., 2021; Twohy et al., 2021).
Focusing on Antarctica proper Shaw (1988) provided a first review of aerosol data up to 1987 concluding that most of the Antarctic aerosol is “deriving from biological processes taking place in the surrounding oceans”. First synoptic efforts to investigate the combination of chemical aerosol data were made at several Antarctic stations (Wolff et al., 1998; Weller et al., 2011). Combining marine data until the end of the 20th century Heintzenberg et al., (2000) derived latitudinal profiles of particle number concentrations and chemical aerosol information over the remote Southern Ocean. Since then a wealth of new aerosol data have been accumulated at ever more stations and on cruises including the first Antarctic Circumnavigation Expedition directed at the pristine aerosol (Schmale et al., 2019). Also, more detailed oceanographic and biological information has been published (Righetti et al., 2019; Hulswar et al., 2021; Kremser et al., 2021). Thus, a new effort to combine the available aerosol data over the pristine southern hemispheric region seems well motivated. This study aims at a synoptic view of available data on latitudinal and seasonal distributions of key physical and chemical aerosol parameters over the pristine Southern Ocean and coastal Antarctica. The combined statistical data set of the present study comprising more than 40 years of data is useful in future validations of simulations of global physical and chemical aerosol models such as GLOMAP (Spracklen et al., 2005), MOZART (Emmons et al., 2010) or UKESM (Sellar et al., 2019). With air-mass back trajectories these distributions are extrapolated over larger geographical regions to identify of potential natural aerosol source areas. Besides minimizing the contributions of continental aerosol sources by means of trajectory filtering we explicitly exclude the discussion of the source of sea salt particles, (Lewis and Schwartz, 2004). Also, we do not discuss the upper Antarctic atmosphere as potential source for the aerosol of the coastal Antarctic environment. Except for the discussion seasonal distributions of parameters with sufficiently large data base we exclude a discussion of temporal developments due to the inhomogeneity and scarcity of available data.
2 Database
Through literature searches and contact with the responsible scientists, aerosol data from 27 research sites and 23 research cruises have been collected in the data base of the present study. The data have been taken from the literature cited in Table T1. Substantial exceptions are the stations Cape Point, South Africa, and the most recent (> 2019) data from Kennaook Cape Grim, Tasmania, G.v. Neumayer, Antarctica, King Sejong, Antarctica, Syowa, Antarctica, and SHIRASE cruises.
The data comprise coastal and island stations from South Africa to Antarctica and cruises from in Atlantic, Australia/New Zealand, Indian, and Pacific sectors of the Southern Ocean to and around Antarctica. Being situated some 200 km inland the Belgian Antarctic station Princess Elisabeth is the only non-coastal site, which is considered in this study because of its relatively low elevation of 1390 m a.s.l., (Herenz et al., 2019). Despite the many technical, logistic and meteorological challenges aerosol measurements have been made in the study region for over 40 years. They started from simple electrical and optical methods in earlier years to the present extremely detailed physical and chemical measurements directed at specific aerosol sources and processes (e.g., Alroe et al., 2020; Brean et al., 2021; Sanchez et al., 2021; Simmons et al., 2021; Twohy et al., 2021). Our minimum requirement for inclusion in the data base was the availability of exact geolocation and sampling times of the measurements or samplings as far back in time as possible.
In Table T1 the details of the sites and cruises are presented together with relevant citations. The map in Figure S1 displays the cruise tracks of all utilized experiments and the locations of the stations of the present study. Most of the stations and cruises lie in the Atlantic and Indian sector of the Southern Oceans whereas the data coverage is sparser in the Pacific sector. In the discussion of meridional distributions of aerosol properties, the question arises to what extent the particular results in the region between –40° and –60° might be determined by a high geographic data density. Figure S2 shows that this indeed is the case for particle number concentration data (due to the long time series at Kennaook Cape Grim, Tasmania) albeit not for chemical aerosol properties. The temporal coverage also is rather uneven because few research or supply cruises operated in the months May through October. We discuss the problem of local and distant contamination of the data in Section 4 of the Supplement.
2.1 Particle number concentrations
Total particle number concentration is a key aerosol property. Any experimental values of this parameter strongly depend on the lower and to some extent on the upper size limit of the specific sampling method or air intake. Initially photo-electric condensation particle counters, (CPC’s), of the Nolan-Pollak-type were utilized (Nolan and Pollak, 1946) in the present study. Gras et al., (2002) determined their lower cut-off-size to 4 nm through intercomparison of condensation nuclei and aerosol particle counters. With the advent of commercially available single-particle CPC’s in the 1980s, the range of utilized aerosol instruments widened rapidly. Cut-off-sizes as low as 3 nm were realized in the TSI-3025 counter (Wiedensohler et al., 1997) and successors, the data of which are named N3 in this study. However, most frequently the data of the present study are based on counters like the TSI 3760 and its successors, having a higher cut-off-size around 10 nm, named N10 in this study. The difference of concurrent measurements with these two types of counters is interpreted as the number concentration between 3 and 10 nm, termed N3–10. In a few experiments CPC data were not available but particle number size distributions were measured (Koponen et al., 2002; Asmi et al., 2010; Huang et al., 2018). In these cases, the size distributions were integrated with the lower cut-off-sizes three and 10 nm to provide comparable results.
2.2 Particle composition
The first chemical information of the aerosol over the Southern Ocean was derived in the 1970ties from bulk high volume samples taken for several hundred hours downstream of inlets with undefined or undocumented characteristics. From the 1990s on well-defined inlets with PM10 (particle diameters less than 10µm) or even PM2.5 (particle diameters less than 2.5µm) characteristics were deployed increasingly even at remote sites (e.g., Keywood et al., 1999). Improved analytical refinements allowed daily low-volume samples. Today aerosol mass spectroscopy (AMS) yields temporal resolutions of less than one hour for a wealth of organic and inorganic particulate components. AMS typically has an upper size limit of one micrometer.
For the sake of a synopsis of all historical information on the natural aerosol over the Southern Ocean we utilized in the present study historical samples without well-defined sampling characteristics as well as later and recent more size-characterized samples. Mass concentrations of the nine most major ions most frequently analyzed by ion chromatography were collected in the data base (sodium, ammonium, potassium, magnesium, calcium, MSA, chloride, nitrate, and sulfate).1 They were complemented with the non-sea salt (nss) parts of sulfate, potassium, magnesium, and calcium when sodium values were available as seawater reference ion (Stumm and Morgan, 1981). AMS data were averaged over one-hour periods. Sub-micrometer or PM2.5-inlet derived data were combined with parallel super-micrometer samples. Nevertheless, there are remaining inherent uncertainties when comparing strongly wind-speed dependent sea salt components taken at different wind speeds and heights over the sea surface or distances from the shoreline with different samplers.
For the interpretation of the aerosol data hourly ten-day air-mass back trajectories were calculated arriving at 300 m over the sampling points. The trajectories cover each hour of the collected physical and chemical aerosol data. Details of the trajectory calculations can be found in Section 3 of the Supplement.
2.3 Dimethyl sulfide
As a by-product of the life cycle of phytoplankton dimethyl sulfide (DMS) is released from the oceans to the atmosphere (e.g., Simó, 2001). After photooxidation DMS provides a major natural source of sulfurous particulate matter in the atmosphere (Charlson et al., 1987). Thus, the spatio-temporal distribution of DMS is of relevance for our study of natural aerosol sources over the Southern Ocean. The climatology of atmospheric DMS is, however, only available at Amsterdam Island (Sciare et al., 2000) and Dumont d’Urville (Preunkert et al., 2007) and we here utilize the distribution of DMS in surface sea water as a proxy of atmospheric DMS. The most recent climatology of this parameter has been published by Hulswar et al., (2021). In this study we compare its monthly distributions in different latitude bands with the collected aerosol data over the Southern Ocean.
An alternative proxy for measured atmospheric DMS concentration could have been modeled air concentrations taken from a global atmospheric model such as presented by Chen et al., (2018). For two reasons we chose not to follow this path. One, the output of an atmospheric model over Southern Ocean and Antarctica suffers from the same meteorological data sparseness as back trajectories. Two, the seawater DMS climatology used in this model does not reflect the recent data improvements, in particular over the Southern Ocean, that the DMS climatology by Hulswar et al., (2021) used in the present study.
3 Discussion of results
The discussion of results proceeds from latitudinal to seasonal distributions and from there to extrapolated data in the form of maps. Despite our efforts to select comparable measurements the content of our database is highly inhomogeneous. Concerning particle number concentrations inlet losses, sensor principles and sensitivities, evaluation algorithms, and particle size limits varied. Concerning particle composition again inlet characteristics and sampling conditions and efficiencies, particle size limits, and analytical procedures varied. Consequently, we consider thorough statistical tests inadequate for the discussion. Instead most results are presented in the form of arithmetic means and medians, the latter complemented by Median Absolute Deviations (MAD) from the medians.
Over the open sea and near ice-free coasts sodium chloride (NaCl) dominates marine particulate mass concentrations. As the source processes of this aerosol components are well understood and incorporated in global models, (Lewis and Schwartz, 2004), we will not discuss it further in detail here. Also, as this mostly super-micrometer-sized component is strongly wind-dependent, samples taken with different samplers on different platforms and at different sites are difficult to compare. We note that on annual average sodium and chloride concentrations of the present data base show prominent maxima of two to six µgm–3 in the latitudinal range of strongest storm activity 30° to 60° south. Annual average and median meridional distributions of Na and Cl are shown in Figure S4, with the caveat of potential chlorine loss affecting the concentrations of Cl as previously seen at numerous remote oceanic regions (Kritz and Rancher, 1980; Ayers et al., 1999) as well as at coastal and inland Antarctica (Maenhaut et al., 1979; Legrand et al., 2017).
Until the advent of in situ aerosol mass spectroscopy (e.g., McMurry, 2000) with temporal resolutions of less than one hour aerosol sampling in pristine environments for chemical analyses often required sampling times of more than 100 hours. Under such conditions at the non-Antarctic coastal (Cape Point, South Africa) or island stations near populated continental regions (Kennaook Cape Grim, Tasmania) very few samples were taken without any influence from the nearest continent, i.e. without any hourly back trajectory hitting the nearest continent during a sampling period. At Cape Point we thus excluded the sulfate results from our study. For the very long chemical time series at Kennaook Cape Grim, MSA data were available. With these data combined with a contamination risk derived from hourly back trajectories during the sampling periods we calculated monthly correction factors for nssSO4 according to a procedure that is detailed in Section 2 of the Supplement.
3.1. Meridional distributions
3.1.1 Particle composition
After sodium chloride, sulfate comprises the second most prominent particulate mass fraction of the aerosol. Expressed as nssSO4 Figure 1 shows annual average and median meridional distributions of this component. While we expect contributions from the oxidation of oceanic DMS-emissions, its general trend of decreasing concentrations with increasing southern latitude corresponds to the picture of decreasing anthropogenic influence on the marine aerosol with increasing distance from the major continental sulfur sources. Non-sea-salt potassium (nssK) is plotted for comparison in Figure 1 because this component also is strongly influenced by continental anthropogenic (or natural) combustion sources (Virkkula et al., 2006b; Hara et al., 2019). The continuous decrease of nssK with increasing southern latitude confirms the nature of a pristine southern marine atmosphere. The differences between means and medians in Figure 1 are worth mentioning. For nssK they imply that the relatively high means are due to rather infrequent events, in the line of the expected sporadic presence of biomass burning plume debris over the Southern Ocean (Giglio et al., 2013). For nssSO4 the means and the high MAD-values in the latitude range –45° to –60° indicate either the possibility of another regional non-combustion source of sulfate or strengthened DMS-emissions.
Figure 1
Annual average (av.) and median (50%) meridional distributions of nssSO4 (NSS), and nssK, both in ngm–3. The error bars represent corresponding MAD values.
In contrast to nssSO4, the annual average meridional distribution of MSA (Figure 2) reveals no decreasing trend with increasing latitudes. These differences in the meridional distributions of the two by-products of DMS-oxidation, already observed during several cruises (Bates et al., 1992; Virkkula et al., 2006a; Virkkula et al., 2006b), confirm the importance of non-DMS source of sulfate over most of the low- and mid- latitude Southern Ocean. In comparison to the distribution of nssSO4 the rather flat, (±20%), annual average meridional distribution of MSA only reveals a weak maximum of 48 ngm–3 in the latitude range –45° to –60° with a shoulder towards Antarctica.
Figure 2
Annual average (av.) and median (50%) meridional distribution of methane sulfonic acid (MSA) in ngm–3. The error bars represent corresponding MAD values.
Whereas this meridional maximum of MSA is co-located with that of nssSO4 with potential biological causes (see also further discussions on nssCa and nssMg), it does not correspond to the distribution of its gaseous precursor DMS. Conversely, annual average and median meridional distributions of DMS in surface sea water in Figure 3, 2 have broad minima between –40° and –60° increasing south of the polar front to extremely high values towards Antarctica. Interestingly however, the region around –45° has been shown to exhibit a hemispheric maximum in plankton species turnover and a minimum in plankton species richness, (Righetti et al., 2019). The same latitude band exhibits maximum net primary production, (NPP), with the highest seasonal variability in NPP in the southern hemisphere, (Righetti et al., 2019; Henley et al., 2020). Furthermore, possibly even more important, the ocean-air flux of DMS is expected to be strengthened in this –40° to –50° latitude belt due to high wind speeds (cf. Bates et al., 1992). Finally, the absence of an increase in MSA <–65° that contrasts with the occurrence of a maximum of the DMSaq precursor suggests important effects of other parameters such as wind speed and atmospheric photochemistry (different atmospheric residence times of DMS and MSA). As discussed in Section 3.2.1, differences can also be observed in the seasonal cycle of DMSaq with respect to those of atmospheric MSA and nssSO4.
Figure 3
Annual average (av.) and median (50%) meridional distribution of dimethyl sulfide (DMSaq, nM) in surface sea water from Hulswar et al., (2021). The error bars represent corresponding MAD values.
The two components nssCa and nssMg in Figure 4 exhibit an anomaly in the latitude range –45° to –60°. Although being abundant in atmospheric dust, these two components likely do not originate there from dust emissions since their high levels of nssCa and nssMg remain when coastal and island data are eliminated from the dataset. The –45° to –60° latitude belt represents the so-called storm track area, where strong emissions of sea-salt aerosols by the bubble bursting mechanism take place. In this latitude range both, averages and medians of nssCa and mean nssMg-data reach absolute maxima (see Figure 4).
Figure 4
Annual average (av.) and median (50%) meridional distributions of nssCa, and nssMg, both in ngm–3. The error bars represent corresponding MAD values. North of –60° median values of nssMg are zero.
High values of nssCa in marine aerosols have been reported from many laboratory and field studies (Hoffman and Duce, 1977; Weisel et al., 1984; Keene et al., 2007; Jayarathne et al., 2016; Salter et al., 2016; Mukherjee et al., 2020). The enrichment relative to the bulk seawater composition is attributed to the divalent ions Ca2+ and Mg2+ residing complexed in the polymer gels of the ocean surface microlayer (OSM), (Chin et al., 1998), fragments of which are emitted together with sea salt emissions, which are highly efficient in the storm region over the Southern Ocean. Mg in the sea salt aerosol can also be enriched by sea salt fractionation on sea ice. Near the Antarctic coast most of sea-salt aerosol, at least in winter, is released from the sea-ice surface as shown in Hara et al. (2020), and references therein. The strength of the OSM is strongly coupled to the seasonally variations of the ecology of the underlying water column (Dreshchinskii and Engel, 2017). For nutrient chemicals maximum enrichments in the OSM have been reported at times of maximum biological activity (Lyons and Pybus, 1980). Note that, contrasting with the case of divalent cations, nssK does not exhibit any anomaly in the southern midlatitude range (see Figure 1). This difference between NssK and NssCa and NssMg is in the line of results from laboratory experiments with sea water yield enrichments showing a far weaker (or an absence of) enrichment for potassium than for divalent cations (Jayarathne et al., 2016).
Ammonia (NH3) and its protonated form ammonium (NH4+) are produced in surface sea water by the biological reduction of nitrate (either directly or via the degradation of biologically synthesized organic nitrogenous material) (Johnson et al., 2008). Large bird colonies on remote islands and coastal regions of the Southern Ocean provide additional strong sources of ammonia (Legrand et al., 1998; Schmale et al., 2013). A ‘co-production’ of NH3 and DMS has been hypothesized by Quinn et al., (1990) and Liss and Galloway (1993), albeit challenged and proposed instead as a DMS-driven ‘co-emission’ of ammonia by Johnson et al., (2008). For the present study, ammonium data are also relevant because its gaseous NH3 precursor has been recognized to act as a facilitating agent in atmospheric new particle formation via the nucleation of sulfuric acid, (e.g., Curtius, 2006).
Figure 5 shows annual average and median high NH4 concentrations co-located with the tropical and mid-southern latitude regions with high NH3-emissions from the ocean in accordance with a review of global experimental and model results (Paulot et al., 2015).
Figure 5
Annual average (av.) and median (50%) meridional distribution of ammonium (NH4) in ngm–3. The error bars represent corresponding MAD values.
In addition to NH4 concentration changes with latitude, the meridional distribution of the NH4/nssSO4 molar ratio provides information on (1) the degree of neutralization of submicron aerosol and (2) the magnitude of NH3 oceanic emissions. Indeed, away from continents where large emissions of calcium carbonates sometimes take place, sea-salt associated cations and ammonium are the most important cations controlling the acidity of atmospheric aerosol. Chemical size distribution have shown that, over the Southern Ocean, ammonium mainly stays together with nssSO4 in submicrometer particles, where the amount of alkaline sea-salt is not high enough to compete with NH3 in neutralizing H2SO4, (Virkkula et al., 2006a; Xu et al., 2013; Weller et al., 2018). In the remote marine aerosol, Quinn et al., (1990) reported mean NH4/nssSO4 molar ratios of ~1.3, indicating only partly neutralized sulfate particles. If confirmed, these findings imply that most NHx (NH3 plus NH4) is there in the form of NH4. Under these conditions, the NH4/nssSO4 molar ratio is sensitive to NH3 emissions and can be used to constrain models dealing with NH3 and DMS marine emissions, (Paulot et al., 2015).
Previous studies of the meridional change of Rm, are very rare and were mainly obtained during short summertime ship traverses in the Southern Ocean, (Paulot et al., 2015). One study reported a gradient of the NH4/NssSO4 molar ratio from 0.06 at 70°S to 0.6 between –60° and –65°, and ∼1.0 north of –60° (O’Dowd et al., 1997). As done by Legrand et al. (2021)in evaluating the state of neutralization of the Antarctic aerosol, we here included MSA to calculate the Rm = NH4/(nssSO4 + MSA) molar ratio. Figure 6 we display mean and median meridional distributions of Rm over the Southern Ocean confirming earlier results of an only partially neutralized submicron aerosol. Due to the large presence of penguin colonies at the site, the strongly locally controlled ammonium data from station Dumont d’Urville (Legrand et al., 1998) have been excluded as an outlier. The most complete particle neutralization by NH3 is observed in the latitude band –30° to –60° where NH4 concentrations are the highest.
Figure 6
Annual average (av.) and median (50%) meridional distributions of the molar ratio Rm = NH4/(nssSO4+MSA). The error bars represent corresponding MAD values.
3.1.2 Particle number
Particle number concentrations N10, and N3–10 are controlled by source and aerosol evolution processes yielding nucleating low-vapor pressure gases such as sulfuric acid from photochemical DMS-oxidation (Hoffmann et al., 2016), potentially also by primary particles from the ocean surface microlayer (Leck and Bigg, 1999). Thus, natural aerosol source processes over the Southern Ocean may control the concentrations of N10, and N3–10. At Kennaook Cape Grim this connection between DMS, nssSO4, MSA, and particle number concentrations has been well established experimentally (Ayers and Gras, 1991; Ayers et al., 1991; Ayers et al., 1995; Ayers et al., 1997). In Figure 7 annual average and median meridional distributions of N10, and N3–10 are collated. Average N10 shows a broad maximum similar to that of MSA with a peak between –45° and –60°, close to the corresponding maximum in the –30° to –45° range that was identified in a previous review, (Heintzenberg et al., 2000). Median N10 has a narrower maximum in the –45° and –60° range with high MAD-values. Median N3–10 values are generally less than 10% of the corresponding N10-values. However, the lowermost latitudinal bin with data between –60° and –75° however, is an exception. Here relatively high average N3–10 values appear, albeit based on a single research cruise in spring 2012, (Humphries et al., 2016).
Figure 7
Annual average (av.) and median (50%) meridional distributions of particle number concentrations >10 nm, (N10, cm–3), and between 3 and 10 nm, (N3–10, cm–3). The error bars represent corresponding MAD values.
The infrequent occurrence of particles below 10 nm in the marine atmospheric boundary layer over the Southern Ocean confirms the results of an earlier general review of the marine aerosol, in which particles below 10 nm were found in only 3% of all data (Heintzenberg et al., 2004). Several models have been advanced to explain the persistent number concentrations of these particles around 300 cm–3 in the marine boundary layer despite low probability of locally formed particles (Pirjola et al., 2000). In the tropical to subtropical region these models involve the strong convective transport of boundary layer precursor gases into the photochemically active upper troposphere, with subsequent particle nucleation (Raes, 1995; Shaw et al., 1998), from which the aged particles subside into the boundary layer. At Kennaook Cape Grim Gras et al., (2009) suggested this subsidence was related to the passage of fronts (post frontal subsidence). In higher latitudes, warm conveyor belts and sub-polar vortices explain the upward transport of particle precursors. McCoy et al., (2021) summarized the present understanding of the different mechanisms of particle formation over the Southern Ocean and provide new experimental evidence for the nucleation in the free troposphere based on the 2018 SOCRATES aircraft campaign. Recent results with more sensitive instrumentation indicate more frequent new particle formation processes in the marine boundary layer than previously seen (Peltola et al., 2022).
3.2 Seasonal distributions
3.2.1 Particle composition
Seasonal distributions of the data are discussed next. All graphs are centered around the austral summer months December/January. The sparseness of aerosol data cover allows only a broad geographical resolution. We divide the Sothern Hemisphere into two regions, the Antarctic region to south of –65° latitude, from the rest of the hemisphere north of –65°. Consistent with regional aerosol studies (Humphries et al., 2016) and rare atmospheric DMS-data of Koga et al., (2014) this latitude was chosen to correspond roughly to the Antarctic circle while enclosing most of the Antarctic coastal and island sites of the present study. Based on the climatological DMSaq of Hulswar et al., (2021) Figure 8 shows the climatological monthly medians ±MAD of DMSaq in surface sea water of the two regions. Whereas monthly winter medians are rather similar in the two regions (~1 nM), monthly summer medians in the Antarctic region are four times higher than further north. As a result, the seasonal cycle in the Antarctic region is more well-marked (a factor of 27 between maximum and minimum values instead of a factor of 3 further north). We note the extremely low winter-MAD-values in the Antarctic region. These spatio-temporal variations of DMS in seawater are controlled by various processes that remain poorly understood including not only phytoplankton biomass but also ecological and biogeochemical processes driven by the geophysical context (e.g., Simó, 2001). As mentioned in Section 3.1.1., the concentrations of atmospheric DMS are not only controlled by DMSaq but also through wind speed affecting the ocean-to-air fluxes.
Figure 8
Monthly median DMSaq in surface sea water (nM) of the Southern Ocean south and north of –65° latitude, based on the climatology of Hulswar et al., (2021). The error bars stand for median absolute deviations. The scales for the two regions were chosen to align January values.
The seasonal distributions of nssSO4 and MSA in Figure 9 to some extent reflect the meridional variation of DMSaq in seasonality, characterized by strong summer maxima including the earlier onset of concentration increases that is also seen in Figure 8. In the Sub-Antarctic and Polar Antarctic zones defined in Arteaga et al., (2020) the plankton blooming phases begin already in July and peak in biomass in November (Llort et al., 2015). Peak summer-values of MSA and nssSO4 are similar in the two regions (50 – 60 ng m–3 and ~300 ng m–3, respectively). In winter MSA exhibits extremely low values in the two regions (1 ng m–3 at >–65, 3 ng m–3 at <–65°), but the summer-to-winter drop of nssSO4 is more significant at high latitudes. One possibility is that the contribution of anthropogenic sulfate is still significant at latitudes > –65°, but is much weaker at < –65°. North of the Antarctic region Figure NSSMSA indicates a season-independent “background level” of some 100 ngm–3 nssSO4 over the pristine Southern Ocean. It is possible that the filtering process described in Section 2 was not sufficient to remove contaminated data, however removing the constraint of requiring at least five days of back trajectories to the nearest continent did not significantly increase this nssSO4 “background level”, albeit with a strong rise in MAD-values.
Figure 9
Top: nssSO4, (NSS, ngm–3) and methane sulfonic acid, (MSA, ngm–3) north of –65° latitude. Bottom: Same but south of –65° latitude. The error bars represent median absolute deviations. The scales for MSA were chosen to align January values of NSS and MSA.
The seasonal distributions of ammonium in Figure 10 also show strong summer maxima in both geographical regions. North of –65°, however, NH4-concentrations rise earlier (November) than further south and exhibit a secondary maximum in fall (March, April). Because of the strong summer NH4-peak one might expect that this maximum causes a maximum in particle neutralization in terms of the molar ratio R. However, this is not the case. Because of the comparable or even stronger summer increase in acidifying sulfuric and methane sulfonic acid particles the aerosol remains acidic in both regions. South of –65° even the annual minimum of R occurs in summer with a median value of R ≈ 0.3 in February. Again, the strongly locally controlled ammonium data from station Dumont d’Urville have been excluded. In this region the occurrence of an NH4 maximum in December compared to a nssSO4-maximum in January is consistent with diatom blooms producing NH3 (Johnson et al., 2008) being one month ahead of phaeocystis blooms producing DMS in the Southern Ocean (Alvain, 2005).
Figure 10
Monthly median ammonium concentrations, (NH4, ngm–3) north of –65° latitude. Bottom: Same but south of –65° latitude. The error bars stand for median absolute deviations. The scales were chosen to align December values in both regions.
The examination of the seasonal cycle of nss-cations included latitudes south of –45° to incorporate the annual meridional maxima shown in Figure 4. In Figure 11, the three nss fractions exhibit different seasonal variations. NssK shows highest values in summer with a first peak in November and the absolute maximum in January. The November peak might suggest a biomass burning source for this species, as discussed in Section 3.1.1. In terms of medians this peak is most clearly expressed. The timing of arrival of biomass burning debris at latitudes <–60° is indeed consistent with black carbon observations made at Neumayer Station, Antarctica, showing a broad maximum in October-November when the long-range transport of biomass burning debris from South America were able to reach coastal Antarctica at the end of the austral winter, when the isolation from mid latitudes of the Antarctic continent ends (Weller et al., 2013; Hara et al., 2019). As reported by Legrand et al.(2021), the arrival of biomass burning debris in spring seems to be a general feature at coastal as well as inland Antarctica. To test this explanation, we eliminated the five-day threshold in air-mass back trajectories, thus allowing air mass transport from nearby continents within five days, albeit with no significant effects on averages and medians of nssK. As alternative or complementary explanation for the course of the nssK seasonal cycle, in particular for the main summer peak, we offer the emissions from the extensive distribution of macroalgae around Antarctica (Wulff et al., 2009) as a contributor to nssK.
Figure 11
Monthly average and median concentrations of nssCa, nssMg, and nssK, in ngm–3 south of –45° latitude. Note the different scales for averages and medians.
On one hand, the summer maximum of nssCa corresponds to its marine biological source in the OSM (cf. section 3.1.1). We note a trend in the monthly averages of nssCa towards the absolute maximum in March that may be caused by dissolution processes in the OSM at the end of the biologically productive period. As the monthly medians of nssCa do not show this trend the trend in averages may be caused by an increasing number of high-value events that are not reflected in the medians. On the other hand, nssMg that also has been found to be enriched in the OSM does not show a clear summer maximum, (with the possible exception of a small December peak). Monthly averages of nssMg in Figure 11 may even indicate a winter maximum. As an aside, total Magnesium does not show a significant seasonal variation either with the exception of March when the absolute maximum of monthly averages occurs with 56 ngm–3.
3.2.2 Particle number
In both regions, the particle number concentrations N3 and N10 in Figure 12 exhibit broad summer maxima, albeit with more structure south of –65° than further north. The October peak in N3 is exclusively due to the SIPEX-II cruise in spring 2012, (Humphries et al., 2016). The high MAD-values in September may indicate sporadic NPF-events in early spring as reported by Giordano et al., Lachlan-Cope et al., (2020), and Hara et al., (2021). Aside from these events the absolute annual maxima south of –65° in both size ranges occur in late summer or fall (February/March). In this region, extremely high MAD-values during the two months of September and April indicate frequent events with exceptionally high number concentrations, mostly in N3 but to some extent visible in N10 as well. There are seasonal changes in the air masses reaching the Antarctic coast elated to the variations in the Zonal Wave 3 (ZW3, Raphael, 2007), related to the Semiannual Oscillation, (SAO, van Loon, 1967). ZW3 is a pattern of three climatological low-pressure centers: over the Amundsen Sea, eastern Weddell Sea, and Indian Ocean south of Australia. The SAO causes a strengthening of the meridional pressure gradient between mid-latitudes and Antarctica in spring and autumn, which results in strengthened transport of airmasses from the lower, biologically active latitudes, towards Antarctica.
Figure 12
Monthly median number concentrations north and south of –65° latitude. Top: (N3, cm–3). Bottom: (N10, cm–3). The error bars stand for median absolute deviations.
3.3 Potential source regions of the natural aerosol over the Southern Ocean
In the last part of the discussion we identify potential source regions of the natural aerosol over the Southern Ocean. As before, even though wind speed to a large extent controls sea-air fluxes, the climatology of DMS in surface seawaters from Hulswar et al., (2021) is used to start the data discussion. Focusing on the biologically and photochemically most active period, we report in Figure 13 the average distribution of DMSaq between October and April. Statistics of the analyzed aerosol parameters during this period are collected in Table T2. Several distinct features of the DMSaq-distribution can be seen. Due to the high productivity in the many polynyas, (Arrigo and Dijken, 2003), high DMSaq-values are found all around the Antarctic coast. Three hot spots with the highest DMSaq-concentrations show up in Weddell Sea, around the Prydz Inlet, and – with absolute maxima – in the largest polynya in the Ross Sea. The High Nutrient Low Chlorophyll (HNLC) region between –40° and –60° that showed up with minimum DMSaq in Figure 3, is visible as a broad belt all around Antarctica with a few weak interruptions that may be due to the high data density on the most frequent shipping routes to Antarctica. High DMSaq-values also stand out in the map in the west of the southwestern corners of both South America and Africa. Near the latter continent the high values even increase into the street between Africa and Madagascar. North of the belt of minimum DMSaq-values the Southern Atlantic and much of the Southern Pacific exhibit low DMSaq-values whereas higher values are reported for the Indian Ocean. These DMSa distributions could contribute to the spatio-temporal distributions of particulate biogenic sulfur- in the marine atmosphere over the Southern Ocean.
Figure 13
Average distribution of DMSaq, (nM), in surface waters of the Southern Ocean during the months October through April.3
Information on the potential aerosol source regions may be inferred from the mapped distributions of air-mass back trajectories from aerosol sampling sites during given sampling times. The maps comprised 14x14 geocells of equal projected size on a stereographical projection of the Southern Hemisphere centered on the South Pole. Hourly hits of back trajectories were counted in each of these geocells. Assuming that the highest values of any aerosol data were most clearly connected to an aerosol source region we summed up trajectory hits in maps for all component values above the respective global 75th percentile. As references we constructed corresponding maps for all component values up to the respective 25th percentile. These two maps for each component comprised 25% each of the data. To compare two maps with different numbers of trajectories we normalized each map with its respective total number of trajectory points. The normalized values in each geocell of the 75%-map (nrT>75%) were then divided by the normalized values of the respective ≤25%-map, (nrT≤25%), forming the ratios termed RX, RX = (nrT>75%/nrT≤25%), with X being any of the studied chemical and physical aerosol parameters. The RX-ratios were plotted in color-scale on component maps shown in Figure 14. RX-ratios slightly above zero (blue) indicate that the back trajectories of data above the 75%-level much less frequently hit the respective geocell than for data at median level or below. Ratio values of one (green) indicate that back trajectories of data above the 75% level equally frequent hit the respective geocell as for data at the 25% level or below. Finally, red-colored geocells indicate that the high level data hit the respective geocell twice as frequently or more often than in the respective reference case. These maps presented below are based on the following statistically constraints:
Figure 14
Ratio RX = (nrT>75%/nrT≤25%) of the relative number of trajectory hits for chemical aerosol sample values over the 75%-percentile to that for sample values ≤25%-percentile for the parameters nssSO4, nssCa, nssK, MSA, nssMg, and NH4 during the months October through April. See text for details about ratio RX.
All back trajectories within any utilized sample time window or hourly particle number data stay at least 120 hours off the nearest non-Antarctic continent.
The positions of the hourly trajectories are traced on maps 120 hours back in time.
North of Antarctica only trajectory points below 1000 m above sea level are considered.
Only geocells with at least 1000 data points in at least 10 different years are accepted. The long sampling times of the chemical parameters implied many more hourly trajectories than the hourly CPC-data in the respective maps. Thus, the year-threshold could be raised to 15 years in Figure 14.
MSA exhibits a large compact potential source region in Figure 14 over Dronning Maud Land and Haakon VII Sea with a small secondary peak over the eastern Dumont D’Urville Sea. It covers a region north of the Antarctic coast in the Atlantic and Indian sectors extending from the Falkland Islands towards Marion Island.
Potential source regions for nssSO4 in Figure 14 are less well defined but more clearly two-pronged than for MSA. Less strongly expressed they cover the same Atlantic and Indian regions as for MSA. Ross Sea and vicinity do appear as potential source of secondary magnitude for both. The latter region also shows up as a secondary potential source for nssCa and nssMg in Figure 14. In the broad belt of reemitted dust between –40° and –60° simulated by Cornwell et al., (2020) with the Community Atmospheric Model, Version 5 (CAM5), several hot emission spots show up in this region. The regions extending over Mawson and Davis Sea and much of the ocean south of Australia this area also are the major potential source region for nssK in Figure 14 with a smaller one over Weddell Sea and Bransfield Strait. Finally, for ammonium in Figure 14 a large ring-shaped potential source region surrounds much of Antarctica between –50° and –65°.
The most pronounced potential source regions for particles larger than 3 nm (N3) is found in Figure 15 in the Indian sector of the Southern Ocean between Marion and Amsterdam Island with a smaller hot spot showing up west of the Ross Ice Shelf (see Figure 15). For recently formed particles between three and ten nm diameter, (N3–10), Figure 15 indicates potential source regions extending from the Indian sector between Marion and Amsterdam Island over Eastern Antarctic to Victoria Land. The lack of source indications for this size fraction over the rest of the Southern Ocean may be simply due to the sparse coverage of concurrent N3 and N10 data. The identified potential source region comprises and confirms the source area over Antarctica identified by Humphries et al., (2016) albeit here with data from at least ten different years with at least 1000 trajectory points per geocell. We note that air masses from a belt reaching from Tasmania and Macquarie Island to Amsterdam Island less frequently led to very high values of (N3–10) as compared to air masses with low number concentrations, again consistent with the findings of Humphries et al., (2016) who measured low particle number concentrations north of the atmospheric polar front. Finally, the potential source region for somewhat aged particles, (N10), in Figure 15 is most strongly expressed over the Antarctic coastal region’s Riiser-Larsen and Cosmonauts Seas with an extension over high Central Antarctica to Victoria Land. This extension can be interpreted as indicating the possibility of a upper tropospheric to stratospheric origin of these particles being carried with katabatic winds towards coastal Antarctica as indicated in the summer map of air parcel origin in Suzuki et al. (2013).
Figure 15
Ratio RX = (nrT>75%/nrT≤25%) of the relative number of trajectory hits for particle number concentrations values over the 75%-percentile to that for sample values ≤25%-percentile for the parameters N3, N3–10, and N10 during the months October through April. See text for details about ratio RX. Circles denote stations, dashed black lines show cruises from which data were used in this study.
4. Summary and conclusions
More than 40 years of aerosol data including concentrations of particle number and of nine major ions collected over the Southern Ocean and coastal stations have been aggregated and filtered with back trajectories to reduce the risk of influence from adjacent continents. Latitudinal, seasonal, and air mass analyses yielded the following results:
Only nssSO4, and to a lesser extent nssK, show a monotonous decrease from high tropical to extremely clean Antarctic levels as might be expected on a course from near-continental to remote polar marine atmospheric conditions.
The latitude range between –45° and –60° stands out with remarkable relative or absolute maxima of particle number and chemical mass concentrations even though the most prominent marine biological aerosol precursor exhibits its zonal minima in this region.
Seasonal statistics of most of the analyzed parameters clearly reflect the marine biological control with broad maxima during austral summer.
The DMS-oxidation products nssSO4 and MSA are suspected to stimulate NPF-events with episodic high concentrations of N3 already in September and October with a subsequent broad summer maximum until February. Consistent with diatom blooms producing NH3, a broad summer maximum also is observed for NH4.
The concentrations of the somewhat larger N10-particles even exhibit their highest values south of –65° in March. During the same month nssCa that we suspect to derive from the OSM also shows highest monthly averages after a structured rise since October from low winter values.
For nssSO4 and MSA, nssCa, nssK, and particle number concentrations the most prominent source regions were found in high DMS-areas close to Antarctica, whereas the potential source regions of nssMg and NH4 were located in part further north over the Southern Ocean. We hypothesize that in part oceanic structures may cause geographic differences in aerosol source regions (see e.g., Deacon, 1982).
Our results suggest two foci of subsequent aerosol studies over the Southern Ocean:
The region between –40° and –60 ° in order to understand the processes leading to the relative maxima in several aerosol parameters, in particular those related to the ocean surface microlayer, and
Aerosol studies south of –65° understand source regions and source processes of new particle formation.
Data Accessibility Statements
Already published data sets utilized in the present study are available in the utilized format from the corresponding author. Unpublished data sets should be requested from the respective principle investigators.
Additional File
The additional file for this article can be found as follows:
Supplementary file
Information about the database, details of data processing, and complementary results. DOI: https://doi.org/10.16993/tellusb.1869.s1
Notes
[2] The complete oceanic coverage of the DMS climatology allows for higher latitudinal resolution than the sparse aerosol data.
[3] Apparent inconsistencies between DMS-distribution, land mask and continental shorelines are due to the retransformation of the original DMS-maps of Hulswar et al., (2021) onto the polar stereographic map of the present study.
Acknowledgements
We gratefully acknowledge the arguments on seasonal meridional circulation changes provided by Vimo Tihma, Finnish Meteorological Institute. The authors would also like to acknowledge the Australian Bureau of Meteorology and CSIRO for their long-term and continued support of the Kennaook Cape Grim Baseline Air Pollution Monitoring Station. RSH wishes to thank the CSIRO Marine National Facility (MNF) for its support in the form of sea time on RV Investigator, support personnel, scientific equipment and data management. We acknowledge the traditional owners of the land and sea where many of these measurements were undertaken. SS wishes to express his gratitude to L. P. Golobokova for her analysis of aerosol samples, as well as to a large number of operators who carried out measurements on different research cruises: Kabanov D.M., Polkin V.V., Turchinovich Yu.S. Gubin A.V., Lubo-Lesnichenko K.E., Prakhov A.N., Sidorova O.R., Terpugova S.A., Vlasov N.I., Zenkova P.N.
Funding Information
Shrivardhan Hulswar and colleagues of the Indian Institute of Tropical Meteorology, Pune, India generously shared their latest DMS-climatology with us before its publication. Aerosol data for the Indian coastal station, Maitri at Antarctica was made available by Vimlesh Pant and Devendraa Siingh of the Indian Institute of Tropical Meteorology, Pune, India which is funded by Ministry of Earth Sciences, Government of India. CPC-data from the Cape Point GAW station, South African Weather Service were contributed by Casper Labuschagne. Marek Kubicki shared the CPC data from the Polish station Arctowski in the South Shetlands with us. Willy Maenhaut of Ghent University kindly provided sample data from Amsterdam Island. We thank Alexander Mangold of the Royal Meteorological Institute of Belgium for providing the particle number concentration data for Princess Elisabeth Antarctica station. These measurements were financed by the Belgian Science Policy office through research contracts BR/143/A2/AEROCLOUD and BR/175/A2/CHASE and BR/175/A2/CHASE-2. Multiple datasets from island and coastal stations were made available by Greg Ayers, CSIRO, Australia, Joe Prospero of the University of Miami, USA, and Rolf Weller, Alfred Wegener Institute, Bremerhaven, Germany. Aerosol data from the first Circumantarctic research cruise ACE and from Bird Island were kindly shared by Julia Schmale, Extreme Environments Research Laboratory, Sion, Switzerland. Young Jun Yoon and Ki-Tae Park provided aerosol size distribution and major ion data collected at the Korean King Sejong Station, Antarctica. One of the authors (AKK) acknowledges the support of the Indian National Science Academy under the INSA Honorary Scientist Program. Silvia Becagli kindly contributed the chemical data from the Italian Antarctic station Mario Zucchelli, the collection of which had been supported by the Italian Ministry of University and Research and Programma Nazionale di Ricerca in Antartide through the Projects “Correlation between biogenic aerosol and primary production in the Ross Sea-BioAPRoS” (grant no. PNRA16_00065-A1).
Competing Interests
The authors have no competing interests to declare.