Spatial distribution of marine aerosols in West Antarctica

The spatial distribution of Na⁺ concentrations across West Antarctica is examined to identify the dominant source regions and pathways of marine aerosols (Fig. 1). The highest mean Na⁺ concentrations occur in the Ross drainage system at Siple Dome (156 μg L^–1), Up-C (91 μgL^–1), 99-1 (59 μgL^–1) and CWA-A (54 μgL^–1), indicating that Na⁺ deposition in the Ross drainage system is dominated by marine aerosols from the Ross Sea transported across the Ross Ice Shelf (Fig. 1; Table 1). In the Pine Island–Thwaites drainage system, Na⁺ concentrations are highest in the eastern sector (01-5 (30 μgL^–1), 01-6 (33 μgL^–1)), indicating that the dominant source region is not the Ross Sea, but rather the Amundsen and Bellingshausen Seas. The ice-divide sites (00-1, RIDS A, 00-4 and 00-5) are most likely influenced by both the Ross and Amundsen–Bellingshausen Sea regions.

Previous studies identify distance inland and elevation as the main controls on the concentration of marine species in snow across Antarctica, noting that a decreasing trend in Na⁺ concentrations vs elevation is more significant than that vs distance inland (Reference Mulvaney and WolffMulvaney and Wolff, 1994; Reference Kreutz and MayewskiKreutz and Mayewski, 1999; Reference KreutzKreutz and others, 1999). An examination of the mean Na⁺ concentration from the 16 ice-core records vs distance inland does not show a trend, even when the drainage systems are considered independently. This may result from difficulties in identifying distance inland, because variations in atmospheric circulation patterns and sea-ice extent can change the relative distance from a location to the marine aerosol source. There is not a trend between Na⁺ concentration and elevation across the study region. However, a decreasing trend in Na⁺ concentrations is evident with increasing elevation when the drainage systems are considered independently (Fig. 3). This relationship does not hold for ice-divide sites, perhaps because of Na⁺ contributions from both the Ross and Amundsen– Bellingshausen Seas. It is difficult to assess the relative importance of distance inland and elevation vs Na⁺ concentration in the study region because distance inland and elevation both vary between sites. For example, in the Ross drainage system, Siple Dome has considerably higher Na⁺ concentrations than the lower-elevation Up-C site, likely because Siple Dome is closer to the coast (Table 1).

Fig. 3. Elevation (m) vs Na⁺ concentrations (μgL^–1) for Ross (a) and Pine Island–Thwaites (b) drainage systems.

Mechanisms controlling marine aerosol production and transport

Factors other than distance inland or elevation contribute to the high Na⁺ concentrations in the Ross drainage system relative to the Pine Island–Thwaites drainage system. Reference Kreutz and MayewskiKreutz and Mayewski (1999) observed that the concentration of sea-salt species at Siple Dome was consistent with that at other Antarctic sites at a similar elevation, but significantly higher than at sites a similar distance inland. Na⁺ concentrations at Up-C, 99-1 and CWA-A in the Ross drainage system are higher than at other sites at similar elevations and shorter distances inland in the Pine Island–Thwaites drainage system (01-6, 01-5, 01-3 and 01-2) (Fig. 1; Table 1). The higher Na⁺ concentrations in the Ross drainage system could be caused by more efficient production or more intense transport of marine aerosols from the Ross Sea region.

Frost flowers

The transport of highly saline frost flowers inland is a possible mechanism leading to high Na⁺ concentrations in the Ross drainage system. The production of frost flowers is dependent upon the rapid formation of new sea ice under cold conditions. As the sea ice freezes, a thermo-molecular pressure gradient transports brine from the ice interior toward the colder surface (Reference Wettlaufer and WorsterWettlaufer and Worster, 1995). Field and laboratory observations have shown that during the initial hours of new sea-ice formation the brine has a salinity of 122% at –8^˚C, which increases to 212% at –20^˚C (Reference RichardsonRichardson, 1976; Reference Perovich and Richter-MengePerovich and Richter-Menge, 1994). The surface brine evaporates, and creates a water-vapor layer in the atmospheric boundary layer in which enhanced crystal growth forms frost flowers (Reference Martin, Yu and DruckerMartin and others, 1996). Surface tension results in transport of the surface brine onto the frost crystals, yielding highly saline frost flowers (Reference Perovich and Richter-MengePerovich and Richter-Menge, 1994). The brine and frost flowers can have salinities three times higher than sea water (Reference Drinkwater and CrockerDrinkwater and Crocker, 1988; Reference Perovich and Richter-MengePerovich and Richter-Menge, 1994; Reference Martin, Drucker and FortMartin and others, 1995). If frost flowers are mobilized into the atmosphere, they present a highly concentrated source of sea-salt particles (Reference Hall and WolffHall and Wolff, 1998). The ideal wind speed for efficient frost-flower transport is not known; however, low winds are insufficient to allow aerosol incorporation of Na⁺ from the flowers, and high wind speeds destroy the flowers (Reference Rankin, Auld and WolffRankin and others, 2000). Usually within 1–2 weeks of formation the structure of the frost-flower layer is modified by strong winds (Reference Drinkwater and CrockerDrinkwater and Crocker, 1988).

The salinity of frost flowers depends on the temperature at which the flowers form (Reference Martin, Yu and DruckerMartin and others, 1996), but generally frost flowers are sodium sulfate-depleted (Reference Rankin, Wolff and MartinRankin and others, 2002). If the ice surface temperature is cold enough, mirabilite precipitates within the ice interior, but at warmer temperatures (–12 to –16^˚C) the mirabilite is precipitated out of a slush layer (Reference Martin, Yu and DruckerMartin and others, 1996). NaCl precipitates below –22^˚C, but because ice surface temperatures this cold are usually associated with multi-year ice, frost flowers depleted in NaCl are not expected (Reference Rankin, Wolff and MartinRankin and others, 2002).

The fractionation process from the precipitation of mirabilite can be expected for all sea-ice production with ice surface temperatures below –8^˚C (Reference Rankin, Wolff and MartinRankin and others, 2002), and under appropriate conditions at certain locations the fractionated frost-flower aerosol could dominate the sea-salt budget (Reference Rankin, Auld and WolffRankin and others, 2000). Reference Rankin, Wolff and MartinRankin and others (2002) calculate that 13% of the samples in an ice core from Dolleman Island on the Antarctic Peninsula have SO₄ ^2–/Na⁺ ratios less than the standard sea-water weight ratio (suggesting a frost-flower source); however, those samples account for 39% of the Na⁺ in the core. Similar analysis is conducted on the ice-core records from this study. The percentage of the total Na⁺ budget that is fractionated is calculated by:

A fractionated sample is defined as having a SO₄ ^2–/Na⁺ weight ratio less than the standard sea-water weight ratio. Mass was estimated for the Up-C core because density measurements were not taken, and the percentage of the Na⁺ budget that the fractionated samples account for at CWA-A is not calculated because density and depth measurements were not taken. Of the total samples, 21% at Siple Dome, 7% at Up-C, 4% at 99-1, and 4% at CWA-A are fractionated, yet this accounts for 40%, 17% and 10% of the total Na⁺ in the Siple Dome, Up-C and 99-1 cores, respectively (Table 1). The samples with SO₄ ^2–/Na⁺ less than the standard sea-water weight ratio account for a disproportionately high percentage of the total Na⁺ budget.

Calculations of fractionated samples may underestimate Na⁺ with a frost-flower source because inputs from marine biogenic or volcanic SO₄ ^2– could increase the ratios above that of standard sea water (Reference Rankin, Wolff and MartinRankin and others, 2002). Long-range transport of marine biogenic SO₄ ^2– from mid-latitude sources can occur throughout the year, but the highest contributions are during summer (Reference MinikinMinikin and others, 1998). Because frost-flower contributions to the SO₄ ^2– budget are greatest during winter, and biogenic inputs are greatest during summer, there is likely a low occurrence of biogenic SO₄ ^2– inputs masking the evidence of frost flowers. Volcanic inputs to the SO₄ ^2– budget more strongly influence the sites with elevations above 1000m (Reference Dixon, Mayewski, Kaspari, Sneed and HandleyDixon and others, 2004). Siple Dome, Up-C and CWA-A, all with elevations below 1000m (Table 1), are the sites with the greatest occurrence of depleted SO₄ ^2–/Na⁺ ratios. This suggests that volcanic inputs to the SO₄ ^2– budget at the higher-elevation sites may be masking evidence of frost flowers. However, volcanic inputs are episodic (Reference Dixon, Mayewski, Kaspari, Sneed and HandleyDixon and others, 2004), so evidence of frost flowers should be apparent during periods with no volcanic inputs.

Periods of SO₄ ^2–/Na⁺ ratios below the standard sea-water weight ratio generally coincide with high winter Na⁺ concentrations (Fig. 4). Analysis of the Siple Dome SO₄ ^2–/ Na⁺ time series indicates fractionated samples for 30 of the 44 winters between 1950 and 1994. Frost flowers may be a significant source of Na⁺ during every winter, but inputs from biogenic or volcanic SO₄ ^2– could mask evidence of the frost flowers. The higher prevalence of fractionated samples at Siple Dome than at the other core sites is likely caused by higher deposition of frost flowers, and/or relatively less nssSO₄ ^2– contributions in winter resulting in fewer masked samples.

Fig. 4. Siple Dome Na⁺ concentrations (μgL^–1) (gray line) and SO₄ ^2–/Na⁺ ratios (black line). SO₄ ^2–/Na⁺ ratios below the standard bulk sea-water ratio (0.25) are black filled, and the gray filled areas are the Na⁺ peaks.

Siple Dome, Up-C, 99-1 and CWA-A (all located in the Ross drainage system) have the highest Na⁺ concentrations in the study area, and the low SO₄ ^2–/Na⁺ ratios at these sites suggest frost flowers as a Na⁺ source (Table 1). Therefore, transport of highly saline frost-flower aerosols into the Ross drainage system may be the cause of the high Na⁺ concentrations at these sites. The size of frost-flower aerosols in part determines the distance the aerosols can be transported. Previous work at Halley research station in the Weddell Sea region indicates that frost-flower aerosols are generally in the super-micron range, but may also be present as very small particles (Reference Rankin and WolffRankin andWolff, 2003). The evidence of frost-flower aerosols at 99-1 and CWA-A indicates that the frost-flower aerosols are transported long distances and to high elevations, and suggests that at these sites the frost-flower aerosols are present as small particles. In contrast, there is minimal evidence of frost flowers in the ice-core records from the Pine Island–Thwaites drainage system (01-5, 01-3, 01-2) (Table 1). The distance inland at 01-5 and 01-2 is less than at Siple Dome, Up-C, 99-1 and CWA-A, and the elevation at 01-5 and 01-2 is similar to that at 99-1, so there must be factors other than distance inland or elevation controlling frost-flower transport into the Ross drainage system. This could be a consequence of different frost-flower production rates in the Ross Sea relative to the Amundsen–Bellingshausen Seas. Alternatively, additional nssSO₄ ^2– inputs in the Pine Island–Thwaites drainage system could be masking the evidence of frost flowers in these ice cores, but this does not explain the lower Na⁺ concentrations in the Pine Island–Thwaites drainage system relative to the Ross drainage system.

Sea ice

Examination of the mean annual Na⁺ distribution map (Fig. 1) suggests that the Ross Sea is the dominant Na⁺ source region for sites located in the Ross drainage system. This, coupled with evidence for frost flowers at Siple Dome, Up-C, 99-1 and CWA-A based on their low SO₄ ^2–/Na⁺ ratios, suggests that the Ross Sea may be a larger source region for frost-flower formation than the adjoining Amundsen and Bellingshausen Seas.

Degrees latitude of sea ice (determined from land to the sea-ice edge) were calculated using mean monthly and annual time series of sea-ice extent in 10⁸-wide sections (Reference Simmonds and JackaSimmonds and Jacka, 1995) from 1973 to 1996 from 130^˚ to 280^˚ to examine the relative extent of sea ice in the Ross, Amundsen and Bellingshausen Seas. Degrees latitude of sea-ice extent for September (when sea ice is at or near its maximum) and the annual mean are greater in the Ross Sea than in the Amundsen and Bellingshausen Seas (Fig. 5). This may in part explain the higher Na⁺ concentrations at Siple Dome and 99-1, because the greater sea-ice extent may enable greater production of frost flowers.

Fig. 5. Degrees latitude of sea-ice extent from the coast to the sea-ice edge for September and the annual average for 10^˚ sections of longitude. Sea-ice extent is greatest in the Ross Sea.

Monthly and annual time series of sea-ice extent from 1973 to 1996 (Reference Simmonds and JackaSimmonds and Jacka, 1995) were tested for correlations with the mean annual Na⁺ time series for all core sites. Significant correlations over a large region and in consecutive months suggest a strong relationship between sea-ice extent and aerosol transport to the core sites. Positive correlations indicate increased sea-ice extent when Na⁺ concentrations are high. Large differences in correlations exist between the sites due to high spatial variability (Figs 1 and 2). Significant positive correlations between Na⁺ concentrations and sea-ice extent exist for Siple Dome (n = 22, p = 0.05, r > 0.36) during winter months (May– July) between 180^˚ and 240^˚ (Fig. 6). The Siple Dome correlations do not exist during all three months across all latitude bands, and the strongest correlations are farther east than expected (220–240^˚) based on the assumption that the majority of Na⁺ at Siple Dome comes from the Ross Sea region. Thus, the strength of the relationship between sea-ice extent and Na⁺ at Siple Dome is not extremely robust. Significant correlations of mean annual sea-ice extent and Na⁺ concentrations exist at RIDS A (n = 23, p = 0.05, r > 0.35) between 240^˚ and 280^˚, at RIDS B (n = 23, p = 0.05, r > 0.35) between 220^˚ and 240^˚, and at CWA-D between 230^˚ and 280^˚ (n = 22, p = 0.05, r > 0.36) (Fig. 7). Reference DixonDixon and others (2005) report significant correlations of ssSO₄ ^2– and sea-ice extent at the same sites. Correlations between sea-ice extent and RIDS A, RIDS B and CWA-D suggest that marine aerosol concentrations in the central part of West Antarctica are linked to sea-ice extent. The correlations for RIDS A and CWA-D are strongest in the region between 240^˚ and 280^˚, suggesting that RIDS A and CWA-D are associated with the Bellingshausen and East Amundsen Seas. The RIDS B correlations are strongest between 220^˚ and 240^˚, suggesting an association with the West Amundsen and Ross Seas. The regions where the correlations are strongest may indicate the dominant source of the aerosols.

Fig. 6. Correlation coefficients (r) of monthly (May–July) sea-ice extent, 1973–96, in 10^˚ sections of longitude and the mean annual Siple Dome Na⁺ record.

Fig. 7. Correlation coefficients (r) of mean annual sea-ice extent, 1973–96, in 10^˚ sections of longitude and the mean annual Na⁺ records from RIDS A, RIDS B and CWA-D.

The production of frost flowers is physically linked with the formation of new sea ice rather than present ice extent, so correlations of new sea ice and mean Na⁺ concentrations were also tested for. New sea ice was determined by taking the difference between the greatest (usually September) and least (usually January) sea-ice extent for each year and 10^˚ -wide sections. These results confirm the findings from the correlations of mean annual sea-ice extent and Na⁺ concentrations: significant correlations of new sea ice and mean annual Na⁺ concentrations exist at RIDS B (n = 23, p = 0.05, r > 0.35) between 240^˚ and 250^˚, and at CWA-D (n = 22, p = 0.05, r > 0.36) between 250^˚ and 270^˚. Increased Na⁺ concentrations during periods of increased sea-ice production and extent may indicate greater production and transport of frost-flower aerosols, but it is not known why Na⁺ concentrations in the central region of the study area are associated with sea ice whereas similar correlations are not found at other nearby cores. Possible mechanisms for the marine-aerosol–sea-ice association include sensitivity to sea-ice duration, overall extent and seasonal or annual variability.

Open-water areas

Frost-flower formation is associated with the development of young sea ice; thus, as sea ice ages and thickens, the production of frost flowers ceases. As a consequence, sea-ice extent may not be the best indicator of frost-flower production. Reference WagenbachWagenbach and others (1998) identified polynyas and leads as potential open-water sources for winter marine aerosols. Leads typically open to widths of tens to hundreds of meters (Reference Perovich and Richter-MengePerovich and Richter-Menge, 1994), and polynyas can be thousands of square kilometers (Reference Arrigo and van DijkenArrigo and Van Dijken, 2003b). These regions could be sources of marine aerosols from both frost flowers and sea spray. The presence of open-water features within the sea ice may allow continual production of frost flowers throughout the year. In these open areas, heat loss to the atmosphere from the relatively warm ocean water reaches a maximum and results in the rapid growth of ice. Katabatic winds can drive extremely high rates of sea-ice production in polynyas because older sea ice is continually blown offshore and replaced by new sea ice (Reference Arrigo and van DijkenArrigo and Van Dijken, 2003b).

Variations in the formation and presence of leads and polynyas in the Ross, Amundsen and Bellingshausen Seas could in part explain why the Ross drainage-site data indicate a frost-flower source when other core sites do not. Currently, there is no available information to indicate that there is greater lead formation in the Ross Sea than in the Amundsen–Bellingshausen region. The Ross Sea does, however, have considerably more open water due to the presence of the Ross Sea polynya. The Ross Sea polynya is the largest in Antarctica (summer area 396 500km², winter area 19 600 km²; Reference Arrigo and van DijkenArrigo and Van Dijken, 2003b). Six smaller polynyas have been identified in the Amundsen– Bellingshausen region (Sulzberger Bay, Hull Bay, Wrigley Gulf, Amundsen Sea, Pine Island Bay, Eltanin Bay), but their total combined area (summer area 103 100km², winter area 12880km²; Reference Arrigo and van DijkenArrigo and Van Dijken, 2003b) is still considerably smaller than that of the Ross Sea polynya. Thus, the Ross Sea polynya may be a significant source of marine aerosols for the Ross drainage system from sea spray and frost flowers. Correlations of the annually resolved Na⁺ records and a time series of the Ross Sea polynya area from 1978 to 1994 (Reference Arrigo and van DijkenArrigo and Van Dijken, 2003a) were examined to look for associations between polynya size and marine aerosol transport, but no significant associations were found. The seasonal variations in the size of the Ross Sea polynya are much greater than interannual variations, and sea-ice (and thus frost-flower) production is controlled more by wind and temperature conditions than by interannual variations in polynya size, which may explain the lack of significant Na⁺-concentration–polynya-size associations.

Atmospheric circulation

Although there is compelling evidence that frost flowers are a major contributor to the Na⁺ budget, sea spray from open-water sources remains an important source of Na⁺. Changes in atmospheric circulation caused by shifts in the position and deepening of low-pressure systems on seasonal and longer timescales alter the pathways that transport marine aerosols inland (Reference Kreutz, Mayewski, Pittalwala, Meeker, Twickler and WhitlowKreutz and others, 2000b).

Wind speeds are strongest in the winter months, with the strongest winds in the Ross drainage region of the study area (Reference KalnayKalnay and others, 1996; Reference King and TurnerKing and Turner, 1997). High Na⁺ concentrations in the winter (Fig. 2) coupled with the higher wind speeds suggest that stronger winds transport more marine aerosols. The dominant wind direction in the Ross drainage system comes from katabatic winds from inland Antarctica that cause consistently more turbulent conditions at the boundary layer. These turbulent conditions ultimately may result in greater production and transport of aerosols from sea spray. Less dominant northerly winds likely transport most of the marine aerosols into the Ross drainage system from the Ross Sea. Increased aerosols from sea spray coupled with greater frost-flower production in the winter months would result in higher Na⁺ concentrations.

The influence of atmospheric circulation on the production of marine aerosols and their transport into West Antarctica is assessed by examining associations between sea-level pressure (SLP) from the US National Centers for Environmental Prediction (NCEP) re-analysis (Reference KalnayKalnay and others, 1996) and the Na⁺ time series. Previous studies found strong associations between SLP and marine aerosol variability (Reference Kreutz, Mayewski, Pittalwala, Meeker, Twickler and WhitlowKreutz and others, 2000b; Reference SouneySouney and others, 2002). To investigate the relationship between SLP and the Na⁺ time series from the ice cores in this study, a method similar to Reference Kreutz, Mayewski, Meeker, Twickler and WhitlowKreutz and others (2000a, Reference Kreutz, Mayewski, Pittalwala, Meeker, Twickler and Whitlowb) is used in which years of high and low marine aerosol transport are defined as being at least ±1σ from the Na⁺ mean (calculated for the period 1948–present) for each record. Low-Na⁺-year NCEP SLP fields are subtracted from high-Na⁺-year NCEP SLP fields (Fig. 8). Negative differences in SLP in the Ross, Amundsen and Bellingshausen Seas for core sites 01-5, 01-2, 00-1, 00-4, 99-1, CWA-D, CWA-A and Siple Dome indicate increased transport of marine aerosols to these sites when SLP is low. This relationship is strongest during September–November, in agreement with Kreutz and others’ (2000a, b) results (Fig. 8). The negative differences in SLP for Ross drainage-system sites (99-1 and Siple Dome) are comparatively small, perhaps because Na⁺ concentrations at these sites are more strongly influenced by frost-flower production than by atmospheric circulation. This method did not result in significant differences in SLP fields for the other core sites (01-3, RIDS A, RIDS B, RIDS C, 00-5, Up-C and South Pole), perhaps because of additional factors controlling Na⁺ deposition at these sites (e.g. blocking high effects, blocking by katabatic winds, core-dating errors). More notably, the sites that do have strong SLP–Na⁺ associations (01-5, 01-2, 00-1, 00-4, 99-1, CWA-D, CWA-A and Siple Dome) are distributed throughout the study area, indicating that SLP influences marine aerosol transport across West Antarctica. Years of high and low Na⁺ concentrations vary between sites, which may result from shifts in the position of the Amundsen Sea low (ASL), a climatological feature associated with cyclonic activity and the transport of moisture and aerosols into West Antarctica. The position of the ASL can migrate ~1400km on an annual basis, moving between a position near the eastern Ross Ice Shelf and the Antarctic Peninsula (Reference Cullather, Bromwich and van WoertCullather and others, 1996). When the ASL is positioned near the Ross Ice Shelf, cyclonic activity in that region may result in increased marine aerosol transport to the Ross drainage system, whereas when the ASL is positioned near the Antarctic Peninsula, cyclonic activity in that region may result in increased marine aerosol transport to the Pine Island– Thwaites drainage system. Thus, in a simplified system a see-saw relationship may exist in which marine aerosol loadings increase (decrease) in the western section of West Antarctica and decrease (increase) in the eastern sector, depending on the positioning of the ASL (Reference Cullather, Bromwich and van WoertCullather and others, 1996). Future research is needed to develop an even more robust deconvolution of the marine aerosol signal and resultant ice-core marine series.

Fig. 8. Mean September–November SLP differences plotted in millibars derived from the US National Centers for Environmental Prediction/US National Center for Atmospheric Research (NCEP/ NCAR) re-analysis. Method used to select years is described in the text.