1 Introduction

Sea ice represents a powerful phenomenon exerting a strong influence on the oceanic, biological and climatic systems, and greatly affects the energy transport at the ocean-atmosphere interface as it shows a much higher albedo compared with open water (Brandt et al., 2005). Satellite observation has revealed an increasing trend in the total sea ice extent (SIE) around Antarctica since the late 1970s (e.g., Parkinson and Cavalieri, 2012). However, regional SIE changes and trends vary within different sectors of the Southern Ocean, ranging from increases, insignificant changes to decreases (Parkinson and Cavalieri, 2012; Jones et al., 2016), largely due to the intricate topographies of coastal Antarctica. Large-scale measurements of Antarctic sea-ice extent are only available from the late 1970s to present, which constrains our understanding of the role of sea ice in climate forcing over longer time scales. To address this, sporadic observations and proxy data such as ice core and marine sediment have been used to infer past sea ice conditions in Southern Hemisphere (Abram et al., 2013).

Previous studies have linked methane sulphonic acid (MSA) (Welch et al., 1993; Curran and Jones, 2000; Curran, 2003), sea salt sodium (Wolff et al., 2006; Severi et al., 2017), and deuterium excess (Sinclair et al., 2014) records in ice core to sea ice state around Antarctica. Unlike the marine sediment representing specific spatial details, a single ice core record can provide general information of the ice conditions in a particular sector (Abram et al., 2013). Generally, sea salt aerosol is generated by bubble bursting and sea spray over open water (Abram et al., 2011; de Leeuw et al., 2011). Large particles deposit rapidly over the ocean, smaller particles could be transported over the continents. As a result, a large amount of sea salt aerosols is transported by the atmospheric circulation and deposit rapidly at inland site far from the coast (Guelle et al., 2001). Therefore, sea salt concentration is suggested to negatively correlate with the distance between ice core locations and open water if open water is the main source of sea salt aerosol. Similarly, the concentration of sea salt is related to the SIE, because the extent of sea ice would increase the distance to the open water source and result in loss of the particle from the ocean during its transportation (Minikin et al., 1994; Mulvaney and Wolff, 1994; Benassai et al., 2005; Röthlisberger et al., 2010).

However, a number of studies queried the assumption of open water source because it is difficult to explain the peak concentration in aerosol and snow during winter half-year (Wolff et al., 2003), which indicates that another important source of sea salt comes from the high salinity frost flowers on sea ice surfaces. However, some laboratory observations (Roscoe et al., 2011) and model experiments (Levine et al., 2014) suggest that frost flowers are unlikely to be the major direct source of sea salt aerosol in Antarctica due to their high mechanical strength, and subsequent lack of observed aerosol production (Yang et al., 2017) even under high wind speeds (Roscoe et al., 2011). In the recent decades, the salty blowing snow lofted from the surface of sea ice is thought an important source of sea salt aerosol to the polar atmosphere (Yang et al., 2008; Dou et al., 2017; Rhodes et al., 2017; Zhang et al., 2018). The source of the sea salts aerosol is complicated, for the progress of its transport and deposition is closely related to the atmospheric circulation, meteorology condition and local geography environment (Mayewski et al., 2017). The real major source of the sea salt content in Antarctic sites is so far still an open and debated question.

In this paper, we assess the reliability of sea salt sodium (ssNa⁺) at Princess Elizabeth Land (PEL) as a proxy for SIE in the SIO. In addition, we aim at discussing the possible drivers of the airmass transport and sea ice change in the past three decades.

2 Materials and Methods 2.1 Sampling Site and Analysis Method

A shallow ice core (12.5 m), designated as '202 km', was drilled in 2017 at a site 202 km distant (71°04'38"S, 77°16'14"E, 2100 m asl) from Zhongshan Station in Princess Elizabeth Land (PEL), East Antarctic by the 33th Chinese National Antarctic Research Expedition (CHINARE, Fig. 1). Since the CHINARE program started in 1996, five auto weather stations (LGB69, PANDA-N, EAGLE, Dome A and PANDA-S) have been installed along the traverse route from Zhongshan Station to Dome A. These weather stations have provided continuous data for the studies of climate changes along Zhongshan-Dome A at East Antarctica. For example, the previous studies manifested the synoptic changes at coastal PEL (Chen et al., 2007; Ma et al., 2010). Based on meteorological records from the LGB69 weather station (about 40 km distant from this shallow core site), the annual mean near surface air temperature at this site is estimated to be -25.6℃.

The shallow ice core was frozenly transported to the State key laboratory of the Cryospheric Sciences (SKLCS) in Lanzhou, China, in 2017. A custom designed lathe (poly-methyl methacrylate) was used to decontaminate the ice core samples. We applied ceramic knives with replaceable blades to remove the potential contamination from the outside portion, and the inner cores were retained. The decontamination procedure was performed in a class 100 laminar flow clean bench below -12℃. A total of 259 samples with 4–5 cm resolution were cut and stored in 125 mL LDPE bottles. The stable water isotopes (oxygen isotopic ratio, δ¹⁸O and hydrogen isotopic ratio, δD) of the melting samples were analyzed using a Picarro L1102-i wavelength-scanned cavity ring-down spectrometer (Picarro Inc., USA) in SKLCS. As done by Du et al. (2018), cations and anions were tested by a Dionex ISC 3000 ion chromatograph (Dionex ISC 3000, Thermo Scientific, USA) with an Ion Pac CS12A column, 20 mmol L^-1 MSA eluent and a cation electrolytically regenerated suppressor (CERS) and an Ion Pac AS11-HC column, 25 mmol L^-1 KOH eluent and an anion electrolytically regenerated suppressor (ASRS), respectively.

Na⁺ content in Antarctic aerosol, snow and ice is dominated by sea salt, and thus could be reasonably regarded as a marker of sea salt (Röthlisberger et al., 2002). The 'pure' sea salt contribution could be more reliably assessed by removing the contribution of mineral dust from Na⁺ content. The total Na⁺ (totNa⁺) content was corrected using non sea-salt Ca²⁺ (nssCa²⁺) as a crustal marker and a simple two-variable, two-equation system allowing the evaluation of the ss-and nss-fractions of both Na⁺ and Ca²⁺:

$ \begin{array}{*{20}{l}} {{{{\mathop{\rm ssNa}\nolimits} }^ + } = {\mathop{\rm tot}\nolimits} {\rm{N}}{{\rm{a}}^ + } - {\rm{nssC}}{{\rm{a}}^{2 + }}/{R_{{\rm{crust}}}}, }\\ \begin{array}{l} {\mathop{\rm nss}\nolimits} {\rm{C}}{{\rm{a}}^{2 + }} = {\rm{C}}{{\rm{a}}^{2 + }} - {R_{{\rm{sea}}\;\;{\rm{water }}}}*{\rm{ssN}}{{\rm{a}}^ + }, \\ \left({{\rm{C}}{{\rm{a}}^{2 + }}/{\rm{N}}{{\rm{a}}^ + } = 0.038 = {R_{{\rm{sea}}\;\;{\rm{water }}}}} \right)\left({{\rm{C}}{{\rm{a}}^{2 + }}/{\rm{N}}{{\rm{a}}^ + } = 1.78 = {R_{{\rm{crust }}}}} \right). \end{array} \end{array} $

The R_crust and R_{sea water} are the mean ratios (w/w) in the Earth crust and in bulk seawater, respectively (Bowen, 1979).

2.2 Reanalysis Data and Climate Indexes

ERA-Interim, covering from 1979 to present, is a global atmospheric reanalysis product generated by the European Center for Medium-Range Weather Forecasts (ECMWF). Many studies contrasted lots of global reanalysis products and concluded that ERA-Interim likely offers the most realistic depiction of tropospheric pressure, wind and precipitation changes in high latitudes of southern hemisphere (Bromwich et al., 2011; Bracegirdle and Marshall, 2012; Bracegirdle, 2013; Wang et al., 2016; Xie et al., 2018), probably benefited from utilizing the four-dimensional variational (4DVar) analysis system which is strongly guided by satellite observations (Dee et al., 2011). The ERA-Interim data obtained from the ECMWF online data server were downgraded from the original model grid to a regular 0.75°×0.75° latitude-longitude resolution.

2.3 Sea Ice Data

SIE is defined as the area of ocean with sea ice concentration (SIC) greater than 15%. This was originally developed for satellite-based passive microwave products to be a robust identifier of ice edges when compared against aircraft observations. For observations, we used the monthly SIE Index, a parameter of sea ice Index products which were obtained from the National Snow and Ice Data Centre (NSIDC) (Fetterer et al., 2017). The latest SIE Index is calculated from gridded SIC data on a nominal 25 km grid derived from two data sets: the Near-Real-Time DMSP SSMIS Daily Polar Gridded, referred to as the NRTSI product in this document, and the SIC Nimbus-7 SMMR and DMSP SSM/I-SSMIS Passive Microwave Data, referred to as the GSFC product.

2.4 Climate Indexes

The Indian Ocean Dipole (IOD) and the Antarctic Oscillation (AAO) indices are used to better examine the possible linkage between SIE and ssNa⁺ at 202 km. The IOD is characterized by the Dipole Mode index (DMI), which is defined as the sea surface temperature (SST) gradient between the western equatorial Indian Ocean (50°–70°E and 10°S–10°N) and the south-eastern equatorial Indian Ocean (90°–110°E and 10°S–0°). The DMI index data since 1870 can be found at http://www.jamstec. go.jp/frsgc/research/d1/iod/iod/dipole_mode_index.html.

The SAM (Southern Annular Mode) index is defined as a mean latitudinal difference of sea level pressure at 40° and 65°S. This climate mode is considered as a prevailing atmospheric circulation in the Southern Hemisphere, representing about 35% of the extratropical Southern Hemisphere climate variability (Marshall, 2003). The monthly SAM index is available from the British Antarctic Survey (http://www.antarctica.ac.uk/met/gjma/sam.html; since 1957).

3 Results and Discussion 3.1 Ice Core Dating

The shallow ice core was dated by annual layer counting, using a multiparametric approach as described in Li et al. (2012). First, the non-sea salt (nss) SO₄^2- concentration is calculated (Fig. 2) using the SO₄^2-/Na⁺ concentration ratio in bulk seawater to identify the signals of volcanic eruptions. Volcanic signatures were extensively used to match different ice core records; volcanic products (mainly ash, dust, tephra particles and SO₂, rapidly oxidized to H₂SO₄) are emitted into the high stratosphere and the troposphere during volcanic eruptions. Influenced by the atmospheric circulation, some products deposited on polar ice sheets via wet and/or dry deposition and preserved in ice and might result in a peak of SO₄^2- concentration in the ice core, which is the so-called nss SO₄^2-. Therefore, the volcanic spikes are more easily distinguished from background values (Udisti et al., 2000) which are contributed by sea salt aerosols (MgSO₄, Na₂SO₄). The 1990s were characterized by the occurrence of the Pinatubo eruption in June 1991 and Cerro Hudson eruption in August 1991, which injected 12–20 Mt (million metric tons, or 1012 g) (Bluth et al., 1992; McPeters, 1993) and 2 Mt of SO₂ (Doiron et al., 1991) into the atmosphere respectively. By comparing the nss SO₄^2- profile from LGB69 ice core, the years within the highest value of nss SO₄^2- were identified as the inferred eruption years 1991/1992.

Second, the ice core was dated based on the well-preserved seasonal cycles of water isotope (δ¹⁸O, δD) and major sea-salt ions (Na⁺). Ice core δ¹⁸O and Na⁺ data are presented in Fig. 2. The age model is established by direct counting of annual isotope and sea salt aerosol cycles; the summer season coincides with δ¹⁸O maxima and Na⁺ minima, while the winter season each year coincides with δ¹⁸O minima and Na⁺ maxima. Finally, the time series were counted to have 27 annual cycles from 1990 to 2016 (Fig. 2) and correspond well with the annual cycles of the well-known ice core LGB69 (Li et al., 2012).

3.2 Sea Ice Extent and Sea Salt Aerosol Proxy

Fig. 3 shows the changes in the SIE over SIO and the total Antarctic continent for the period 1990–2016. For the total Antarctic sea ice, the SIEs range from 11.61×10⁶ km² in 2016 to 13.14×10⁶ km² in 2014. Linear regression is applied to the time series of the Antarctic total SIE and Southern Indian Ocean SIE (here after SIE_SIO, see bold lines in Fig. 3). The trends in the annual SIE of the whole Antarctica and Indian Ocean are positive, suggesting that in the studied period there is a general increasing trend in the SIE at the Southern hemisphere (Parkinson and Cavalieri, 2012). Most previous studies showed that the general increase in SIE at continental scale in Antarctica is mostly driven by the Ross Sea and the Indian Ocean (Parkinson and Cavalieri, 2012).

The ssNa⁺ record (Fig. 2, top panel, black line) shows a negative correlation with the annual mean SIE of the whole Antarctic (R = -0.5, N = 27, P < 0.01). The correlation with the SIE in the Indian Ocean sector is stronger (R = -0.6, N = 27, P < 0.01). Furthermore, we found that the most significant correlation occurred in winter season (June, July, August, JJA).

To verify the relationship between the ssNa⁺ and SIE and discuss the possible mechanism, the spatial correlation between ssNa⁺ and SIC and Geopotential Height (GPH) at 850 hPa during the winter (JJA) are highlighted in Fig. 4. The comparison with SIC (Fig. 4A) shows that the strongest anti-correlation occurs in the western Indian Ocean (30°–60°E), along the sea ice edge where sea ice variability is the greatest. The correlation with the GPH (Fig. 4B) shows a large positive feature spread over the central Indian Ocean from 60°–90°E. Due to the geography of PEL, the pressure, precipitation and temperature of this region are particularly sensitive to the strength of the Southern Indian Ocean Low (SIOL; Xiao et al., 2004), a quasi-stationary atmospheric system consistent with the SAM. It is evident that the ssNa⁺ concentration is largely affected by the strength and position of the atmospheric circulation center. To investigate the atmospheric circulation's effect on the ssNa⁺ concentration at this site, the composite analysis was applied to the meridional wind and GPH fields via calculating the difference between the years of high ssNa⁺ concentration value(1990, 1991, 1992, 2002, 2016) and low ssNa⁺ concentration value(1999, 2000, 2010, 2012, 2013, 2014). The composite analyses indicate that there is an anticyclone in coastal Antarctica in 60°–90°E (Fig. 4C) in high ssNa⁺ years, consistent with a significant southward wind in 30°–70°E and a significant northward wind near 90°E (Fig. 4D), which contribute to the poleward transport of marine airmass from central Indian Ocean. These features suggest the sea salt aerosol in the shallow core at this site probably comes from the open water and is closely related to sea ice edge in the SIO, particularly the central and western SIO. More southerly (northerly) winds would increase (decrease) ice transport from the south (Holland and Kwok, 2012) and cause atmospheric cooling (warming) from increased (decreased) advection of cold polar air masses (Liu et al., 2004; Turner et al., 2009) resulting in decreased (increased) airmass transport from the open water.

In addition, the major ions (Na⁺ and Cl^-) along the transect from Zhongshan to Dome A, East Antarctica, demonstrate that there is a significant decrease from 100 km to the coast (Du et al., 2018), suggesting that the ssNa⁺ concentration is negatively correlated with the distance to the open water. From the 27 years records in this shallow core, we inferred that the concentration of sea salt decrease as the amount of sea ice increases, because the expanding extent of sea ice would push the open water source further, increasing the amount of loss during transport (Minikin et al., 1994; Mulvaney and Wolff, 1994; Benassai et al., 2005; Röthlisberger et al., 2010). It is reasonable that the transport of ssNa⁺ is related to the formation and variability of the sea ice in SIO sector. Thus, the ssNa⁺ can be used as a potential proxy for reconstructing the sea ice extent in the SIO at least for the recent decades.

3.3 SIE Reconstruction and Validation

Fig. 3 shows the annual mean SIE over SIO and total Antarctic, and the ssNa⁺ concentration over the 27-year long record. According to the significant negative correlation between the ssNa⁺ concentration and SIE, we compared the annual mean log(ssNa⁺) with the SIE_SIO (see Fig. 5A). The value of the Pearson's correlation coefficient (R) is -0.62 (> 99% confidence level). Significance was calculated using two-tail t-test via the measurement of effective degree of freedom. The relationship between the SIE and the log (ssNa⁺) can be described by a simple linear regression according to the equation:

$ S I E_{-} S I O=-0.41 * \log \left[\mathrm{ssNa}^{+}\right]+0.26. $

The regression model based on the ordinary least square method explained 36.3% (after adjusting the degrees of freedom) of the total variance of the instrumental annual mean SIE_SIO from 1990 to 2016. The transfer function was significant (F = 15.84, P < 0.001) (Fig. 5). No significant autocorrelation was detected in this model. The above results suggest that the regression is valid to manifest the relationship between the annual SIE_SIO and the ssNa⁺ concentration. The uncertainty in the reconstructed SIE_ SIO series is mainly associated with the uncertainties in the model and the ice core dating error. For this approach, the uncertainty was defined as two times the standard deviation of the residuals of the SIE reconstruction from the SIE during the calibration interval (Fig. 5B).

According to this equation, we used our yearly-averaged ssNa⁺ to calculate the modeled SIE over SIO for each year from 1990 to 2016. The profile of this modelled SIE is shown as a blue line in Fig. 5B. We observed a general agreement between the reconstructed and satellite-derived SIE. Most variations can be depicted in our simple linear model, only several high frequency variations are not captured accurately (e.g., the 1994 and 2007 SIE peaks).

Severi et al. (2017) showed that the ssNa⁺ record in Talos Dome which comes from sea ice region rather than the open water can be used as a proxy for sea ice extent reconstruction in Ross Sea and Western Pacific Ocean for the period 1979 to 2003 as well. Their results highlighted the largest positive SIE anomalies since the late 1990s and agreed with the increased sea ice in SIO observed from this study. However, the sea salt aerosol source in the two regions are different due to the geophysical environment and local atmospheric circulation.

3.4 Possible Effect of Climate Modes

The Indian Ocean Dipole can excite Rossby wave trains over the eastern tropical Indian Ocean and modulate the cyclonic activity close to sea ice zone from the Indian Ocean to the west Ross Sea (Saji et al., 2005; Cai et al., 2011; Nuncio and Yuan, 2015; Ekaykin et al., 2017) Fig. 5C shows the variations of the annual IOD index (DMI) during the studied period, highlighting a general agreement between the IOD and our reconstructed SIE_ SIO. The IOD index significantly correlated with the SIE in the SIO (R = 0.33, P < 0.1) and ssNa⁺ concentration in 202 km (R = -0.38, P < 0.05). However, the correlation coefficient between the IOD and sea ice changed to non-correlated if we replaced the reconstructed SIE with the observed SIE series. It is suggested that the IOD contributes more to sea salt transport and less to sea ice in this region.

The composite of the SIC anomaly based on the satellite observation suggests that sea ice in central SIO (50°– 80°E) increases (decreases) in the years of positive (negative) IOD (Fig. 6A), consistent with the positive correlation showed in time series. Fig. 6B shows differences in the composites of 850 hPa meridional wind and wind field during the positive and negative IOD phases. The results indicate that the positive IOD produces a cyclonic circulation in SIO (Fig. 6B), accompanied with a northward meridional wind over 30°–70°E which contributes to the sea ice expand and a southward meridional wind in 80°– 100°E which suppresses the sea ice expand. On the other hand, even though there is a significant southward wind occurring near 90°E which favors airmass transport from open water to PEL in positive phase of IOD, the cold and strong katabatic wind from inland would resist the airmass incursion and result in the low ssNa⁺ concentration in the site 202 km distant from coast.

The circulation variability of the lower atmosphere in the Southern Hemisphere which has a direct effect on airmass transport and sea ice condition in Antarctic is generally triggered by the annular mode in the pressure field anomaly at various time scales, from seasonal to inter-annual (Thompson and Wallace, 2000). This mode has been referred to as the SAM and is significantly related to the strength of the westerlies and storm track around Antarctica (Baldwin, 2001). Variations in the phase of the SAM cause an exchange of mass between the Antarctic and midlatitudes. In the positive (negative) phase, sea level pressure anomalies are negative (positive) at high (mid) latitudes, resulting in a stronger (weaker) polar jet. The dataset described in Marshall (2003) is used to discuss SAM's effect on sea ice changes and sea salt aerosol transport. The correlation coefficient of the SAM index with SIE_SIO is 0.45 (P < 0.05) and that with the observed SIE is 0.47 (P < 0.05) during the past 27 years. Meanwhile, the SAM index is negatively correlated (R = -0.45, P < 0.05) with the ssNa⁺ concentration in the shallow core.

The composite analyses suggest that the significant increase of sea ice concentration (Fig. 6C) in the whole SIO is probably attributed to the strengthened westerlies and the katabatic wind (Fig. 6D) when the SAM is positive, and vice versa. Holland and Kwok (2010) demonstrated that sea ice concentration trends surrounding Antarctica are largely driven by near-surface wind changes associated with climatological lows in the circumpolar pressure trough surrounding Antarctic. Fig. 6D shows the stronger westerlies are associated with the weak even disappeared SIOL and northward (katabatic) wind anomaly in the eastern and western of SIO. Katabatic wind events occur year round, but are greatly enhanced when cyclones move into the region, typically from the west (Turner et al., 2009; Wang et al., 2014). Strong katabatic wind area associated with the cyclone-induced strong southerly winds can cause accumulated and extensive westward drift of sea ice (Wang et al., 2014), and thus have great potential to generate high sea ice concentration center on the western side of 120°E. In addition, the increased equatorward heat flux and equatorward Ekman drift associated with the positive SAM (Hall and Visbeck, 2002; Fan et al., 2014) would increase transport of cold waters to the northward and cause an increase of sea ice cover in SIO.

4 Conclusions

In this paper, we presented an analysis of the recent variability in ssNa⁺ concentration from a new shallow ice core extracted in site 202 km away from Zhongshan station in PEL, East Antarctica. Based on the remarkable seasonal variations of water isotope and sea salt ions, the core was dated to cover 27 years from 1900 to 2006. The correlation analysis and atmospheric circulation pattern suggest that the sea salt aerosol in this region mainly comes from the open water over central and western SIO and is closely related to the sea ice changes in SIO; greater sea ice extent would lead to lower sea salt at this site, and vice versa. Therefore, the ssNa⁺ concentration at this site from 1990 to 2016 is proved to be a potential proxy to reconstruct the annual SIE in the SIO, at least for recent decades. Based on the statistically significant negative linear relationship between the ssNa⁺ and the satellite-retrieved SIE_SIO, we reconstructed the SIE profile using our ssNa⁺ record for a 27-year long period. The significant relationship manifests the potential of this site to record variations in sea ice and sea ice-related parameters, and to use ssNa⁺ as a robust sea-ice tracer from the ice core drilling site.

The sea ice extent and concentration and the progress of the airmass or moisture transport are highly related to the changes of Southern Hemisphere atmospheric circulations which are regularly defined as the climate mode indices. The annual mean SIE in SIO is significantly and positively correlated with the annual mean IOD index and SAM index. The cyclonic circulation occurring in positive IOD produces a northward meridional wind in central SIO which contributes to the sea ice expand. The effects of the SAM are stronger than those of the IOD. During the past 27 years since 1990 in this study, the sea ice increase in SIO is probably attributed to the positive phase shift of SAM in the 1990s (Yang and Xiao, 2018). The positive SAM is associated with the strengthened westerlies and decreased poleward pressure gradient and thus diminishes the transport of heat and moisture which leads to increased sea ice in SIO and decreased sea salt aerosol transport from open water.

It is known that sea-salt aerosol originated from the sea ice surface which depends on the meteorological conditions also has a strong influence on the ssNa⁺ concentration in coastal Antarctica. Current and future efforts in sea salt aerosol modeling are urgent to improve the understanding of the information hidden in the ssNa⁺ ice core records.

Acknowledgements

This work was supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (No. XDA19070103), the National Natural Science Foundation of China (Nos. 41425003, 41701071), the National Key Research and Development Program of China (No. 2018 YFC1406100), and the CAS 'Light of West China' Program. Thanks are to logistics support of Mrs. Fuhai Wei, Xu Yao and Nan Zhang.