Introduction
Stable isotope records of snow and ice cores in Antarctica were widely used to infer past local temperature variations and moisture origin (Dansgaard, Reference Dansgaard1964; Landais and others, Reference Landais, Barkan and Luz2008; Steen-Larsen and others, Reference Steen-Larsen2011; Xiao and others, Reference Xiao2013). The isotopic composition in snow, namely the ratio between its heavy and light water molecules, is expressed as δD or δ18O relative to the Vienna Standard Mean Ocean Water (V-SMOW) reference (δD and δ18O, defined as δD = (([HD16O]/[H216O]) sample / ([HD16O]/[H216O]) V−SMOW −1) × 1000; δ18O = (([H218O]/[H216O]) sample/ ([H218O]/[H216O]) V−SMOW-1) × 1000) (Craig, Reference Craig1961; Dansgaard, Reference Dansgaard1964; Masson-Delmotte and others, Reference Masson-Delmotte2008; Steen-Larsen and others, Reference Steen-Larsen2011). More recent ice core studies in Antarctica have been dedicated to obtaining the longest possible climate records in the Polar Regions (e.g., Jouzel and others, Reference Jouzel2007), or characterizing at high-resolution structure of climate variability during glacial (Masson-Delmotte and others, Reference Masson-Delmotte2003; EPICA Community Members, 2006; Stenni and others, Reference Stenni2010) or interglacial periods (Pol and others, Reference Pol2011). Ongoing efforts are directed at documenting the regional variability of climate in various sectors of Antarctica and the related trajectories of air mass transportation and moisture origin. Triple oxygen isotopes (16O, 17O and 18O) and double hydrogen isotopes (1H and 2H) and the second-order parameters (d-excess and 17O-excess, calculations follow the equations d-excess = δD-8 × δ18O and 17O-excess = 106 (ln(δ17O/1000 + 1)-0.528 × ln(δ17O/1000 + 1)), respectively) were widely involved to study the temporal and spatial variations of water isotopes (Dansgaard, Reference Dansgaard1964; Masson-Delmotte and others, Reference Masson-Delmotte2003; Barkan and Luz, Reference Barkan and Luz2007; Hou and others, Reference Hou, Li, Xiao and Ren2007, Reference Hou, Wang and Pang2013; Landais and others, Reference Landais, Barkan and Luz2008; Xiao and others, Reference Xiao2013; Benetti and others, Reference Benetti2014; Pang and others, Reference Pang2015, Reference Pang2019). The isotopic content is strongly influenced by fractionation processes encountered by the air mass throughout its history, such as evaporation conditions, mixture of moisture sources and distillation of the air mass (Dansgaard, Reference Dansgaard1964; Masson-Delmotte and others, Reference Masson-Delmotte2003, Reference Masson-Delmotte2008; Steen-Larsen and others, Reference Steen-Larsen2015). The variations of the relationship between stable isotopes and the temperature arise from changes in ocean evaporation conditions such as sea surface temperatures (SST), relative humidity and ocean surface water isotopic composition (Bonne and others, Reference Bonne2019). When only the equilibrium isotopic fractionation occurs, a mean slope of 8.0 for δD/δ18O of the precipitation will be presented. However, during non-equilibrium phase occurs, such as ocean surface evaporation processes, a kinetic fractionation effect is added to the equilibrium fractionation (Craig and Gordon, Reference Craig, Gordon and Tongiogi1965), which causes a larger kinetic fractionation for H218O than for HDO due to their different molecular diffusivities. As a result, d-excess is strongly imprinted by the kinetic effect. Therefore, the calculation of the d-excess will provide independent isotopic information related to the initial air mass evaporation conditions at the ocean surface (mainly SST and relative humidity) (Merlivat and Jouzel, Reference Merlivat and Jouzel1979; Bonne and others, Reference Bonne2019).
The mean slope between δD and δ18O during the equilibrium fractionation (8.0) also known as the mean meteoric water line (MWL) was widely used to study the fractionation of the water isotopes in the worldwide precipitation (Dansgaard and others, Reference Dansgaard1964). However, a lower mean MWL slope (7.75) was detected on Antarctica ice sheet (Masson-Delmotte and others, Reference Masson-Delmotte2008) and showed great spatial variations (Xiao and others, Reference Xiao2013; Li and others, Reference Li2016). The evaporation, transportation, condensation and the post-depositional processes on water vapor have influences on both the evolution of the water stable isotopes and MWL slopes (both for the equilibrium fractionation and kinetic fractionation) in Antarctic snowpack (Masson-Delmotte and others, Reference Masson-Delmotte2008). The influencing factors at the deposition site mainly include the meteorological and topographical conditions (Xiao and others, Reference Xiao2013), precipitation type and intermittency (Cuffey and Steig, Reference Cuffey and Steig1998; Helsen and others, Reference Helsen2006; Laepple and others, Reference Laepple2018), redistribution of snow (Ekaykin and others, Reference Ekaykin, Lipenkov, Barkov, Petit and Masson-Delmotte2002; Ekaykin and Lipenkov, Reference Ekaykin and Lipenkov2009; Laepple and others, Reference Laepple2016; Münch and others, Reference Münch, Kipfstuhl, Freitag, Meyer and Laepple2016), cyclones and cyclone paths (Qin and others, Reference Qin, Petit, Jouzel and Stievenard1994), surface and interstitial sublimation and re-condensation processes (Ding and others, Reference Ding2010, Reference Ding2011; Steen-Larsen and others, Reference Steen-Larsen2013; Xiao and others, Reference Xiao2013; Madsen and others, Reference Madsen2019), etc. Calculations of the d-excess based on the routine MWL slope would result in great discrepancies in different sites. By now, few systematic studies on the spatial variations on MWL slopes were executed in Antarctica (Masson-Delmotte and others, Reference Masson-Delmotte2008). In order to accurately interpret the stable water isotope records from deep ice cores, it is critical to evaluate the spatial and temporal distribution of MWL slopes, which needs to be documented from modern measurements and processes studies (Xiao and others, Reference Xiao2013).
In addition, the influences of the fractionation on the initial deposition signals of the water isotopes and the alteration of climatic information were another research topic for inland Antarctic regions (Hoshina and others, Reference Hoshina2014, Reference Hoshina, Fujita, Iizuka and Motoyama2016; Casado and others, Reference Casado2016, Casado, Reference Casado2018; Ritter and others, Reference Ritter2016). During the past decades, there existed some opposing views on what drives that deposition of water isotopes were due to climate variations or post-deposition noise. Some sites seemed to be greatly influenced by the signal (e.g., Dome F, Dome C and Vostok) while other sites were not (e.g. DML and South Pole) (Steen-Larsen and others, Reference Steen-Larsen2014; Münch and others, Reference Münch, Kipfstuhl, Freitag, Meyer and Laepple2016). To make these controversial debates clear, more inland and coastal sites should be included and detailed studies on fractionation effects on temporal isotopic evolution process should be executed.
Here we present a modern process study on spatial and temporal variations of water stable isotopes (δ18O and δD) in the snow along a traverse from coast to interior plateau in eastern Antarctica, and two topics are mainly involved: (1) spatial distribution of the MWL slopes at different sites with varied environmental and geographical conditions along the traverse and (2) the temporal variations of the MWL slopes in different depth of the snow pits and ice core and the possible causes for their spatial and temporal variations.
Samples and data
Samples and analysis
Dome A is located at the highest point of Eastern Antarctica ice sheet (4093 m) and 1248 km from the nearest coast (Fig. 1) (Hou and others, Reference Hou, Li, Xiao and Ren2007; Reference Hou, Wang and Pang2013; Xiao and others, Reference Xiao2008; Reference Xiao2013). According to in-situ measurement at the Dome A station (Hou and others, Reference Hou, Li, Xiao and Ren2007; Xiao and others, Reference Xiao2008; Ma and others, Reference Ma, Bian, Xiao, Allison and Zhou2010), the mean annual temperature at Dome A is −58.5°C, the accumulation rate is 0.023 m a−1 w.e. and the mean wind velocity is 2.2 m s−1. These values are the lowest among the stations in Antarctica. During the 29th Chinese National Antarctica Research Expedition (CHINARE) inland expedition from coastal Zhongshan Station (ZSS, 69°37′31″S, 76°37′22″E) to inland Dome A (80°25′01″S, 77°06′58″E) in the austral summer in 2012/13, 13 snow pits and 115 surface samples were approximately evenly captured along the transect (Fig. 1 and Table 1) (Li and others, Reference Li2016). The depth of the snow pits varied from 2 to 3 m and the period covered changed from ～4 years at the coastal region to ～36 years at the interior regions. For the inland two snow pits (29-M and 29-L, Table 1), the mean sampling resolution was 3 cm, amounted to 0.3–0.4 year per sample depending on the density. The sampling depth for the surface samples along the traverse was 3–5 cm and the time covered varied from ～1 year in the interior regions to only the summer precipitation episodes at the coast. During the 32nd CHINARE inland expedition in 2015/16, another coastal snow pit (32-A) with larger depth (3 m) and higher sampling resolution (～4 cm per sample) was dug. During the same expedition, an ice core (CA2016-75) with the length 33.24 m were drilled (75 km from the coast) in a dry hole with a mechanical drill. The length, weight and diameter of each piece of ice core were measured just after its extraction.
Surface sampling sites along the CHINARE inland traverse between Zhongshan Station and Dome A. Red dots show the snow pit locations, the blue dotting line shows the surface sample sampling sites and the green column shows the location of the coastal ice core (CA2016-75).
Statistics of the slopes of δD and δ18O in snow samples along the ZSS-Dome A traversea
a For the detailed data and information, please refer to Fig. S1 and Table S1 to S16 in the supplementary materials.
All the samples were transported frozen to Lanzhou and all the analyzing procedures were executed at the State Key Laboratory of the Cryospheric Science (SKLCS) in Lanzhou China under 100–1000 class clean environment. Cations were analyzed on a Dionex ISC 3000 ion chromatograph (Dionex ISC 3000, Thermo Scientific, USA) using an Ion Pac CS12A column, 20 m mol L−1 MSA eluent and a cation electrolytically regenerated suppressor (CERS). The anions were analyzed on a Dionex ISC3000 ion chromatograph (Dionex ISC3000, Thermo Scientific, USA) using an Ion Pac AS11-HC column, 25 m mol L−1 KOH eluent, and an anion electrolytically regenerated suppressor (ASRS). The detection limit, defined as 3 times the standard deviation of the baseline noise, is ～1 ng g−1 for all major ions. The stable isotope (D and 18O) compositions of all samples were analyzed by a liquid-water isotope analyzer (DLT 100, Los Gatos, USA) based on off-axis integrated cavity output spectroscopy (OA-ICOS) at the SKLCS. The isotopic ratios are expressed in per mil (‰) units relative to Vienna Standard Mean Ocean Water (V-SMOW) (Xiao and others, Reference Xiao2013; Li and others, Reference Li2016). The accuracies of δD and δ18O measurements are ±0.6 and ±0.2‰, respectively.
The linear second-order parameters on water isotopes of d-excess and the logarithmic d-excess (d-ln) are both adopted in this study to show their spatial and temporal variations and the related influencing factors and the calculation of d-ln is based on the equation: d-ln = ln(1 + δD) + 0.0285 (ln(1 + δ18O))2–8.47 ln(1 + δ18O) (Uemura and others, Reference Uemura2012; Markle and others, Reference Markle2017).
Ancillary data
The coastal ice core (CA2016-75) was dated by the annual layer counting from the seasonal variation profiles of marine ions (Na+, SO42−) and black carbon (BC). The separated dating results on Na+ and BC are well consistent with the uncertainty of ±2 years for 102 years (1915–2016 A.D.). The calculated annual mean accumulation rates on the ice core is ～0.180 m a−1 w.e., met well with the previous results (Ding and others, Reference Ding2011). The two inland snow pits (29-M and 29-L) were dated according to the volcanic deposit signals (Li and others, Reference Li2014) and the accumulation rate data. Because of the relative calm wind condition, the influence of the drifting snow was negligible at Dome A. The calculated accumulation rates for the whole duration (37 years for 29-M and 36 years for 29-L) are consistent with the measurements by the automatic weather station (AWS) (Ma and others, Reference Ma, Bian, Xiao, Allison and Zhou2010).
Results and Discussions
Spatial distribution of slopes along the transect
δD and δ18O both show decreasing trends while d-excess increases with the distance from the coast along the ZSS to Dome A traverse (Xiao and others, Reference Xiao2013; Pang and others, Reference Pang2015, Reference Pang2019; Li and others, Reference Li2016). The mean slope of δD with the surface temperature was 6.4 ± 0.2‰ per °C, like the average for East Antarctica (Masson-Delmotte and others, Reference Masson-Delmotte2008; Xiao and others, Reference Xiao2013). The mean MWL slope in surface snow samples in this study (7.78 ± 0.04, R 2 = 0.99) is little larger compared with the values measured before (7.5 ± 0.1, R 2 = 0.99) along the same traverse (Xiao and others, Reference Xiao2013). We speculate the different sampling depth is responsible for the discrepancy. Our sampling depth (3–5 cm) is less compared with Xiao's study (5 cm) (Xiao and others, Reference Xiao2013).
According to the spatial variations of the accumulation rates and the impurities in snow (Ma and others, Reference Ma, Bian, Xiao, Allison and Zhou2010; Ding and others, Reference Ding2011; Xiao and others, Reference Xiao2013; Li and others, Reference Li2014, Reference Li2016; Shi and others, Reference Shi2019; Ma and others, Reference Ma2020), the ZSS-Dome A traverse can be divided the into three different sections, the coastal section (0–400 km), the intermediate section (400–900 km) and the interior section (900–1248 km). The mean slopes between δD and δ18O in the coastal, intermediate and interior sections are 8.18 ± 0.15 (R 2 = 0.99, n = 30), 8.03 ± 0.09 (R 2 = 0.99, n = 50) and 7.79 ± 0.10 (R 2 = 0.99, n = 35), respectively. The general decreasing trend of the MWL slopes in three sections may be caused by the increasing distance from the source regions and intensified fractionation effects (Masson-Delmotte and others, Reference Masson-Delmotte2008; Xiao and others, Reference Xiao2013; Pang and others, Reference Pang2015, Reference Pang2019).
To get a detailed spatial distribution pattern of the slopes along the traverse, the MWL slopes in the 13 snow pits were calculated (Fig. 2). In general, the slopes have a decreasing trend with the increasing distance inland. The snow pits located near the coast always show relatively higher slopes (Table 1) and the highest slope occurs in snow pit 29-C (8.61 ± 0.26, R 2 = 0.98, n = 20) and this value was significantly larger than the global mean MWL (8.0) (Dansgaard and others, Reference Dansgaard1964) and also larger than the mean slope of the Antarctic snow (7.75 ± 0.02, R 2 = 0.998) (Masson-Delmotte and others, Reference Masson-Delmotte2008) (Fig. 2). Coastal regions of Antarctic ice sheet usually receive moisture from adjacent offshore water while the interior regions get sources from the distant open sea (Reijmer and others, Reference Reijmer, Van den Broeke and Scheele2002; Masson-Delmotte and others, Reference Masson-Delmotte2008; Sodemann and Stohl, Reference Sodemann and Stohl2009). The higher MWL slopes at the coastal section may indicate less fractionation during the evaporation and the transportation processes (Hou and others, Reference Hou, Wang and Pang2013; Xiao and others, Reference Xiao2013). In addition, the high accumulation at coastal section can cause a quick bury of the deposited snow, which is of benefit to stop (or reduce) the surface sublimation and condensation processes (Ding and others, Reference Ding2010). The slopes in the intermediate section (400–900 km) generally show a ‘V’ spatial pattern and the slopes decrease at the initial part potentially implies a decrease of water vapor sources from the adjacent coastal ocean. However, after ～700 km, the slope shows an increasing trend till the ～900 km inland. We speculate that the intensified wind speed, special surface topography and the post-depositional effects may have influences to some extent (Xiao and others, Reference Xiao2013; Münch and others, Reference Münch, Kipfstuhl, Freitag, Meyer and Laepple2016). The most interior snow pit at Dome A (29-M) has the lowest slope (7.24 ± 0.11, R 2 = 0.98, n = 100), much lower than the global mean MWL and the Antarctic snow samples. Another interior snow pit 29-L also shows a low slope value (7.84 ± 0.10, R 2 = 0.98, n = 100). The distance between the two sites is 148 km and they share similar environmental conditions (Ren and others, Reference Ren and Qin1995, Reference Ren2010).
Spatial distribution of the MWL (meteoric water line) between δD and δ18O in 13 snow pits (29-M to 29-A, black square) and surface snow (red square) along the traverse from Zhongshan Station to Dome A together with the d-excess (bottom) and d-ln (middle). Different sections of the traverse are divided by the vertical dashed lines and the mean values of the parameters in different sections are shown in squares. The horizontal dotted lines show the global mean meteoritic water line (8.0, Dansgaard, Reference Dansgaard1964) and the mean slope in Antarctic snow (7.75, Masson-Delmotte and others, Reference Masson-Delmotte2008).
In the previous studies, most interior snow samples usually show the lowest water isotope ratios (Masson-Delmotte and others, Reference Masson-Delmotte2008; Xiao and others, Reference Xiao2013; Li and others, Reference Li2016). The water isotope ratios at the snow pit 29-M vary between −394.80 and −485.96‰ with the mean −450.93 ± 16.25‰ for δD and between −63.25 and −50.67 ‰ with the mean −58.46 ± 2.22‰ for δ18O. These values are the lowest in Antarctica up to date. Two potential reasons can be put forward to be mainly accountable for the lowest slope at the Dome A region. The first is the moisture source. The moisture arriving in the interior region is mainly from the low to mid-latitudes open oceans and has experienced long-distance transportation in the upper troposphere or lower stratosphere, and thus more intensive fractionation effects may occur during the long pathway (Xiao and others, Reference Xiao2013). The second is that at Dome A, low temperature together with the special precipitation form and accumulation have an intensifying effect on the surface evaporation and re-condensation of the water vapor at this highest location of Eastern Antarctica (Fujita and others, Reference Fujita and Abe2006; Ding and others, Reference Ding2015). Clear sky precipitation (diamond dust) is particular in interior regions on the Antarctic ice sheet and can occupy a large fraction of the total accumulation (Hou and others, Reference Hou, Li, Xiao and Ren2007). During the days with weak wind and higher temperature at the snow surface compared to the air, the vapor pressure of the air mass close to the snow surface can reach or even exceed saturation, resulting in frost flower growth at the snow surface (Hou and others, Reference Hou, Li, Xiao and Ren2007; Xiao and others, Reference Xiao2013). According to Hou and others (Reference Hou, Li, Xiao and Ren2007), the frost flower can comprise ～7% of the whole accumulation at Dome A. The diamond dust and frost flowers both present a large specific surface area (300–590 cm2 g−1) and hence are favorable to intensify the surface sublimation and re-condensation processes of water vapor. During the sublimation of ice grains and re-condensation of water vapor onto ice grains under extremely low temperatures, the isotopic fractionations will occur in the porous snow layers (Town and others, Reference Town, Warren, Walden and Waddington2008; Wang and others, Reference Wang2012; Hoshina and others, Reference Hoshina, Fujita, Iizuka and Motoyama2016; Casado and others, Reference Casado2016, Casado, Reference Casado2018; Madsen and others, Reference Madsen2019).
Comparison between the slopes and the mean d-excess values in the 13 snow pits shows that they have a generally opposite spatial variability (R = −0.71, P < 0.01, N = 13), the slopes decrease with the distance inland while the d-excess increase simultaneously (Fig. 2). They also show some different variation patterns at the coastal and the intermediate sections. The d-excess value in Antarctic snow is mainly controlled by the relative humidity, sea surface temperature and wind velocities at the moisture source regions (Hou and others, Reference Hou, Wang and Pang2013; Bonne and others, Reference Bonne2019) and also influenced by the condensation conditions at the deposition site (Markle and others, Reference Markle2017). The slope between δD and δ18O is mainly controlled by the climatic and meteorological conditions of snowfall (air temperature, relative humidity, wind velocity, etc.) (Masson-Delmotte and others, Reference Masson-Delmotte2008). Therefore, the difference between the slope and the d-excess may indicate that the local climatic and meteorological conditions had significant influences on the fractionation processes of the water isotopes. One point should be pointed out that the relation between the slopes and the source region conditions is mainly based on distillation modeling and future observations are in great demand to testify that (Masson-Delmotte and others, Reference Masson-Delmotte2008).
The lognormal statistical on deuterium excess (d-ln) can effectively reduce the influences from the non-source factors and is regarded as an optimal index to study the climatic variations at the moisture source regions (Uemura and others, Reference Uemura2012; Markle and others, Reference Markle2017). We made some comparisons among the slopes, d-excess and the d-ln values along the traverse (Fig. 2). The d-ln clearly indicates a consistently increasing trend inland, implying a constant shift of different source regions and this is consistent with previous results (Masson-Delmotte and others, Reference Masson-Delmotte2008; Xiao and others, Reference Xiao2013). The difference between the d-ln and the slopes may be mainly caused by the influences from the local environment, especially on the intermediate section.
Temporal variations of water isotopes
The mean water isotopes (δD and δ18O) values in the 13 snow pits are systematically lower than the surface snow in the same areas, implying depleted isotope signals exist in the snow pits (Xiao and others, Reference Xiao2013; Li and others, Reference Li2016). We further investigate the temporal variations of the fractionation between δD and δ18O in the snow pits and ice core. The temporal variations of the slopes in the two inland snow pits (29-M and 29-L) are firstly investigated. The two pits are vertically divided into three parts (1 m resolution for 29-M and 0.83 m for 29-L). It shows a distinct increase (K = 0.26, P = 0.15) of the slope with depth in the 29-M (6.99 ± 0.30, 7.14 ± 0.14 and 7.51 ± 0.11 for the top, middle and bottom sections, respectively), but 29-L shows an opposite variation pattern (K = −0.29, P = 0.23) (8.01 ± 0.18, 7.91 ± 0.15 and 7.44 ± 0.18 for the top, middle and bottom sections, respectively). This difference between the two snow pits should reflect the different fractionation processes at the two sites. During expeditions inland by the CHINARE, surface frost flower has been only seen at the 1150–1248 km on the traverse and no frost flower layer had been detected at 29-L. The fractionations of the water isotopes will sustain in the snow layer during the snow metamorphism and cause an increase of the isotopic composition in the upper layer. Isotope exchange and diffusion within the porous matrix of the snowpack affect snow isotopic composition (Langway, Reference Langway1970; Johnsen, Reference Johnsen1977; Ebner and others, Reference Ebner, Steen-Larsen, Stenni, Schneebeli and Steinfeld2017; Casado and others, Reference Casado2018) and this procedure will be sustained until the firn transformation into ice. The different fractionation procedures between HDO and H218O in the snowpack may be the causes for the increase of the slopes at the upper layers at 29-L (Masson-Delmotte and others, Reference Masson-Delmotte2003). The inverse variations of the slopes at 29-M should be caused by the intensified fractionation effects on the frost flower with much lower slopes. The mixing effects within the surface snow will decrease the mean slope in the upper snow layers and the mixing effect decrease with depth can cause the slope increase. Because no surface frost flower is developed, no specific fractionation mechanism exists at the surface of 29-L, so the slope vertical variation shows an opposite pattern to 29-M，i.e. the slope decreases with depth at 29-L.
The vertical temperature gradients in different layers of the snow pits may be another important factor for the fractionation. The temperature gradient can cause the convection of the water vapor in the firn layers. Especially during the autumn and winter seasons, when the surface is significantly cooled down while the deeper layer still kept a relatively higher temperature, the water vapor sublimates at the lower depth can move upward and re-condenses at the shallower layers accompanied by the kinetic fractionation (Masson-Delmotte and others, Reference Masson-Delmotte2008). The formation of depth hoar in the snow layers also has a great influence on the fractionation of water isotopes, especially at the sites in the inland with low accumulations (Satake and Kawada, Reference Satake and Kawada1997; Neumann and Waddington, Reference Neumann and Waddington2004; Neumann and others, Reference Neumann, Waddington, Steig and Grootes2005). The widespread depth hoar in East Antarctic snow layers during the winter is a good proof for the interstitial sublimation and re-condensation processes (Qin and others, Reference Qin and Ren1992, Reference Qin, Ren, Wang, Petit and Jouzel1993; Ekaykin and others, Reference Ekaykin and Lipenkov2009). However, no depth hoar layers were observed in our snow pits since all the snow pits were dug in the summer season.
Previous studies on the variations of water isotopes with depth in the inland Antarctic snow pits (e.g. Vostok, Dome F, Dome C) showed a mean 20 cm cycle, and these intervals were ascribed to accumulation adjusting, wind erosion and post-depositional redistribution, which can significantly alter the initial seasonal depositional signals and climatic information (Hoshina and others, Reference Hoshina2014, Reference Hoshina, Fujita, Iizuka and Motoyama2016; Casado and others, Reference Casado2016; Laepple and others, Reference Laepple2016; Casado, Reference Casado2018). We check the deposition signals of δ18O in the snow pits along the traverse and find a similar variation pattern (Fig. 3). The two interior snow pits display the cycle periodicity varied between 16 and 51 cm with the mean value 32.38 ± 11.36 cm at 29-M and between 10 and 34 cm with the mean value of 23.89 ± 8.95 cm at 29-L. Moreover, both the snow pits show enlarged cycle periodicity with the depth accompanied by the decreasing variation amplitudes, implying that the smoothing effects of the isotopes is intensified with depth. The smoothing effects on the water stable isotopes with the depth can also be seen at the coastal and intermediate sections (Fig. 3). 29-G shows an extreme smoothing trend in the bottom part, suggesting that besides the interstitial sublimation and re-condensation effects, the intensified wind erosion and mixing effects on the snow have great influences on the variations of the isotopes at this site. Sodium (Na+) is usually regarded as a good indicator of the seasonal variations of the marine incursion, but its seasonal signals seem to be smoothed at the inland snow layers with similar depth intervals to the water isotopes (Fig. 3). Opposite variation patterns between the Na+ and the water isotopes always exist generally in our inland snow pits and this phenomenon seems widespread among the eastern Antarctica (Hoshina and others, Reference Hoshina, Fujita, Iizuka and Motoyama2016). However, at the coastal snow pits, such as 32-A, Na+ has clear seasonal variations but the water isotopes do not.
Temporal variations of the δ18O (black) and Na+ (red) in different snow pits along the traverse between ZZS and Dome A.
Because of the smoothing influences on the water isotopes in the inland snow pits, we select the ice core (CA2016-75) at the coastal region to study the annual variations of the water isotopes and the potential influencing factors (Fig. 4). The CA2016-75 ice core drilling site has a relatively high accumulation rate (～0.18 m a−1 w.e.) and thickness of annual layers vary between 30 and 40 cm, so the annual mean depth usually has seven–ten samples and is sufficient to study the seasonal variations. Based on the dating results, the annual mean water isotopes and the second-order proxies (d-excess and d-ln) are calculated (Fig. 4). The seasonal variations of the δ18O and δD can be detected for the top 10 m depth (since ～1990s). The smoothing effect is significantly below 10 m depth, eliminating the annual and seasonal signals. From this phenomenon, we can find the post-depositional sublimation and re-condensation of the water vapor is also significant in the coastal region, even for the sites with the high accumulation rates.
Temporal variations of the δD (red), δ18O (blue), d-excess (pink), d-ln (green) in the CA2016-75 ice core. The vertical dotted lines show the dating results of the ice core and the thick solid lines show the 10-point smoothing results (red for δD, blue for δ18O, pink for d-excess and green for d-ln, respectively).
The MWL slope for the whole ice core (8.54 ± 0.11) is similar with the coastal 29-A and 32-A snow pits and a decrease of the slopes with depth is also detected (8.56 ± 0.08 for the upper 10 m and 8.37 ± 0.09 for the lower 23.24 m, Table 1). The temporal variability of water isotopes at CA2016-75 ice core is different from the Law Dome ice core (DE08-2) which shows significant seasonal cycles till 150 m depth (Masson-Delmotte and others, Reference Masson-Delmotte2003). Higher accumulation for DE08-2 ice core site (between 1.00 and 1.50 m a−1 w.e.) and lower wind speed may be the dominant factors for the good preservation of the seasonal cycles of water isotopes in the ice core (Masson-Delmotte and others, Reference Masson-Delmotte2003).
Variations of water isotope ratios (δD, δ18O, Fig. 4) show a general flat pattern between 1910s and mid-1980s, after that significant increases during 1980s–90s (K = 2.89, P < 0.0001, N = 92 for δD and K = 0.36, P < 0.0001, N = 92 for δ18O, respectively) and the ratios stay at a relatively higher level till the sampling time. The d-excess and d-ln show similar temporal variations for the whole duration (mid-1910s–2015A.D.) (Fig. 4). They both show relatively flat variations between 1910s and 1990s, and significant decreasing trends are followed till the mid-2000s and then a quick increasing trend till the sampling time. The reasons for the variations and variability of the isotopes and the second-order parameters (d-excess and d-ln) are beyond the scope of this paper and more detailed discussions will be included in the following studies.
Conclusions
Spatial and temporal distributions of fractionation slopes between δD and δ18O in the surface snow, snow pits and ice core samples along the traverse from ZSS to Dome A were calculated. Three spatial sections were divided along the traverse, the low slope existed at the interior section and high values at the coastal region, the intermediate section showed large variations implying that complex mechanisms were involved in the fractionation processes and more efforts are needed in the future. The lowest slope existed at Dome A and was much lower than the mean MWL, it was speculated that the long-range transportation of moisture to this interior plateau and the particular precipitation and accumulation styles at Dome A are the dominant influencing factors. Inland snow pits showed significant smoothing effects on the water isotopes as ever been found in other inland sites (Vostok, Dome C, Dome F, etc.) in eastern Antarctica and the original deposition signals of the water isotopes has been significantly altered. Moreover, not only the inland snow pits, the coastal snow pits in our study were also found for the post-depositional smoothing effects. The persistent sublimation, migration, and re-condensation of the water vapor in the snow (firn) layers caused by the deviations of the temperature in different depths and the wind sweeping effects may be the main influencing factors. The slopes of the isotopes in different depth changed significantly. Variations of the slopes with depth at Dome A snow pit (29-M) showed opposite distribution pattern to 29-L and the existence of the frost flower or not may be mainly accountable for the discrepancy. Coastal snow pits and ice cores showed different smoothing effects on the snow impurities which presented significant seasonal variations and multi-year oscillation. More efforts on other water isotopes tracers (e.g. δ17O and tritium) in the near future are in demand to study the sources of the moisture along the traverse between ZSS and Dome A and their fractionation procedures under different environmental conditions and also for their relationships with the climatic variations.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/jog.2021.5
Acknowledgement
We thank all the members who participated in the 2012/13 and 2016/17 CHINARE field campaigns for sample collection. Special thanks should also be given to analyzing personnel on the water isotopes in the State Key Laboratory of the Cryospheric Sciences. Dr Xiaoqing Cui, Ms. Xiaoxiang Wang, Yan Liu and Tingting Su and Yuman Zhu from the cold and arid region environmental and engineering research institute, Chinese Academy of Sciences who completed parts of the analysis. This work is financially supported by the Innovative Research Group, the National Natural Science Foundation of China (No. 41671063, 41830647, 41625012, and 41961144028), the State Key Laboratory of Cryospheric Sciences (supporting fund No. SKLCS-ZZ-2018-01) and the Youth Innovation Promotion Association, Chinese Academy of Sciences.
Open access
