An East Antarctic, sub-annual resolution water isotope record from the Mount Brown South Ice core

The Mount Brown South (MBS) Ice CoreThe ice core siteThe MBS ice core site is located near the boundary of Princess Elisabeth and Wilhelm II Land in East Antarctica, 380 km inland and to the west of Australia’s Davis Station. It is located in the Indian Ocean sector of coastal East Antarctica, a region poorly represented by high-resolution ice core records, limited to several short cores (up to 350 years) in Princess Elizabeth Land17 and the Law Dome ice core18 (Fig. 1). The Law Dome record spans up to 90,000 years and is available at annual or sub-annual resolution for the past 2000 years. It has been widely used to reconstruct regional climate variability of coastal East Antarctica19 as well as hydroclimate variability in both southwest Western Australia5,20,21 and eastern Australia22,23 due to synoptic-scale weather and climate links between coastal East Antarctica and Australia24.Fig. 1Map of Antarctica and location of prominent ice core drilling sites.The MBS ice core site was selected following an extensive site-selection study12 to fill a critical gap in the current network of ice core records. Seven criteria were considered when selecting the ice core site including (i) at least 1000-year-old ice at depths of 300 m; (ii) summer surface temperatures <0°C to minimise melt; (iii) high-resolution (sub-annual) requiring accumulation of >250 mm a−1 (iv) minimal surface reworking; (v) located on a ridgeline or dome; (vi) teleconnection with mid-latitude climate; (vii) complementary to existing records. The final site, Mount Brown South, meets all of these requirements and expands the current network of high-resolution temperature and hydroclimate proxies. An annually resolved chronology for the MBS ice core, MBS2023, was developed in Vance et al.13 using annual layer counting and alignment to known volcanic horizons.Drilling and samplingDuring the 2017/18 austral summer field season, a 295 m long ice core with a diameter of 98 mm was drilled from 4 m below the surface using a Hans Tausen drill25. Wet drilling started at a depth of 93 m (driller’s depth) using Estisol 140 as drilling liquid. The core was cut in 1 m sections (ice core bags) in the field and transported to the ice core facilities of the Institute for Marine and Antarctic studies (IMAS) in Hobart, Australia for further processing. Visual features in the ice such as breaks, wind crusts and bubble free layers were thoroughly logged prior to further sectioning. Following the cutting scheme shown in Fig. 2, samples were prepared at a temperature of −18°C using a trace-clean band saw. Two rectangular rods measuring 34 × 34 mm were cut for the purpose of CFA and for trace ion chemistry analysis. The two wedges on each side of the square rods were collected for tephra and discrete water stable isotope analysis. The latter was sampled at a resolution of 30 mm.Fig. 2Cross section of the MBS ice core showing the cutting scheme of the ice core samples.The CFA square rods were subsequently transported to Copenhagen, Denmark for further sample preparation and subsequent chemical and water isotope analysis. Typically, 1-2 mm from the ends of the CFA rods were carefully shaved off in order to minimise contamination of the chemical impurity analysis. For rods with slightly slanted ends the removal of material was adjusted and documented accordingly. For the same reason, we removed internal breaks and then shaved off and flattened the ice on each side of the break. The lengths of all the ice pieces were carefully documented, prior as well as after the processing of breaks and edges of the CFA rods.Stable Isotope Analysis – Cavity Ring Down SpectroscopyTwo versions of the Picarro water isotope analyzer (Picarro L-2130i and L-2140i) were used for the analysis of the discrete and continuous samples. The Picarro L-21xxi water isotopes analyzers are Cavity Ring Down Spectrometers (CRDS), operating in the near-IR region at 7200 cm−1 and both operate on the same spectroscopic principles. The analyzer uses a continuous emission diode laser source coupled to a high-finesse, 3-mirror, V-shaped cavity. Analysis takes place in the gas phase at a cavity pressure of 67 mbar (50 Torr) and a temperature of 80°C. Precise control of the cavity’s pressure and temperature (≈2mbar, 1mK, 1-σ) is key for optimal instrument performance. This is achieved by the instrument’s control software continuously adjusting the position of two proportional valves up- and downstream of the optical cavity maintaining stable pressure and by driving heat elements installed in a well insulated and ventilated box that contains the optical cavity. Temperature and pressure stability is also facilitated by maintaining a stable laboratory temperature and a stable gas flow at the inlet of the instrument. The sample flow through the optical cavity is \(30\,{\rm{cc}}\,{\min }^{-1}\).Discrete samples water isotope analysis; 0-104 m depthThe upper 100 m of the ice core were analysed at the Australian Antarctic Division (AAD) Ice Core laboratory facilities at IMAS, Hobart (4.0–74.2 m) and the Australian National University (ANU), Canberra (74.2–100.0 m), for interlaboratory comparison purposes using a discrete sampling scheme. The discrete ice core samples were stored frozen at IMAS, Hobart after processing. For the section analysed at the ANU, Canberra, samples were transported frozen and remained frozen until analysis. The ice samples were melted at room temperature, whereafter 200 μL were transferred into 2 mL screw thread glass vials (Fisherbrand 033918) fitted with 300 μl polyspring inserts (Thermo Scientific C4010-S630) and sealed with silicone/PTFE septum caps (Thermo Scientific THC09-15-1178). The ice samples at IMAS were defrosted in a fridge at 4°C overnight. Aliquots of 250 μL were transferred into 2 mL snap top glass vials (Agilent 5182-0546) fitted with 250 μL pulled point glass vial inserts (Agilent 5183-2085) and sealed with PTFE/red silicone snap caps (Agilent 5182-0564).Instrumentation used for discrete samplesStable water isotope analysis on the discrete samples was performed using two variants of the Picarro L21xx-i analyzer with similar performance characteristics with respect to precision and sample handling. At IMAS, Hobart, a Picarro L2130-i stable water isotope analyzer was used, while the L2140-i variant was used at ANU, Canberra. At both IMAS, Hobart, and ANU, Canberra, the Picarro analyzers were coupled to the Picarro A0211 vaporizer and the Picarro A0325 autosampler. Pressurized dry air (\(\left[{{\rm{H}}}_{2}{\rm{O}}\right] < 2\,{\rm{ppmv}}\)) was used as carrier gas and mixed with the injected sample in the vaporizer’s sample cell at a temperature of 110°C, regulated by means of PID control. The water sample was injected using a 10 μL syringe ( SGE 10R-BT-GT-0.47C) with an injection volume of 1.8 μL liquid water at IMAS, Hobart and 1.5 μL at the ANU, Canberra. After an equilibration time of 60 s in the vaporizer’s cell, the vapor-air mixture was gradually introduced into the optical cavity for spectroscopic analysis. The instruments comprising each experimental setup as well as the specifications of the autosampler injection methods are given in Tables 1 and 2 respectively.Table 2 Autosampler injection method for the Picarro L2130-i (analyses performed at IMAS, Hobart) and the Picarro L2140-i (analyses performed at ANU, Canberra).Run configuration for discrete analysesAnalysis of discrete samples in Hobart and Canberra followed similar protocols with small deviations. At both facilities, an analytical sequence consisted of calibration sequences bracketing “unknowns”, with isotopic values expressed as per mil (‰) and relative to the Vienna Standard Mean Oceanic Water (VSMOW). A typical analytical sequence using the AAD Picarro L2130-i consisted of 30 “unknowns”, while a sequence at the ANU, Canberra consisted of 20 “unknowns”.In Hobart, three in-house calibration standards were used (tap water, Dome Summit South (DSS) and Lambert Glacial Basin (LGB)), as well as a laboratory reference water sample or check standard, MBS, treated as an unknown (Table 3). The check standard (MBS) was analysed after 10 unknown samples to monitor run stability. Analysis was performed in “High-Precision” mode with methods that enabled analysis of a single injection in ~8 mins. A typical analytical run with calibration standards and 30 “unknowns” was completed in just over 24 hours. The analysis of each standard (including the check standard) consisted of 8 injections, with the first 4 injections discarded in order to account for memory effects. The “unknowns” were analysed sequentially with increasing depth to reduce the memory effects. Each “unknown” measurement consisted of 4 injections with the first injection discarded.Table 3 ANU local standards used for the discrete analyses.In Canberra, we used two in-house calibration standards (DSS and LGB), as well as a check standard, MBS, treated as an unknown. The check standard (MBS) was analysed between unknown samples #10 and #11 to monitor run stability, as well as at the mid-point of each calibration sequence. The analysis of each standard (including the check standard) consisted of 20 injections, with the first 13 injections discarded in order to account for memory effects. The “unknowns” were analysed sequentially with increasing depth to reduce the memory effects associated with large isotopic steps. Each “unknown” measurement consisted of 10 injections with the first four discarded. Analysis at the ANU, Canberra was performed in the Picarro “High-Precision” mode, which utilises injection times of ~9 min (total time ~1.5 hours per sample). A run consisted of several analytical sequences and thus lasted several days.Continuous Flow Analysis (CFA)The more recent development of high-resolution Continuous Flow Analysis (CFA) techniques based on the continuous melting of ice core samples26,27,28,29 has yielded unprecedented records of chemical impurities30,31,32,33. We continuously melt an ice core rod over a custom designed and manufactured heated melt plate29 and further debubble and distribute the liquid sample stream to an array of analyzers. This approach has the potential to yield multi-component, very high resolution data sets on a common depth scale with a massively improved throughput compared to the laborious relatively low-resolution analysis performed in the traditional way using ion chromatography or inductively coupled plasma mass spectrometry.Laser spectroscopy techniques employing Cavity Enhanced/Cavity Ring Down Spectroscopy (CRDS) configurations have made it possible to measure trace gas concentrations and stable water isotope ratios in a continuous flow mode with minimal sample preparation34,35,36. Interfacing these techniques to existing CFA systems allowed for continuous, high-resolution records of CH437,38 as well as water stable isotope records (δ18O, δD)39,40,41,42 in ice cores with significant improvements in precision, accuracy, sample consumption and throughput.The Copenhagen CFA systemThe Copenhagen CFA system is optimized for high depth resolution capable of analysing an array of chemical parameters such as conductivity, insoluble dust, Ca2+, \({\,\mathrm{NH}}_{4}^{+}\), Na+ and pH29,32,43. Following a modular approach, more components can be incorporated in the synchronous analysis. The system utilizes an aluminum melter with an outer sample square ring measuring 34 × 34 mm and an inner sample square ring measuring of 26 × 26 mm (Fig. 2 in29). The melt rate is controlled by the melter’s temperature maintained by four 125 W Mickenhagen heat cartridges in combination with a J-type thermocouple and a PID–loop control unit (dTRON 308, Jumo). A draw wire sensor incorporating an optical encoder (WayCon SX-80) records the melt speed with a sensitivity γ = 25 counts mm−1.During the campaign the nominal melting speed of \(3.5\,{\rm{cm}}\,{\min }^{-1}\) yielded a clean, central line sample flow of \(24\,{\rm{ml}}\,{\min }^{-1}\) maintained by a peristaltic pump (Ismatec IPC-16). This sample contained ≈10% air due to the air bubbles occluded in the ice core. Removal of the air occurred in a homemade Polyether-Ether-Ketone (PEEK) flat triangular-cell debubbler following a simple buoyancy principle (Fig. 3 in29). Downstream of the debubbler, the air-free flow equalled \(4\,{\rm{ml}}\,{\min }^{-1}\) of which \(0.4\,{\rm{ml}}\,{\min }^{-1}\) was pumped further for water isotope analysis. The liquid water-air mixture with a flow of \(20\,{\rm{ml}}\,{\min }^{-1}\) was forwarded to a gas extraction membrane (3M, Liqui-Cel MM-0.75 × 1 Series) for further gas-liquid separation and subsequent CH4 analysis. A more thorough presentation of the chemical analysis of the MBS core accompanied by the adjoined data records will follow in a separate publication. The dataset can be accessed at the Australian Antarctic Data Center44.Continuous flash vaporisation for water isotope analysisIn Fig. 3 we present the water isotope analysis segment of the CFA system. We follow a slightly modified version of the CFA-water isotope system presented in14,39 in which a microflow of liquid water is continuously injected and flash-vaporised in a PID-regulated 170°C oven (T2 in Fig. 3) to which we introduce a stream of dry air (<30 ppmV) with a flow of \(\approx 50\,{\rm{cc}}\,{\min }^{-1}\). The liquid microflow is in the order of 2 \({{\rm{\mu lmin}}}^{-1}\) and is achieved by sampling a higher flow of 0.4 \({\rm{ml}}\,{\min }^{-1}\) with a fused silica capillary (ID 50 μm) in a tee-split (T1 in Fig. 3) configuration. The high flow of 0.4 \({\rm{ml}}\,{\min }^{-1}\) is maintained by the peristaltic pump P1 (Ismatec ISM597D) that pulls the sample through V1, a 6-port selection valve (Vici C25-3186EMH). The latter is used to choose between the CFA sample line or glass containers with water isotope standards or a glass container with background deionised water. We used amber glass 250 ml bottles (Duran GL45, 250 ml, S/N: 818063604) with GL-45 thread safety caps (JR-S-11011) and inlet solvent stainless steel filters (IDEX A-302) with a porosity of 10 μm. Upstream of V1, a parallel configuration of two inline filters (IDEX A-313) fitted with stainless steel 20 μm porosity frits (IDEX A-120) ensures that the CFA sample stream is free of particles that can deteriorate the smooth operation of the selection valve or the oven capillary. Downstream of the vaporisation oven T2, the vapour sample is introduced into an open split (T3 in Fig. 3) consisting of a stainless steel 1/16 inch tee-split (Swagelok SS-100-3), through which the analyzer pulls the necessary flow of \(30\,{{\rm{cc\; min}}}^{-1}\).Fig. 3Block diagram of the water isotope analysis part of the CFA system. For a complete description of the system’s components, the reader is referred to Section Methods.Our flash-evaporation approach has been shown to yield a vapor sample flow that is not affected by isotopic fractionation issues, thus allowing for a very stable performance with the possibility to also average over longer times in order to achieve very high signal-to-noise ratios. The practically zero-dead volume flow system minimises sample dispersion such that ultra high resolution measurements are possible39,45. For optimal results we operate the spectrometer in the [H2O] range of 10,000-23,000 ppmV. Tuning of the [H2O] level can be achieved with a needle valve installed on the dry air line, which in combination with the open split T3, disturbs neither the sample flow nor the pressure in the optical cavity.The MBS CFA campaignThe analysis of the MBS core took place in Copenhagen, at the CFA lab of the Ice and Climate group at the Niels Bohr Institute over 13 working days during October 2019. A total of 195.386 m of ice was processed through the CFA system equivalent to 201.634 m of ice core length when including the pieces of ice removed during the processing of the breaks. Depthwise, the record spans the range 93.082-294.712 m corresponding to the age interval 1736-873 CE13. In Fig. 4 we show the processing throughput during the campaign.Fig. 4Analysis production during the MBS CFA campaign spanning 13 working days (09/10/2019 – 25/10/2019).CFA isotope processing sequenceFrom the water isotope point of view, a typical CFA processing day consists of the Idle, Calibration and Run modes (Fig. 5). In the Idle, mode the selection valve V1 connects to the DI water container feeding the system overnight. These sections were used in order to evaluate the performance of the system with respect to stability, accuracy and long term averaging, by means of Allan variance tests14,46 (section Technical Validation).Fig. 5Example of the sequence of events during an analysis day of the MBS CFA campaign. We plot the raw δD signal (left axis) and the water concentration levels in the laser spectrometers optical cavity during a period of 24 hours on the 22nd of October. The pale red, blue and yellow areas of the plot mark the IDLE, VSMOW-SLAP calibration and ice core run phases of the day. The pale green areas correspond to the time when the isotope system was on IDLE mode while the chemical analysis part of the CFA system was being calibrated.The processing day starts with a VSMOW-SLAP calibration (section “Cal” colored light blue in Fig. 5) a procedure that involves three local water isotope standards, with the middle one being used as a “check” standard for estimation of the systems accuracy. Each standard water is run for 20 min resulting in a calibration procedure that lasts 60 min in total. The whole sequence can be initiated and monitored remotely via a homemade Python-based valve control and data acquisition script.When on “Run” mode the system is measuring ice core sample and the selection valve V1 is connected to the “CFA” sample port. As seen in Fig. 5, a processing day contains two sample runs each consisting of 9 ice core bags equivalent to ≈9 m of ice. In the particular example of Fig. 5 from the 22nd of October, the first run (Run #224) consists of the ice core bags 224-232 and the second run (Run #233) consists of ice core bags 233-241. A necessary calibration procedure for the chemical analysis results in a period of Idle mode between the two runs that can last 1-2 h (green areas in Fig. 5).Data analysis – Discrete samplingVSMOW-SLAP calibration – Discrete analysisThree local standards calibrated against the IAEA reference waters SLAP and VSMOW were used to calibrate the raw δ18O and δD signals to the VSMOW-SLAP scale. All of the in-house standards consist of bulk Antarctic meltwater samples collected from three locations in Antarctica: Dome Summit South (DSS), Lambert Glacial Basin (LGB) and Mount Brown South (MBS). (Table 3). We follow the IAEA recommended procedure11 and use the most enriched (DSS) and depleted (LGB) local standards to build a two-point fixed calibration line as: $$\delta {\prime} =\alpha \,\delta +\beta $$
(1)
where δ’ represents the calibrated signal and the slope α and the intercept β are given by: $$\alpha =\frac{\delta {{\prime} }_{{\rm{DSS}}}-\delta {{\prime} }_{{\rm{LGB}}}}{{\delta }_{{\rm{DSS}}}-{\delta }_{{\rm{LGB}}}}$$
(2)
$$\beta =\delta {{\prime} }_{{\rm{DSS}}}-\frac{\delta {{\prime} }_{{\rm{DSS}}}-\delta {{\prime} }_{{\rm{LGB}}}}{{\delta }_{{\rm{DSS}}}-{\delta }_{{\rm{LGB}}}}\,\,{\delta }_{{\rm{DSS}}}$$
(3)
The additional “MBS” standard was analysed several times during each run sequence in order to assess the internal accuracy and precision of each analytical run after calibration to the VSMOW-SLAP scale as described above.Data analysis – CFAVSMOW-SLAP calibration – CFA analysisReporting of the δ18O and δD signals on the VSMOW-SLAP scale requires a calibration step similar to the one taken for the discrete data and described in section. Following an approach similar to14,39 we used three local water isotopic standards calibrated against the primary IAEA reference waters VSMOW and SLAP (Table 4) in order to (i) produce a two fixed-point calibration line defined by the two extreme local standards “–40” and “–22” and (ii) estimate the system’s accuracy by comparing the value of the middle “NEEM” standard post calibration with its nominal value (Table 4). Using the prime notation for the calibrated δ signals, raw δ measurements are calibrated on the VSMOW-SLAP scale according to the calibration line in Eq. (4)$$\delta {\prime} =\alpha \,\delta +\beta $$
(4)
where the slope α and the intercept β are given by: $$\alpha =\frac{\delta {{\prime} }_{{\rm{-22}}}-\delta {{\prime} }_{{\rm{-40}}}}{{\delta }_{{\rm{-22}}}-{\delta }_{{\rm{-40}}}}$$
(5)
$$\beta =\delta {{\prime} }_{{\rm{-22}}}-\frac{\delta {{\prime} }_{{\rm{-22}}}-\delta {{\prime} }_{{\rm{-40}}}}{{\delta }_{{\rm{-22}}}-{\delta }_{{\rm{-40}}}}\,\,{\delta }_{{\rm{-22}}}$$
(6)
From Eqs. (4)–(6) we calculate the VSMOW calibrated value of the middle “check”standard, which is then compared to the nominal VSMOW value of the “check” standard water (Table 4).Table 4 Copenhagen local standards used for the MBS record VSMOW-SLAP calibrations.Depth assignmentAssigning the depth on each CFA run requires the alignment of the wire depth encoder signal with the δ18O and δD signals. The WayCon SX-80 wire encoder delivers a digital signal with a resolution γ = 25 countsmm−1. The melted length as a function of time can then be calculated by the integration of the differential counts signal \(c{\prime} (t)\) as $$l(t)={\int }_{{t}_{i}}^{{t}_{f}}\,{\gamma }^{-1}c{\prime} (t)\,dt$$
(7)
where ti and tf denote the start and end of the CFA run. Following the detailed logging of the CFA sample rods with respect to internal breaks’ positions and lengths as well as the top depths of every ice core bag it is possible to reconstruct the depth vector z(t) of each CFA run as a function of elapsed measurement- and Unix Epoch time. In order to transfer the depth scale on the water isotope data, we first locate the beginning and the end of each CFA run in the δD signal. Due to the isotopically heavier value of the background DI water relative to the isotopically light Antarctic sample, CFA run segments are easily distinguishable. We search for the minimum and maximum values of a smoothed version (10-point Bartlett window) of the derivative of the δD signal corresponding to the beginning and end of CFA run respectively. Finally, linear interpolation of the z(t) signal on the time vector of the isotope signal \(\delta {\rm{D}}(t{\prime} )\) sets the isotope measurements on the depth scale of the ice core. In Fig. 6 we show an example of locating the beginning and end of CFA run 224 based on the δD signal and its derivative. We also show the result of the melted length l(t) calculation based on Eq. (7) without (solid line) and with (dashed-dotted line) the core breaks taken into consideration.Fig. 6δD of Run 224. The run contains the ice core bags 224-232 and spans ≈9 m. The smoothed derivative of the δD signal (blue curve) is used for the identification of the beginning and the end of the run. With solid black, we plot the run length signal as measured with the wire optical encoder and with the dashed line we show the length signal corrected for the logged internal breaks.Data analysis: The age scaleAn age scale has been developed for the MBS ice core record, hereafter referred to as MBS2023, resulting in a chronology spanning 1137 a (873–2009 CE)47. The construction of the chronology utilises a layer-counting approach, in combination with aligning volcanic horizons by comparing the non-sea sulfate signal (\({{\rm{nssSO}}}_{4}^{2-}\)) in the MBS, Law Dome48, Roosevelt Island Ice Core49 as well as the West Antarctic Ice Sheet (WAIS) Divide chronology, WD201431. Annual layers are identified and counted visually in the sea salt and non-sea sulfate signals, the former yielding maxima in austral winter and the latter in austral summer. This type of variability results additionaly in signal peaks for the sulfate to chloride ratio during the austral summer50. Austral summer water isotope maxima (δ18O and δD) are also used in the counting of the annuals. For a detailed description of the dating methods the reader is referred to the dedicated paper by Vance et al.13.