2000 years of annual ice core data from Law Dome, East Antarctica

. Ice core records from Law Dome in East Antarctica collected over the last four decades, provide high resolution data for studies of the climate of Antarctica, Australia and the Southern and Indo-Paciﬁc Oceans. Here we present a set of annually dated records of trace chemistry, stable water isotopes and snow accumulation from Law Dome covering over the period from -11 to 2017 CE (1961 to -66 BP 1950), as well as the level 1 chemistry data from which the annual chemistry records are derived. Law Dome ice core records have been used extensively in studies of the past climate of the Southern Hemisphereand 5 (cid:58) , (cid:58)(cid:58) as (cid:58)(cid:58)(cid:58)(cid:58) well (cid:58)(cid:58) as (cid:58) in large scale data syntheses and reconstructions in a region where few records exist, especially at high temporal resolution. This dataset provides an update and extensions both forward and back in time of previously published subsets of the data, bringing them together into a coherent set with improved dating to enable continued use of this record. The data are available for download from the Australian Antarctic Data Centre at https://doi.org/10.26179/5zm0-v192

Further visits to this site for collecting ice cores is expected as part of an ongoing monitoring project, with subsequent updates to this record when ice cores become available for analysis. These regular visits to update the record by collecting shallow cores extends the period which overlaps with modern satellite observations, ensuring the record stays up to date and more data is available for calibration of palaeoclimate reconstructions.
This publication provides a reference point for the full suite of data, with links to versioned and updated data publicly available for download from the Australian Antarctic Data Centre at https://doi.org/10.26179/5zm0-v192 :::::::::::::::: .
Note that in the text of this manuscript we use years CE for dates as the more natural dating scale, and to be consistent with the 70 naming of some cores using the year or season in which they were drilled. The datasets available for download provide years in both CE and BP 1950 scales and we use BP 1950 (i.e. years before 1950 CE) for plotting of data in figures here.
The drill sites of the four cores included in this composite record are all located within approximately 1 km of each other, with their relative locations shown in Fig. 1(b). Future visits to the site are planned to ensure the record continues to be updated. A summary of the details of the drilling campaigns is found in Table 1, with more detailed description of the original drilling site :::::::: published by Morgan et al. (1997).  Dating and age horizon uncertainties 2.2 ::::: Dating :::: and ::: age ::::::: horizon :::::::::::: uncertainties The DSS record has been dated using seasonal variations in the deposition of species to define calendar year boundaries, commonly known as annual layer counting. The stable water isotope record is the primary seasonal indicator, with a well-defined maximum of January 10±2.2 days (van Ommen and Morgan, 1997). Confirmation of dating is provided by summer peaking 85 ::::::::::::: summer-peaking : hydrogen peroxide where available, and other chemical markers, such as summer peaking methanesulfonic acid (MSA), non-sea salt sulphate (nssSO 2− 4 )and : , sulphate/chloride (SO 2− 4 /Cl − ) ratio, and the winter peaking :::::::::::: winter-peaking sea salt species (chloride (Cl − ), sodium (Na + ), magnesium (Mg 2 +)). Layer counting of the DSS record between 1300 to 1995 CE was previously completed by Palmer et al. (2001b) and verified and extended to cover over 2000 years by Plummer et al. (2012). The DSS record currently spans -11 to 2017 CE with the addition of the DSS1617 core which is dated unambiguously 90 from 1989 to 2017 CE. The timescale from Plummer et al. (2012) has been updated by refining the annual layer placement, assisted by new water isotope analysis, and by applying a stricter error counting protocol. The number of years counted has not changed, however the error estimate has increased in the oldest part of the record to +20/-7 at -11 CE, where this date may be up to 20 years older or 7 years younger (-31 to -4 CE) as shown in Fig. 2. Where evidence for a year horizon was not clearly identifiable in the primary seasonal indicators, : or only weakly supported by confirmatory species, it was not counted.

95
When the majority of (but not all) seasonal indicators show evidence of a year these were counted. Both cases contribute to the total uncertainty estimate. The asymmetrical uncertainty is biased toward under-counting, reflecting increasing likelihood of encountering years with unclear signals with the decreasing layer thickness at depth. DSS is a long record from a high accumulation site (0.69 my −1 ice equivalent using an ice density of 917 kgm −3 ) and maintains 6-8 chemistry samples per year at -11 CE; enabling development of a long, accurate annual layer-counted chronology. Intra-annual dating of individual 100 sample depths is determined by interpolation between the layer counted annual horizon depths.
We take a conservative approach to calculating the age horizons and uncertainties. The depth of the top of each 1 m length of drilled core is locked to the top of the bag as recorded in the drilling logs, so uncertainty from the sample resolution is limited only within each 1 m length and errors in depth that may arise ::::: arising : from missing or badly fragmented core segments are not cumulative. The sample resolution of chemical measurements is generally much greater than the water isotope sample 105 resolution, hence the larger ::::::::: uncertainty : value is used.
The DSS chronology presented here has been dated using annual layer counting only; without reference to externally dated reference events (e.g. volcanic events) and is independent of other ice core age scales. There are differences when compared to other ice core chronologies, however those differences are within our error bounds. To assist users to include DSS data in large-scale reconstructions or regional synthesis products, we have provided a translation of the age of the volcanic events used 110 in the PAGES Ant2k compilation (Ahmed et al., 2013) to the WD2014 (Sigl et al., 2016)   The depths of the year boundaries for the 2000 years of data, produced using annual layer counting methods as described above, are contained in the file titled "DSS_2k_age_horizons.csv". The column headings for this file are described in  Table 2. Column headings for the file "DSS_2k_age_horizons.csv" which contains the depths of year boundaries produced by annual layer counting methods. Year horizons by depth are provided along with the specific core used for that year in the compilation. Accumulated minimum and maximum errors in age are provided, calculated using the method described in Section 2.2.

115
The level 1 datasets have undergone calibration, quality control and post-processing of the raw instrument measurements.
The stable water isotope (δ 18 O) record is sampled at 10-50 mm resolution, with finer physical sampling as annual layers thin with depth. Level 2 annual mean data is provided ::: (see :::: 4.1), however the level 1 datasets of stable water isotopes are not provided at this time but will be published in the future with a manuscript currently in preparation. To assist with dating, sections of cores were measured for hydrogen peroxide at an average resolution of 50 mm as described by van Ommen and Transparent colours indicate where data overlapping the other cores exists, but is not included in the compilation where due to better quality core or data being available.
up to 60 samples per year in the near-surface firn. Due to a sampling error, a section of DSSMain from 251-273 m (1611-1568 CE) was sampled for chemistry at 100 mm resolution, however isotope and peroxide measurements are available at 50 mm resolution to maintain dating integrity. Beyond 402 m, sample resolution changed to 30 mm and beyond ::::: below 578 m (979 140 CE), resolution changed to 25 mm to offset the effects of layer thinning on seasonality.

Trace Ion Chemistry
Ice core samples have been analysed using ion chromatography for chloride (Cl − ), sodium (Na + ), magnesium (Mg 2+ ), cal- and methanesulfonic acid (MSA − ). Analytical methods have been modified and updated over the course of the analysis, and are summarised as follows (see also   (1992) were discussed by Palmer et al. (2001b), and showed no significant difference. A comparison of the technique used by Plummer et al. (2012) and Curran and Palmer (2001) Table 3. Resolution and methods used for trace chemistry analysis on the 4 ::: four : different cores used in the compilation. †A sampling error resulted in the larger sample size in this section.
A number of species that were measured on parts of the core are not included in this dataset. MSA was not measured on a significant portion of the record. Additionally, deeper sections of the MSA record suffer from unquantified losses from storage prior to analysis (Roberts et al., 2009). The potassium and calcium records are incomplete and suffer issues with poor detection 155 in some older analyses. Additionally, some sections of the DSSMain calcium record appear to have been affected by dust from storage in a concrete floored freezer. Due to these quality concerns, the potassium and calcium records are not discussed further.
For the remaining datasets we provide , for each chemical species, : : the concentration for each sample used in the composite record, an ID corresponding to the drilled core it was obtained from, the top and bottom depths of each sample and the 160 corresponding age. Summary statistics for the level 1 trace chemistry species are included in Table 4, with the corresponding histograms shown for each species shown in Fig. 4. These show that the the distributions of these species are generally non-normal. In further analysis (such as shown later in this paper in 4.3) : the concentrations are log-transformed to partially compensate for the long-tail of the distributions.
The level 1 trace chemistry dataset provides a depth and age for the top and bottom of each sample as well as concentrations 165 of the measured ions. These are contained in the file titled "DSS_2k_chemistry_level1.csv", with column headings provided for reference here in Table 5.     Table 5. Column headings Level 1 trace ion chemistry data file.

Density
The firn densification in this region of Law Dome was established from density measurements taken from several cores in the 175 vicinity of DSS. Samples were prepared from dry drilled cores or from melt free interior of thermally drilled cores and were either machined to cylinders on a lathe or pressed (cookie cutter) from softer firn to a precise volume.
An empirical fit to the firn density profile with depth was presented in van Ommen et al. (1999). Specifically the density (ρ, in kg m −3 ) is fitted as a compound function of depth (z, in m) as (1)

180
The density observations and empirical fit as a function of depth (see Eqn. 1) are plotted in Fig. 5, showing good agreement between the two.

Ice Equivalent Depths
For some derived data products (e.g. snow accumulation history) it is useful to be able to remove the effects of firn densification.
One common method is to utilise "ice equivalent depths" that represent the depth of a column numerically compressed to the 185 same density as glacial ice (density ρ ice = 917 kg m −3 ) and same cross-sectional area, with the same mass as the firn column.
Algebraically, the ice equivalent depth (Z) is given by The level 2 datasets are derived from the level 1 datasets above. Those presented here have largely been previously published or 190 included in large data compilations but are now updated, with improved dating of the core. Where there are differences between the versions, these are remarked upon here. These level 2 derived datasets can be considered as examples of best practise for utilising the level 1 data. A plot of the timeseries for all level 2 datasets is shown in Fig. 6.
Two CSV format text files are provided for the Level 2 data sets, "DSS_2k_winter_centred.csv" and "DSS_2k_summer_centred.csv" which contain the annually averaged datasets as described here in Section 4. Column headings are provided for reference here 195 in Tables 6 and 7. These datasets are provided together to ensure that a consistent age scale across all the derived datasets.

Annual stable water isotopes
The DSS annual stable water isotope presented here updates and extends the stable isotope record included in Emile-Geay et al. (2017) and Stenni et al. (2017) using the DSS1617, DSS99, DSS97 and DSSMain cores (see Fig. 3), and currently spans the period -11 to 2016 CE. The previously published annual composite record for DSS, used in Antarctic temperature 200 reconstructions (Ahmed et al., 2013;Emile-Geay et al., 2017;Stenni et al., 2017), included other short cores which are now no longer used as the time periods they covered are now superseded by the longer DSS1617. Also, further water isotope analysis has extended the annual stable water isotope for the DSSMain record to cover -11 to 174 CE. The record presented here includes some changes from the previous records due to the new cores, minor changes in the year boundaries and the new analysis data.

Annual trace ions 210
Annual average :::: trace ::: ion :::::::::::: concentration values are presented here, and the calculation method is species dependent. A wintercentred average is used for winter-peaking species (e.g. sea salts) with the boundaries set at the beginning and end of each calendar year. The summer-centred value is calculated from mid-year to mid-year, and used for the nominally summer-peaking species nitrate and non-sea-salt sulphate. This was done to reduce edge effects where small differences in year boundary placement could have strong effects on summer-peaking species. The non-sea-salt sulphate concentration is calculated according to 215 the method of Palmer et al. (2001a) and has a fractionation correction applied to minimize non-physical negative values. The distributions of the annually averaged records is shown in Fig. 8.

Seasonal sea salts 220
Characteristics of seasonal cycles of the different species of ions from the Law Dome ice core has been studied over short time periods (Curran et al., 1998), while annual sea salt data has been used previously for high-precision dating of volcanic events (Palmer et al., 2001b), and used for palaeoclimate studies of atmospheric circulation (Souney et al., 2002).
Three seasonal sea salt records were developed from aggregating the level 1 sodium record for climate proxy and reconstruction purposes in Vance et al. (2013Vance et al. ( , 2015 and further used Crockart et al. (2021); Vance et al. (2021). Specifically these 225 three records include a prescribed summer season of December to March inclusive mean chloride concentrations, hereafter the 'summer sea salt record', as well as a semi-annual aggregation into a defined warm (December to June inclusive) and cool (July to November inclusive) mean salt concentration. The summer sea salt record was developed as a proxy for rainfall variability in eastern Australia as well as a Western Tropical Pacific ENSO proxy over the Common Era, and is provided as log-transformed   Ocean (Udy et al., 2021). The warm and cool season mean sea salt concentrations were similarly log-transformed and provided input timeseries for reconstructing Pacific Ocean decadal variability, specifically the Interdecadal Pacific Oscillation index (Parker et al., 2007;Vance et al., 2015Vance et al., , 2016. These datasets have been extended and undergone improved quality control procedures, as detailed in Vance et al. (2021) and adopted here, which extended the IPO reconstruction to span the Common Era.  These seasonal records were derived by binning the quality controlled, dated sea salt data from the level 1 datasets into 12 months per year. The salt concentrations were log-transformed to normalize (as sea salt data from Law Dome displays long tails due to high frequency aerosol generation events from synoptic activity in the southern Indian Ocean). After binning, timeseries of seasonal averages (December-March, December-May and June-November) were produced with the following caveats. The level 1 sea salt records were examined in conjunction with the stable water isotope records and field and laboratory logbooks 240 detailing core morphology and ice sample cutting. In some instances, the binned sea salt data was compromised due to either  missing ice core material from core breaks or shattered sections, or analytical problems resulting in no data or suspected contamination. If this had occurred, the data in question was removed from the analysis. Where more than one month of data was compromised for the summer sea salt record, or two months for the warm and cool season records, we elected to not include that season in our published timeseries. This results in missing values in the seasonal sea salt timeseries of summer,

Annual snow accumulation
Snow accumulation is derived using the year horizons obtained through the layer-counted dating. This record has been previously published in Roberts et al. (2015) and included in PAGES 2k snow accumulation data compilation .

250
This update includes the newer DSS1617 core data and the improved dating. We assume steady state for depth profiles of both density (Sorge's law) and the vertical strain rate. To compensate for firn densification effects, the depths for year boundaries have been converted to ice equivalent depths using Equation 2. Annual snow accumulation rates were then estimated using the power-law vertical strain rate method of Roberts et al. (2015), based on the assumption of no long-term trend in snow accumulation rate. The resulting snow accumulation rate is shown in Fig. 18, with the long term mean snow accumulation rate 255 of 0.691±0.004 IE m y −1 .

Summary
In this paper we have presented the suite of data covering the last 2000 years from the Law Dome ice cores. This set has been updated from previously published results with improved and extended annual dating. We provide quality controlled trace ion chemistry datasets as well as derived, annualised products.

260
This dataset can be used for high resolution studies of southern hemisphere climate drivers at annual to seasonal scales Our aim for this paper has been to provide the community with the most complete and consistent suite of Law Dome for the past 2000 years, and examples of best-practise methods for generating derived datasets such as the annual means. The data sets here supersede all previous releases. This dataset is useful for future high resolution studies of palaeoclimate proxies as has been useful in the past. Future users of the data should take the following summary of the points into consideration 1. The dating presented has uncertainties due to occasionally unclear seasonal markers. The dating is biased towards undercounting of years with unclear boundaries.
2. As this dataset is dated through annual layer counting, independent of reference to volcanic or solar activity ties, there is an offset in the chronology compared to other ice core chronologies. A translation in ages of volcanic ties is provided to facilitate the inclusion of this data where a common age scale is required. 3. While we provide all the level 1 chemistry data, it is more useful to consider annualised values for the data, either summer or winter centered as we have described depending on the particular analyte. Future users of the data remain free to define seasonal boundaries as best suits their particular needs. We encourage all potential users of the data to contact us if they have any further queries about how best to use the data.