Description: Marine sedimentary archives are routinely used to reconstruct past environmental changes. In many cases, bioturbation and sedimentary mixing affect the proxy time-series and the age-depth relationship. While idealized models of bioturbation exist, they usually assume homogeneous mixing, thus that a single sample is representative for the sediment layer it is sampled from. However, it is largely unknown to which extent this assumption holds for sediments used for paleoclimate reconstructions. To shed light on 1) the age-depth relationship and its full uncertainty, 2) the magnitude of mixing processes affecting the downcore proxy variations, and 3) the representativity of the discrete sample for the sediment layer, we designed and performed a case study on South China Sea sediment material which was collected using a box corer and which covers the last glacial cycle. Using the radiocarbon content of foraminiferal tests as a tracer of time, we characterize the spatial age-heterogeneity of sediments in a three-dimensional setup. In total, 118 radiocarbon measurements were performed on defined small- and large-volume bulk samples ( ∼ 200 specimens each) to investigate the horizontal heterogeneity of the sediment. Additionally, replicated measurements on small numbers of specimens (10 × 5 specimens) were performed to assess the heterogeneity within a sample volume. Visual assessment of X-ray images and a quantitative assessment of the mixing strength show typical mixing from bioturbation corresponding to around 10 cm mixing depth. Notably, our 3D radiocarbon distribution reveals that the horizontal heterogeneity (up to 1,250 years), contributing to the age uncertainty, is several times larger than the typically assumed radiocarbon based age-model error (single errors up to 250 years). Furthermore, the assumption of a perfectly bioturbated layer with no mixing underneath is not met. Our analysis further demonstrates that the age-heterogeneity might be a function of sample size; smaller samples might contain single features from the incomplete mixing and are thus less representative than larger samples. We provide suggestions for future studies, optimal sampling strategies for quantitative paleoclimate reconstructions and realistic uncertainty in age models, as well as discuss possible implications for the interpretation of paleoclimate records.

Description: Radiocarbon ages measured on different sample types from the marine sediment core OR1-1218-C2-BC which was retrieved from the southern South China Sea using a box corer. Onboard the ship, the core was separated into several sub-cores of which nine sub-core were analysed. Bulk samples contain 1/3 of 200 individual foraminifera (~800 µg) picked from the nine sub-cores at discrete depths (0 - 2, 6 - 8, 10 - 12, 16 - 18, 22 - 24, 28 - 30 and 32 - 34 cm.). Additional small number samples are based on 5 specimens with ten replications for two depths (6 - 8 and 36 - 37 cm) from sub-core 1. The data are used to estimate the sediment accumulation rate and the vertical extent of the sediment mixing. Additionally, the data are used to assess the intensity of the sediment mixing and the three-dimensional age-heterogeneity within the core.

Description: Dataset containing surface snow measurements of snow specific surface area (SSA), snow density and snow accumulation. Surface samples were taken from the surface 2.5cm of snow. SSA measurements were determined using an Ice Cube measuring device (Zuanon, 2013). Snow density was measured from the SSA samples with a fixed volume. Snow accumulation describes the change in surface height at each sample site. All measured parameters have 10 daily samples taken at 10m intervals over a 90m transect. Sampling was carried out daily between May and August of 2016-2019, at approximately 24hr time intervals. All measurements were taken at the East Greenland Ice Core Project site (EastGRIP) situated in the accumulation zone of the Greenland Ice Sheet.

Description: Throughout the summer seasons 2017 to 2019 snow profiles were taken repeatedly. The sample positions are aligned along a 300 m wind-parallel transect at the EastGRIP ice core deep drilling site. The snow was collected at 6 positions with a 20 – 50 m spacing. Snow was sampled using carbon fiber tubes of 1-m length that were pushed gently into the snow. A maximum compaction of 1 cm was observed during extraction. The snow cores were carefully removed from the carbon fiber tubes on the cutting table. The core was then cut into slices of 1 cm thickness for the upper 10 cm and 2 cm thickness for the lower 90 cm. The samples were placed into Whirl-Pak® bags and closed airtight. The samples were shipped frozen to the Alfred-Wegener-Institut and stored at -25 °C. Prior to measurements, the samples were melted in the sample bags at room temperature. For the measurement of the isotopic composition, the instrument Picarro L2130-i were used. The measurement set-up followed the Van-Geldern protocol (Van Geldern and Barth, 2012). Each sample was injected 4 times unless otherwise noted in the comment column. As a measure of accuracy, we calculated the combined standard uncertainty (Magnusson, et al., 2017) including the long-term reproducibility and bias of our laboratory by measuring a quality check standard in each measurement run and including the uncertainty of the certified standards. The combined uncertainty for δ18O is 0.11 ‰ and for δ2H is 0.8 ‰. Deuterium excess is calculated as both 1) d = dD - 8*d18O; (Merlivat and Jouzel, 1979) and 2) dln = ln(dD + 1) - 8.47(ln(d18O+1)) - 28.5(ln(d18O+1))2; (Uemura et al., 2012).

Description: Snow samples were taken on a daily basis along a 100 m wind-parallel transect at the EastGRIP ice core deep drilling site. The snow was collected in the morning at 11 positions with 10 m spacing into four cumulative samples – each for one depth interval. The depth intervals are top 0.5 cm , top 1 cm , top 2 cm and top 5 cm . Each day, undisturbed snow was sampled and the exact sample location marked to avoid sampling disturbed snow during the next sampling event. The samples were shipped frozen to the Alfred-Wegener-Institut and stored at -25°C. Prior to measurements the samples were melted in the sample bags at room temperature. For the measurement of the isotopic composition the instruments Picarro L2120-i and Picarro L2140-i were used. The measurement set-up followed the Van-Geldern Protocol. Each sample was injected four times and the standard deviation is computed. We calculate the average over all the standard deviations as a measure of uncertainty. We find this average to be 0.01 permil for δ18O (with stdev 0.01) and 0.08 permil for δD (stdev 0.1). The maximum standard deviation within the data set was found to be 0.07 for δ18O and 0.9 for δD. As a measure of accuracy the off-set between the defined and measured value of the quality check standard for each measurement run is provided. We calculate the average of this off-set for the whole data set and obtain a value of -0.06 permil for δ18O (stdev 0.02) and -0.69 permil for δD (stdev 0.18). The maximum off-set found within the data set was -0.1 permil for δ18O and -1.12 permil for δD.

Description: Snow samples were taken on a daily basis along a 100 m wind-parallel transect at the EastGRIP ice core deep drilling site. The snow was collected in the morning at 11 positions with 10 m spacing into four cumulative samples – each for one depth interval. The depth intervals are top 0.5 cm , top 1 cm , top 2 cm and top 5 cm . Each day, undisturbed snow was sampled and the exact sample location marked to avoid sampling disturbed snow during the next sampling event. The samples were shipped frozen to the Alfred-Wegener-Institut and stored at -25°C. Prior to measurements the samples were melted in the sample bags at room temperature. For the measurement of the isotopic composition the instruments Picarro L2120-i and Picarro L2140-i were used. The measurement set-up followed the Van-Geldern Protocol. Each sample was injected four times and the standard deviation is computed. We calculate the average over all the standard deviations as a measure of uncertainty. We find this average to be 0.01 permil for δ18O (with stdev 0.01) and 0.08 permil for δD (stdev 0.1). The maximum standard deviation within the data set was found to be 0.07 for δ18O and 0.9 for δD. As a measure of accuracy the off-set between the defined and measured value of the quality check standard for each measurement run is provided. We calculate the average of this off-set for the whole data set and obtain a value of -0.06 permil for δ18O (stdev 0.02) and -0.69 permil for δD (stdev 0.18). The maximum off-set found within the data set was -0.1 permil for δ18O and -1.12 permil for δD.

Description: Snow samples were taken on a daily basis along a 100 m wind-parallel transect at the EastGRIP ice core deep drilling site. The snow was collected in the morning at 11 positions with 10 m spacing into four cumulative samples – each for one depth interval. The depth intervals are top 0.5 cm , top 1 cm , top 2 cm and top 5 cm . Each day, undisturbed snow was sampled and the exact sample location marked to avoid sampling disturbed snow during the next sampling event. The samples were shipped frozen to the Alfred-Wegener-Institut and stored at -25°C. Prior to measurements the samples were melted in the sample bags at room temperature. For the measurement of the isotopic composition the instruments Picarro L2120-i and Picarro L2140-i were used. The measurement set-up followed the Van-Geldern Protocol. Each sample was injected four times and the standard deviation is computed. We calculate the average over all the standard deviations as a measure of uncertainty. We find this average to be 0.01 permil for δ18O (with stdev 0.01) and 0.08 permil for δD (stdev 0.1). The maximum standard deviation within the data set was found to be 0.07 for δ18O and 0.9 for δD. As a measure of accuracy the off-set between the defined and measured value of the quality check standard for each measurement run is provided. We calculate the average of this off-set for the whole data set and obtain a value of -0.06 permil for δ18O (stdev 0.02) and -0.69 permil for δD (stdev 0.18). The maximum off-set found within the data set was -0.1 permil for δ18O and -1.12 permil for δD.