Core MD95-2042 alkenone and GDGT data: This dataset provides the following information for core MD95-2042: depth, age, summed OH-GDGT, iGDGT, and di-unsaturated and tri-unsaturated C37 alkenone concentrations, OH-GDGT-based, iGDGT-based, and alkenone-based paleothermometric indices, GDGT-2/GDGT-3 ratio, and biomarker-based sea surface temperature (SST) and 0‐ to 200‐m sea temperature (subT; gamma function probability distribution for target temperatures with a = 4.5 and b = 15) estimates. Sediment samples were taken every 5 cm from core MD95-2042 and homogenized before lipid extraction. The lipid extracts were splitted into two fractions: one for alkenone analysis by gas chromatography coupled to a flame ionization detector, and the other for GDGT analysis by high-performance liquid chromatography coupled to mass spectrometry. All GDGT analyses were done in duplicate. The 1σ analytical uncertainties from 37 replicate analyses of the core catcher sample from core MD95-2042 are 0.007 (0.4 °C) for RI-OH, 0.008 (0.2 °C) for RI-OH′, 0.003 (0.2 °C) for TEX86, 0.238 for GDGT-2/GDGT-3, and 0.010 (0.26 °C) for UK′37. RI-OH′-SST estimates are from the following global calibration: SST = (RI-OH′ + 0.029)/0.0422 (Fietz et al., 2020). RI-OH-SST estimates are from the following global calibration: SST = (RI-OH − 1.11)/0.018 (Lü et al., 2015). TEX86H-SST estimates are from the following regional paleocalibration: SST = 68.4 × TEX86H + 33.0 (Darfeuil et al., 2016). UK′37-SST estimates are from the following global calibration: SST = 29.876 × UK′37 − 1.334 (Conte et al., 2006). Bayesian calibrations were also used for TEX86-SST and TEX86-subT estimates (BAYSPAR; Tierney & Tingley, 2014, 2015) and for UK′37-SST estimates (BAYSPLINE; Tierney & Tingley, 2018). Alkenone data covering the 160–70 and 70–0 ka BP periods are from Davtian et al. (2021) and Darfeuil et al. (2016), respectively. GDGT data covering the 160–45 ka BP period are from Davtian et al. (2021). The age model of core MD95-2042 for the 160–43 and 43–0 ka BP periods was obtained by tuning to Chinese speleothems (Cheng et al., 2016) and by recalibrating existing 14C ages with the Marine20 calibration curve (Heaton et al., 2020), respectively. MIS, Marine Isotope Stage; GDGT, glycerol dialkyl glycerol tetraether; and N/A, not available.
Greenland atmospheric temperature record: This dataset consists in a composite Greenland atmospheric temperature record, which was built with the following records: the GISP2 atmospheric temperature record by Kobashi et al. (2017) for the 10–0 ka BP period, the NGRIP atmospheric temperature record by Kindler et al. (2014) for the 120–10 ka BP period, and the NEEM atmospheric temperature record by NEEM community members (2013) for the 129–120 ka BP period. The NEEM temperature anomalies obtained by NEEM community members (2013) were shifted by –31 °C to obtain absolute air temperatures. The employed age model is the one of Davtian and Bard (2023) for Greenland and Antarctic ice-core records.
Antarctic δ18Oice and atmospheric temperature stacks: This dataset consists in two stacks of three Antarctic records (EDC, EDML, and WD), one for δ18Oice and the other for atmospheric temperature: both stacks are provided with their stacking uncertainties. To build the Antarctic δ18Oice stack, the Antarctic δ18Oice records were resampled every 10 years before centering to zero means and normalization to unit standard deviations over the 140–0 ka BP period (68–0 ka BP for WD). To optimize the continuity between the portions with and without the WD ice core, the Antarctic δ18Oice records were centered to zero means over the 68–67 ka BP period. The resulting Antarctic δ18Oice records were then averaged and stacking uncertainties were calculated as the pooled standard deviation of the stacked Antarctic δ18Oice records divided by the square root of the number of stacked Antarctic δ18Oice records. The final Antarctic δ18Oice stack, expressed in ‰, has the same standard deviation as the δ18Oice record from EDML over the 140–0 ka BP period, and has a zero mean over the 1–0 ka BP. The Antarctic atmospheric temperature stack was built like the Antarctic δ18Oice stack, except that the Antarctic δ18Oice records were corrected for seawater δ18Oice variations before conversion into atmospheric temperature. The employed age model is the one of Davtian and Bard (2023) for Greenland and Antarctic ice-core records.
|Date made available||30 Jan 2023|