Quantifying Benthic δ18O Lags Across Termination 1: A Probabilistic Approach Based on Radiocarbon and Benthic δ18O Chronologies

Abstract

Abstract Temporal offsets (“lags”) between benthic δ18O (δ18Ob) signals in different locations are not only a source of age model uncertainty during δ18Ob stratigraphic alignment but also provide an opportunity to improve reconstructions of deep ocean circulation change during Termination 1 (T1). While methods based on the visual identification of identical δ18Ob features have previously been used to estimate lags, here we present a novel method to calculate δ18Ob lags and their statistical uncertainties by subtracting a core's radiocarbon age model from an age model based on its δ18Ob alignment to a target stack. This method produces lag estimates as a function of time with statistical uncertainties. As an example of this approach, we calculate lags for a depth transect of 12 cores in the Brazil Margin, whose lags were described qualitatively by Lund et al. (2015, https://doi.org/10.1002/2014PA002657). Our new methodology supports a previously described lag between mid‐depth (1,802–2,296 m) and deep (2,500–2,951 m) water; specifically, we find a statistically significant lag across this boundary throughout T1, with a maximum of 2.08 kyr (95% credible interval: 1.36–2.85) at 14 ka BP. We also identify statistically significant lags between intermediate (1,105–1,627 m) and mid‐depth cores that were not previously described. Furthermore, we demonstrate that the Brazil Margin δ18Ob lags during T1 are inconsistent with a tracer transport model under modern circulation pathways (Gebbie, 2012, https://doi.org/10.1029/2011PA002273) that fits an Atlantic‐Pacific δ18Ob lag.