UC Santa Barbara

Accurate and precise age models are essential to interpret the climate signals preserved in ocean sediment cores, yet age models are often constructed with only a single dating technique. Radiocarbon dating directly dates discrete sediment layers but suffers from low resolution and is restricted to the last 55 ka BP, while benthic δ18O stratigraphic alignment is vulnerable to temporal offsets between δ18O signals, or lags. Here I present a new Bayesian software package to construct ocean sediment core age models by statistically combining direct age information, such as radiocarbon data, with probabilistic benthic δ18O stratigraphic alignment. The software package also offers a novel stack construction algorithm capable of building regional and continuous benthic δ18O stacks. I also develop a novel method to calculate time-dependent benthic δ18O lags complete with statistical uncertainty estimates. Specifically, lags are calculated by subtracting radiocarbon age models from benthic δ18O age models and are relative to the target stack. The uncertainty from each age model is incorporated into the lag calculation via the subtraction of Markov Chain Monte Carlo Samples. I apply this method to calculate lags for 33 Atlantic sediment cores, the results of which indicate three distinct regions based on the timing of benthic δ18O change. I find statistically significant leads in the intermediate West Atlantic, statistically significant lags in the deep South West and Abyssal North West Atlantic, and I find that cores in the East Atlantic share the timing of δ18O change with the target stack. In addition, I calculate the regional lag between the Iberian Margin and Eastern Equatorial Pacific and I find that it is 1.5 kyr smaller than the original 3.9 kyr estimate. While lags will help future investigators during age model construction, I also interpret them to reflect asynchronous surface signals, transit times from the surface to the deep ocean, and water mass geometry changes during Termination