Modeling the Influence of Past Ocean Circulation Change on the Lags between Atlantic Benthic Oxygen-isotope Records

Abstract:

Timeseries of benthic oxygen isotope (d18O) measurements reflect a combination of changes in (1) ocean surface boundary conditions, (2) the distribution of water masses, and (3) the turnover rate of these water masses. Our goal is to detect past circulation changes of type (2) and (3) above, but detection of these changes through d18O lags requires (1), ocean surface boundary conditions, to be changing at the same time. To understand how to decompose (1), (2), and (3), we use idealized cases with known ocean circulation to model d18O timeseries on the Atlantic seafloor, from which d18O leads and lags are calculated. We find that d18O lags are most informative when the d18O signal-to-noise ratio is high, for example, during glacial terminations. Non-informative d18O lag estimates result from the ill-defined nature of lags when surface boundary conditions are slowly changing. When sampling informative times, the majority of the lag variance in a constant circulation experiment is related to the median age of seawater, or the time for a 50% response to surface boundary conditions to occur. Additional experiments explore the effects of non-uniform surface boundary conditions and changing water mass boundaries. The results of idealized cases are compared to the observed lags inferred from a Bayesian analysis of 35 Atlantic cores during Termination 1 (Rand et al., AGU Fall Meeting, 2020) to assess which ocean circulation changes are detectable.