A Bayesian Approach for Reconstructing the Past Ocean Circulation from a Limited Number of Sediment-Core Radiocarbon Measurements

Abstract:

Paleoceanographers are faced with the problem of making inferences about the ventilation of the ocean in the past from localized benthic and planktonic radiocarbon measurements obtained from a small number of sediment cores, which leads to an underdetermined problem. With the goal of moving beyond testing the null hypothesis that the sediment core data are consistent with the modern circulation we seek to reconstruct the most probable paleocirculation based on our knowledge of ocean dynamics and available constraints from sediment-core radiocarbon records. We propose a Bayesian inversion approach in which we use a modern circulation estimate constrained by modern radiocarbon data to define the mean of the prior probability distribution for the unknown paleocirculation. The approach resolves the indeterminacy of the inverse problem by choosing a paleocirculation that is minimally different from the modern circulation while still being consistent with the available sediment-core radiocarbon records. In this talk we will present the general formulation of the method as well as various approximations to reduce the computational challenge.