A Bayesian Approach to the Estimation of Radiocarbon Calibration Curves: The IntCal09 Methodology

Abstract

This article presents a new approach to the construction of radiocarbon calibration curves. The Bayesian methodology was developed specifically to facilitate construction of the 2009 updates to the internationally agreed 14C calibration curves known as IntCal09 and Marine09. The curve estimation approach taken uses Markov chain Monte Carlo sampling, specifically a Metropolis-within-Gibbs sampler, which offers improved flexibility and reliability over the approaches used in the past. In particular, the method allows accurate modeling of calibration data with 14C determinations that arise from material deposited over several consecutive calendar years and that exhibit complex uncertainty structures on their calendar date estimates (arising from methods such as wiggle-matching and varve counting).