The Bayesian Approach to Radiocarbon Calibration Curve Estimation: The IntCal13, Marine13, and SHCal13 Methodologies

This article outlines the Bayesian models and methods used to facilitate construction of the 2013 internationally agreed radiocarbon calibration curves known as IntCal13, Marine13, and SHCal13. The models build on those used for the 2004 and 2009 estimates of the curves and, as in 2009, arc implemented using Markov chain Monte Carlo sampling, specifically a Metropolis-within-Gibbs sampler. In addition to the data structures accounted for within the 2004 and 2009 models, the approach outlined here also allows for: the presence of additional uncertainty that the data providers have been unable to quantify; tree-ring data that derive their calendar age from wiggle-matching (in addition to ring counting); varve-counted data that exhibit zero increase in calendar age error between 2 or more consecutive layers; and any data source for which we have dependent calendar age uncertainties.