IBIS: an Integrated Bayesian approach for unique Initial thorium corrections and age-depth models in U-Th dating of Speleothems

Proxies recorded in speleothems offer insight into past patterns of rainfall, wildfires, vegetation, and other components of the terrestrial climate system. For these records to be effective they must be accurately and precisely dated. Uranium-thorium (U-Th) dating is commonly applied to create chronological models for speleothems, including stalagmites, from present-day to ~650 ka. Making accurate and precise corrections for initial thorium (i.e., 230Th that is not a result of in-situ decay of uranium within the sample) can be the limiting factor in building robust speleothem age models. Two methods are broadly used to correct for initial thorium: (1) direct estimation using isochron techniques; or (2) assuming a constant, model value of the initial 230Th/232Th ratio. However, isochron approaches are resource-intensive, while corrections made using a single, assumed initial thorium value with arbitrary uncertainty may fail to characterize significant variations in initial thorium, leading to inflated age uncertainties and potentially inaccurate ages, thereby limiting speleothems that can be utilized for climate reconstruction. IBIS is a Bayesian model that uses a probabilistic approach to constrain the initial thorium isotope composition of each sample, thereby facilitating the derivation of accurate U-Th ages and errors and robust age-depth models for speleothems. The Bayesian model proceeds in two stages - first estimating U-Th ages and unique initial thorium compositions under a stratigraphic order constraint (Part 1), then fitting those ages with an age-depth model (Part 2). To test the validity of the approach, we show the efficacy of IBIS when applied to case studies from the published literature that exhibit rapidly changing growth rates, uranium loss, growth hiatuses, and very young (