USING VARIED DATA TYPES TO ASSIGN REALISTIC AGE UNCERTAINTIES TO STRATIGRAPHIC SEQUENCES AND PROXY RECORDS

Accurate age-depth models for proxy records are crucial to inferring changes to the environment through space and time. Yet, traditional methods of constructing these models assume unrealistically small age uncertainties. Existing Bayesian models simultaneously account for absolute ages, relative ages (superposition) and variations in sedimentation rate. However, existing Bayesian age-depth models do not account for many geologic complexities including magmatic crystal populations, detrital zircon ages, magnetostratigraphy, and chemostratigraphy, each with their own distinct probability distributions. We modified a commonly used Bayesian age-depth model, BChron, to allow the specification of the wide variety of these probability distributions to produce more realistic age-depth uncertainties.

For example, geochronology on single mineral crystals, nearly always reveals dispersion arising from a variety of physical mechanisms (inheritance, crystal residence times, daughter isotope loss), which complicates the interpretation of eruptive age. Allowing the model to explore the full variance present in magmatic crystal populations, rather than a subset of ages, permits outliers that conflict with superposition to be rejected. Other examples include minimum depositional ages (one-sided constraints), magnetic reversals and chemostratigraphic information. These types of data help condition dated horizons and create a more realistic age depth model.

In general, model age uncertainties increase in areas that lack closely spaced dated horizons, leading to considerable age uncertainties for discrete proxy measurements that are not co-located with dated horizons. We also demonstrate a simple Monte Carlo method to propagate age model uncertainties onto proxy measurements to build a continuous record to allow robust estimates of environmental change.