Bayesian population correlation: A probabilistic approach to inferring and comparing population distributions for detrital zircon ages

This is a Preprint and has not been peer reviewed. The published version of this Preprint is available: https://doi.org/10.1016/j.chemgeo.2019.03.039. This is version 1 of this Preprint.

Downloads

Download Preprint

Supplementary Files
Authors

Alexander Robert Tye , Aaron Wolf , Nathan Niemi 

Abstract

Populations of detrital zircons are shaped by geologic factors such as sediment transport, erosion mechanisms, and the zircon fertility of source areas. Zircon U-Pb age datasets are influenced both by these geologic factors and by the statistical effects of sampling. Such statistical effects introduce significant uncertainty into the inference of parent population age distributions from detrital zircon samples. This uncertainty must be accounted for in order to understand which features of sample age distributions are attributable to earth processes and which are sampling effects. Sampling effects are likely to be significant at a range of common detrital zircon sample sizes (particularly when n < 300).

In order to more accurately account for the uncertainty in estimating parent population age distributions, we introduce a new method to infer probability model ensembles (PMEs) from detrital zircon samples. Each PME represents a set of the potential parent populations that are likely to have produced a given zircon age sample. PMEs form the basis of a new metric of correspondence between two detrital zircon samples, Bayesian Population Correlation (BPC), which is shown in a suite of numerical experiments to be unbiased with respect to sample size. BPC uncertainties can be directly estimated for a specific sample comparison, and BPC results conform to analytical predictions when comparing populations with known proportions of shared ages. We implement all of these features in a set of MATLAB® scripts made freely available as open-source code and as a standalone application. The robust uncertainties, lack of sample size bias, and predictability of BPC are desirable features that differentiate it from existing detrital zircon correspondence metrics. Additionally, analysis of other sample limited datasets with complex probability distributions may also benefit from our approach.

DOI

https://doi.org/10.31223/X51G8K

Subjects

Geochemistry, Probability, Statistical Methodology, Tectonics and Structure

Keywords

Geochronology, Bayesian, Density estimation

Dates

Published: 2021-06-19 08:30

License

CC BY Attribution 4.0 International

Additional Metadata

Conflict of interest statement:
None

Data Availability (Reason not available):
https://github.com/alextye/BPC

Add a Comment

You must log in to post a comment.


Comments

There are no comments or no comments have been made public for this article.