Constraining Density Anomalies in the Interior of Planetary Bodies from Gravity Data Using Bayesian Inference and Voronoi Cells
Introduction: Global anomalies in the interior of a planet reveal information about its evolution. For example, the location and shape of Large Low Shear Velocity Provinces (LLSVPs) on Earth have been used as constraints for mantle evolution models [1, 2]. However, LLSVPs have first been detected using seismic data, which is not available in sufficient quantity for other planets [3]. Instead, gravity data must be used to search for equivalent features in planetary interiors [4]. However, gravity inversions are non-unique, and the epistemic uncertainty (uncertainty due to model assumptions) is rarely quantified [5]. We describe here a gravity inversion method that addresses these issues and provides a measure of confidence in gravity-derived interior models. Method. The approach advocated here is to conduct a Transdimensional Hierarchical Bayesian Object-Oriented Gravity Inversion (THeBOOGIe). The central idea of this method is to use gravity data of a planetary body as input in a Markov chain Monte Carlo (McMC) algorithm that generates many models of the interior density distribution, each parameterized as a collection of finite-size objects. Confidence in the shape or magnitude of density anomalies depends on how consistent these objects are throughout the different models.