An Algorithm for Thermodynamic Parameter Optimization: Application to the Martian Mantle

Abstract

The compilation of thermodynamic models for geophysical applications is such a tedious and complex process that it is generally impractical for researchers to refit parameters in existing models in light of new constraints. To mitigate this difficulty, we develop a Bayesian algorithm that permits the modification of a thermodynamic model to account for additional observational constraints. This algorithm can be applied to any thermodynamic dataset and can utilize a wide variety of experimental constraints. To demonstrate the applicability of the algorithm it is used to revise the Stixrude and Lithgow-Bertelloni (2011, https://doi.org/10.1111/j.1365-246x.2010.04890.x), whole-mantle terrestrial thermodynamic model, using phase equilibrium constraints provided by Bertka and Fei (1997, https://doi.org/10.1029/96jb03270), for the more iron-rich compositions that are thought to be relevant to the Martian mantle. The revised thermodynamic model provides a more reliable prediction of phase equilibria in the Martian mantle. Seismic properties are calculated in an internally self-consistent manner along hot and cold areotherms to constrain the upper and lower bounds of these properties for different bulk silicate Mars compositional models.