Predicting complex mineral compositions in Gale crater, Mars based on CheMin instrument X-ray diffraction and machine learning

Session: Advancing Mineral Science and Exploring Planetary Surfaces: In Honor of MSA Dana Medalist, Elizabeth B. Rampe (Posters)

Presenting Author:

Shaunna Morrison

Authors:

Morrison, Shaunna M¹, Eleish, Ahmed M.², Prabhu, Anirudh³, Pan, Feifei⁴, Downs, Robert T⁵, Rampe, Elizabeth B⁶, Bristow, Thomas F.⁷, Hazen, Robert M.⁸, Blake, David Frederick⁹, the, CheMin team¹⁰

(1) Earth and Planetary Sciences, Rutgers University, Piscataway, NJ, USA, (2) RPI, Troy, NY, USA, (3) Earth and Planets Laboratory, Carnegie Science, Washington, DC, USA, (4) RPI, Troy, NY, USA, (5) Geosciences, University of Arizona, Tucson, AZ, USA, (6) NASA Johnson Space Center, Houston, TX, USA, (7) NASA Ames Research Center, Moffett Field, CA, USA, (8) Carnegie Institution for Science, Earth and Planets Laboratory, Washington, DC, USA, (9) NASA Ames Research Center Exobiology, Moffett Field, CA, USA, (10) NASA, Mars Science Laboratory, USA,

Abstract:

The CheMin X-ray diffractometer onboard the NASA Mars Science Laboratory (MSL) Rover, Curiosity, performs X-ray diffraction (XRD) experiments on martian rock and sediment collected in Gale crater. To gain a greater understanding of the formational conditions and geologic history of Gale crater and the minerals found therein, the CheMin team developed a crystal-chemical method to predict limited chemical compositions of the minerals observed in the CheMin samples (e.g., Fe-Mg olivine, Na-Ca plagioclase, K-Na alkali feldspar, Mg-Fe-Ca pyroxene, alunite-jarosite) [1-3]. Previous methods were chemically constrained due to the complexity of the crystal chemical relationships beyond three chemical components. However, in this study we adapt a machine learning technique, Label Distribution Learning (LDL) [4], to predict multicomponent chemical compositions of Gale crater mineral phases, thereby allowing for more detailed petrologic interpretation of the geologic history of the martian surface [5-7].

 

LDL is a novel framework for classification problems with small datasets and has been widely applied to facial recognition problems such as age estimation. In this study, we adapt the LDL algorithm such that it can predict chemical elements (labels) and their abundances (degrees) for each martian mineral sample, based on crystallographic parameters. We evaluate performance using distance and similarity between label distributions as well as mean square error and also compare the results to traditional machine learning methods.

 

This work is expanding our understanding of Gale crater, Mars, and the processes that formed the terrestrial bodies in our solar system.

 

References:

[1] Morrison et al. (2017) Am Min, 103(6): 848-856

[2] Morrison et al. (2017) Am Min, 103(6): 857-871

[3] Morrison et al. (2024) Minerals (Submitted)

[4] Geng (2016) IEEE Transactions on Knowledge and Data Engineering, 28(7), 1734-1748

[5] Morrison et al. (2019) Goldschmidt abstract

[6] Eleish et al. (2025) Computers & Geoscience (In prep)

[7] Morrison et al. (2025) JGR: Planets (In prep)