Outlier Effects on LIBS Geochemistries

Predicting geochemistry in geological samples using laser-induced breakdown spectroscopy (LIBS): Effects of compositional and textural outliers.

Jack D. Henry, Kirsten L. Siebach, M. Darby Dyar, Kate H. Lepore, & Cai R. Ytsma (2026)

Spectrochimica Acta Part B: Atomic Spectroscopy, volume 235, article 107376

‍This is the first journal publication resulting from the collaboration with Rice University Ph.D. student Jack Henry, advisor Kirsten Siebach, and the Mount Holyoke College team (me, Darby, and Kate).

Jack is interested in applying realistic constraints on the predictions of Mars geochemistries generated by the ChemCam and SuperCam instruments. Though the science team calibrations are based on homogeneous, pressed rock powders, samples on Mars are of varying size and form and may contain compositions outside the scope of the calibration standards.

Jack has expanded on the data Kate, Darby, and I collected on the SuperLIBS and ChemLIBS instruments to better understand how different powder grain sizes, mixtures of samples, and inclusion of mineral data can affect model accuracies.

For this project, I advised Jack and Kirsten on methods for testing existing models, generating new models to incorporate these data, and contextual interpretation of results. I also performed data engineering tasks to cumulate, clean, process and perform multivariate regression on these data using custom pipelines, which generate user-friendly reports and visualisations.

Previous
Previous

LIBS Geochemistries: Showcase

Next
Next

Portable XRF Geochemistries