Linear Modeling of Spectra of Fine Particulate Materials: Implications for Compositional Analyses of Primitive AsteroidsOPEN ACCESS 

Vanessa C. Lowry, Kerri L. Donaldson Hanna, Humberto Campins, Neil Bowles, Victoria E. Hamilton, Eloïse C. Brown

Earth and Space Science
First published: 04 March 2022


“Key Points

  • Fine particulate primitive asteroid analog spectra were modeled using weighted least squares and fine particulate spectral library
  • Model underestimated abundances of spectrally dominant mineral phases, minor phases were not modeled, and phases not present in mixtures were selected
  • Linear modeling does not accurately estimate abundances when modeling less than 90-µ-sized particle materials”

“In this study, we applied a sum to one constraint weighted least squares (STO WLS) model to thermal infrared (TIR) spectra of a suite of primitive asteroid analogs spectrally and volumetrically dominated by fine particulates (<38 μm). Since coarse particulate emissivity spectra combine linearly, deriving compositions from these are fairly straightforward. Across the TIR it is not as straightforward for fine particulate emissivity spectra due to the nonlinear behavior that arises from volumetric scattering between particles when the particle size becomes comparable to the wavelength of light. Using a WLS model, mixed spectra may be deconvolved into areal percentages of each end member which we assume corresponds to their volume percentages. We used a spectral library of pure mineral spectra to model TIR spectra of a suite of physical mixtures and meteorites obtained under ambient (Earth-like) and simulated asteroid environment (SAE) conditions collected for the OSIRIS-REx team (Donaldson Hanna et al., 2021, The STO WLS model underestimated the modal abundances of the dominant mineral phases by ∼25.0 ± 0.1% on average over the full spectral range and ∼22.0 ± 0.4% over a limited spectral range in all of the ambient and SAE physical mixture and meteorite spectra. Minor phases present in the mixtures and meteorites (abundances ≤5%) were typically not modeled. Also, phases not present in the mixtures were commonly selected (∼2 phases on average with an average abundance of ∼25.2 ± 1.9%) to get the best mathematical fit.”