GSA Connects 2021 in Portland, Oregon

Paper No. 108-1
Presentation Time: 1:35 PM


HAMILTON, Victoria, Southwest Research Institute, 1050 Walnut St, Ste 300, Boulder, CO 80302-5142

Factor analysis and target transformation (FATT) is a multivariate technique for recovering end-member spectra from datasets containing linear mixtures of materials at the scale of the spectral measurement, where the abundances of the mixture components vary independently between measurements [e.g., Malinowski, 1991; Bandfield et al., 2000; 2002]. A benefit of this approach relative to linear least squares modeling is that it does not rely on the direct identification of a component spectrum via a library of candidate materials and can be used to reduce or eliminate contributions from noise. As such, FATT can be used in situations where the end-member spectra may be unknown or may not be adequately represented by available spectral libraries. FATT has been used successfully with laboratory spectra of field samples [Bandfield, 2002] and with remote sensing data for surface-atmosphere separation [Smith et al. 2000], atmospheric component identification [Bandfield et al. 2000], and identification of constituent lithologies and minerals in geologic targets [e.g., Glotch et al., 2006; Hamilton and Ruff, 2012; Thomas et al., 2017; Amador et al., 2018].

In this presentation I discuss the application of FATT to thermal infrared spectral data collected at the asteroid Bennu by the OSIRIS-REx Thermal Emission Spectrometer (OTES). Based on visual inspection and simple spectral indices, two end-member spectral shapes have been identified on the surface of Bennu in OTES global mapping data (40 m/spot) [Hamilton et al., 2021]. The difference between these two spectral shapes is attributed primarily to the relative abundance of a fine particulate component within the field of view rather than compositional variation. All variations at this spatial scale appear to be explainable, for the most part, as linear mixtures of these two components. However, based on albedo variability observed at much smaller spatial scales (cm to m) in visible images, it is plausible that there is more than one lithology present on Bennu. Along with additional data acquired by OTES during reconnaissance mapping, sample collection rehearsals, and the sample collection event, which offer increasingly greater spatial resolution (to decimeter scales), FATT may enable the detection of compositional variation not yet detected by other methods.