EFFECTS OF MAGMATIC EVOLUTION, OXIDATION, AND CRYSTALLINITY ON SPECTRAL VARIABILITY IN A MARS-ANALOG VOLCANIC SYSTEM

Igneous volcanic surfaces are common and varied throughout the terrestrial planets. Often, orbital spectroscopy is the primary or only method for determining surface mineralogy, which in turn can provide insight into petrologic processes and place constraints on interior evolution and chemistry. Magmatic evolution and detection of primary minerals in visible/near-infrared (VNIR) and thermal infrared (TIR) spectra of igneous rocks are well-studied; however, natural volcanic systems also exhibit variability in crystallinity, oxidation, and eruption-related alteration (e.g. hydrothermalism). The extent to which spectroscopy can be used to identify these characteristics alongside each other in volcanic provinces is thus a key question in interpreting volcanic processes from orbit.

To test how volcanic processes alter VNIR and TIR spectra, we collected over 140 rock samples from 12 mapped volcanic units at the North and Middle Sister Volcanoes in Oregon. 43 samples were chosen as representative of the variability in crystallinity, oxidative state, bulk alteration, and magmatic evolution present at North and Middle Sister. These samples range from basaltic andesite to rhyolite, are variously oxidized, and include lava flows and tephra. We measured powders, cut faces, and split faces of these samples using VNIR (0.3-2.5 µm) spectroscopy, TIR (8-40 µm) spectroscopy, and quantitative x-ray diffraction.

VNIR spectra are sensitive to aqueous alteration and Fe-bearing phases, but cannot detect Fe-poor primary silicates, and thus show only minor correlations with magmatic evolution or crystallinity. Instead, the strengths of absorptions due to primary minerals appear partly controlled by the presence of matrix opaques. TIR deconvolutions correctly predict bulk composition and rock type (e.g. basaltic andesite), plagioclase composition, and % crystallinity for relatively crystalline unoxidized samples, but oxidized and poorly crystalline samples are more variable, and in some cases these two properties can cause non-unique spectra. Furthermore, oxides that dominate VNIR spectra are modeled below the detection limit in TIR. To fully characterize the eruptive history of a volcanic province, we find that both VNIR and TIR are required to avoid issues of non-uniqueness and insensitivity in both techniques.