Deriving chemical trends from thermal infrared spectra of weathered basalt: Implications for remotely determining chemical trends on Mars

Variations in chemical composition over a planetary surface can be used to study petrologic and aqueous alteration processes. The desire for such data on Mars has prompted investigators to derive chemistry from models of Thermal Emission Spectrometer data. Although chemistry derived from thermal infrared spectral models is reportedly reliable for unaltered igneous rocks, the martian surface has experienced chemical weathering, which can adversely affect models. Here, we examine weathered basalts from Baynton, Australia, for which chemical weathering trends have been previously characterized, to test how well chemistry and chemical trends can be determined from TIR spectra of weathered rocks. The mineralogy of variably weathered rocks was derived from TIR spectra by linear mixing, and major-element chemistry was calculated from those mineral models. Derived chemistries and trends were compared to those measured by X-ray fluorescence. TIR spectroscopy is sensitive to weathering products in weathering rinds because the products are present in a coating geometry, making it a useful technique for remotely detecting weathered surfaces on planetary surfaces such as Mars. This sensitivity results in significant modeled abundances of weathering products (>80% of all phases) from TIR spectra of weathered Baynton surfaces, despite evidence from microscopy and X-ray diffraction showing that igneous minerals dominate the weathering rind. Measured chemical weathering trends show loss of MgO, CaO, Na₂O, and K₂O and relative enrichment in Al₂O₃ and FeO_T. The modeled trends are similar to the measured trends, but a closer look at the modeled oxide abundances demonstrates that most oxides (i.e., alkalis, SiO₂, and FeO_T) are not well modeled, especially for weathered surfaces. The reasons for this are: (1) non-linear mixing and the presence of secondary coatings causes the overestimation of secondary phases in spectral models, and (2) spectral libraries generally lack poorly crystalline and amorphous secondary phases that are common in weathering rinds so that crystalline phases such as phyllosilicates are selected. The martian surface has likely been weathered less pervasively than the Baynton rocks and, therefore, weathering products may be dominated by poorly crystalline and amorphous phases, rather than crystalline phyllosilicates. Adding these phases to spectral libraries could improve bulk chemistry derived from the martian surface; however, if the secondary phases are present in a coating geometry, the derived chemistry will reflect the composition of the coating, and it may be difficult to infer the chemistry of the parent rock.