To date, the Diviner compositional investigation has been conducted using Diviner’s three narrow band “8 µm channels” to model the position of the silicate Christiansen feature (CF), the position of which is sensitive bulk silica content. In this work, we extend Diviner spectral analyses to include an additional broadband mid-infrared channel centered at 18 µm. Incorporation of this channel into compositional analyses enables the use of spectral indexing and spectral mixture analysis techniques that were not previously available using 3-point spectra. Our initial work has focused on demonstrating the utility of these techniques for an olivine-bearing region of the Moon southeast of Aristarchus crater.
Locations on the Moon with olivine-bearing, pyroxene-free lithologies have been detected by both the Moon Mineralogy Mapper (M3) and Spectral Profiler (SP). These areas are generally associated with crater ejecta on thin portions of the lunar crust. While Diviner is capable of detecting pure olivine using the three “8 µm channels”, very few of the sites identified by M3 and SP as “olivine-rich” look different from average lunar mare composition in Diviner 3-point emissivity data. This suggests that either (1) olivine abundances at these sites are quite low, or (2) that a high abundance of olivine relative to plagioclase is necessary to produce CF values that are distinguishable from average lunar mare.
To address this issue, we acquired a set of simulated lunar environment (SLE) spectra of olivine/anorthite mixtures. That work showed that for the CF position of an anorthite-forsterite mixture to be elevated above average mare (~8.3 µm), there must be greater than about 90 weight percent forsterite present. It is clear from this work that non-linear spectral mixing makes it difficult to determine the presence of olivine from the spectral shape of channels 3-5 alone. However, these spectra show that with increasing olivine content, there is a concavity inversion between Diviner channels 4-6. We developed a concavity index using channels 4, 5, and 6. We made use of this index to map the presence of olivine southeast of Aristarchus, and the distribution of olivine-bearing surfaces mapped by this technique corresponds well with that mapped by M3.
The addition of a fourth channel to Diviner compositional analysis also enables rudimentary spectral mixture analysis (SMA). While the fine-grained nature of the lunar regolith likely precludes linear unmixing, spectral mixture analysis may prove to be a useful method to determine the presence and distribution of major silicate phases on the lunar surface. For this analsyis, we used laboratory spectra of anorthite, forsterite, and augite acquired under SLE conditions. We then used factor analysis and target transformation to determine scene-derived spectral endmembers from the laboratory data. The SMA results generally correlate well with the spectral index mapping, although the modeled olivine distribution along the Aristarchus rim is not as strong as that seen in the index map. Future work will focus on correlation of deconvolution and index mapping methods and incorporating an additional Diviner broadband channel (band 7, centered at ~31 µm) into our analyses.