Dr. Katherine Shirley (she/her)

Abstract:

Introduction: During the Reconnaissance phase of NASA’s OSIRIS-REx mission, the Thermal Emission Spectrometer (OTES) acquired high–spatial resolution emissivity spectra over Bennu’s four prospective sampling sites [1, 2]. We analyse the calibrated OTES dataset archived in the Planetary Data System [3] to quantify compositional and mineralogical diversity across the original four candidate sample sites (Nightingale, Kingfisher, Osprey, and Sandpiper) and to explore possible drivers of Bennu’s surface heterogeneity, including implications for Bennu’s mineralogy and space-weathering history.  Figure 1: Site-Averaged Emissivity Spectra with Annotated Band Parameters Methods: Calibrated emissivity spectra (5.7-100 µm) were linked to corresponding OCAMS imagery [5] to place the thermal infrared measurements in geological context, by cross-referencing observation times. For every spectrum we derived four diagnostic band parameters: Christiansen Feature (CF), silicate stretching band, silicate bending band and spectral slope, following the methods outlined in [6]. Each site contains thousands of spectral observations (site-averaged for visualization in Figure 1). The corresponding band parameters were compared using three statistical models: Principal Component Analysis (PCA) [5], k-Nearest Neighbors (KNN) [7], and Analysis of Variance (ANOVA) [8]. The three methods compare the mean and variance of each individual observation per site, considering how the in-group variance (i.e. the spread within all observations of a single site) compares to the out-group variance (i.e. the spread from other sites).  Results: Significant differences in emissivity spectra emerged among the four sites. PCA indicated that the first three components explain 85.5% of spectral variance, distinguishing Kingfisher as notably unique, with Sandpiper and Osprey exhibiting the greatest similarity. The KNN analysis corroborated PCA findings, reaching optimal classification accuracy (47%) at k = 21. ANOVA highlighted significant variability among the sites, especially in the spectral slope parameter (F = 762.8), suggesting differences in particle size distribution and space weathering could be driving factors in the detected heterogeneity [9]. Band ratio analyses provided additional insight into site-specific mineralogical distinctions, notably the relationship between silicate features and aqueous alteration indicators [10].  Figure 2: Distributions of Band Parameters by Site Discussion: Variability in spectral parameters aligns with documented particle size frequency distributions and known space weathering spectral types across Bennu’s surface [9]. Nightingale, the mission’s selected sample site, captures representative global characteristics, contrasting with Kingfisher’s distinct compositional and physical attributes, potentially related to differences in Fe/Mg content and degree of aqueous alteration [10].  Conclusion: Integrative use of multiple statistical approaches confirms the compositional and physical diversity of Bennu's surface, as seen through the four prospective sites. These analyses provide a framework for interpreting returned sample data and offer insights into the connections between mineralogy, particle size, and space weathering processes on small airless body surfaces.  References: [1] Lauretta D. S. et al (2021) Sample Return Missions. [2] Hamilton V. et al. (2021) A&A (Vol. 650). [3] Christensen, P. R. et al. (2019) NASA Planetary Data System [4] Christensen P. R. et al. (2018) Space Science Reviews (Vol. 214, Issue 5). [5] Rizk B. et al (2018) Space Science Reviews (Vol. 214, Issue 1). [6] Xie B. et al (2022) Minerals (Vol. 508, Issue 12). [7] Kramer O. (2013) Intelligent Systems Reference Library (13-23). [8] Sawyer S. (2009) Journal of Manual & Manipulative Therapy. [9] Clark B. E. et al (2023) Icarus (Vol. 400). [10] Bates H. et al (2020) MaPS (Vol. 55, Issue 1). 

Abstract:

The ultramafic and silicic rocks on the lunar surface have played an important role in expanding our knowledge regarding its thermal and magmatic evolution. The surface identification and quantification of these rocks on the global scale can significantly improve our understanding of their spatial extents, relationships and formation mechanisms. Christiansen feature positions using Diviner data have aided in global identification and mapping of relatively silica-rich and silica-poor lithologies on the lunar surface. We have used laboratory thermal infrared spectra of silicic and ultramafic rocks to analyze the variation in Christiansen feature in simulated lunar environment. We have characterized the absolute bulk silica content of the rocks and minerals and their Silica, Calcium, Ferrous iron, Magnesium index. We find that they are linearly correlated to the Christiansen feature despite particle size variations. Furthermore, we find that the Christiansen feature shifts toward longer wavelengths with increase in ilmenite content in the ilmenite-basalt mixtures. We have explored the effect of instrument's spectral band position on the accuracy of the parabolic method that is currently used for the estimation of Christiansen feature position from Diviner data. We find that this method performs poorly for the estimation of the Christiansen feature for ultramafic and silicic rocks and minerals/mineral mixtures. We propose using a machine learning algorithm to estimate the Christiansen feature with higher accuracy for all kinds of silicate compositions on the Moon. This method will lead to increased accuracy in absolute quantification of bulk silicate composition of the lunar surface at varying spatial scales.

Abstract:

A laboratory study was performed using the Visible Oxford Space Environment Goniometer in which the broadband (350–1250 nm) bidirectional reflectance distribution functions (BRDFs) of two representative Apollo regolith samples were measured, for two surface roughness profiles, across a range of viewing angles—reflectance: 0–70°, in steps of 5°; incidence: 15°, 30°, 45°, and 60°; and azimuthal: 0°, 45°, 90°, 135°, and 180°. The BRDF datasets were fitted using the Hapke BRDF model to (1) provide a method of comparison to other photometric studies of the lunar regolith and (2) to produce Hapke parameter values which can be used to extrapolate the BRDF to all angles. Importantly, the surface profiles of the samples were characterized using an Alicona 3D® instrument, allowing two of the free parameters within the Hapke model, φ and θ ¯ $$ \overline{\theta} $$ , which represent porosity and surface roughness, respectively, to be constrained. The study determined that, for θ ¯ $$ \overline{\theta} $$ , the 500–1000 μm size‐scale is the most relevant for the BRDF. Thus, it deduced the following “best fit” Hapke parameters for each of the samples: Apollo 11 rough— w $$ w $$ = 0.315 ± 0.021, b $$ b $$ = 0.261 ± 0.007, and h S $$ {h}_S $$ = 0.039 ± 0.005 (with θ ¯ $$ \overline{\theta} $$ = 21.28° and φ = 0.41 ± 0.02); Apollo 11 smooth— w $$ w $$ = 0.281 ± 0.028, b $$ b $$ = 0.238 ± 0.008, and h S $$ {h}_S $$ = 0.032 ± 0.006 (with θ ¯ $$ \overline{\theta} $$ = 13.80° and φ = 0.60 ± 0.02); Apollo 16 rough— w $$ w $$ = 0.485 ± 0.155, b $$ b $$ = 0.155 ± 0.083, and h S $$ {h}_S $$ = 0.135 ± 0.007 (with θ ¯ $$ \overline{\theta} $$ = 21.69° and φ = 0.55 ± 0.02); Apollo 16 smooth— w $$ w $$ = 0.388 ± 0.057, b $$ b $$ = 0.063 ± 0.033, and h S $$ {h}_S $$ = 0.221 ± 0.011 (with θ ¯ $$ \overline{\theta} $$ = 14.27° and φ = 0.40 ± 0.02). Finally, updated hemispheric albedo functions were determined for the samples, which can be used to set laboratory measured visible scattering functions within thermal models.