Figure 22. Calculated SiO2 vs. MgO + FeO for all accepted model runs for (a) MDI and (b) SoF, displayed as heatmaps. Dark red color indicates a higher concentration of points, while white means there are no points in that space. Mineral endmembers are displayed as colored circles, while the best fit model run for each MDI and SOF is a star.
8 Conclusions and Implications
Multiple CRISM false-color image data show that the linear Bagnold dunes in Gale Crater have a reddish-brown color, while sand sheets to the south of the dunes are darker and lack the reddish-brown color. The traverse of the Curiosity rover has led to observations of the linear dunes at Mt. Desert Island (MDI), and a sand sheet to the south at the Sands of Forvie (SoF). SSA spectra retrieved from CRISM scene FRT00021C92 show that the difference in the false-color images can be attributed to a long wavelength (1.7 to 2.5 µm) rise in the MDI sand spectrum that is absent in the SoF spectrum. Instead, the SoF sand spectrum is characterized by a broad ~2.2 µm absorption feature.
To classify the potential mineralogic controls on these different spectral features, we performed checkerboard unmixing of the SSA image cube to extract the extreme spectral endmembers from both sand deposits. These spectra were modeled using carefully selected laboratory spectra and Hapke (2012) theory to invert for mineral abundances and grain size information. Model results, including inversions to equivalent APXS oxide compositions, show that spectral endmembers from both sites are dominated by basaltic glass (i.e., an amorphous phase) and labradorite or other spectrally equivalent feldspar. The broad 1 µm feature in both MDI and SoF is primarily controlled by olivine, with some contribution from basaltic glass. Inclusion of a pigeonite spectrum explains the increase in SSAs with increasing wavelength for the MDI spectral endmember, while the absence of pigeonite and presence of augite explains the deep, broad ~2.2 µm absorption for the SoF spectral endmember. Modeling results also suggest that the SoF deposit has coarser grains than the MDI deposit, consistent with measurements from MAHLI data (Weitz et al., 2022). Importantly, modeling reflectance data displayed the non-uniqueness of deriving chemical composition from CRISM data and highlighted the importance of statistical analysis of the data.
The primary emphasis of the paper was to retrieve the mineralogy and grain sizes for MDI and SoF and to compare to observations acquired by Curiosity, but there are also implications for aeolian processes. First, we note that the two sand bodies are only ~2.5 kilometers from one another. Even given their proximity, they have subtle yet distinctly different mineralogy and grain sizes. We speculate that the differences are a consequence of the topographic settings intensifying wind sorting effects over time. The Bagnold linear dunes are not encumbered by topography, but rather are migrating from NE to SW associated with the dominate wind direction (Sullivan et al., 2022). On the other hand, the SoF is bordered on the south and southwest by topographic obstacles, specifically the hills associated with Mount Sharp and the scarp that defines the edge of the Greenheugh pediment (Figure 23). The sands in SoF are thus unable to migrate in an unencumbered manner. We infer that the finest grained sands preferentially migrate onto the pediment surface, leaving behind overall coarser material and a slightly different mineralogy as compared to the Bagnold dune sands. The remaining grains left behind can become segregated by wind conditions saltating smaller grains but causing creep of larger grains. This can lead to coarsening of the ripples, ultimately causing the formation of coarse-grained megaripples as observed in SoF (Lapotre et al., 2018). This cycle causes the variation seen in the spectral data, that is, SoF contains larger ripples with concentrated coarser grains while MDI lacks this coarsening and has smaller ripples.