Earth and Space Science Open Archive This preprint has been submitted to and is under consideration at Journal of Geophysical Research - Planets. ESSOAr is a venue for early communication or feedback before peer review. Data may be preliminary.Learn more about preprints preprintOpen AccessYou are viewing the latest version by default [v1]Orbital and In-Situ Investigation of the Bagnold Dunes and Sands of Forvie, Gale Crater, MarsAuthorsEleanor LouiseMorelandRaymond E.ArvidsonRichard V.MorrisThomasCondusMadison NicoleHughesCatherine M.WeitziDScott J.VanBommelSee all authors Eleanor Louise MorelandCorresponding Author• Submitting AuthorRice Universityview email addressThe email was not providedcopy email addressRaymond E. ArvidsonWashington University in St. Louisview email addressThe email was not providedcopy email addressRichard V. MorrisNASA Johnson Space Centerview email addressThe email was not providedcopy email addressThomas CondusWashington University in St. Louisview email addressThe email was not providedcopy email addressMadison Nicole HughesWashington University in St. Louisview email addressThe email was not providedcopy email addressCatherine M. WeitziDPlanetary Science InstituteiDhttps://orcid.org/0000-0002-4646-0825view email addressThe email was not providedcopy email addressScott J. VanBommelWashington University in St. Louisview email addressThe email was not providedcopy email address

The Mars Reconnaissance Orbiter Compact Imaging Spectrometer for Mars (CRISM) covers the spectral range from 0.362 to 3.92 μm with a midafternoon local solar time data acquisition. For equatorial to midlatitudes, depending on the season and surface materials, wavelengths longer than ∼2.65 μm exhibit spectral radiances on sensor that include sunlight and thermal‐emission related terms. We developed a radiative transfer based neural network approach to model both solar and emitted terms in which surface kinetic temperatures are retrieved for each image pixel, together with single scattering albedo spectra, over the full CRISM wavelength range. We applied the method to along‐track oversampled scene FRT00021C92 over Glen Torridon within Gale Crater, where the Curiosity rover traversed and acquired remote sensing and in‐situ data. Synergistic analysis of orbital and rover‐based data, coupled with laboratory analyses of ferric‐rich smectites, provide a self‐consistent set of results for the presence of desiccated nontronite associated with Murray formation mudstones exposed as periodic bedrock ridges located just to the south of Vera Rubin ridge. The desiccated nature is consistent with Curiosity's CheMin data, which for Glen Torridon drill samples indicate an abundance of nontronite having a collapsed structure resulting from loss of interlayer H 2 O.

The Bagnold linear dune field investigated by Curiosity at Mount Desert Island (MDI) is north of the ∼5.5 km high Aeolis Mons mound in Gale crater. False‐color images (RGB: 2.496, 1.802, and 1.237 μm, respectively) generated from Mars Reconnaissance Orbiter Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) data show the dune field has a reddish‐brown color. The Sands of Forvie (SoF), located ∼2.5 km to the southeast of MDI, is darker and lacks the reddish‐brown color. Single scattering albedo (SSA) spectra retrieved at 12 m/pixel using along‐track oversampled CRISM observation FRT00021C92 show a long wavelength (1.7–2.5 μm) increase in albedo for MDI dunes. For the same wavelength interval, SoF is characterized by a broad ∼2.2 μm absorption feature, consistent with color differences between the two deposits. Checkerboard un‐mixing of the SSA image cube was used to isolate spectral endmembers within the MDI and SoF deposits. Radiative modeling of these CRISM spectral endmembers using Hapke (2012, https://doi.org/10.1017/CBO9781139025683) theory implies finer grain sizes, more pigeonite, plagioclase, and olivine, and less basaltic glass and augite for MDI as compared to the SoF deposit. These results are consistent with Curiosity‐based observations that MDI contains smaller ripples with overall finer grains, whereas SoF has large ripples with coarser grains on the crests. Although these sand deposits are only located ∼2.5 km away from one another, wind and local topographic controls are interpreted to have modulated grain sizes and mineralogy.

Introduction: A widespread surface deposit of crystalline hematite in the Sinus Meridiani region of Mars was revealed by the Mars Global Surveyor Thermal Emission Spectrometer (TES) [1, 2]. This discovery led to the selection of Meridiani Planum (in southern Sinus Meridiani) as the landing site for the Mars Exploration Rover Opportunity because the hematite may have had an aqueous origin [3]. Ubiquitous hematite-rich concretions were observed by the rover, confirming the orbital signature. The concretions have been concentrated on the surface as a wind-blown lag deposit by weathering of the sulfate-bearing sandstones of the Burns formation [e.g., 4, 5]. In this abstract, a link is established between areal abundance of hematitic concretions as seen by Opportunity, and the 0.86 μm band depth in Mars Reconnaissance Orbiter Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) observations of Meridiani Planum. The 0.86 μm band minimum is due to a crystal field transition of ferric iron, and is diagnostic of crystalline hematite [6]. This band minimum is shifted towards slightly longer wavelengths in the CRISM data because of the mixing of basaltic material and nanophase ferric oxides, and has also been seen in Opportunity Pancam spectra [7, 8]. With the positive correlation between rover and orbital data established we then extend mapping of the concretion abundances across Sinus Meridiani. Data Processing: The WUSTL processing pipeline [9] was used to retrieve surface single scattering albedos (SSA) from CRISM S data (0.362 to 1 μm) [10]. Dust and ice aerosols were modeled, and a modestly backscattering Hapke function was used as the surface boundary condition. CRISM data, which have pixel sizes ranging from 18 to 36 m for targeted observations, are regularized to account for the instrument spatial and spectral transfer functions and extract the best SSA values [11, 12]. The 0.86 μm band depth spectral parameter (BD860_2) [13] was then generated and map-projected. Percentage of concretion cover was calculated from segmentation of the concretions in Hazcam and Pancam calibration target images. These data were utilized for consistent views of the surface throughout Opportunity’s mission. The segmentation takes advantage of the large, round shapes of the concretions, and the percent coverage values are plotted along the traverse at every site. Results and Discussion: At Endeavour crater, there are correlatively high BD860_2 values and concretion cover percentages within the Botany Bay rim gap (Fig. 1a). The high values are interpreted to be due to wind blowing up and out of the crater, concentrating the relatively large and dense concretions at the surface [14]. A similar pattern has been observed at Iazu crater, directly to the south [15]. The high BD860_2 values and concretion cover percentages are also associated with dark wind streaks emanating from craters, such as Victoria (Fig. 1b). Basaltic sands sourced from the interior of this crater are blown up and out by southeasterly winds [16]. A BD860_2 spectral parameter map of Sinus Meridiani has been produced using high quality CRISM mapping survey images at 100 or 200 m/pixel (Fig. 3). The improved spatial resolution of CRISM relative to TES (3 x ~6 km/pixel) [1] and OMEGA (2-5 km/pixel for global mapping) [17] allows for more detail to be seen, such as the mantle of BD860_2 values around the ejecta deposit of Bopolu crater. Future Work: We will continue to create BD860_2 spectral parameter maps from CRISM mapping surveys to fill in gaps of coverage in Sinus Meridiani. BD860_2 maps for all targeted CRISM observations across the region will be created and analyzed together with HiRISE, CTX, and CRISM-derived thermal inertia [18]. A campaign to acquire additional targeted S band CRISM observations and HiRISE images for select locations in the region will be requested. The products will then be analyzed to understand the stratigraphic and erosional contexts for the concretion abundances.

Introduction: In this abstract we present the use of a machine learning neural network approach, Separating Temperature and Albedo by Neural Networks (STANN), to retrieve both surface single scattering albedos (SSA) and kinetic temperatures (T) using CRISM hyperspectral imaging data as an example application. The issue is that the retrieval of both SSA and T from emission spectra at thermal wavelengths (e.g., THEMIS and TES), and from mixed solar and thermal wavelengths (e.g., CRISM and OMEGA), is an underdetermined and ill-posed problem. We use the pipeline processing developed by us over the past few years to model atmospheric effects (gases and aerosols) with DISORT [1] and the Hapke function [2] for surface scattering and emission, followed by application of a log maximum likelihood approach to retrieve denoised and sharpened SSA spectra and images [3]. We now show how to retrieve both solar and thermal effects in this pipeline using a neural network approach, using Mars Reconnaissance Orbiter CRISM scene FRT0000B6F1 covering Mount Sharp as an example. In a companion abstract Condus et al. [4] show application of our pipeline to THEMIS and TES data. Mathematical Model: We use the function f to describe our overall approach. For the j band whose wavelength is λj, the radiance at this band rj can be computed based on the surface temperature T, the spatial geometric information l, and the single scattering albedo sj as ( , , , ) j j j r f s T l   .

MARS. Y. Liu 1 , T. D. Glotch 2 , N. A. Scudder 2 , M. L. Kraner 2 , T. Condus 2 , R. E. Arvidson 3 , E. A. Guinness 3 , M. J. Wolff 4 , and M. D. Smith 5 , 1 Southwest Research Institute, San Antonio, TX (yang.liu@swri.edu), 2 Department of Geociences, Stony Brook University, Stony Brook, NY, 3 Department of Earth and Planetary Sciences, Washington University in St. Louis, St. Louis, MO, 4 Space Science Institute, Boulder, CO, 5 NASA Goddard Spaceflight Center, Greenbelt, MD.

Introduction: The Compact Reconaissance Imaging Spectrometer for Mars (CRISM) is a hyperspectral imager that maps the Martian surface at ~20 m/pixel from 362 to 3920 nm at 6.55 nm/channel. This data provides high-resolution compositional and geological information on surface features [1]. A range of compositional data products based on spectral band analysis and indexing is already well-established [2], but mineral abundances can be difficult to determine from these data. High spatial and spectral resolution make it possible to perform quantitative spectral unmixing to determine the abundance and spatial distribution of surface minerals. Reflectance spectra are highly non-linear in the visible-near infrared (VNIR) wavelength range. For this reason, most linear unmixing studies have focused on the mid-infrared wavelength region. However, single scattering albedo (SSA) spectra add linearly in the VNIR range [3] and can be derived either from optical constants n and k or reflectance, making single scattering albedo a valid unmixing parameter. Here we introduce a linear unmixing model for single scattering albedo spectra derived from CRISM data.