We designed a laboratory visible-to-near-infrared (VNIR) hyperspectral experiment to test the effectiveness of factor analysis/target transformation for detecting minerals mixed with Mars Global Simulant-1 (MGS-1). The purpose of this experiment is to test for true positive, true negative, false positive, and false negative results from application of factor analysis/target transformation methods and determine the parameters that dictate good versus bad algorithm performance. Gypsum, calcite, montmorillonite, nontronite, and kaolinite were each mixed with MGS-1 at abundances of 1%, 2.5%, 5%, 10%, 20%, and 50%. The mixtures were placed in 2.5 x 2.5 x 1 cm sample trays and imaged using a Headwall Imaging Spectrometer with a spectral range of 0.9-2.6 mu m, 8.98 nm spectral sampling, and 0.34 mm/pixel spatial resolution. These images include thousands to tens of thousands of hyperspectral pixels covering each individual mixture tray. Full-image factor analysis/target transformation (FA/TT) and Dynamic Aperture Factor Analysis/Target Transformation (DAFA/TT) were applied to these data to detect the minerals mixed with MGS-1. The results demonstrate that factor analysis/target transformation is prone to both false positive and false negative detections, but in certain applications-including DAFA/TT-it can be useful for highlighting spectrally interesting areas in hyperspectral images for follow-up investigation. The results presented here demonstrate that applications of factor analysis/target transformation to VNIR hyperspectral datasets should be used to highlight small outcrops and/or weak spectral signals in pixels for follow-up investigation. This emphasizes the need for supporting evidence to be obtained-in addition to factor analysis/target transformation-before interpretations of planetary surface processes should be made.

Carbonate minerals have been detected in Jezero crater, an ancient lake basin that is the landing site of the Mars 2020 Perseverance rover, and within the regional olivine‐bearing (ROB) unit in the Nili Fossae region surrounding this crater. It has been suggested that some carbonates in the margin fractured unit, a rock unit within Jezero crater, formed in a fluviolacustrine environment, which would be conducive to preservation of biosignatures from paleolake‐inhabiting lifeforms. Here, we show that carbonate‐bearing rocks within and outside of Jezero crater have the same range of visible‐to‐near‐infrared carbonate absorption strengths, carbonate absorption band positions, thermal inertias, and morphologies. Thicknesses of exposed carbonate‐bearing rock cross‐sections in Jezero crater are ∼75–90 m thicker than typical ROB unit cross‐sections in the Nili Fossae region, but have similar thicknesses to ROB unit exposures in Libya Montes. These similarities in carbonate properties within and outside of Jezero crater is consistent with a shared origin for all of the carbonates in the Nili Fossae region. Carbonate absorption minima positions indicate that both Mg‐ and more Fe‐rich carbonates are present in the Nili Fossae region, consistent with the expected products of olivine carbonation. These estimated carbonate chemistries are similar to those in martian meteorites and the Comanche carbonates investigated by the Spirit rover in Columbia Hills. Our results indicate that hydrothermal alteration is the most likely formation mechanism for non‐deltaic carbonates within and outside of Jezero crater.

Gale crater features abundant outcrops of phyllosilicates, sulfates, and other minerals that are under investigation by Curiosity [e.g. 1] following their observation in CRISM images [2]. These analyses by the rover are enabling ground truthing of minerals identified from orbit, especially those acquired by the CRISM instrument. Recent advances in image processing [3] and hyperspectral mapping [4], coordinated with mineral detections on the surface, are facilitating improved characterization of aqueous and igneous minerals across the Gale crater region. Here we present the results of mapping phyllosilicates, sulfates, other hydrated phases, and pyroxene using CRISM data of Mount Sharp at Gale crater. Methods: We used the Parente group algorithm for simultaneous atmospheric correction and denoising of CRISM images in the 1.0-2.6 μm spectral range that removes the majority of the residual atmospheric bands and spurious noise [3]. Several CRISM images have been processed using this technique for the Gale crater region, including images FRT000095EE, FRT000B6F1, FRT0000C0EF, and FRT0000C518 shown in Figs 1-2. We applied a new mapping algorithm using hyperspectral components in the feature extraction to discriminate among spectral types [4]. This technique applies Generative Adversarial Networks (GANs) [5] to learn the diagnostic spectral features needed for discriminating among spectral types using hyperspectral components in the feature extraction, to extend the CRISM mineral parameters developed using a spectral band, ratio, or slope [6]. This new GAN method is highly effective in identifying promising locations in the

Introduction: We are testing advanced algorithms for mineral detection and abundance determination in the laboratory using mineral mixtures of known abundance and particle size. We measure these mixtures using a visible to short-wave infrared (i.e. 0.4-2.6 μm) hyperspectral imager [1]. Over the past decade, new approaches for mineral detection and abundance have emerged [2, 6] that are very promising, but have not been rigorously tested in laboratory settings. Specifically, laboratory testing is required to evaluate the fidelity of detection algorithms with imaging spectrometer data that exhibit signal to noise, structured noise, and other properties characteristics of imaging spectrometer data sets (e.g. CRISM, M3). In this phase of our ongoing investigations, we are presenting the experimental design in a companion abstract [1] and initial results of mineral detection methods (Factor Analysis/Target Transformation, FA/TT) and abundance determination with a Hapke nonlinear mixture model. Data Acquisition: The experimental design and measurements are presented in [1]. A specially designed sample tray is loaded with binary mixtures of a target mineral mixed with the Exolith Mars Global Simulant (MGS-1) [4]. The loaded sample tray is measured with the Brown University Visible-Shortwave Infrared Headwall Imaging Spectrometer. The imaging spectrometer consists of a visible-near infrared and a shortwave infrared component, bore-sighted through the same fore optics. Together they provide optical observations across the wavelength range 400 – 2600 nm [1]. The spectrometer is fixed 25 cm above a translation tablegiving a pixel size of <0.071 mm in the visible-near infrared (VNIR) (0.38-1.01 μm) and ~0.339 mm in the shortwave infrared (SWIR) (0.95-2.6) μm. We have collected imaging spectrometer data of over 8 mixture suites, but are presenting the preliminary results of just the mixtures of selenite (gypsum) and MGS-1. Analysis Methods: (1) Mineral Abundance At VNIR-SWIR wavelengths, multiple scattering leads to nonlinear spectral mixing systematics [5, 6, 7]. Linearization is achieved by converting reflectance to singlescattering albedo [6]. Here we apply the most simple nonlinear spectral mixture analysis [8] where the endmember spectra are known and generated from the imaging spectrometer data. Mineral abundance is a fundamental goal of our work and we will employ more advanced approaches [e.g. 1] in future work. In this test, endmembers for selenite and MGS are extracted from the imaging spectrometer data and used to unmix the scene along with a neutral 100% white endmember. The preliminary results are shown in Figure 1 and Table 1. The simplified Hapke model [8] is able to estimate the abundance of selenite in binary mixtures with MGS to within 5% for mixtures with 5% or more of selenite. The mixtures with <5% selenite are not distinguishable from the blanks (pure MGS-1). We will compare this performance with other target minerals such as carbonate and serpentine.

This study exploits recent advances in image calibration and feature extraction techniques for analysis of hyperspectral images acquired by the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) to characterize subtle geologic outcrops at the border of phyllosilicate-bearing and sulfate-bearing regions of Mars. Specifically, a unique spectral “doublet” feature at 2.21-2.23 and 2.26-2.28 μm is isolated to characterize salty regions that may represent a changing climate on Mars. The martian locations exhibiting these spectral features are identified and compared with terrestrial settings with similar geologic compositions.

Introduction: Mt. Sharp in Gale crater is a ~5 km thick stratigraphic section, the lower portion of which includes a >200 m thick sequence of mudstones deposited in a lacustrine environment [1] that have been explored by the Curiosity rover. One of the most prominent features observed from orbit in Mt. Sharp is a dark section of strata (Fig. 1) previously mapped by [2] as the middle member of the lower formation of the Mt. Sharp group. This zone is underlain and superposed by lighter-toned strata. We use newly processed visible-near infrared CRISM spectral data to examine the mineralogy of these zones and the transitions between them in greater detail, with the goal of providing insight into whether these transititions are most consistent with diagenetic or primary depositional processes. Importantly, we demonstate that the tonal and mineralogical changes across these boundaries appear to be present throughout Mt. Sharp, thus they are likely to be traversed by the Curiosity rover in the future.

CRATER USING ADVANCED DATA PROCESSING TECHNIQUES FOR CRISM HYPERSPECTRAL IMAGING DATA. M. Parente1, R. Arvidson2, Y. Itoh1, H. Lin3, J. F. Mustard4, A.M. Saranathan1, F.P. Seelos5, J. D. Tarnas4. 1Dept. of Electrical and Computer Engineering, University of Massachusetts Amherst, MA 01003 (mparente@ecs.umass.edu). 2Dept. of Earth and Planetary Sciences, Washington University in St. Louis, St. Louis, MO. 3Institute of Geology and Geophysics, Chinese Academy of Sciences 4Dept. of Earth, Environmental and Planetary Sciences, Brown University, Providence, RI. 5Johns Hopkins University Applied Physics Lab, Laurel, MD.

Introduction: The Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) [1] on board of Mars Reconnaissance Orbiter has made significant contributions on revealing many detailed aspects of surface mineralogy of Mars, owing to significant efforts made on calibration and atmospheric correction to retrieve accurate surface reflectance. However, the atmospheric correction algorithm used in the current CRISM data processing pipeline, so called “volcano scan” [2, 3, 4], still leaves some residuals despite of its improvements [5, 6, 7]. In addition, the processed spectra are still sometimes corrupted with moderate to large noise that needs to be addressed. Our previous works [8, 9] tackled the issue of the residual caused by volcano scan using an unmixing technique and improved the signal quality for variety of images. We made a significant upgrade; our new method now considers severe noise and atmospheric correction and de-noising are simultaneously performed. The presence of water ice aerosol is also taken into consideration. Our new method is applicable to wide range of scenes including noisy ones acquired at IR high detector temperature. Methodology of simultaneous atmospheric correction and de-noising: As in our previous work [9], it is assumed that the light propagation through the atmosphere is modeled by the Beer-Lambert model and that the surface reflectance is approximately modeled by the corrupted linear spectral mixing model [10] in the logarithmic domain. The estimation of atmospheric transmission is performed by a model inversion formulated as a minimization problem, in which a modified version of sparse unmixing with adaptive background [10] is simultaneously conducted to estimate the surface reflectance and model parameters. The sparse unmixing allows us to select endmembers/phases contributing to each spectral signal from a large spectral library and estimate their fractional abundances. In order to take various noise with different magnitude into consideration, we formulate the unmixing in a robust way using `1-norm. The algorithm to solve the minimization problem is iterative and large noise are detected and removed at each step in the algorithm using hardthresholding on the residual of modeling. In order to ensure the method to work on variety of scenes with little adjustment, many phases are included for the computation of modeling surface spectra. The 686 candidate phases (endmembers) are selected from CRISM spectral library, RELAB spectral database [11], U.S. Geological Survey (USGS) spectral library (splib06) [12], and CRISM Type Spectra Library [13]. The transmission spectra in the ADR are also used for the initialization of the transmission spectrum to stabilize the optimization algorithm. The absorption efficiency spectra of water ice are also calculated using Mie theory from the optical constants in the Grenoble Astrophysics and Planetology Solid Spectroscopy and Thermodynamics (GhoSST) database1 and stacked to the library to compensate the absorption by water ice aerosol. Currently, its scattering is not considered. Note that a scaling parameter for the transmission spectrum is also obtained for each pixel as a byproduct, which is then used for removing the effect of transmission by applying the BeerLambert law. The computation is performed column-by-column due to wavelength shift caused by smile effect. The uniformity of the atmospheric transmission over each column is required. Our method works on the wavelength region over 1.0-2.6μm. Comparison of corrected spectra: We applied our algorithms on the CRISM images including the ones around the final candidates of the Mars2020 landing site, North East Syrtis, Jezero crater, and Columbia hills. The computational time for processing each image was about one and half to two hours in our environment. Fig. 1 shows some comparisons of I/F spectra corrected by our algorithm (red) and ones corrected by volcano scan method using the CRISM Analysis Toolkit (CAT) software (version 7.3.1) (blue). The empirical selection of the transmission spectra is used for CAT and photometric correction is not applied. Overall, spiky features on the CAT-corrected spectra are removed on the ones corrected by our method. These spiky features may be artifacts created in the CRISM processing pipeline (including atmospheric correction) or random noise. A depression around 2.0μm in Fig. 1(c) that seems to be the bowl-shape artifact [14] and a triplet residual in Fig. 1(a) are also clearly removed. Furthermore, absorption features such as hydration bands at 1.4 and 1.9μm are much more clearly retrieved. Fig. 1(a) shows our corrected spectrum shows much clear Alsmectite bands at 2.2μm. Our corrected spectra tend to exhibit more small fluctuations, which can be greatly minimized when spatially averaged (Fig. 1(f)), confirming that the small zig-zags on our corrected spectra are not atmospheric residual, but small random noise. Fig. 2 shows another comparisons where the significant contribution of water ice aerosol is observed. The contribution is confirmed by the kink at 1.5μm and an

MINERALS AT MAWRTH VALLIS. J. K. Miura1, J. L. Bishop2, J. M. Danielsen2,3, A. M. Sessa4, Y. Itoh5, M. Parente5, J. J. Wray4, and G. A. Swayze6. 1Brown University (Providence, RI: jasper_miura@brown.edu), 2SETI Institute (Mountain View, CA: jbishop@seti.org), 3San Jose State University (San Jose, CA), 4Georgia Institute of Technology (Atlanta, GA), 5University of Massachusetts (Amherst, MA), 6US Geological Survey (Denver, CO).

Introduction: Mt. Sharp in Gale crater contains ~5 km of strata that may span the transition in Mars’ climate from conditions favoring clay formation to those favoring sulfate formation [2,6,7,8]. Continued in situ investigation of these units by the Curiosity rover in the coming months and years will provide a detailed view of the nature of this mineral transition and associated processes. Morphologic and mineralogical attributes that occur throughout Mt. Sharp are of interest, as these imply spatially extensive processes that occurred within Gale, not simply local conditions. One such feature is transition in tonality of strata [2] that overlie the lacustrine sequence observed thus far by Curiosity [1]. Here we present mineral maps (Fig. 1) of newly processed CRISM images to examine the broad context of the mineralogical stratigraphy of Mt. Sharp and the nature of this lightdark-light transition in particular (Fig. 2). We find that this transition corresponds to the changes in mineralogy, primarily sulfate hydration state, and appears to be present throughout Mt. Sharp. In situ observations are necessary to determine whether this transition is primary or diagenetic, and the Curiosity rover is likely to cross this transition as it moves stratigraphically up through the sulfate section that superposes the clay-bearing unit (Fig. 1b). Methods: Stratigraphic variations in mineralogy are documented using nine CRISM images (Fig. 1a) processed using a corrupted linear spectral mixing model (CLMM), which reduces the effects of instrument noise and atmosphere [3,4]. Mineralogical mapping is performed using CRISM spectral parameters [5], and mineral detections are verified by manual inspection of individual pixels. HiRISE and CTX data are also used to assess the geologic context and geomorphology. Results: Sulfate spectra (Fig. 2) are consistent with Mg sulfate, and both monohydrated (most consistent with kieserite) and polyhydrated (PHS) forms are present. The stratigraphic boundary between sulfate hydration states is sharp at the orbital scale. The entire mound appears to be affected by a throughgoing transition in sulfate hydration state and/or composition (Figs. 1,2). The lower light-toned region is spectrally dominated by PHS; in some locations it is spectrally mixed with Fe-Mg smectite and in some localities it transitions laterally to clay-dominated spectral signatures but is not spectrally mixed (Fig. 1 b,c). This light-toned zone transitions to the dark toned zone that contains kieserite, which is then overlain by the more spectrally pure PHS. This upper light-toned PHS area hosts the “marker bed” [2]. Discussion and implications: The light-dark-light tonality transition observed around Mt. Sharp corresponds to the same mineralogical transition, suggesting that this is a coherent package of strata that is enriched in sulfates with lesser clay that exists throughout the entire mound. Because of their high solubility, it is unlikely that the Mg sulfates represent a detrital component. As presumably authigenic phases, they represent chemical precipitates formed by primary (direct precipitation from water column) or diagenetic (early or late) processes. Determining whether they are primary or diagenetic requires roverscale observations of whether the transitions conform to stratigraphic boundaries, the textural attributes of