45th Lunar and Planetary Science Conference (2014) 1431.pdf

QUANTITATIVE COMPOSITION AND GRANULOMETRY OF AEOLIAN BEDFORMS IN ENDEAVOUR AND GALE CRATERS INFERRED FROM VISIBLE NEAR-INFRARED SPECTRA. Mathieu G.A. Lapotre1, Bethany L. Ehlmann1,2, Raymond E. Arvidson3. 1Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA, USA. 2Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA, 3Department of Earth & Planetary Sciences, Washington University in St. Louis, MO, USA.

Introduction: Modern Mars is a wind world. Its ing Spectrometer for Mars (CRISM) visible near- surface hosts a variety of aeolian features, such as line- infrared spectra (VISIR). The goal of this study is to ar, barchan and star dunes, ripples, granule ripples, compare inversions made from orbit to ground truth yardangs and ventifacts [1]. Even though active sand provided by instruments aboard Opportunity at En- transport was observed at the surface [2], it is not clear deavour Crater, Terra Meridiani and Curiosity in Gale whether all of the preserved aeolian bedforms are ac- crater. tive. In particular, transverse aeolian ridges have been We use Hapke’s bidirectional reflectance spectros- suggested to be remnant dunes that formed under past copy theory [6] to invert for optical constants of miner- climatic conditions [3]. als from laboratory spectra [e.g., 7, 8]. These are used Sand transport is largely controlled by the size and to compute single scattering albedos of mineral the density of the grains [4]. Moreover, dunes and rip- endmember components of varying grain sizes. We use ples form in unimodally distributed sand particles from an atmospheric radiative transfer approach, DISORT different instabilities, and the wavelengths of these [9], to correct the CRISM spectra for the effects of the different bedforms do not have the same dependence Martian atmosphere. Examination of spectra corrected on grain density. In contrast, granule ripples typically using the standard volcano scan method [10] showed have bimodal grain size distributions [5]. A quantita- undesired atmospheric absorptions and albedo varia- tive inversion of the mineralogy and granulometry of tions due to uncorrected-for aerosols. Consequently, aeolian bedforms would thus provide great insights on single scattering albedo spectra were retrieved by sim- their modern dynamics, but also potentially on Martian ultaneously modeling atmospheric gases, aerosols, and paleoclimates and distances of transport from the sedi- surface scattering and thermal emission, using the ment source areas. Hapke bidirectional scattering function [11]. Input spectra were typically averages of 10s of pixels and, prior to modeling, input spectra were smoothed to re- move spurious single channel excursions from the con- tinuum. We chose the mineral endmembers to include from a careful qualitative inspection of the spectra to identi- fy obvious absorption features present. These com- bined with common minerals found in basalt were used to construct a parsimonous minimum set for each site (Table 1). We also included an amorphous phase (ba- saltic glass, relatively flat and dark spectrum) in some of the endmember suites. Mineralogy and granulometry of bedforms were determined by use a downhill sim- plex algorithm to invert for their abundances and grain sizes, optimizing for the lowest RMS error between measured and modeled single scattering albedo . Merid1 olivine (Fo91), pigeonite, augite, labra- (n=6) dorite, magnetite, hematite Merid2 olivine (Fo91), pigeonite, augite, labra-

(n=7) dorite, magnetite, hematite, basaltic glass Figure 1: (top) Barchan dunes in Endeavour crater (HiRISE Gale1 olivine (Fo60), pigeonite, augite, labra- image ESP_032006_1775). (bottom) Dark dunes in Gale (n=5) dorite, magnetite crater (HiRISE image PSP_009294_1750). Gale2 olivine (Fo60), pigeonite, augite, labra- Methods: We invert for mineral abundances and (n=6) dorite, magnetite, basaltic glass grain sizes of aeolian dunes in Endeavour and Gale Table 1: Endmembers used to model the VISIR spectra of craters (Figure 1) from Compact Reconnaisance Imag- aeolian bedforms at Endeavor and Gale craters. 45th Lunar and Planetary Science Conference (2014) 1431.pdf

grain sizes using five and six endmembers, with and without amorphous phases (Table 1). The fit is slightly better when an amorphous phase is included in the modeling, and the best fit abundance for the amorphous phase is 24 wt%. At Gale in a sand ripple, the presence of a large portion of X-ray amorphous material was detected by Curiosity (27±14 wt% according to [13] and up to 45 wt% according to [14]). The best fit abundances as calculated with the amorphous phase, rescaled to modeled crystalline phases only, are com- pared with those measured by APXS and CheMin [14] on Curiosity at Rocknest (Table 3). Figure 2 shows a comparison between the CRISM spectra and the best fit models at Meridiani and Gale.

Ol Low Ca Px High Ca Px Plag. Gale1 4.4 45.1 0.1 50.4 Gale2 4.2 20.1 32.4 43.3 Rocknest13 24.4 15.1 15.9 44.5 Table 3: Inverted mineral abundances (wt%) compared to those determined by APXS and CheMin. These numbers were renormalized to only take into account the modeled Figure 2: CRISM and best fit spectra from barchans dunes crystalline phases. in Endeavour crater (Merid1 and Merid2, Figure 1 top) and dark dunes in Gale crater (Gale1 and Gale2, Figure 1 bot- Implications: Mineral abundances can be inverted tom). within reasonable errors if the effects of the atmos-

phere are properly accounted for. DISORT corrected Results: Endeavour. We invert the CRISM spec- spectra yield mineral abundances and grain sizes that trum (from CRISM frame FRT00019E9C) using six are in very good agreement with those observed by and seven endmembers (without and with an amor- Opportunity at Meridiani. phous phase, Table 1). Not including hematite leads to Nevertheless, olivine abundances seem to be con- large amounts of magnetite to compensate for the low sistently underestimated. Moreover, the suite of albedo. The best fit abundances as calculated without endmembers used in the modeling, and in particular the the amorphous phase, rescaled to Fe-bearing mineral use of an amorphous phase, may significantly alter the abundances only, compare well with those measured in modeled mineralogy. Discriminating between two Berry soils by the Mossbauer spectrometer [12] on endmember suites is not trivial as the solution to the Opportunity in the surroundings (Table 2). Berry soils inverse problem is highly non-unique. are surface lag deposits, and were observed at the crest

of ripples on sols 367-373. The inverted grain sizes References: [1] Ward A.W. (1985) Journal of Geophysical Re- (D ≈ 1400 and 870 μm) compare very well to esti- 50 search, 90, B2, 2038–2056. [2] Bridges N.T. et al. (2012) Nature, mates made using the Microscopic Imager (D ≈ 1100 50 485, 339–342. [3] Zimbelman J.R. (2010) Geomorphology, 121(1- μm) [5]. 2), 22–29. [4] Bagnol R.A. (1941) ISBN: 9780486141190. [5] Ol Px Mt Hem Jerolmack D.J. et al. (2006) Journal of Geophysical Research, 111, Merid1 9.1 12.6 1.4 77 E12S02. [6] Hapke B. (1981) Journal of Geophysical Research, 86, Merid2 1.1 61.2 12.0 25.7 12 B4, 3039–3054. [7] Clark, R.N. et al. (2007), USGS Digital Spectral Berry 15.6 17.8 3.3 63.3 Library splib06a, U.S. Geol. Surv. Data, 231. [8] Ehlmann, B.L., Table 2: Inverted mineral abundances (wt%) of Fe-bearing J.F. Mustard, F. Poulet, T. Hiroi. LPSC 42, 2011, abs. #1704. [9] phases compared to those determined by Mossbauer spec- Stamnes K.S. et al. (1988) Applied Optics, 27, 2502–2509. [10] 12 trometry (Berry – Moessberry subclass). These numbers Murchie S.L. (2009) Journal of Geophysical Research, 114, were renormalized to only take into account the modeled Fe- E00D07. [11] Arvidson et al. (2014), Science (in press), January 24 bearing phases. 2014 issue. [12] Morris R.V. et al. (2006) Journal of Geophysical Research, 111, E12S15. [13] Bish D.L. et al (2013), Science, 341. Results: Gale. We invert spectra (from CRISM [14] Blake D.F. et al. (2013) Science, 341. frame FRT0000B6F1) for mineral abundances and