Portal de Periódicos da CAPES

Temporal observations of bright soil exposures at Gusev crater, Mars

2011; American Geophysical Union; Volume: 116; Linguagem: Inglês

10.1029/2010je003683

2156-2202

M. S. Rice, J. F. Bell, E. A. Cloutis, J. J. Wray, K. E. Herkenhoff, R. Sullivan, J. R. Johnson, R. B. Anderson,

Space Exploration and Technology

Journal of Geophysical Research: PlanetsVolume 116, Issue E7 Free Access Temporal observations of bright soil exposures at Gusev crater, Mars M. S. Rice, M. S. Rice [email protected] Department of Astronomy, Cornell University, Ithaca, New York, USASearch for more papers by this authorJ. F. Bell III, J. F. Bell III Department of Astronomy, Cornell University, Ithaca, New York, USASearch for more papers by this authorE. A. Cloutis, E. A. Cloutis Department of Geography, University of Winnipeg, Winnipeg, Manitoba, CanadaSearch for more papers by this authorJ. J. Wray, J. J. Wray Department of Astronomy, Cornell University, Ithaca, New York, USASearch for more papers by this authorK. E. Herkenhoff, K. E. Herkenhoff Astrogeology Science Center, U.S. Geological Survey, Flagstaff, Arizona, USASearch for more papers by this authorR. Sullivan, R. Sullivan Department of Astronomy, Cornell University, Ithaca, New York, USASearch for more papers by this authorJ. R. Johnson, J. R. Johnson Astrogeology Science Center, U.S. Geological Survey, Flagstaff, Arizona, USASearch for more papers by this authorR. B. Anderson, R. B. Anderson Department of Astronomy, Cornell University, Ithaca, New York, USASearch for more papers by this author M. S. Rice, M. S. Rice [email protected] Department of Astronomy, Cornell University, Ithaca, New York, USASearch for more papers by this authorJ. F. Bell III, J. F. Bell III Department of Astronomy, Cornell University, Ithaca, New York, USASearch for more papers by this authorE. A. Cloutis, E. A. Cloutis Department of Geography, University of Winnipeg, Winnipeg, Manitoba, CanadaSearch for more papers by this authorJ. J. Wray, J. J. Wray Department of Astronomy, Cornell University, Ithaca, New York, USASearch for more papers by this authorK. E. Herkenhoff, K. E. Herkenhoff Astrogeology Science Center, U.S. Geological Survey, Flagstaff, Arizona, USASearch for more papers by this authorR. Sullivan, R. Sullivan Department of Astronomy, Cornell University, Ithaca, New York, USASearch for more papers by this authorJ. R. Johnson, J. R. Johnson Astrogeology Science Center, U.S. Geological Survey, Flagstaff, Arizona, USASearch for more papers by this authorR. B. Anderson, R. B. Anderson Department of Astronomy, Cornell University, Ithaca, New York, USASearch for more papers by this author First published: 27 January 2011 https://doi.org/10.1029/2010JE003683Citations: 18AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Abstract [1] The Mars Exploration Rover Spirit has discovered bright soil deposits in its wheel tracks that previously have been confirmed to contain ferric sulfates and/or opaline silica. Repeated Pancam multispectral observations have been acquired at four of these deposits to monitor spectral and textural changes over time during exposure to Martian surface conditions. Previous studies suggested that temporal spectral changes occur because of mineralogic changes (e.g., phase transitions accompanying dehydration). In this study, we present a multispectral and temporal analysis of eight Pancam image sequences at the Tyrone exposure, three at the Gertrude Weise exposure, two at the Kit Carson exposure, and ten at the Ulysses exposure that have been acquired as of sol 2132 (1 January 2010). We compare observed variations in Pancam data to spectral changes predicted by laboratory experiments for the dehydration of ferric sulfates. We also present a spectral analysis of repeated Mars Reconnaissance Orbiter HiRISE observations spanning 32 sols and a textural analysis of Spirit Microscopic Imager observations of Ulysses spanning 102 sols. At all bright soil exposures, we observe no statistically significant spectral changes with time that are uniquely diagnostic of dehydration and/or mineralogic phase changes. However, at Kit Carson and Ulysses, we observe significant textural changes, including slumping within the wheel trench, movement of individual grains, disappearance of fines, and dispersal of soil clods. All observed textural changes are consistent with aeolian sorting and/or minor amounts of air fall dust deposition. 1. Introduction and Background [2] The Mars Exploration Rover (MER) Spirit has excavated subsurface deposits of sulfate- and/or silica-rich materials in eighteen locations in the Columbia Hills of Gusev crater, Mars. These deposits occur as anomalously high albedo soil exposures within the rover's wheel tracks, with white and/or yellow hues that vary over small length scales. Most of these deposits have been brought up from depths of ∼10 cm by the dragging motion of Spirit's inoperative right front wheel, its actuator having failed on sol 779 of the mission. In situ measurements with Spirit's Alpha Particle X-ray Spectrometer (AXPS) have revealed that these soils have the highest sulfur concentrations (up to 38 wt.% SO3) of any materials yet observed by either of the MER rovers [Ming et al., 2006, 2008; Arvidson et al., 2010]. Mössbauer (MB) spectrometer data suggest the presence of ferric-bearing sulfates [e.g., Gellert et al., 2006; Morris et al., 2006, 2008], and the soils' distinctive Panoramic Camera (Pancam) and Miniature Thermal Emission Spectrometer (Mini-TES) spectra are consistent with a heterogeneous mixture of hydrated ferric sulfates [e.g., Johnson et al., 2007; Lane et al., 2008; Parente et al., 2009]. [3] At the soil exposure called Gertrude Weise, the highest albedo soil observed by Spirit, APXS measurements revealed a nearly pure silica composition (∼98 wt.% SiO2 when corrected for dust contamination) and minor TiO2 [Squyres et al., 2008]. Mini-TES measurements of this soil are consistent with the presence of hydrated amorphous silica [Squyres et al., 2008; S. W. Ruff et al., Characteristics, distribution, and significance of opaline silica in Gusev crater, manuscript in preparation, 2010], and a ∼1 μm absorption feature in Pancam spectra indicates the presence of H2O and/or OH [Rice et al., 2010a]. While Gertrude Weise is the only nearly pure silica soil yet discovered, the sulfate-bearing exposures at regions called Tyrone and Ulysses also contain a component enriched in SiO2 [Squyres et al., 2008; Wang et al., 2008; Arvidson et al., 2010; S. W. Ruff et al., manuscript in preparation, 2010]. [4] The mineralogy, geochemistry, spatial variability, and geologic setting of the bright subsurface soils suggest that they likely formed in a hydrothermal environment from fumarolic condensates, precipitation from geothermal waters, and/or leaching of local basaltic rocks [e.g., Squyres et al., 2008; Yen et al., 2008; Morris et al., 2008]. The time of hydrothermal activity within Gusev crater, when the sulfate- and silica-rich soils would have formed, is not well constrained; however, the processes that have sorted, transported and modified the bright soil material may be ongoing. Indeed, the layered structure of the Ulysses soil, where soluble ferric sulfate species appear to be segregated below a layer with a less soluble silica-rich component, all below an insoluble, cemented hematite and calcium sulfate surface layer, suggests an ongoing pedogenic modification from downward migration of soluble materials by gravity-driven water [Arvidson et al., 2010]. [5] Some ferric sulfate species that have been proposed as possible components of the Spirit bright soil exposures, such as ferricopiapite [Fe2/32+Fe43+(SO4)6(OH)2 · 20(H2O)] and fibroferrite [Fe3+(SO4)(OH) · 5H2O] [Johnson et al., 2007; Lane et al., 2008; Parente et al., 2009], are known to be unstable under current Martian surface conditions [Chipera et al., 2007; Cloutis et al., 2008; Freeman et al., 2009; A. Wang et al., Ferric sulfates on Mars: Analysis of Pancam spectral data from the Spirit Rover supported by laboratory investigations, submitted to Journal of Geophysical Research, 2010]. If the relative humidity environment in the near subsurface differs from that of the surface (i.e., is buffered by hydrated minerals or ground ice), it is possible that buried soils are not in equilibrium with surface conditions at the Spirit site, and that some minerals could undergo dehydration and/or phase changes after exhumation by the rovers' wheels. [6] Spectral changes are known to accompany mineralogic changes among ferric sulfates and hydrated silica phases, resulting in modifications to their visible color [Cloutis et al., 2008; Rice et al., 2010b; A. Wang et al., submitted manuscript, 2010]. Wang et al. [2008] suggested that such modifications occurred for the Tyrone soil, and they described changes in the blue-to-red (432 to 753 nm) spectral slope in Pancam observations spanning a period of ∼150 sols. Textural changes, such as shrinkage and cracking, are also expected to accompany the dehydration of hydrated ferric sulfates based on laboratory experiments [e.g., Cloutis et al., 2008]. [7] To test the hypothesis that the Spirit bright soils have undergone mineralogic changes upon exposure to the Martian surface, we have performed a detailed analysis of Pancam multispectral images at the four sites where repeat observations have been made: Tyrone (sols 790–1062), Gertrude Weise (sols 1158–1198), Kit Carson (sols 1864–1866), and Ulysses (sols 1888–2132). We include a multispectral analysis of repeated Mars Reconnaissance Orbiter (MRO) High Resolution Imaging Science Experiment (HiRISE) observations of these soils for sols 1935 to 1967 and compare the analyses to Pancam observations. We also compare temporal observations in both the Pancam and HiRISE multispectral data to the variations predicted from laboratory experiments on ferric sulfate minerals exposed to current Martian surface conditions. [8] For the Pancam images with high enough spatial resolution to resolve soil texture (at Kit Carson and Ulysses), we describe textural changes observed with time and discuss whether they are indicative of mineralogic changes, aeolian sorting, and/or dust deposition within the soil trenches. Although detailed studies of the wind-driven mobility of basaltic sand and air fall dust have been performed at Gusev crater [e.g., Greeley et al., 2006; Sullivan et al., 2008], no assessment has yet been made of the mobility of the bright soils, which we address here. We also use the available repeated observations of these soils from Spirit's Microscopic Imager (MI) to observe and document additional textural changes, allowing us to characterize temporal changes in the bright soils from the microscopic to orbital scale. 2. Methods Pancam Observations 2.1.1. Pancam Instrument and Calibration [9] The Pancam instrument consists of two cameras at a 30 cm stereo separation, each using a 1024 × 1024 pixel charge-coupled device (CCD) detector with 0.27 mrad per pixel resolution [Bell et al., 2003, 2006]. Pancam's 13 narrowband geology filters cover 11 specific wavelengths in the visible and near infrared (432 to 1009 nm; Table 1). Some of the images used in this study were acquired using lossy wavelet-based ("ICER") compression [Maki et al., 2003]. Based on prelaunch tests, compression effects on radiometric precision at the typical compression bit rates employed using the ICER compressor were estimated to be less than 1% [Bell et al., 2006]. Table 1. Pancam Filter Data Filter Effective Wavelengtha (nm) Band Passa (nm) Camera L7 432 32 LEFT R1 436 37 RIGHT L6 482 30 LEFT L5 535 20 LEFT L4 601 17 LEFT L3 673 16 LEFT L2 753 20 LEFT R2 754 20 RIGHT R3 803 20 RIGHT R4 864 17 RIGHT R5 904 26 RIGHT R6 934 25 RIGHT R7 1009 38 RIGHT a From Bell et al. [2003]. [10] We use near-simultaneous observations of the Pancam calibration target, as well as prelaunch calibration and modeling, to derive estimated reflectances of the scene relative to the standard reflectance materials on the calibration target [Bell et al., 2003, 2006]. To correct for dust contamination of the calibration target, Pancam data are calibrated using a two-layer radiative transfer model [Sohl-Dickstein et al., 2005; Bell et al., 2006; Kinch et al., 2007]. The Pancam reflectance products are called "IOF" images, where IOF (also known as the "radiance factor" or "I over F") [Hapke, 1993], is defined as the ratio of the bidirectional reflectance of a surface to that of a normally illuminated, perfectly diffuse surface. Dividing the Pancam IOF images by the cosine of the solar incidence angle at the time of each observation gives the relative reflectance R* [Reid et al., 1999; Bell et al., 2006], also known as the "reflectance factor" or "reflectance coefficient" [Hapke, 1993]. Bell et al. [2006] have estimated the relative filter-to-filter uncertainties in R* to be 1–5%, and the absolute reflectance levels to be accurate to within ∼10%. Because diffuse component corrections are relatively minor at the solar incidence angles of our data set (6°–40°; Table 2) [Johnson et al., 2006], we did not correct for diffuse illumination. Table 2. Pancam Full Filter Imaging Sequences Used for Spectral Characterization of Bright Soilsa Sol Siteb Positionb Sequence ID Local True Solar Timec Imaging Duration (s) Incidence Anglec (deg) Emission Anglec,d (deg) Phase Anglec (deg) Sols After Exposure Taue Saturated Filtersf Tyrone 790 126 142 P2531 11:56:05 236 26.9 28.2 41.5 8 0.342 L67 864 128 0 P2547 12:37:47 231 38.3 5.8 64.0 82 0.256 L6 922 128 0 P2552 12:02:10 288 39.9 5.4 71.6 140 0.295 L6 959 128 0 P2560 12:07:35 249 38.9 5.4 70.7 177 0.302 — 982 128 0 P2566 12:10:36 211 37.0 5.4 70.5 200 0.312 — 1005 128 0 P2576 12:23:12 353 34.7 5.4 68.3 223 0.240 — 1036 128 20 P2585 13:03:39 200 33.1 6.1 60.2 254 0.381 — 1062 128 115 P2596 12:30:12 412 25.1 5.6 70.5 280 0.883 — Gertrude Weise 1158 128 1318 P2581 12:29:46 274 7.3 12.0 74.9 10 0.800 — 1187 129 112 P2533 12:16:34 261 6.4 19.6 64.6 39 0.939 — 1198 129 140 P2539 12:11:16 394 7.4 42.2 55.1 50 0.943 — Kit Carson 1864 136 614 P2562g 12:32:30 183 10.1 48.0 31.9 3 1.187 — 1866 136 614 P2555 12:43:18 286 12.3 57.3 20.7 5 1.103 — Ulysses 1888 137 130 P2559 14:01:10 878 30.0 65.6 42.1 2 0.875 — 1892 137 178 P2560 13:07:52 254 18.7 54.6 34.3 6 0.823 — 1894 137 182 P2560 13:04:35 252 18.1 54.6 34.1 8 0.792 — 1897 137 195 P2562 12:24:41 262 11.6 58.3 24.6 11 0.768 — 1933 137 249 P2382h 13:32:11 4372 23.8 58.5 40.4 47 0.470 — 1982 137 249 P2547 13:15:13 254 18.3 66.3 26.3 96 0.377 R1 2019 137 249 P2392h 11:41:49 4360 6.1 58.7 32.8 142 1.452 L567R1i a Feature names are informal and not formally accepted by the International Astronomical Union. b The surface coordinate frames utilized by MER [Maki et al., 2003]. c At the starting time of the Pancam observation and for the center of the image. d The INSTRUMENT_ELEVATION parameter stored in the image label. e Visible optical depth observed with Pancam's L8 filter [Lemmon et al., 2004]. f Where clusters of pixels in the raw image equal 4095 DN. g This sequence includes a reduced filter set: L257R12467. h These sequences contain five pointings to cover the extent of the soil exposure. i The saturated pixels in this sequence were over reflective portions of the spacecraft, not the soil exposures. 2.1.2. Extraction of Pancam Spectra [11] We have acquired visible to near-infrared (Vis-NIR) spectra of the soil targets by manually selecting pixels from common regions of interest (ROIs) in the right and left camera data sets and averaging the R* values of those regions for each filter. We have chosen ROIs for each target that include as many pixels as possible (to minimize instrumental artifacts and statistical noise), excluding shadowed regions. We also exclude pixels that are potentially approaching saturation to ensure that we only use data within the demonstrated linearity of the Pancam instrument [Bell et al., 2003]. For the Tyrone, Kit Carson, and Ulysses soils, we have selected ROIs from both the "yellow" and "white" hue separations (indicated by the yellow and black outlines, respectively, in Figures 2, 6, 9 and 12). In all images, we also extracted representative spectra from other scene elements (undisturbed dusty soil and disturbed dark soil, indicated by the red and brown outlines). For the blue (432 and 436 nm) and red (753 and 754 nm) stereo filters, we have used the R* values acquired by the left camera (432 and 753 nm). The error bars represent the variance among the selected ROI pixels, rather than from the formal instrumental noise (which is generally much lower) [Bell et al., 2006]. 2.1.3. Comparisons of Spectral Parameters [12] To quantify spectral variations with time, we have chosen four spectral parameters to compare between image sequences (Table 3). Three of these parameters characterize distinctive regions of the Vis-NIR spectra of ferric sulfates: (1) the 753 to 432 nm ratio, which quantifies the overall "redness" of the sample's color; (2) the 535 nm band depth, which indicates the strength of Fe3+ absorptions near ∼550 nm; and (3) the 864 nm band depth, which quantifies the diagnostic Fe3+-related absorptions at ∼850–900 nm. Strong 535 nm and 864 nm band depths have been used in previous studies to classify the Pancam spectra of sulfate-rich soils at Gusev crater [Farrand et al., 2008; Parente et al., 2009]. We have also acquired laboratory spectra of four candidate ferric sulfates during long-term exposure to Martian surface conditions in order to characterize the temporal behavior of these spectral parameters for specific minerals (described in section 2.4). Table 3. Spectral Parameters Used in This Study Parameter Description 535 nm band depth 1 – (R*535/[(0.573 × R*432) + (0.427 × R*673)]) 753 nm to 432 nm ratio (R*753)/(R*432) 864 nm band depth 1 – (R*864/[(0.387 × R*753) + (0.613 × R*934)]) 934 to 1009 nm ratio (R*934)/(R*1009) [13] The fourth spectral parameter that we monitor is the 934 to 1009 nm ratio. This parameter has previously been used to detect silica-rich materials along the rover's traverse in Gusev crater [Wang et al., 2008; Rice et al., 2010a]. A high 934 to 1009 nm ratio (or steeply negative 934 to 1009 nm slope) is attributed to a combinational mode of H2O (2v1 + v3) and/or an OH overtone (3v) near ∼970–1000 nm, and its magnitude is sensitive to the amount of OH/H2O present in the mineral and/or adsorbed on mineral grains [Rice et al., 2010a]. For a detailed explanation of the attribution of a negative 934 to 1009 nm slope to water and/or OH, as opposed to other factors that can affect the spectral slope (such as viewing geometry, dust coatings or an uncorrected diffuse component), we refer the reader to Rice et al. [2010a]. Because the depths of all H2O and OH absorption features in silica decrease during dehydration experiments [e.g., Cloutis et al., 2008; Rice et al., 2010b], an observed decrease in the magnitude of the 934 to 1009 nm ratio with time would be consistent with the soils dehydrating upon exposure. [14] We estimate the uncertainty in the spectral parameters by propagating the uncertainty in R* through the calculations shown in Table 3 by standard error analysis (e.g., the fractional uncertainty in the 753 to 432 nm ratio is calculated as the sum in quadrature of the fractional uncertainties in R*753 and R*432). 2.1.4. Correction for Minor Dust Contamination of Camera Optics [15] All images used in this study exhibit some effects attributed to minor heterogeneous dust contamination on the front sapphire windows of the Pancam instruments. The effect is too small to be meaningful in individual filter images, but it is enhanced in the ratio and band depth maps used in this study, where top-to-bottom brightness gradients of up to 5–10% have been observed in some scenes. This gradient does not affect our spectral comparisons when the ROIs are selected from the same region of the CCD in each image (as is the case for the Tyrone and Ulysses observations), but it must be accounted for when we compare spectra from different locations in each image. [16] In this study we account for dust contamination effects on the optics by considering the spectral parameters as ratios of the bright soil targets to dusty, undisturbed soils imaged in similar regions of the field of view (FOV), rather than as absolute values. We assume that the spectra of undisturbed, dust-covered surface soils should not change between image sequences under the same or nearly the same lighting conditions. Because undisturbed soils are primarily of basaltic composition [e.g., Sullivan et al., 2008; Morris and Klingelhöfer, 2008] and have likely been exposed to surface conditions for an extensive time, we assume that they are in equilibrium and do not undergo mineralogic changes during our short observation periods. Thus we can enhance our ability to detect changes in the sulfate- and silica-rich soils by observing them relative to unchanging, undisturbed materials. The uncertainty in these ratios is estimated by propagating the uncertainties in the values of the spectral parameters for the bright and undisturbed soils, which is a conservative but prudent overestimate of the true instrumental error. 2.1.5. Coregistration of Multiple Image Sequences [17] To monitor pixel-to-pixel changes between Pancam image sequences, we have coregistered the image sequences from Tyrone that were acquired from nearly identical positions. During Spirit's second winter campaign, the rover was parked at Low Ridge in view of the Tyrone soil exposure at Low Ridge (Figure 1) and made five Pancam 13-filter observations from sols 864–1005 (Table 2). We have shifted each of those image sequences by an integer number of pixels in the x and y directions to coregister all images for direct spectral comparisons. In all cases the required shifts were small (<30 pixels, <3% of the FOV), and thus there were no significant variations in emission angle geometry among these observations. The effect of coregistration on radiometric precision is minor given that the shifts were integer-only and the images were not resampled. Tyrone was the only soil target from which multiple Pancam observations were acquired from similar enough perspectives to perform a coregistration. Figure 1Open in figure viewerPowerPoint Spirit traverse map of the Home Plate vicinity (sols 743–2185) showing locations of the soil exposures discussed in this work and the Spirit rover embedded in the Ulysses soil as of sol 2185. Traverse is shown over a subframe of HiRISE image ESP_0013499_1650_red. [18] To illustrate trends in temporal variations of the spectral parameters listed in Table 3, we have calculated the percent change of each parameter for each pixel in the coregistered Tyrone image set. For every pixel, we have performed a linear regression fit to the spectral parameter versus sol data, and we calculate the percent difference between the values of the linear function at the first and last sols of observation. We use the resulting "percent change maps" in addition to spectra from selected ROIs (section 2.1.2) to look for systematic spectral variations relative to the rest of the scene in the Tyrone observations. HiRISE Observations 2.2.1. HiRISE Instrument and Calibration [19] Orbital monitoring provides an independent view of the spectral evolution of bright soils exposed by Spirit, and enables monitoring to continue even after the rover has driven out of Pancam imaging range. Specifically, the MRO HiRISE camera [McEwen et al., 2007] has a sufficiently small pixel scale (∼26.5 cm at the latitude of the Columbia Hills) to enable detection of the largest exposures of bright soils. HiRISE has three color filters: blue-green (BG; λeff = 502 ± 157 nm), red (RED; λeff = 686 ± 267 nm), and near-infrared (IR; λeff = 878 ± 143 nm) [McEwen et al., 2007; Delamere et al., 2010], allowing detection of color changes over a limited portion of the spectrum. BG and IR data can only be acquired for a central strip covering 20% of each HiRISE RED image. In this study we compare the two observations with color coverage of the Home Plate vicinity during the period over which the Kit Carson soils were visible from orbit (Table 4). Table 4. HiRISE Observations of Kit Carson Soils Image ID Sol Local True Solar Time RED Pixel Scale (cm) IR/BG Pixel Scale (cm) Incidence Angle (deg) Emission Angle (deg) Phase Angle (deg) Atmospheric Opacity (τ) ESP_013499_1650 1935 14:39 26.9 53.8 38.7 10.1 48.5 0.51 ESP_013921_1650 1967 14:28 26.5 26.5 35.8 8.5 27.4 0.39 [20] The current state of radiometric calibration for HiRISE was recently described by Delamere et al. [2010]. Uncertainties in absolute I/F values were estimated as ±20% absolute, whereas relative errors within a HiRISE image are typically only ∼2% (∼0.5% within a single CCD channel). This high relative precision can be exploited for change detection studies by observing not just the change in absolute I/F of a feature of interest, but also the change in I/F ratio between a feature of interest and an undisturbed homogeneous area of adjacent terrain within the same CCD channel. This technique has been used successfully to monitor color and albedo changes of 20–60% attributed to sublimation of subsurface water ice exposed by impact cratering [Byrne et al., 2009]. [21] To correct for the spectral contribution from atmospheric aerosols, which can either brighten or darken the apparent I/F of a given pixel depending on the surface albedo, we use the method described by Portyankina et al. [2010], which has previously been tested on the set of images used in this study and demonstrably reduces I/F discrepancies between them. We have performed our analysis with and without the Portyankina et al. dust correction, and find that the correction changes relative I/F values in the RED and IR filters by <1% and in the BG filter by 3–5% for both observations. The estimated added uncertainty in I/F due to the dust correction is ∼1%. 2.2.2. Comparison of HiRISE Color Data [22] We resampled each HiRISE image (Table 4) to 50 cm/pixel so that they could be compared directly. This pixel size is large compared to the bright soil exposures, suggesting that most or all "bright soil" pixels are actually spatial mixtures of bright soil plus adjacent undisturbed soils and/or rocks (which are universally darker). We therefore measured I/F only for the brightest pixel within each soil exposure in a given image in order to minimize spectral contamination from other materials. There was no saturation in any HiRISE band for the soils observations. To reduce the effects of differing illumination and observation geometry between images (Table 4), we divided these bright soil I/F values by those extracted from nearby areas of the surface that appear relatively uniform and were not disturbed by Spirit. We tested several of these denominators (each an average of ∼100 pixels) to confirm that our results do not depend on the choice of denominator. [23] We measured the BG, RED and IR relative I/F for the Kit Carson, Gertrude Weise and Tyrone soils, which had been exposed to the surface for 74, 787 and 1153 sols, respectively, when the first HiRISE observation was made. For additional points of comparison, we extracted single-pixel I/F values for four nearby bright "calibration" outcrops undisturbed by Spirit, which would not be expected to undergo intrinsic spectral changes. We also measured the ratios of relative RED to relative BG for comparison with Pancam red to blue ratios (753 nm to 432 nm), as well as relative IR to relative RED. Microscopic Imager Observations [24] The Microscopic Imager (MI) is a fixed focus camera mounted on the instrument arm, with the same 1024 x 1024 CCD as the other MER cameras [Herkenhoff et al., 2003]. The MI acquires panchromatic images at a scale of 31 microns/pixel over a broad spectral range (400 to 700 nm). The MI acquires images using solar or diffuse skylight illumination of the target surface. The MI focal section merges used in this study combine the best focused parts of images acquired at multiple distances from the target, separated by the 3 mm depth of field of the MI. Some MI images were merged with Pancam color data using the approach described by Herkenhoff et al. [2006]. Details of all MI observations used in this study are provided in Table 5. Table 5. Microscopic Imager Observations Soil Exposure Name Target Name Sol Starting Image ID Relation to MB Sols After Exposure Kit Carson/John Wesley Powell John Wesley Powell 1863 2M291757231 post MB 3 Ulysses Sackrider 1922 2M296987659 post MB 36 1925 2M