Geological characterization of remote field sites using visible and infrared spectroscopy: Results from the 1999 Marsokhod field test

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 106, NO. E4, PAGES 7683-7711, APRIL 25, 2001

Geological characterization of remote field sites using visible and infrared spectroscopy: Results from the 1999 Marsokhod field test

Jeffrey R. Johnson, 1 Steven W. Ruff, 2 Jeffrey Moersch, 3'4 Ted Roush, 3 Keith Horton, s Janice Bishop, 3 Nathalie A. Cabrol, 3 Charles Cockell, 3 Paul Gazis, 3 Horton E. Newsom, 6 and Carol Stoker 3

Abstract. Upcoming Mars Surveyor lander missions will include extensive spectroscopic capabilities designed to improve interpretations of the mineralogy and geology of landing sites on Mars. The 1999 Marsokhod Field Experiment (MFE) was a Mars rover simulation designed in part to investigate the utility of visible/near-infrared and thermal infrared field spectrometers to contribute to the remote geological exploration of a Mars analog field site in the California Mojave Desert. The experiment simultaneously investigated the abilities of an off-site science team to effectively anfilyze and acquire useful imaging and spectroscopic data and to communicate efficiently with rover engineers and an on-site field team to provide meaningful input to rover operations and traverse planning. Experiences gained during the MFE regarding effective communication between different mission operation teams will be useful to upcoming Mars mission teams. Field spectra acquired during the MFE mission exhibited features interpreted at the time as indicative of carbonates (both dolomitic and calcitic), mafic rocks and associated weathering products, and silicic rocks with desert varnish-like coatings. The visible/near-infrared spectra also suggested the presence of organic compounds, including chlorophyll in one rock. Postmission laboratory petrologic and spectral analyses of returned samples confirmed that all rocks identified as carbonates using field measurements alone were calc-silicates and that chlorophyll associated with endolithic organisms was present in the one rock for which it was predicted. Rocks classified from field spectra as silicics and weathered marlcs were recognized in the laboratory as metamorphosed monzonites and diorite schists. This discrepancy was likely due to rock coatings sampled by the field spectrometers compared to fresh rock interiors analyzed petrographically, in addition to somewhat different surfaces analyzed by laboratory thermal spectroscopy compared to field spectra.

1. Introduction

The 1999 Marsokhod Field Experiment (MFE) provided an opportunity to test the suitability of rover-borne visible/near- infrared and thermal infrared field spectrometers to contribute to the remote geological exploration of a Mars analog field site near Silver Lake, California, in the Mojave Desert. The planned Mars Surveyor lander missions will carry spectrometers capable of determining the mineralogy of the Martian surface in unprecedented detail [e.g., Squyres et al., 1999]. The MFE described here was the first rover mission simulation to include extensive visible and infrared

spectroscopic capabilities. It was therefore directly relevant

•United States Geological Survey, Flagstaff, Arizona. 2Department of Geology, Arizona State University, Tempe. 3NASA Ames Research Center, Moffett Field, California. 4Now at this address: Department of Geological Sciences, University of Tennessee, Knoxville. 5Hawaii Institute of Geophysics and Planetology, University of Hawaii, Honolulu. 6Institute of Meteoritics and Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque.

Copyright 2001 by the American Geophysical [Jnicm

Paper number 1999JE001149. 0148-0227/01/1999JE00114959.00

to the upcoming Mars experiments while also building on experiences learned from previous rover field experiments [e.g., Arvidson et al., 1998].

Stoker et al. [this issue] discuss this field site's relevance as an analog for potential Mars Surveyor program landings sites and the overall experimental objectives of the field test. These objectives include evaluating the ability of an off-site science team to (1) effectively use imaging and spectroscopic data obtained at a remote field location to interpret the geology and mineralogy of the area and (2) to communicate efficiently with rover engineers and the on-site field team regarding data collection and rover operations such as traverse planning.

To understand the mineralogy and geology of the field site, color images of the region surrounding the rover obtained with a stereo camera were first used to identify rocks of interest on the basis of color, morphology, and location relative to the lander and its manipulator arm. Point spectra obtained with both a visible/near-infrared (VIS/NIR) spectrometer (400-2500 nm) and a thermal infrared spectrometer (-8-14 I. tm; 1250-715 cm -•) provided independent and complementary information on the presence and/or type of carbonates, clays, mafic minerals, and rock coatings or desert varnish [cf. Guinness et al., 1997]. The visible/near-infrared spectrometer also contributed information on ferric mineralogy and hydrated minerals,

7683

7684 JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

while the thermal infrared spectrometer was also sensitive to quartz and other silicate minerals and the effects of particle size variations [e.g., Clark, 1995; Salisbury et al., 199!]. Interpretations of the mineralogy of the field site using these data were integral to formulating hypotheses regarding the rock types populating the region near the rover. This information was used by the geology/geomorphology team members to interpret the geology and stratigraphy of the field site [DeHon et al., this issue; Grin et al., this issue]. As the rover moved from one location to the next and obtained

additional images and spectra, the off-site science teams' confidence and understanding of the geologic history of the region incrementally improved.

We present here an overview of the instrumentation used, examples of point spectra, and acquisition of spectral image cubes obtained with both point spectrometers. We also present interpretations of the mineralogy and subsequent classifications of rock types made by the off-site team using only spectral data acquired during the main MFE mission. These preliminary interpretations are then compared to postmission ground-truth laboratory spectroscopic and petrologic analyses of returned samples. This is followed by a commentary on the utility of the instrumentation and techniques that were used during the field test and suggestions for future rover mission field tests and Mars rover missions.

2. Instrumentation and Methodology 2.1. Pancam Simulator

The Pancam Simulator (PCS) consists of two color stereo cameras with three CCD chips, one for each primary color (red, green, blue). A stereo jig allowed the two color cameras to be precisely aligned in pitch, roll, and vergence. The horizontal (8 ø) and vertical (11 ø) fields of view (FOV) could be extended by obtaining mosaics of multiple images and assembling them into larger images. The cameras mimic the vision characteristics of humans in that they are located about 1.8 m above the ground, have a 25 cm stereo baseline, and have a 0.25-0.33 mrad resolution typical of the foveal region of the human eye. The stereo baseline is verged and focused at 5 m, and a combination of these settings yield the best resolution for objects on the ground about 3 m in front of the rover (see Stoker et al. [this issue] for specifications of the PCS color cameras). Figure 1 shows a portion of the full PCS mosaic taken at the MFE "landing site" (site 1).

2.2. Visible/Near-Infrared Spectrometer

The FieldSpecFR TM (Analytical Spectral Devices, Inc. (ASD)) is a fiberoptic spectrometer operating over the 350- 2500 nm wavelength range. This device uses three detectors operating over three wavelength domains. In the 350-1000 nm region (VNIR) a fixed grating is used to disperse the wavelengths across a Si-photodiode detector array. In the 1000-1800 nm (SWIRl) and 1800-2500 nm (SWIR2)regions rotatable gratings are used to disperse the wavelengths onto InGaAs detectors. Instrument preparation consisted of optimization, which set the gain and integration time of the VNIR region and the gain and offset of the SWIRl and SWIR2 regions. Although the fiberoptic cable was mounted to the Marsokhod rover, the spectrometer and its controlling

laptop computer were not. This required the operator to manually attach the fiberoptic cable, collect the data, and then disconnect the cable from its mount on the rover.

After optimization a dark current measurement was made for subsequent processing, and additional dark current measurements were made at various times throughout a data collection process. At the beginning of each data collection sequence, another optimization was performed and the reflectance of a white diffuse reflectance standard panel (spectralon) was obtained at a distance of approximately 15- 30 cm. Ten scans were averaged for both sample and reference measurements. The dark current was subtracted

from both spectra, and the ratio of the sample to reference was calculated for immediate display. Subsequently, each spectrum obtained in a sequence was automatically ratioed to this standard measurement to produce relative reflectances, which were written to the computer hard disk. The files produced by the spectrometer were automatically interpolated by the data collection software to a resolution of 1.0 nm over the entire spectral domain, although the true resolution of the instrument varied from 3 nm at 700 nm to 10 nm in the 900-

2500 nm region (ASD, Portable analyzers, spectrometers, and spectroradiometers, 1998 (available on the Web at http://www.asdi.com/prod/)). Further reduction of the spectra are discussed by Gazis and Roush [this issue].

The reflectance measurements were collected in two

different modes. One interfaced the fiberoptic cable with a 1 o FOV foreoptic telescope (•-17 mrad, or 1.7 cm at a viewing distance of 1 m) attached to the same pan and tilt platform as the PCS and aligned with the left camera. In this mode, natural (ambient) light was used to illuminate the target. As such, the spectrometer acted as a proxy for the Mars Surveyor 2003/Athena rover miniature Thermal Emission Spectrometer (mini-TES) instrument (albeit working at different wavelengths), which will also be bore-sighted to a stereo camera [e.g., Squyres et al., 1999; Morris et al., 1998]. In the second mode the fiberoptic cable was interfaced with a cylindrical casing containing an artificial light source that was placed directly over the sample. This "high intensity reflectance probe" (ASD, 1998) allowed direct illumination and collection of a sample spectrum (with a FOV of-3 cm) without significant influence from reflected irradiance or significant atmospheric interference. These spectra obtained under artificial illumination most closely approximate an absolute measurement of the sample reflectance.

In the VIS/NIR spectra, distinctive water vapor absorptions exist near 1350-1450 and 1800-1950 nm (Figure 2). In the artificial-light spectra these bands are mainly due to the sample itself because the short atmospheric path length between the sample and fiberoptic cable (<15 cm) results in minimal contamination from the atmosphere. Differences in detector dark current drift at •-1000 nm (between the end of the VNIR and beginning of the SWIRl wavelength regions) and at -•1800 nm (between the end of the SWIRl and beginning of the SWIR2 wavelength regions) have been corrected by scaling both the VNIR and SWIR2 regions to the SWIRl region. Additional reflectance differences are likely due to the different viewing geometry and fields of view of the two measurement techniques.

An additional mode of data collection permitted acquisition of hyperspectral image cubes under natural- lighting conditions. The pan/tilt platform was used to raster

JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7685

Figure 1. Monochrome version of a Pancam simulator (PCS) mosaic of site 1 looking toward Boba Fett and Acorn Hills, with rocks Solo, Lando, and Emperor (at base of cairn) labeled. Rover solar panels are visible at base of image.

individual point spectra across a given scene to create either 5x5 pixel or 10xlO pixel images. Plate 1 shows an example of this product (displayed as an enhanced color image) obtained for a horizon scene including a portion of the Boba Fett Hills region and nearby sky.

Multiple sources of uncertainty during spectral acquisition made it impossible to produce absolute reflectance measurements. Such sources include differences in viewing geometry, distances between sample and reference measurements, and environmental effects between the time

that the reference and sample measurements were obtained. For example, although the reference may have been located in

the vicinity of the sample, it may have been oriented quite differently relative to the illumination source and to the portion of the subsequent sample that was measured. An example of changing environmental effects would be the presence of clouds or shade during collection of reference data that was absent when the sample data were acquired, or vice versa (the average time gap between sample and standard measurements was 10 rain for the individual measurements

but over 1 hour for the image cube sequences). The spectra presented here should be interpreted as accurate in relative reflectance but not in absolute reflectance values.

In addition, the natural light spectra in Figure 2 are of

7686 JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

1.0

0.9

0.8

0.6 o

ß -> 0.5

0.4

o 0.3

n' 0.2

0.1

T106 (gravel)

Artificial light Natural light

.0 • I , 1 • I , I 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

Wavelength (nm)

Figure 2. Comparison of example visible/near-infrared (VIS/NIR) spectra (400-2500 nm) obtained under natural and artificial lighting conditions. Differences in viewing geometry, distance, and fields of view of the measurements techniques account for the spectral differences observed (see text).

lower quality than can be acquired by the ASD instrument under optimum observing conditions, particularly in the SWIR regions [e.g., Gilmore et al., 1999]. Although most spectra were obtained between 1200 and 1500 local time, the maximum sun elevation in February at the MFE site was only -40 ø (compared to -80 ø in June), and intermittent cloudy conditions prevailed, both of which likely decreased the quality of the data. Also, an extra 1 m length of fiberoptic cable was required to attach the foreoptic of the spectrometer onto the Pancam platform. This added cable length may have decreased the signal quality from the SWIR2 detector (ASD, 1998), although further tests would be required to quantify this effect.

A typical data collection sequence began after target requests received from the off-site team members were transmitted to the rover engineering team and then to the on- site field team. These requests included target context image scenes extracted from PCS mosaics, from which the field team made monochrome hardcopy images using a printer at the field base in order to identify the targets in the field. The standard practice was to also obtain bore-sighted PCS images of the area sampled by each spectral measurement. At the end of each uplink cycle the data were translated from binary to ASCII formats, brought back to the field base and placed on a computer for downlink to the science team.

2.3. Thermal Infrared Spectrometer

The Designs and Prototypes gFTIR field spectrometer is a portable, battery-powered Fourier transform infrared (FTIR) spectroradiometer that operates in the 8-14 gm (1250-750 cm -•) wavelength range at a resolution of 6 cm '• [cf. Korb et al., 1996; Hook and Kahle, 1996; Crowley and Hook, 1996; Horton et al., 1998]. The instrument consists of an optical head housed in a hard case that provides temperature stabilization via a thermoelectric cooler, a small Cassegrain reflector foreoptic, and a data collection/instrument control laptop computer. The instrument was not directly mounted to the Marsokhod rover. Rather, the foreoptic and optical head were bolted to a pan/tilt platform on a tripod, which was placed at a fixed offset position next to the rover for data collection. The laptop computer was mounted in an aluminum briefcase and connected to the optical head by a 5 m electronic cable. The foreoptic gave the instrument an instantaneous FOV (IFOV) of 15 mrad full width half maximum (FWHM), or 15 cm at a range of 10 m. The interferometer consists of two KBr prisms separated by a small air gap. Optical coatings are applied to the prisms instead of using component mirrors and a beamsplitter. A small laser diode provides accurate wavelength calibration. Input and output ZnSe lenses are used to provide a straight optical path from the foreoptic to the detectors. The detectors

JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7687

Plate 1. Near-infrared 10xl0 pixel image cube constructed using the VIS/NIR field spectrometer. The context scene is shown in the right image (taken in later afternoon after the image cube was acquired); the image cube is displayed as enhanced color using red (670 rim), green (530 rim) and blue (443 rim) wavelengths.

are liquid nitrogen-cooled in a dewar. In the field, power was supplied to the optical head through a portable generator. The instrument temperature control was activated approximately an hour before measurements were taken in order to provide adequate thermal equilibration.

Production of a calibrated radiance spectrum of a sample required four separate measurements: a spectrum of a cold blackbody, a spectrum of a hot blackbody, a spectrum of a diffuse reflector, and the spectrum of the target. "Hot" and "cold" for the blackbodies are defined as temperatures that bracket the temperature of the target. These blackbody spectra were used to convert raw data numbers from the spectrometer to radiances. The blackbodies used for the Marsokhod field test were two 60x60 cm boards, painted with two different types of paint: one that is black at visible wavelengths and one that is white at visible wavelengths. The difference in visible albedos of these two blackbodies allowed

them to equilibrate at temperatures bracketing the temperatures of natural targets when exposed to sunlight. The diffuse reflector used was another 60x60 cm board that was

covered with crinkled aluminum foil. Acquiring a spectrum of this board provided a spectrum of the downwelling sky radiance incident upon the target, which was removed during calibration of the spectra to emissivity. All three calibration boards were placed in the vicinity of the target to minimize pathlength differences to the spectrometer.

The foreoptic was aimed using an eyepiece on the optical head case that could be slid into the beam. Spectra were collected via a control program on the laptop computer. Sixteen coadded scans were usually taken for each acquisition. A handheld broadband radiometer was used to measure the radiant temperature of each calibration board and the target during data collection. Approximate air temperature readings were also recorded during data collection. Because this instrument is fairly sensitive to sky conditions, even scattered cloud cover present more than -20 ø above the horizon or near the Sun introduced atmospheric

bands in the spectra that were difficult to compensate for and made spectral interpretations difficult.

During the Marsokhod rover test the field team for the thermal infrared (TIR) spectrometer consisted of two operators. Similar to VIS/NIR spectral acquisition, target requests from the off-site team members were first identified in the field (using hardcopy images of PCS scenes), and spectra were acquired of the targets and calibration boards. The data were brought back to the field base for calibration [cf. Horton et al., 1998] and transmitted back to the science team at the end of the cycle. In order to minimize disruption of other rover operations and contamination of the natural scene by calibration boards, etc., data acquisition took place after the rover completed all its activities at a given stop. The rover's command sequence was placed in a variable-length hold while spectra were being acquired and then restatted after spectral acquisition was complete.

The number of spectra that could be acquired during one cycle depended on the number of targets for which spectra were desired. About 15 min were required to set up the instrument, and about 1 hour was required to acquire eight targets (including three calibration spectra for each) from that fixed location. The total time taken between identification

and transmission of target requests by the off-site team to acquisition, calibration, and return transmission of target spectra by the on-site team was about 3 hours for a typical cycle containing 15 targets. The science team prioritized their targets to allow for the possibility of insufficient time to complete all requests.

The day before the nominal mission began, successful tests were conducted to acquire a hyperspectral image cube from near the rover's initial location using the TIR spectrometer in combination with a pan/tilt tripod mount. Figure 3 shows the context scene for the image cube location, an example band of the image cube at 10 lain, and a map of the 11.3 lam (885 cm -•) carbonate absorption band depth.

7688 JOHNSON ET AL.' REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

"p leoshoreline"

t

Sites 1,2, Arroyo

a

c d

b

Plate 2. (a) SPOT panchromatic image (510-730 nm) of the Marsokhod field test site in Silver Lake, California area (10 m pixel-•), showing location of site I ("landing site"), site 2, site 3 (Arroyo), and site 4 ("paleoshoreline"). (Scene 545-279; acquired November 4, 1990. Image courtesy of R. Arvidson.) (b) Three-band false color composite image of field test area derived from thermal infrared multispectral scanner (TIMS) images where band 5 (10.2-11.2 lum) is displayed as red, band 3 (9.0-9.4 lum) is displayed as green, and band 1 (8.2-8.6 lum) is displayed as blue. Color scheme is such that bright red indicates high quartz content, bright green indicates carbonate, and cyan indicates mafic or quartz-poor material. (TIMS data provided by R. Arvidson.) (c) Ratio of (band 2)/(band 4) Landsat TM data of field area (560/830 nm) in which carbonates are dark and mafic and/or varnished rocks are bright. (TM data provided courtesy of R. Arvidson.) (d) Color composite of three Landsat thematic mapper (TM) band ratios: red (band 5/band7; 1650/2210 nm); green (band 3/band 1; 660/485 nm); and blue (band 2/band 4; 560/830 nm). In this color scheme, clays/carbonates are red; mafic rocks and iron oxides are blue-green.

JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7689

7690 JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

2.4. Methodology of Rover Data Acquisition

Rover operations for the Marsokhod field test began on February 8, 1999, and terminated on February 16, 1999 [Stoker et al., this issue]. Table 1 shows a summary of the rover positions and spectra that were acquired on each day of operation. For the VIS/NIR spectrometer, all image cubes (and the disturbed "paleoshoreline" soil at site 4) were obtained only under natural-lighting conditions, while other samples most often were measured using at least the artificial- light source. Images of rocks and rover location panoramas can be found in other papers in this issue. Also noted in Table 1 are samples collected for ground-truth analyses.

At site 1, a 360 ø PCS mosaic was obtained on February 8 and downlinked on the first day of operation. This provided the science team with a complete view of the "landing site" area, from which the targets shown in Table 1 were identified for spectral acquisition, high-resolution color PCS imaging, and sample retrieval. Future rover traverses were also planned using this mosaic. The uplink commands associated with the traverse to site 2 on February 9 included acquisition of high-resolution single-camera PCS color images directly in front of the rover in addition to stereo monochrome

navigation camera images [cf. Stoker et al., this issue]. The PCS images were obtained so that color images of rocks within reach of the manipulator arm could be immediately downlinked for identification of spectral candidate targets and/or sample retrieval in time for the next uplink opportunity. However, the mosaic was targeted too closely to the rover and imaged mostly rover components. A three- times subsampled stereo PCS panorama of a portion of the site 2 region was also obtained. This was used in combination with a navigation camera mosaic to identify rocks and gravels near the rover that were targeted for spectral

acquisition on February 10-12, including acquisition of spectral image cubes (Table 1).

The methodology of first acquiring images before making decisions regarding spectral acquisition proved to be inefficient because of the short time periods between acquisition of downlinked images and uplink opportunities. The task of identifying intended targets with sufficient accuracy to permit rover engineers and field team members to make the identifications in the field was challenging but manageable. The more critical problem was supplying the rover and field team with this information within time

constraints imposed by uplink windows and adequate field environmental conditions (i.e., sufficient sunlight for spectral acquisition and rover power). As such, the rover remained at site 2 longer than desired by the off-site and field teams in order to allow for acquisition of the requested imaging and spectral data sets. Some of this time was used to downlink large data-volume image mosaics that had been obtained and stored previously.

The above method was altered during the drive to site 3 on Feburary 13. The inclusive uplink commands for this sequence required the rover to drive to site 3 as well as perform imaging and point spectroscopy of the near-field surface before and after disturbance of the materials by the rover wheels. The exact spot for this "scuff" test was chosen blindly under the assumption that much of the matedhal in the arroyo comprising site 3 would be similar. Images of the area obtained before and after the scuff test documented the

conditions of the rocks and soils. Also, the final uplink command in this sequence directed the PCS to take images of the arroyo out to the horizon for later analysis by the geology/geomorphology team.

The method of obtaining images and spectra "blindly" at the end of each rover traverse to a new location was used

Table 1. Spectra Acquired During Marsokhod Field Test

Date in 1999 Rover Position VIS/NIR (ASD) TIR (D&P)

Feb. 6-9 site 1: landing site Rocks R2D2,* Tarken,* Solo,* Lando,* Luke,* Yoda,* Vader,* Jawa,* Valentine,* C3PO,* Jabba*

Vegetation Chewie

Feb. 10-11 site 2: cairn Image cubes Acorn Hills (5x5) Boba Fett Hills (10x 10) Near-field rocks (5x5)

Rocks

Emperor, T105, T106 (gravel) * Feb. 12 site 2: cairn Rocks

Exfoli ated/endolithic,* Tubular*

Feb. 13-14 site 3: arroyo Soils Undisturbed,* disturbed*

Feb. 15 site 4: Image cubes "paleoshoreline" near-field rocks: Left, Right of rover (5x5)

Feb. 16 site 4: disturbed "paleoshoreline" "paleoshoreline" material (natural light only)

Rocks

R2D2,* Lando,* Luke,* Valentine*

Image cube Acorn Hills region (8x15)

Horizon spots Acorn Hills (left/right spots) Boba Fett Hills (left/right spots)*

Rocks

Emperor,* T103,* T105, T106 (gravel) *

Rocks

Exfoliated/endolithic,* Tubular*

Soils

Undisturbed,* disturbed* Undisturbed "paleoshoreline" material*:

left, right of rover disturbed "paleoshoreline"

material

ASD, Analytical Spectral Devices, Inc.; D&P, Designs and Prototypes. *Field sample acquired for ground-truth laboratory measurements.

JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7691

again during the traverse to site 4 on February 15. Here, visible/near-infrared spectral image cubes were acquired to the left and right of the new rover location and downlinked immediately for use and analysis pertinent to subsequent uplinks. A partial PCS panorama was obtained and downlinked on February 16 along with portions of a 360 ø PCS color mosaic.

The final series of rover operations on February 17 required the manual intervention of a field team member to disturb one of the locations for which an image cube had been acquired at site 4 so that both point spectrometers could obtain spectra of disturbed soil to compare to spectra obtained previously of the undisturbed soil/gravel. The final imaging sequence obtained PCS high-resolution images of the disturbed soils to complement the bore-sighted images obtained previously.

2.5. Multispectral Data

In addition to "descent images" of the field site obtained from a helicopter [see Stoker et al., this issue; DeHon et al., this issue], the science team also was supplied with orbital

and airborne multispectral data sets that covered the field test and surrounding areas. In Plate 2a a panchromatic SPOT context image of the field area is shown with the sites shown in Table 1 designated. Plate 2b is a three-band false color composite image produced from airborne thermal infrared multispectral scanner (TIMS) data. Bands 5, 3, and 1 are displayed as red, green, and blue, respectively, to enhance the discrimination of the various compositional units in the scene [e.g., Kahle and Goetz, 1983; Gillespie et al., 1984; Sabine et al., 1994]. In this scheme, carbonate, mafic, and quartz-rich units are readily distinguished. The TIMS data set was particularly valuable during the mission as a means of establishing the regional geology and mineralogy of the field site, including possible sedimentary features such as paledshorelines and source outcrops for gravels found in the arroyos. Such information was critical for rover planning, particularly during discussions regarding the desire to characterize the immediate vicinity near a given rover location versus efficient exploration of new and different geological terrains and lithologies. Landsat thematic mapper (TM) single-band images were available during the field test as GIF images, but the actual data sets were obtained after the

' I ' I ' I ' I ' I ' I ' I ' I ' I ' I ' .•h ' I • I ' I 0.4 Varnished

13.3

13.2

Lando " :' '" I 0.1 - Dolomite "..:' - •' I Desert varnish - ......... Calcite

o.o , • , I , I , I , • , I , I , I , I ' I '

Weathered Mafic rocks, Class 1 Weathered Mafic rocks, Class 2 0.5 - -

0.4

0.3

0.2

- - R2D2 0.1 • • ......... Amphib01e • ;. Phlogopit• Liza(,dite , , 0.0 I , I , I I • I t , I , I

400 aoo • aoo • •00 aooo a400 400 aoo •aoo • •00 aooo a400

Wavelength (nm) Wavelength (nm) Fi•ur• 4. •epresentatiw VIS•I• field spectra of rock t•e classes (collected with a•ificial-li•ht source) compared to scaled and ve•ically offset USGS laboratory spec•a of minerals [cf. •l•r• • •l., l PP•], •howin• •imilar special features, pa•iculady in the •000-•500 nm m•ion. •aboratory sampl•s (and •ain sizes) are: Calcite-1 WS•?• (8•% 4•0 pro; 15% 1• pro); Dolomite-1 RS10• (•85 pro); Dese• vamish-• ANPP0-14 (coating); Amphibole NMNR?8• (•0 pro); Phlo•opite RS• (undocumented); and •iza•dite-• N•R•4•87 (1•

7692 JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

O• o •

•• o

o •

+•v

•+• + •-•

• V

¸

JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7693

N N

7694 JOHNSON ET AL.' REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

N

N

<

A

• N•

JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7695

7696 JOHNSON ET AL.' REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

0.1

0.0 I I • I • I I , I , I , I , I • I 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

Wavelength (nm)

Figure 5. VIS/NIR spectra (obtained with the artificial-light source) for rocks comprising the Carbonates-1 class. All were interpreted as dolomitic rocks, although Jawa exhibits slightly longer wavelength absorptions near 2330 nm, indicative of a combination of calcite and dolomite.

mission ended. Shown in Plates 2c and 2d are a single-ratio gray scale image (560/830 rim) and a three-ratio color composite image from TM data in which carbonates and mafic/varnished rocks are discriminated throughout the field site.

3. Results and Interpretations

3.1. Visible/Near-IR Spectra

Although over 200 VIS/NIR spectra were acquired during the field test, most were obtained under natural-lighting conditions and were prohibitively difficult to interpret for mineral identification because of their poor signal to noise (as discussed by Gazis and Roush [this issue]). However, 20 VIS/NIR spectra were obtained under artificial-lighting conditions with adequate signal to noise (Table 1) and were capable of providing mineralogic information. Figure 4 shows representative spectra of VIS/NIR spectra compared to

carbonates composed dominantly of dolomite (exhibiting a diagnostic absorption band at 2320 nm) with some trace calcite (2340 nm band) (cf. Gaffey, 1986; 1987). The Carbonates-2 class contains additional clays and organic contaminants, including chlorophyll. The spectrum of the "exfoliated/endolith" rock in Figure 6 shows the best evidence of all the rocks measured of a 675 nm absorption band due to chlorophyll interpreted to exist owing to endolithic organisms (cf. Newsom et al., this issue).

The Weathered/Mafic rock classes show evidence for

amphibole and phlogopite (Figures 7 and 4) and serpentine minerals such as lizardite in the near-infrared (Figures 8 and 4). The Vamished/Silicic class (Figure 9) shows similarities to desert varnish deposited on sandstone (Figure 4). Thermal infrared spectra (see below) were used to infer the silicic composition of these rocks.

A comparison between undisturbed and disturbed soil matehals was conducted near the arroyo at site 3 and is shown

U.S. Geological Survey (USGS) laboratory spectra of in Figure 10. Here the gravels in the undisturbed material minerals [ Clark et al., 1993 ] (available at http://speclab.cr.usgs.gov). The similarities between the field spectra and minerals (and coatings) shown in Figure 4 suggested that five distinct rock types were sampled by the artificial-light VIS/NIR spectra. These are categorized in Table 2 and shown in Figures 5-9 below.

Figures 5 and 6 show spectra interpreted to represent

were interpreted to contain a major carbonate (dolomite) component. Upon disturbance of this material, the spectral contrast of the dolomite feature was reduced and additional

near-infrared absorptions were introduced indicative of clays and other weathering products brought up from the subsurface.

Another experiment conducted with the VIS/NIR spectra

JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7697

' I ' I ' I ' I I I' ' I ' ' I ' I ' I

I • • i V I

0.3 / ,/' -

0.2 F / ,-' .... Solo ,/ ----- Exfoliated/endolith /

0.1 /' Emperor

400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 Wavelength (nm)

Eisare 6. VIS•IR spectra (obtained with the a•ificial-li•ht source) for rocks comprisin• the Carbonates-2 class. All were interfered as dolomitic rocks, with some additional clay and/or organic si•atures, pa•icularly the chlorophyll feature near 675 nm [cf. Newso• et •l., this issue].

was an autonomous detection of carbonates using the ASD spectrometer's computer [cf. Kruse et al., 1993]. This type of automated detection will be important in upcoming missions when data compression and decision-making algorithms provide the opportunity to reduce downlinked data volumes by determining what information is worth returning to Earth for further analysis. This algorithm worked relatively well during the field test, although it was not used substantially as part of rover operations and management of camera sequences. Gazis and Roush [this issue] describe this algorithm, its implementation during the field test, and results.

3.2. Thermal Infrared Spectra

Seventeen TIR spectra were acquired during the field test of both near-field rocks and of the distant ridges (500 m away) nicknamed Boba Fett and Acorn Hills. Field spectra representative of the rock classes (see Figure 4) are compared to laboratory emission spectra of minerals [Christensen et al., 2000] in Figure 11. The characteristic absorption centered near 11.3 •tm (886 cm -•) for calcite and 11.2 gm (896 cm -•) for dolomite suggests that the dominant carbonate rock type in Figure 12 is dolomite. Although the TIR spectra are noisy owing to environmental variations during acquisition, they nonetheless consistently exhibit the dolomite feature. The

T106 (gravel) spectrum is particularly complicated due to a combination of environmental conditions and the multiple lithologies that fell within the field of view of the instrument (e.g., quartzitic and mafic rocks in addition to carbonates?).

Figure 13 shows spectra interpreted to be weathered basalts, with the T103 and Tubular rocks showing a prominent absorption near 9.48 •tm (1055 cm-•), consistent with serpentine and/or smectite clay. The R2D2 spectrum exhibits less spectral contrast and less well-defined spectral features, making it difficult to classify. Figure 14 shows

1

exampses of spectra interpreted to represent quartzitic rocks that contain spectral features indicative of microcline feldspar (---9-10 gm; 11 l 1-1000 cm-•), quartz (---8-9 gin; 1250-1111 crn-•), and possibly clays (peak near 9.5 gm; 1055 cm -•) that are either surface dust coatings or part of a discontinuous desert varnish coating [cf. Christensen and Harrison, 1993; Rivard et al., 1993] (Table 2, Figure 11).

Figure 15 shows a spectral comparison of undisturbed and disturbed materials at the site 3 arroyo region. Here a dolomite signature in the undisturbed surface spectrum is lost upon disturbance, and quartzitic/feldspar spectral signatures are likewise changed with the addition of subsurface clays in the disturbed surface (similar to the spectral features in Figure 13). Figure 16 shows a similar comparison of undisturbed/disturbed materials at the "paleoshoreline" region

7698 JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

0.5

ß 0.4

o

->- 0.3

0.2

Tarken

.... Vader

----- Tubular

..................... T103

0.0 • I • I • I • I I • I • I I • I , I 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

Wavelength (nm)

Figure 7. VIS/NIR spectra (obtained with the artificial-light source) for rocks comprising the Weathered/Mafic-I class. All were interpreted to contain mafic minerals such as phlogopite and/or amphibole but may also exhibit a varnished or coated surface (Figure 4).

at site 4. In the undisturbed spectra, quartzitic components mixed with clays are interpreted to be present in varying amounts. However, in the disturbed spectrum the spectral features and overall spectral contrast have markedly decreased owing to the effects of fine-grained particle coatings 6n the disturbed materials [cf. Salisbury et al., 1991, 1994; Johnson et al., 1998].

Figure 17 shows portions of a PCS mosaic of the Acorn Hills and Boba Fett with two circled locations representing -1 ø FOVs (slightly larger than that of the TIR instrument). These areas were targeted for TIR observations to determine if compositional information could be obtained even through large atmospheric path lengths and with nonoptimal calibration conditions. Figure 18 shows the resulting spectra presented as apparent emissivity (i.e., not corrected for environmental conditions) because the standard data reduction algorithms for spectra obtained of nearby targets did not work as well when applied to spectra of distant targets. Nonetheless, it can be observed that a carbonate (calcite) feature at 11.3 gm (886 cm -•) exists in the "right" Acorn Hills spectrum but not in the "left" spectrum, which is consistent with the PCS mosaic (Figure 17) that shows a higher albedo region in the right targeted area compared to the left. Further, both the Boba Fett Hills spectra show carbonate

features, although the left spot shows a calcite signature compared to a dolomite signature for the right spot.

3.3. Laboratory Results From Ground Truth Samples

3.3.1 Petrologic results. At least one sample from each of the five rock spectral classes, all undisturbed and disturbed soils, and the exfoliated/endolithic rock were sent for commercial thin-section preparation, petrographic analysis, and photomicrography. All samples were impregnated in epoxy prior to standard thin-section preparation. A summary of the petrographic results is provided in Table 2.

In these analyses, clay minerals common in altered rocks were identified only by optical methods. The term "clay" is used in Table 2 to denote fine-grained phyllosilicates in general, probably dominated by kaolinite, chlorite, smectite, and mixed layer illite/smectite. The term "sericite" is applied to fine-grained colorless phyllosilicates that show upper second-order maximum interference colors. These could

include muscovite, illite, paragonite, lepidolite, margarite, clintonite, pyrophyllite, and talc. The term "opaques" is used to refer to all materials opaque (and sometimes semiopaque) to transmitted light. The term "FEOH" is used to indicate fine-grained, yellowish to reddish brown, earthy materials of varying opacity in transmitted light. FEOH is probably

JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7699

I I I I I I I I ' I ' I

0.5

0.4

0.3

0.2

I I • I • I • I , I • I • I I

600 800 1200 1400 1600 1800 2000 2200 2400

Wavelength (nm)

0.1 -

0.0 4OO

R2D2

..... Jabba

1

1000

Figure 8. VIS/NIR spectra (obtained with the artificial-light source) for rocks comprising the Weathered/Mafic-2 class. Both were interpreted to contain mafic weathering products such as serpentine or lizardire (Figure 4).

mostly Fe oxyhydroxides but may sometimes include sphalerite, realgar, orpiment, jarosite, a number of Mn oxyhydroxides, and organic matter. For all samples, mineral abundances are visual estimates. For multilithologic materials (soils, etc.), mineralogy, textures, and alteration are described only for the dominant lithology.

3.3.2 Spectroscopic results. Spectra of selected rock samples fi'om the field were measured in the laboratory at Arizona State University using a Nicolet 670 spectrometer configured for emission measurements [Ruff et al., 1997]. These spectra were measured over the range of-2000 to 200

-1

cm with 4 cm 4 resolution. The calibrated emissivity spectrum of each sample was then deconvolved according to the technique of Ramsey and Christensen [1998] to produce an estimate of the type and abundance of the minerals present (Table 3). For better comparison with the field spectra, no sample preparation was performed. All laboratory spectra were taken from approximately the same natural surfaces as those measured in the field. Figure 19 shows laboratory emissivity spectra of four calc-silicate rocks, all of which exhibit the characteristic dolomite absorption to varying depths, indicative of carbonate abundance (see Table 2). Figure 20 shows spectra of other silicic and mafic rocks exhibiting spectral features indicative of quartz, feldspar,

clays, and carbonate components (Table 3; Figure 11). Figure 21 shows emissivity spectra of samples retrieved from the left and fight portions of the Boba Fett Hills region for which long-distance spectra were acquired (Figure 18). Consistent with the field spectra and petrologic analyses (Table 2), the left sample shows a more calcite-rich spectrum, and the fight sample is more dolomitic.

Because the VIS/NIR spectra were obtained under artificial illumination that approximated an absolute measurement of the sample reflectance, no laboratory VIS/NIR spectra were acquired of the returned samples.

3.3.3 Comparison to interpretations of field spectroscopy. Mineralogical interpretations of field spectra correlated rather well to laboratory petrologic and spectral analyses of the returned samples (Tables 2 and 3). Rocks identified as carbonates on the basis of field spectra alone were all petrographically categorized as calc-silicates (Lando, Luke, Emperor, Endolith/exfoliated), including the spectra of the Boba Fett Hills obtained on the horizon. Further, discrimination between dolomite and calcite made on the basis of field spectra for these samples was also confirmed by petrographic and laboratory spectral analyses (Table 3).

The arroyo and paleoshoreline field interpretations agreed moderately well. with the petrologic analyses (laboratory

7700 JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

0.5

c

0.4

o

ß 0.3

0.2

0.1

Valentine

----- Yoda

..... T105

0.0 , I I I ? I , t I • I , I • I 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

Wavelength (nm) Figure 9. VIS/NIR spectra (obtained with the artificial-light source) for rocks comprising the Varnished/Silicic class. All were interpreted to be dominated by desert varnish, although Yoda has some near-infrared features similar to features of the Weathered/Mafic- 1 rocks.

spectra were not obtained for these samples), more so for the undisturbed than the disturbed samples. One reason for this difference is likely related to the reduced spectral contrast in disturbed soils due to fine-grained coatings, which makes deconvolution of spectral components more uncertain [Ramsey and Christensen, 1998].

Moderate to poor agreements between field interpretations and laboratory analyses were found for Valentine, Tubular, T103, and particularly for R2D2. Valentine indeed exhibited a coated surface overlying a silicate-rich rock, but one that was an altered monzonite and not a granite or sandstone as hypothesized from field spectra. Tubular and TI03 were metadiorites, which is somewhat consistent with the interpretation of "weathered mafic" rocks with amphibole and/or a weathered basalt, although both rocks did not contain any serpentine minerals as suggested from field spectra.

interpretation of the VIS/NIR spectra, whereas the TIR spectrum lacked significant spectral features and was difficult to interpret.

One reason for the poorer identifications of Valentine, Tubular, T103, and R2D2 may have resulted from the effects of discontinuous coatings, particularly on Valentine and R2D2 (Table 2). Deconvolution results from laboratory TIR spectra (Table 3) also show poorer correlations to .the petrologic analyses, particularly for R2D2, which may imply the coatings were thick enough to obscure the underlying mineralogy (i.e., greater than ---100 microns) [e.g., Christensen and Harrison, 1993]. Alternatively, the deconvolution algorithm's effectiveness could have been compromised by low spectral contrast and perhaps the fine-grained nature of the sample surface [Ramsey and Christensen, 1998]. Further, the surfaces measured for field and TIR laboratory spectra

The field spectra interpretation of R2D2 as a weathered were natural (weathered), whereas petrologic analyses were mafic rock was incorrect for this calc-silicate sample, based on thin sections of the rock interiors. These two although serpentine minerals suggested by the field data were methods measure quite different materials for these rock found during petrologic analysis, as were small amounts of types. The lack of such coatings on the calc-silicate rocks dolomite suggested by the TIR field spectra. However, in this allowed better identification of their true mineralogies. case the rock sampled by the field team was not the dark rock Discrepancies exist between laboratory and field TIR identified as R2D2 by the off-site team on the basis of PCS spectra, particularly R2D2, T103, and Tubular rocks (compare images. The preconception that R2D2 was a dark rock biased Figure 13 to Figure 20), which may have resulted from the

JOHNSON ET AL.' REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7701

0.5

0.4

0.3

0.2

0.1

I " I ' I ' I ' I ' I ' I ' I ' I ' I

Undisturbed scuff area

Disturbed scuff area

O0 t I t I , I , I , I , I , I , I t I , I ß

400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

Wavelength (nm) Figure 10. VIS/NIR spectra (obtained with the artificial-light source) for an undisturbed gravel area near the arroyo compared to the same location after disturbance. In the disturbed spectrum the depth of the dolomite band is reduced (compared to the undisturbed spectrum) and spectral features attributable to clays and other weathering minerals derived from the underlying fine-grained soils are apparent.

different fields of view afforded by each instrument (-10-15 cm for the field spectrometer versus -1 cm for the laboratory measurements). Variable amounts of discontinuous surface coatings could have been sampled by each measurement. Alternatively, the small sizes of these samples (less than the field of view of the field spectrometer) make it likely that the field spectra included several nearby rocks in addition to the single rock measured in the laboratory.

4. Conclusions

4.1. Summary of Results

In preparation fbr upcoming Mars rover deployments as part of the Mars Surveyor program, the 1999 Marsokhod Field Experiment provided a valuable test of the usefulness of both visible/near-infrared and thermal infrared field

spectrometers. Using data similar to what will be available from upcoming Mars surface missions (combined color stereo imaging, orbital multispectral data, and "field" spectra), the off-site science team was better able to understand the local

and regional lithology, geomorphology, and geology at each

rover location and of the field site in general. These types of observations on a Mars surface mission would help constrain the geologic history and its implications for a landing site region, such as whether environmental conditions were optimum in the past (or present) for the formation of certain minerals (e.g., carbonates, clays, rock coatings) or the degree to which volcanism and/or tectonism influenced the

assemblage of minerals present in the lithologies inferred at the landing site.

For the visible/near-infrared data, spectra acquired under natural-lighting conditions were of low signal to noise and were contaminated significantly by atmospheric absorptions, particularly in the near-infrared. However, high signal-to- noise spectra acquired using an artificial light source that blocked surrounding skylight from a given sample were extremely useful in identifying differences in mineralogy. The thermal infrared data were also subject to environmental conditions (e.g., cloud cover) but nonetheless provided sufficiently high quality data to allow detailed mineral discrimination that proved critical to understanding the composition of both local and regional rock types.

7702 JOHNSON ET AL.' REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

Wavelength (pm) Wavelength (pm) 8.5 9 9.5 10 11 12 8 8.5 9 9.5 10 11 12

I I I 1.0

0.9

0.8 Dolomite

i i I I I I I

:,-,,..,,,'•"• Varnished ........ :.._........• ß '"',, \ silicate .- ............... • "I

,

ß

', ........ " Valentine 0.7

0.6

0.9

O.8

I

0.7

0.6 1300

......... Calcite

i i

',,, ,,,.-' . .

Tubular 1 - ...... Sm'eotite

Serpentine i

Weathered Mafic rocks, Class 1

Microcline

......... Quartz

R2D2 Illite

....... Dolomite

Weathered Mafic rocks, Class 2

1200 1100 1000 .1 ,•00 1300 1200 1100 1000 . 1 900 Wavenumber (cm) Wavenumber (cm)

8OO

Figure 11. Representative thermal infrared (TIR) field spectra of rock type classes compared to laboratory spectra of minerals [cf. Christensen et al., 2000] with grain sizes > 710 pm. Quartz, calcite, and dolomite spectra have been scaled to more closely match field spectra.

A technique of building spectral image cubes using point spectrometers was successfully tested with both spectrometers and should be considered as a potential data product from similar point spectrometers on future missions (as is planned for the mini-TES instrument [Squyres et al., 1998, 1999]). In addition, the carbonate detection algorithm described by Gazis and Roush [this issue] shows promise as an efficient means of automated carbonate identification and could prove useful in reducing data volumes on future missions if carbonates are present. We note that currently planned rover missions will not carry a VIS/NIR point spectrometer and instead will rely on multispectral or broad-band imaging systems in this wavelength region. However, the MFE results presented here demonstrate that the complementary capabilities of both VIS/NIR and TIR spectrometers should improve the detectability of important mineralogical variations.

Spectra from both field instruments independently exhibited spectral features indicative of carbonates (both dolomitic and calcitic), mafic rocks and associated weathering products, and silicic rocks with desert varnish-like coatings.

The visible/near-infrared spectrometer also detected organic compounds such as chlorophyll from endolithic organisms. Comparison of undisturbed and disturbed soils and gravels showed variations related to differences in composition and particle size.

Mineralogical identification using the field spectra was most successful in identifying carbonate rock types. The validation of these identifications for Boba Fett Hills strongly emphasizes the ability of thermal infrared spectra to correctly identiœy subtle differences between adjacent lithologic units even when measurements are acquired under nonoptimum conditions (i.e., viewing the horizon through a substantial atmospheric path length). Silicic and mafic rock types were identified with mederate success, partially hampered by the obscuring effects of rock coatings and weathering rinds on the measured field spectra. These results are consistent with interpretations made recently using laboratory Raman spectroscopy of samples collected at the Silver Lake site [dolliffet al., 1999)] as part of the Field Integrated Design and Operations (FIDO) field tests [Arvidson et al., 1999]. Newsom et al., [this issue] also discuss extensively

JOHNSON ET AL.' REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7703

1.00

0.95

0.90-

0.85 -

Wavelength (pm) 8 8.5 9 9.5 10 11 12

•"I L •,..•, il;•,,,,.• "j. I Io

' ,,j , / - ! T106 •'

• ..t• •i V It I ----' Exfoliated rock (•-•dolith) • . ,,,,d•l! • ___ Lando - t/• .... Luke Emperor

----- T106 (gravel)

0.80 ' ' ' ' 1300 1200 11 O0 1000 900 800

Wavenumber (cm '•) Figure 12. TIE field spectra interpreted as carbonate rocks, showing absorption at l 1.2 pm (896 cm-•), indicative of dolomite (Figure 11). T106 (gravel) spectrum is most affected by environmental conditions during acquisition but may also be influenced by other lithologies in the scene (quartz, basalt).

1.00

0.95

0.90

•' 0.85

._• o.8o E

0.75 -

0.70 -

0.65 -

Wavelength (ILtm) 8 8.5 9 9.5 10 11 12

R2D2

T103

Tubular

0.60 ' ' ' • ' "

1300 1200 1100 1000 • 900 800 Wavenumber (cm')

Figure 13. TIR field spectra interpreted as weathered basaltic rocks. R2D2 may alternatively be a weathered dolomitic rock, as it is distinct from T103 and the tubular rock, which both show absorptions near 9.48 pm (1055 cm-]), suggestive of serpentine and clays.

7704 JOHNSON ET AL.' REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

Wavelength 8 8.5 9 9.5 10 11 12 I I I I I I

1.00

0.95

0.90

0.85

0.80 0.75

0.70 -

0.65 TV•l•tlne , ; 0.60 - ' ' • 1300 1200 1100 1000.1 900 800

Wavenumber (cm)

Figure 14. TIR field spectra interpreted as quartzitic rocks with microcline feldspar (-•9-10 gm; 1111-1000 cm-•), quartz (-•8-9 gm; 1250-1111 cm-•), and possibly clays (peak near 9.5 gm; 1055 cm 4) associated with a discontinuous desert varnish coating (compare Figure 11).

Wavelength (gm) 8 8.5 9 9.5 10 11 12

I i i I I I

1.00 0.95

'• 0.90 LU

0.85 -

0.80 •- • -- 1300 1200 800

Arroyo undisturbed Arroyo disturbed

i I I

1100 1000 1 900 Wavenumber (cm')

Figure 15. TIR field spectra for undisturbed and disturbed materials near the arroyo at site 3. Note decrease in spectral contrast in the disturbed soil and a loss of spectral features associated with dolomite (11.2 •tm; 896 cm-l), microcline feldspar (•-9-10 •tm; 1111-1000 cm4), and quartz (---8-9 •tm; 1250-1111 cm 4) (Figure 11).

JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7705

1.00

0.95

0.90

0.85

0.80

0.75

0.70

0.65

Wavelength (jm) 8.5 9 9.5 10 11 12

I I I I I L, v"' • •1

,.

......... Disturbed

--- o-- Undisturbed--left

Undisturbed--right '

0.60 ; ' ' • ' 1300 1200 11 O0 1000 1 900 800

Wavenumber (cm') Figure 16. TIR spectra for undisturbed and disturbed materials to the left and right of the rover at site 4 near the "paleoshorelines." Note decrease in spectral constrast in the disturbed soil and a loss of spectral features associated With dolomite (11.2 gm; 896 cm-•), microcline feldspar (-9-10 !.tm; 1111-1000 cm-•), and quartz (-8-9 pm; 1250- 1111 cm -•) (Figure 11).

confirmation of the presence of chlorophyl! in the 4.2• Comments and Suggestions for Future Rover Field Endolith/exfoliated carbonate rock via laboratory analyses. Tests This constitutes the first remote detection of such a biomarker

via spectroscopic measurements [cf. Stoker et al., this issue]. Admittedly, the possible presence of metamorphic rocks

(marble, schist, altered monzonite) was not given equal consideration as igneous or sedimentary lithologies by the off- site mineralogy team. This was likely due to an experiential bias on the part of the team members influenced by the dominant presence of igneous and sedimentary regions on Mars and other terrestrial planets. Expanding such a viewpoint to include metamorphic lithologies was one of the enlightening final lessons of the MFE experience for the off- site teams.

Aside from the important geoscience issues that were addressed by analyzing spectroscopic orbital and field data, other observations made during the field test centered on examining the information exchange between different mission operation groups and the dynamics between them [cf., Thomas et al., this issue]. For example, one critical component of the experiment was to demonstrate to an off- site science team how their scientific goals and intended activities for the rover (and field team personnel) must be compromised by the realities of teleoperating an instrument under limitations imposed by finite bfind width (data return)

The main benefit of using the field spectrometers was their and limited time for rover engineers and field operators to ability to provide information relevant to understanding the prepare and execute operational sequences with the rover. variations in mineralogy and, by inference, rock types at a The off-site team thus obtained valuable experience with (1) given rover location. This information was mandatory in data sharing among the science team members (using local deciding the overall exploration strategy for the rover. By database management systems at NASA Ames [Stoker et al., incorporating obr estimates of rock types and their this issue]); and (2) the feasibility of different communication distributions, the geology/geomorphology team improved methods among science team members, rover engineers, and their understanding of the regional geology and history of the field operators during mission operations to achieve the landforms present in the field area [DeHon et al., this issue; desired science goals. Grin et al., this issue]. However, it should be noted that there From the standpoint of the mineralogy team, several was still a difference in opinion among team members important findings resulted from the field test relevant to regarding the advantages of completely characterizing the future rover simulations and Mars lander missions. Foremost mineralogy and geology at one rover location versus the was the difficulty in communicating effectivelyto the on-site benefits of driving the rover to many locations and exploring field team the samples for which spectra were desired. In in more of a reconnaissance mode. some cases, requested rocks had been moved accidentally by

7706 JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

:!'•'•.i•':: ':" ........................... : ........................ •'•::":":"•'?:•: ............................. *<;;: :.'.' ............................................. 'i•½i": ;'i..' ;½::• ...... ::!i;i::.: "..

..

ß ;.7 :<<::; ..... '• '•' '.'.,.;iz ::: ..... ß

::•'•::•

.

. ...

.

JOHNSON ET AL.' REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7707

Wavelength (pm) 8 8.5 9 9.5 10 11 12

t I I I I i i ••., 1.00 tJ .. .. ,•* ( x ø•' '"•'

o..o , - • Acorn Hills--Left • ----- Acorn Hills--Right ............... :---• Boba Fett Hills-Left

• .... Boba Fett Hills-Right

o.8s • •

0.80 ' • • • -"

1300 1200 11 O0 1000 1 900 800 Wavenumber (cm')

Figure 18. TIR apparent emissivity spectra of targeted regions on Acorn and Boba Fett Hills (see Figure 17). Carbonate features are seen in the left Acorn Hills and both Boba Fett Hills spectra, although the spectra are noisy owing to the long atmospheric path length.

1.00

0.95

.j 0.90 0.85

Wavelength (pm) 8 8.5 9 9.5 10 11 12

I I I i • i i

-

----- Lando _ • Luke - ---- Emperor y '

0.80 -" ; I [ •

1300 1200 11 O0 1000 1 900 •00 Wavenumber (cm')

Figure 19. Laboratory TIR spectra for calc-silicate rocks, showing diagnostic carbonate band, the depth of which is correlated to carbonate content.

Table 3. Summary and Comparison of Field and Laboratory Analyses

Sample Field VIS/NIR Field TIR Lab TIR* Petrology

Valentine desert varnish, overlying microcline-dominated silicic (sandstone?) granite rock

30% K-spar altered monzonite 12% plag 43% K-spar 5% quartz 40% plag 45% clays* 8% quartz 8% other silicates 4% biotite

Tubular phlogopite, amphibole, weathered basalt, with varnished? serpentine

35% plag altered metadiorite 10% amphibole schist 10% pyroxene 40% plag 5% quartz 37% actinolite 40% clays* 10% bio

5% quartz

T103 phlogopite, amphibole, weathered basalt, with 25% amphibole altered diorite varnished? serpentine 10% serpentine 27% plag

10% plag 27% amphibole 10% pyroxene 27% epidote 5% carbonates 10% clays

35% clays

R2D2 serpentine/lizardite plagioclase, serpentine, 40% dolomite calc-silicate dolomite 50% clays* 68% calcite, 10%

10% other silicates dolomite, 10% olivine

5% serpentine

Lando dolomite dolomite 90% dolomite altered calc-silicate 10% clays* 90% dolomite, 5 %

chlorite

Emperor dolomite + clays/organics? dolomite 85% dolomite altered calc-silicate 5% plagioclase 80% dolomite 10% other silicates 10% calcite

10% olivine

Luke dolomite + clays/organics? dolomite 85% dolomite altered calc-silicate 15% clays 66% dolomite

15% calcite

15% serpentine

Endolith/ dolomite + clays/organics? dolomite 95% dolomite altered calc-silicate exfoliated chlorophyll 5% other silicates 91% dolomite

5% olivine

Arroyo, undisturbed dolomite dolomite, quartz, NA dolomite, plag, k-spar feldspar epidote

Arroyo, clays + other weathering clays? +- weathered NA 33% clay disturbed products silicates 20% plag

15% quartz 15% calcite

5% dolomite

5% K-spar

Paleoshoreline, NA quartz, feldspar, clay NA quartz, feldspar, undisturbed carbonate cement Paleoshoreline, NA clays? +- weathered NA 28% K-spar disturbed silicates 22% plag

20% clay 15% quartz 15% calcite

Boba Fett, left NA calcite 85% calcite calc-silicate 5% dolomite 81% calcite

10% clays* 10% phlogopite Boba Fett, right NA dolomite 80% dolomite calc-silicate

20% clays* 83% dolomite 8% olivine

5% calcite

* Type and areal percent mineralogy were determined using deconvolution algorithms of Ramsey and Christensen [ 1998]. *"Clays" includes hydrous-alumino-silicates such as dickite, hectorite, kaolinite, montmorillonite, Fe-smectite, halloysite, and illite.

JOHNSON ET AL.' REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7709

1.00

0.95

0.90

0.85-

O.80 -

0.75 t 0.70 !

1300

Wavelength (pm) 8 8.5 9 9.5 10 11 12

1 I I I I i i /

R2D2

----- T103

....... Tubular

Valentine

800 ! I I I

1200 1100 1000 1 900 Wavenumber (cm')

Figure 20. Laboratory TIE spectra. R2D2 was misclassified as a weather mafic rock but is in fact a calc-silicate T103 was an altered diorite, Tubular was a metadiorite schist, and Valentine was an altered monzonite (Table 2).

1.00

0.95

0.90

0.85

Wavelength (pm) 8 8.5 9 9.5 10 11 12

I I I I I I I

Boba Fett Hills-Left • Boba Fett Hills-Right

0.80 , I [

13•)0 1200 1100 1000 1 900 800 Wavenumber (cm')

Figure 21. Laboratory TIR spectra for Boba Fett Hills "left" and "right" samples showing diagnostic carbonate bands that distinguish calcite ("left") from dolomite ("right"), consistent with both field and petrographic results.

7710 JOHNSON ET AL.' REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST

movements of on-site team members and/or transportation and nightly securing of the Marsokhod rover. In other cases, annotated images transmitted to the field team that documented requested targets were difficult to interpret because of the lack of color hardcopy printing capability in the field. This was the reason for the misidentification of

R2D2 that led to subsequent confusion regarding combined interpretations of PCS and field spectra for this rock. Second, the difficulty in correctly positioning the rover arm onto intended targets (a required element for sampling rocks in future missions) in a timely fashion was not appreciated and became a major time requirement when developing sequence uplink strategies. Third, necessary practical experience was gained regarding how best to analyze a given rover location in an efficient and "standard" manner. Once at a new rover

location, it became apparent that limitations due to finite data volumes and limited uplink/dowlink opportunities could be offset by abilities to autonomously choose rock (or soil) targets worthy of spectral analysis [cf. Gazis and Roush, this issue].

One method developed during the course of this experiment was to incorporate preliminary imaging and spectral measurements of near-field targets (i.e., targets within the reach of the manipulator arm for sampling) as part of the uplink sequence of driving to a new rover location. Considering the prohibitively large data volumes and downlink times required for acquisition of complete color stereo panoramas, obtaining images and spectra at such "blind" locations immediately upon reaching a new rover location appears to be one strategy worth further testing. The establishment of automated detection algorithms also would allow for more informed choices than "blind" measurements

would otherwise offer [cf. Gazis and Roush, this issue] and reduce downlink volumes further.

Principal recommendations from the mineralogy team to improve efficiency and precision of the MFE operations include (1) better communication between off-site and on-site team members regarding location and acquisition of requested rock targets; (2) better record keeping and summary reporting between what was requested for rover activities in uplink plans versus activities and data sets that were actually accomplished and downlinked; and (3) better communication between off-site teams regarding geologic interpretations of the landing site using the combined strengths of field and "orbital" spectroscopy, color imaging, and morphologic analyses. For example, the geology/geomorphology teams used preliminary interpretations of the field and TIMS spectra from the mineralogy team to hypothesize that basalts were the dominant "dark" material comprising portions of the hills surrounding the landing site [DeHon et al., this issue]. Petrologic analyses and a subsequent field site visit by members of this team showed that these dark rocks were

predominantly mafic metamorphic rocks. While the original hypothesis of basalts did not conflict greatly with early interpretations of the spectroscopic and morphologic results, more consistent and detailed discussions between these team

members may have resulted in consideration of alternative hypotheses. However, one practical experience from the MFE that is likely shared with other limited-duration planetary missions is that detailed scientific analyses often receive lower priority than the primary goal of the mission, namely, continuous and efficient operation of the science instruments

so as to obtain the greatest amount of data during the lifetime of the mission.

Acknowledgments. We thank the rover engineering team and support staff that operated in the field and at NASA Ames for their dedicated efforts, without which the 1999 Marsokhod Field Experiment would not have occurred. TIMS, SPOT, and Landsat TM data were provided by R. Arvidson (Washington University) as part of a collaborative arrangement between the Marsokhod and FIDO science teams, both working at the Silver Lake area. We thank G. Swayze, K. Tanaka, M. Chapman, and an anonymous reviewer for their constructive comments. Any use of trade, product, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. government.

References

Arvidson, R.E., et al., Rocky 7 prototype Mars rover field geology experiments 1. Lavic Lake and Sunshine Volcanic Field, California, J. Geophys. Res., 103, 22,671-22,688, 1998.

Arvidson, R., P. Backes, E. Baumgartner, D. Blaney, L. Dorsky, A. Haldemann, R. Lindemann, P. Schenker, and S. Squyres, FIDO: Field-test rover for 2003 and 2005 Mars sample return missions, Lunar Planet. Sci., XXX, abstract 1201, 1999.

Christensen, P.R., and S.T. Harrison, Thermal infrared emission spectroscopy of natural surfaces: Application to desert varnish coatings on rocks, J. Geophys. Res., 98, 19,819-19,834, 1993.

Christensen, P.R., J.L. Bandfield, V.E. Hamilton, D.A. Howard, M.D. Lane, J.L. Piatek, S.W. Ruff, and W.L. Stefanov, A thermal emission spectral library of rock-forming minerals, J. Geophys. Res., 105, 9,735-9,739, 2000.

Clark, R.N., Reflectance Spectra, in Rock Physics and Phase Relations: A Handbook of Physical Constants, AGU Ref Shelf vol. 3, T.J. Ahrens, pp. 178-188, AGU, Washington, D.C., 1995.

Clark, R.N., G.A. Swayze, A.J. Gallagher, T.V.V. King, and W.M. Calvin, The U.S. Geol. Survey, Digital Spectral Library: Version 1:0.2 to 3.0 microns, U.S. Geol. Surv. Open File Rep. 93-592, 1340 p., 1993.

Crowley, J.K., and Hook, S.J., Mapping playa evaporite minerals and associated sediments in Death Valley, California, with multispectral thermal infrared images, J. Geophys. Res., 101, 643-660, 1996.

DeHon, R., et al., Remote observation of the geology and geomorphology of the 1999 Marsokhod test site, J. Geophys. Res., this issue.

Gaffey, S.J., Spectral reflectance of carbonate minerals in the visible and near infrared (0.35-2.55 microns): calcite, aragonite, and dolomite, Amer. Mineral., 71, 151-162, 1986.

Gaffey, S.J., Spe:tral reflectance of carbonate minerals in the visible and near infrared (0.35-2.55 urn): Anhydrous carbonate minerals, J. Geophys. Res., 92, 1429-1440, 1987.

Gazis, P., and T. Roush, Autonomous identification of carbonates using near-IR reflectance spectra during the February 1999 Marsokhod field tests, J. Geophys. Res., this issue.

Gdlespie, A. R., A. B. Kahle, and F.D. Palluconi, Mapping alluvial fans in Death Valley, California, using multichannel thermal infrared images, Geophys. Res. Lett., 11, 1153-1156, 1984.

Gilmore, M.S., R. Castao, T. Roush, E. Mjolsness, T. Mann, R.S. Saunders, B. Ebel, and E. Guinness, Effects of distance and azimuth on spectroscopic measurements at Silver Lake, CA, Lunar Planet. Sci., XXX, abstract 1886, 1999.

Grin, E., et al., Geological analysis of the Silver Lake Marsokhod field test site from ground truth sampling and mapping, J. Geophys. Res., this issue.

Guinness, E.A., R.E. Arvidson, I.H.D. Clark, and M.K. Shepard, Optical scattering properites of terrestrial varnished basalts compared with rocks and soils at the Viking Lander sites, J. Geophys. Res., 102, 28,68%-28,703, 1997.

Hook, S.J., and A.B. Kahle, The micro Fourier transform interferometer (gFTIR) - A new field spectrometer for acquisition of infrared data of natural surfaces, Remote Sens. Environ., 56, 172-181, 1996.

Horton, K.A., J.R. Johnson, and P. G. Lucey, Infrared measurements of pristine and disturbed soils, 2, Environmental effects and field data reduction, Remote Sens. Envrion., 64, 47-52, 1998.

Johnson, J.R., P.G. Lucey, K.A. Horton, and E.M. Winter, Infrared

JOHNSON ET AL.: REMOTE MINERALOGY-1999 MARSOKHOD FIELD TEST 7711

measurements of pristine and disturbed soils, 1, Spectral contrast differences between field and laboratory data, Remote Sens. Environ., 64, 34-46, 1998.

Jolliff, B., A. Wang, K. Kuebler, and L. Haskin, Raman analysis of weathered rocks from the FIDO Mars rover test site, Silver Lake, California, Lunar Planet. Sci., XXX, abstract 1512, 1999.

Kahle, A. B., and A. F. H. Goetz, Mineralogic information from a new airborne thermal infrared multispectral scanner; Science, 222, 24-27, 1983.

Korb, A.R., P. Dybwad, W. Wadsworth, and J.W. Salisbury, Portable FTIR spectroradiometer for field measurements of radiance and emissivity, Appl. Opt., 35, 1679-1692, 1996.

Kruse, F.A., A.B. Lefkoff, and J.B. Dietz, Expert system-based mineral mapping in northern Death Valley, California/Nevada, using the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS), Remote Sens. Environ., 44, 309-336, 1993.

Morris, R.V., et al., Analyses of Martian surface materials during the Mars Surveyor 2001 mission by the Athena Instrument payload, Lunar Planet. Sci., XXIX, abstract 1326, 1998.

Newsom, H., J.L. Bishop, C.S. Cockell, T.L. Roush, and J.R. Johnson, The search for life on Mars in surface samples: Lessons from the 1999 Marsokhod rover field experiment, d. Geophys. Res., this issue.

Ramsey, M. S. and P.R. Christensen, Mineral abundance determination: Quantitative deconvolution of thermal emission spectra, J. Geophys. Res., 103,577-596, 1998.

Rivard, B., S.B. Petroy, and J.R. Miller, Measured effects of desert varnish on the mid-infrared spectra of weathered rocks as an aid to TIMS imagery interpretation, IEEE Trans. Geosci. Remote Sens., 31,284-291, 1993.

Ruff, S.W., P.R. Christensen, P.W. Barbera, and D.L. Anderson, Quantitative thermal emission spectroscopy of minerals: A laboratory technique for measurement and calibration, J. Geophys. Res., 102, 14,899-14,913, 1997.

Sabine, C., V.J. Realmuto, and J.V. Taranik, Quantitative estimation of granitoid composition from thermal infrared multispectral scanner (TIMS) data, Desolation Wilderness, northern Sierra Nevada, California, J. Geophys. Res., 99, 4261-4271, 1994.

Salisbury, J.W., L.S. Walter, N. Vergo, and D.M. D'Aria, Infrared (2.!-25 ram) Spectra of Minerals, 267 pp., The Johns Hopkins Univ. Press, Baltimore, Md., 1991.

Salisbury, J.W., Wald, A., and D'Aria, D.M., Thermal-infrared remote sensing and Kirchhoffs law 1. Laboratory measurements, J. Geophys. Res., 99, 11,897-11,911, 1994.

Squyres, S.W., et al., The Athena Mars rover science payload, Lunar Planet. Sci., XXIX,, abstract 1101, 1998.

Squyres, S.W., et al., The Mars 2001 Athena Precursor Experiment (APEX), Lunar Plan. Sci., XXX, abstract 1672, 1999.

Stoker, C. et al., The 1999 Marsokhod rover mission simulation at Silver Lake, California: Mission overview, data sets, and summary of results, d. Geophys. Res., this issue.

Thomas, G., M.K. Reagan, E.A. Bettis, N.A. Cabrol, and A. Rathe, Analysis of science team activities during the 1999 Marsokhod Rover Field Experiment: Implications for automated planetary surface exploration, J. Geophys. Res., this issue.

J. Bishop, N.A. Cabrol, C. Cockell, P. Gazis, T. Roush, and C. Stoker, NASA Ames Research Center, Moffett Field, CA 94035.

K. Horton, Hawaii Institute of Geophysics and Planetology, University of Hawaii, Honolulu, HI, 96822.

J.R. Johnson, Astrogeology Team, United States Geological Survey, 2255 North Gemini Drive, Flagstaff, AZ 86001. (j rj ohns on•us gs. go v)

J. Moersch, Department of Geological Sciences, University of Tennessee, Knoxville, TN 37996-1410.

H.E. Newsom, Institute of Meteoritics and Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque, NM 87131.

S.W. Ruff, Department of Geology, Arizona State University, Temp, AZ 85287.

(Received July 21, 1999; revised September 22, 1999; accepted September 28, 1999.)