Mars Pathfinder spectral measurements of Phobos and Deimos: Comparison with previous data

JOURNAL OF GEOPHYSICALRESEARCH,VOL. 104,NO. E4, PAGES9069-9079,APRIL 25, 1999

Mars Pathfinder spectral measurementsof Phobosand Deimos: Comparison with previous data ScottMurchie, l Nick Thomas, 2 DanielBritt,3 KenHerkenhoff, 4 and JamesF. Bell III • Abstract.The Imagerfor Mars Pathfinder(IMP) acquiredfourspectraof partsof the sub-Mars hemispheres of PhobosandDeimos.The measured regionof Phobosis expectedto be a mixture of the two spectralunitsidentifiedon that satellitefrom Phobos2 data,andthe IMP spectraof Phobosare intermediateto the two unitsas expected.The derivedgeometricalbedois consistent with the valuefor thatpartof PhobosdeterminedfromViking imagery.The IMP spectrumof Deimosis generallyconsistent with previousmeasurements acquiredfromthe groundandfromthe HubbleSpaceTelescope(HST), but the signal-to-noise ratio is lowerthanthat of the Phobos data.The spectralcontrastbetweenthe two moonsis similarto that seenin HST andPhobos2 data. Mars Pathfinder measurements therefore substantiate recent results which indicate that

PhobosandDeimosare not, aspreviouslybelieved,analogous to C-type asteroids.They also providesomeindicationsfor an absorptionnear700 nm, perhapslike that seenin otherlowalbedoasteroids.Both Martian moonsareredderthanmostasteroids,andmostcloselyresemble two analogmaterialsbelievedto haveundergonevery dissimilarhistories:primitiveD-like asteroids, andhighlyspace-weathered, mafic-richassemblages, suchasarepresentin lunarmare soils.

1. Background The spectralreflectanceproperties of Phobos and Deimos are a key piece of evidencefor the compositionand origin of the Martian satellite system.After the Mariner 9 and Viking missions,existing visible-wavelength imageryrevealed little or no evidencefor color variation on the moons [Veverka and Duxbury, 1977]. Ultraviolet to near-infrared spectra of the satellites, assembledfrom coverageof different geographic regions by different instruments at different wavelengths, yielded compositespectra analogous to low-albedo, C-type asteroids[Pang et al., 1978, 1983; Pollack et al., 1978]. These observations,combinedwith an unexpectedlylow density of

Subsequently, new spectralobservationsand reanalysisof Viking resultshaveprovenlargely inconsistentwith the Casteroid paradigmfor the Martian satellites. Spectral data coveringthe wavelengthsbelow 1000 nm are summarizedin Plate 1. The spectralpropertiesof Phobos are better known than those of Deimos, thanks largely to the information returnedby the Phobos 2 spacecraft[Avanesovet al., 1991; Murchie et al., 1991; Murchie and Erard, 1996]. The trailing

hemisphere wasobservedby the CombinedPhotometer and Radiometerfor Mars (KRFM) UV-visible point spectrometer,

the Imaging Spectrometerfor Mars (ISM) near-infrared spectrometer, andthe Videospectrometric Camera(VSK) twocolor imagerwith visible (400-600 nm)and near-infrared the satellites near2 g/cm 3 [Duxburyand Callahan,1982, (800-1100 nm) filters. The leading hemispherewas covered 1989], fueled major inferencesabout the Martian satellites, including (1)compositionally homogeneoussurfaces;(2)a compositionlike carbonaceouschondrite meteorites,inferred from the moons' putative spectral similarity to C-type asteroids and low metamorphic grade carbonaceous chondrites; and therefore(3) probably an origin by captureof asteroids

from the outer main belt. This view

of the nature

of

Phobos and Deimos, along with its numerous selfcontradictions,is summarizedby Thomaset al. [1992], Burns [1992], and Bell et al. [1993].

only by KRFM and VSK. Phobos

2

data

show

that

satellite

to

contain

two

fundamentalcolor units, whose spatial distributions within the region observedby Phobos 2 are shown in Plate 2 [Murchie and Erard, 1996]. The two units differ in their

geographic distributions, their albedo patterns, their relationshipbetweencolor and albedo, and their ultraviolet spectral properties.A "bluer unit" is associatedwith the interior and proximal ejecta of the large crater Stickney.

Within the bluer unit the albedo range is large, varying from someof the darkest materialson the moon (in relatively gray •Applied Physics Laboratory, TheJohns Hopkins University, Laurel, materialwest of Stickney) to the brightest (relatively blue material on the east flank of Stickney). Color and visibleMaryland. 2MaxPlanck Institute forAeronomy, Katlenburg-Lindau, Germany. wavelengthalbedoare stronglycorrelatedin the bluer unit, so 3Lunar andPlanetary Laboratory, University of Arizona, Tucson. that the brigbest regions are also the bluest. The bluer unit's 4jetPropulsion Laboratory, Pasadena, California. 5Center for Radiophysics andSpaceResearch, CornellUniversity,dropoffin reflectancefromvisible to ultraviolet wavelengths Ithaca, New York.

Copyright1999by the AmericanGeophysicalUnion.

is -•10%. In contrast,a "redderunit" coversthe remainderof the satellite's surface,distal from Stickney. Its albedo is more

homogeneous on a largescalethanin the bluerunit. However,

Papernumber98JE02248.

at the scale of kilometer-sized craters, albedo variations are

0148-0227/99/98JE-02248509.00

commonplacebut, unlike in the bluer unit, occur without 9069

9070

MURCHIE

ET AL.'

PHOBOS'

AND

DEIMOS'

SPECTRAL

PROPERTIES

,,,I,,

0.10-

Deimosleading..' Deirnos trailing

0.08-

.'

ß

ß• ß ß.; ....:.. .o..•";:

'--. .... ...

;.•:

.'..'.

ß ?"" '"' ,.. ;,,..... ,.•..'.-,

' ß Phobos trailing

4,ß

0.06-

Phobosleading

0.04-'

? Deirnos trailing, HST Deimos leading, HST Deimos leading, telescopic Phobos leading, Mariner 9 Phobos leading, Phobos 2 Phobos leading, HST Phobos trailing, Phobos 2:

ß

=,

..,

ß a ß --

0.02.,

,=

0.00

'

200

'

'

I

'

400

'

'

!

'

'

600

'

i

800

'

'

'

!

Plate 1. Comparison of post-1989 data for Phobos and Deimos. Phobos' trailing hemisphereis represented by a regionof that hemispherecovered by Phobos 2 data. Phobos' leadinghemisphereis representedby a Phobos2 measurement of a relativelylow-albedopart of the hemisphere,hemisphereintegratedMariner 9 UVS data, and hemisphere-integrated HST data. Deimos is representedby hemisphere-integrated HST and terrestrial ground-based data. Ground-based and HST data are scaled to the Viking clear-filter albedo of the measuredregions of the satellites [Thomas et al., 1996; SimoneIll et al., 1998].

'

1000

Wavelength in Nanometers

1.2

Vis/NIR

253 Mathilde (C-type)

Phobos(globallyD- or T-like)

0.6

Plate 2. (top) Viking imagemosaicof Phobos, and color ratios (in falsecolor) of the regions observed by Phobos 2. Redder false colors indicate redder color properties,and bluer false colors indicate bluer color properties.The red dashedline is the approximateboundary of "redder"and "bluer"spectralunits; the blue dashedline is the best estimateof the boundary in the opposite hemisphere,based on Viking clear-filter albedoproperties.The white and yellow boxesoutline regionsviewed by IMP, which are expectedto be a mixtureof the two units. The map is in simple cylindrical projection,centeredat the sub-Mars point (0øN, 0øW). (bottom)Comparison of color properties of Phobos and the C-type asteroid 253 Mathilde. The wavelengthsusedfor the vis/NIR ratio are 510 and 910 nm for the Phobos 2 images,and 550 and 900 nm for the NEAR imagesof Mathilde.Falsecolorsare scaledas in the map at top.

MURCHIE

ET AL.'

PHOBOS'

AND

attendant changes in color. The ultraviolet dropoff is much largerthan in the bluer unit, >20%. Viking clear-filter albedo variations [SimoneIll et al., 1998] closely follow the color variationsobservedin Phobos2 data. The boundary between the two spectral units in the anti-Mars hemisphere was observed by Phobos 2. In the sub-Mars hemisphere the boundarycan only be estimatedbasedon albedo variations in Viking imagery(Plate 2). Hubble SpaceTelescope(HST) data are available for the leading hemisphere[Zellner and Wells, 1994], and agree closely with Phobos 2 measurements of that region as shown by Murchie and Erard [1996]. The spectral properties of Phobos contrast strongly with thoseof the only C-type asteroid visited by a spacecraft,253 Mathilde [Veverka et al., 1997]. Near-Earth Asteroid Rendezvous(NEAR) imagesof Mathilde (Plate 2)reveal a surface with a visible/near-infrared (vis/NIR) color ratio approximatingthat of the least red part of Phobos, and with very uniformcolor propertiesdespitethe occurrenceof several large, Stickney-likecratersthat penetrateto depth. Mathilde's visible-wavelength albedo (0.045+0.01) is also significantly lower, only about two-thirds that of Phobos [Veverka et al., 1998]. Additional

information

on Phobos'

redder unit comes from

Phobos2 ISM infraredspectra[Bibringet al., 1990,Langevin, 1991; Murchie and Erard, 1996]. At 1000-3000 nm the spectral continuum is comparably red to that at visible wavelengths, and there is no evidence of a 3000-nm absorption due to H20 (Figure la). The shapeof the spectrum near 1000 nm exhibits spatially coherent variations, which appearconsistentwith variations in the depth of a shallow absorption perhaps due to olivine and/or pyroxene (Figure lb).

DEIMOS'

SPECTRAL

PROPERTIES

9071

[1991] and by the HST [Zellner and Wells, 1994] showed Deimos to be much redder than carbonaceous

chondrites

or C-

type asteroids,and similar to the putatively moreprimitive Dtype asteroids [Bell et al., 1989a]. Furthermore,the spectral properties of the leading and trailing hemispheres are significantly different. The albedo properties of Deimos (Figure 2), as revealed by Viking clear-filter (-540 nm) imagery, are highly heterogeneousand suggestiveof exposure of higher-albedo material by mass wasting [Thomas et al., 1996]. Infrared spectraacquiredby Bell et al. [1989b] did not reveal the 3000-nm absorption expected if there were a significant content of chemically bound water in the surface material; they concluded that a "dry" (analogous to the C5 chondrite Karoonda) rather than "wet" carbonaceous chondritecomposition(like the C10rgueil) was likely. In Plate 1, post-1989 spectral measurements of comparable portionsof the two moonshave been integrated,to show the best pre-MarsPathfinderinformationon the moons' spectral propertiesand spectralheterogeneity.Data that are in relative units, e.g., HST spectra,are scaled to Phobos 2 data using Viking clear-filter albedo measurementsof the measured regions [Thomas et al., 1996; Simonelli et al., 1998]. Phobos' bluer unit and both hemispheresof Deimos have grosslysimilar spectralproperties,including spectral slope at visible wavelengths and magnitude of the ultraviolet reflectance falloff. On these materials, albedo variations are

associatedwith large topographic features, and small craters lack brightenedrims. Phobos'redder unit is distinct fromthe other materials on the satellites: only this unit exhibits bright-rimmed craters,and its spectrum"crossesover" that of the other materials. None of the materials exposed on the satellite surfacesare consistent spectrally with carbonaceous Observationalcoverage of Deimos is less extensive. Diskchondrites, or with C-type asteroids in general. Rather, at integrated spectra of the leading and trailing hemispheres visible and near-infrared wavelengthsthey most resembletwo obtainedfrom a ground-basedtelescopeby Grundy and Fink very dissimilar analog materials [Murchie and Erard, 1996]:

I,,

i,l,,,,

I,,,,

I,,,,

I,,,,

I,,,,

I

Very weak

absorption

.

m

m

mm

m

Insignificant H2 0 absorption

o

m

m

m m

m

500

1500

2500

3500

Wavelength in Nanometers Figure la. Phobos2 ISM near-infraredreflectancespectraof the redderunit on Phobos.

9072

MURCHIE ET AL.' PHOBOS' AND DEIMOS' SPECTRAL PROPERTIES

1.01

,',,1',,'1''''1,,,,I,,,,I,,,,I,III

1.00 0.99

o o

0.98

o

O0

0.97 o

0.96

o

o l

o

o o

0.95

o

* Ratioed spectrum fit with 3rd order polynomial

0.94

0.93

8OO

1000

1200

1400

Wavelength in Nanometers Figure lb. Ratio of spectrawith greaterand lessercurvaturesnear 1000 nm, suggestinga very weak absorption consistent with a mariemineralsuchas olivineor pyroxene.

ßD-type and relatedasteroids,thoughtto be compositionally 1991]. Murchie and Erard [ 1996] showedthat, basedon the similar to C-types but even moreprimitive and rich in red spatialpatternof spectralheterogeneities revealedby Phobos organic materials; and space-weathered,marie mineral-rich 2, theC asteroid-likespectrum of Phobos[Panget al., 1978, assemblages suchas lunarmaresoils,which resemble D-type 1983] mighthave arisenby the mergingof measurements of asteroidsspectrallybut are chemicallyevolved. differentwavelengthrangesthat samplespectrallydistinct Two sourcesof uncertainty in the quantification of the partsof the satellite.In contrast, the very un-C-like composite satellites'spectralpropertieshave createdquestionsabout spectrain Plate I are derivedonly fromdata that samplethe even the most basic spectral attributes of the Martian

sameor similar geographic regions. The second source of

satellites.The first source of uncertainty arises from the mergingof spectraldataacquiredby differentinstruments.In thepast,thishasbeendonewithout regardto the geographic regionscoveredin the data[e.g.,Panget al., 1978]. However, the imagingresultsfromPhobos2 show clearlythat at least Phobosis highly heterogeneous spectrally[Avanesoveta!.,

uncertainty,data calibration,has proven to be a particular issuewith Phobos2 data.This is due in partto problemswith datareductionprocedures performedon spectrashownin some publications.Ksanfomality and Moroz [1995] calibrated KRFM measurements of Phobos againsta bright region of Mars, whose propertiesare well known from telescopic

Figure2. Vikingimagemosaic of Deimos, with shadingadjusted to matchthespatialdistribution of Viking clear-filter (-540-nm)albedo.Whiteandstippledboxesoutlinetheregionsviewedby IMP. Themapis in simplecylindricalprojection,centeredat the sub-Marspoint(0øN,0øW).

MURCHIE

E

ET AL.'

2.2

AND

DEIMOS'

SPECTRAL

PROPERTIES

'

•

1.8 '

• 2.0 •

=

KRFM [Murchieand Erard, 1996] :

o

KRFM [Ksanfomality and Moroz,1

--

:

: -

_o : -o 1.4-

: -

1.6'

(p

:

. :

03

ß

'

1.0'

-

•

'=

•=.

-,

(•

9073

i i i i I i i i i I i i i i I i i i i I i i i i

o

o. '-

PHOBOS'

0.8'

O

.

ß

-

n-

200

= 0.6

..,

ß

HST [Zellner and Wells, 1994]

••1,•,,i,•',,i,•,•1,,,, 300

400

500

600

-

700

Wavelength in Nanometers Figure 3, Phobos 2 spectraof a region representativeof the leading hemisphere,comparedto HST diskintegratedmeasurements of the leadinghemisphere[Zellner and Wells, 1994]. The spectrumof Ksanfomality and Moroz [1995] was calibratedagainstMars usingdata which approachedsaturation,in a brightnessrange to which the instrumentrespondednonlinearly. The spectrumof Murchie and Erard [1996] was calibrated againstMars using data in a brightnessrangeto which the instrumentrespondedlinearly. measurements. Although the procedure is sound, a typical resultant spectrum(covering a region shown in Phobos 2 imagerdatato be representativeof the leading hemisphere)is wildly inconsistent with HST measurements (Figure 3). In a detailed analysis of sourcesof calibration uncertainty in the Phobos2 data, Murchie and Erard [1996] showed that many of the KRFM Mars observations (including those used by Ksanfomalityand Moroz) are in a brightnessrange in which the KRFM instrument approaches saturation and responds nonlinearly. This has the effect of artificially inflating the "calibrated"Phobosspectrumat long wavelengths,making the satellite appear extremelyred. (This affliction is not an issue with the VSK or ISM instrument.) Murchie and Erard completeda parallel procedureof calibratingdatafrom all three Phobos 2 instruments to Martian standard regions that the instrumentsobserved,exceptbeing careful to use nonsaturated spectra, and they derived spectra consistent with HST measurements to within the uncertaintyin the data (Figure 3).

whether color and albedo are related throughout these surfaces.In previous data, a weak relationship of color and albedo may be present (Figure 4). Imager for Mars Pathfinder (IMP) spectrataken at different rotational phases, showing materialsof differentalbedo,canbe instrumentalin testing the hypothesisof a relationshipbetween color and albedo. Third, Phobos 2 data revealed evidence for a weak absorption in the redder unit centered near 1000 nm, consistentwith the presenceof Fe-containingminerals(Figure lb). However, prior to Mars Pathfinder, there was no measurement diagnostic of a 1000-nm absorption for Deimos. In addition, the only well-calibrated measurementswith

unbrokenwavelengthcoveragefrom 500 to 900 nm cover one hemisphere of Deimos [Grundy and Fink, 1991]. This wavelength range contains absorptions due to ferric ironcontainingsilicates as are present,for example,in the spectra of low-gradeC2 carbonaceouschondrites[Gaffey, 1976] and manylow-albedo astereoids[Vilas, 1994]. By virtue of their Mars Pathfinder measurements of Phobos and Deimos continuous wavelength coveragefrom 440 to 1000 nm, IMP address three major objectives. The most important is spectral measurementscomplement the post-1989 data characterization of the basic spectralcharacterof the satellites, describedabove and allow morethorough assessment of the using a single instrumentcovering single geographicregions occurrenceand identity of Fe-mineral absorptions. The simultaneously at all observed wavelengths. Such basic occurrenceof Fe mineralabsorptionsis an importanttest for observations are neededto confirmthe differencein properties the two divergent compositional interpretations outlined indicated by recent measurementsfrom the properties above:a ferric-containing silicateabsorptionwould suggest suggestedby Viking-era measurements,and to provide an an affinity with primitive materialsthat contain such minerals, independentvalidation of previously published post-1989 such as low matemorphicgrade carbonaceouschondrites; spectralmeasurements. occurrenceof a pyroxeneor olivine absorptionwould indicate Second,in Phobos' bluer unit, color appears correlated a more evolved composition, perhaps a higher-grade with albedo,with higher albedos correspondingto less red carbonceous chondriteor a pyroxene-richassemblage suchas colors. This is of significance becausecolor and albedo are crustalmaterialejectedfrom Mars by largeimpacts. correlatedin mariemineral-richassemblageswhich are spaceweatheredto varying degrees,for example,lunar mare soils, 2. IMP Spectral Data and Calibration and such assemblagesare a possible spectral analog to the Martian satellites. Spectrally, Phobos' bluer unit is similar to The investigation strategy for observing the two moons both hemispheres of Deimos,and this raises the question of from Mars Pathfinderincludedmultiple measurements by IMP

9074

MURCHIE

ET AL.: PHOBOS'

AND

DEIMOS'

SPECTRAL

PROPERTIES

0.07

ß-

0.06

ß r.0.56 EB

'E

=LO05

004 o003

0.02

--

ß

- HST IMP

0.01

0.00-,, 0.050

:

Both data sets

I, ,,, 0.055

I,,,,

0.060

I, ,,, 0.065

--

I,,,,-

0.070

0.075

Viking Clear Filter Albedo Figure 4. Comparison of Viking clearfilter albedowith spectralslopeat 440-750 nm,fromIMP and HST data coveringDeimos and the bluer unit of Phobos. at different rotationalphases,coveringparts of the two moons with different albedo and spectral properties. During its nighttime observations(completedsuccessfullybeforefailure of the spacecraftpower system in Septemberand October 1997), IMP did acquiretwo measurements of parts of the subMars hemisphereof eachmoon [Smithet al., 1997a] (Table 1). Three of the four measurements were acquiredthroughall 12 of the IMP "geology" filters, in both eyes of the stereo camera [Smithet al., 1997b]. Measurementsat 440, 670, and 965 nm are duplicated in both eyes, allowing testing of the eyes'

representthree measurement"firsts" in the study of small bodies in the solar system: 1. For the first time, a single multispectral instrument measured both Martian moons under similar viewing conditions,over the full spectralrangeof 440-1000 nm.

2. This spectralrange overlaps that of most spectraldata acquiredpreviouslyby differentinstruments, thusservingas a

basisfor validating previousdata sets. 3. For the first time the 900- to 1000-nm wavelength region was measuredwith high precision for Deimos and the bluer intercalibration. One of the measurements of Deimos was unit of Phobos, allowing characterizationof the two moons' acquiredonly in right-eye filters (440-750 rim, 965 nm). All Fe-mineralabsorptionfeatures. observations consisted of 64 x 64 pixel subframesof the full Detailed descriptionsof data acquisition, processing,and 248 x 256 pixel images.To conserveddownlink, 2 x 2 pixel calibration are given in a companionpaper by Thomaset al. blocks were summed onboard, and data were compressed [this issue]. Both satellites were measured nearly losslessly. simultaneously with multiple standardstars,which were used For each moon, the two observations largely overlapped both to solve for atmosphericopacity and to check instrument spatially, but becauseof differentillumination geometriesat calibration.The datawere correctedto geometricalbedo using each observation the covered areas are not identical. On a triaxial ellipsoidshapemodel [Burns, 1992] and a published Deimos the observedregion is in the lowest albedopart of the phase function for the satellites [Klaasen et al., 1979]. satellite (Figure 2), similar to the leading hemisphere for Thomas et al. [this issue] show that the illumination of the which previousobservationsare available. On Phobos (Plate satellites is primarily by the sun with only a negligible 2), the region observedby IMP is expected,based on Phobos contributionof Marsshine, so that the acquiredspectramay be 2 imagedata, to be a mixture of the bluet and redder spectral comparedwith spectraof wholly solar-illuminatedbodies. units. The measuredregion contains the easternrim and flank One particular problem with IMP measurementsof the of Stickney,the highest albedo part of the satellite. Although satellites is their small apparentangular diameters,-•3 pixels the acquireddata do not samplethe full spectralheterogeneity (-•3 mrad)in the caseof Phobos and -•1 pixel (-•1 mrad)in the observed on Phobos, they are sufficient to corroborate case of Deimos. This becomes an issue because of the smallprevious measurementsand to place limitations on the scalestructureof the IMP CCD. Each pixel on the CCD is 23 x presenceand characterof Fe mineral absoptions. They also 23 grnin pitch, of which 17 x 23 grn is optically active. The

Table 1. Summaryof IMP SpectralMeasurements of PhobosandDeimos Date

Moon

Longitudeat

Aug. 30, 1997 Sep. 12, 1997 July 7, 1997

Phobos Phobos Deimos

Equator 300ø-92øW 324ø-90øW 308ø-86øW

July18, 1997

Deimos

270ø-89øW

PhaseAngle

Wavelengths

29ø 54ø 43ø

12 12 12

11ø

6

MURCHIE

ET AL.' PHOBOS'

AND

DEIMOS'

SPECTRAL

PROPERTIES

9075

remainderof eachpixel consistsof a part of an antiblooming

errors due to structure of the IMP

channel,which is optically inactive. Thus, for each Phobos or Deimosimage,a differentfraction of the light incident on the CCD may be digitized dependingon where on the CCD the

measurement consisted of an averageof threeexposures.

satellite imagefalls. For Phobos, which is more nearly an extendedsource,this is lessof a concern,but it is significant for the~l-pixel imagesof Deimos.Forthe multispectralimager ontheNEAR spacecraft, whichhasa similarCCD, the infiight measurement precision for point sources (analogous to Deimos)is ~12% [S. Murchie et al., Inflight calibration and performance of the NEAR Multispectral Imager,submittedto Icarus, 1998]. Pre-flightanalysisof the IMP CCD suggests the blur circle for a near-pointsourcelike Deimosis large enough at longer wavelengths (near 2 pixels) to minimize effectsof CCD nonuniformity,and that a smallerblur circle in shorter-wavelengthimages could lead to several percent uncertainty in measuredsignals. To minimize measurement

3. Results and Comparison to Previous Data Figure 5 compares IMP spectraof Phoboswith composite Phobos2 spectraof the redderand bluer units [Murchie and Erard, 1996]. The compositePhobos 2 spectraare available only for restricted locations, in which coverageby the

instruments overlapped. As it happened, the coverageoverlap in the bluerunit is in a particularly low-albedo location, and

the coverageoverlap in the redder unit is in a region somewhatdarkerthan is typical. In contrast,the geometric albedoof the region measuredby IMP is relatively high becauseit includes the bright region on the easternflank of the craterStickney(Figure 5a) [SimoneIllet al., 1998]. When scaledto unity at a 480-nm wavelength commonto both data

[[[[[][[][[[[[[[[[[[[[[[[[]][[[n][lllll • Phobos 2, bluer unit ß Phobos 2, redder unit

0.12

0.10-

---•

CCD, each satellite

,n

I--•

Vikingalbedo,IMP region I

=

IMP, 300ø-92øW IMP, 324ø-90øW .

0.08...

..

0.06n,

.

0.04-

3rd and 4th order polynomial fits

..

"''l""l'"'l''"l""l''"l""l'"'

300

5OO

7OO

9OO

11oo

Wavelength in Nanometers

a

1.8 I''n'ln''"ln'nnln'''lnnn'l''',lllllJllll

u) _• ,0--Phobos 2,bluer unit

o

• 1 6

Phobos2, redder unit

'• o1.4 E = IMP, 300ø-92øW '•'• _• [] IMP, 324ø-90 øW -Z

1.2-

I

Q)

-

r,.) (::)

-

,-OOl •-

•

ß

0- .

_.e mO.8Q) C:)

'

n",- O.6 -

3rd and 4th order polynomial fits ''"l''"l""l'"'l''"l""

300

500

I''"l""

700

900

11oo

Wavelength in Nanometers Figure5. IMP spectraof Phobos, compared to Phobos2 spectraof the bluerandredderspectralunits which bothoccurin themeasured area.(a) Geometric albedo.Thevalueexpected fromVikingclearfilter albedoof the measured areasis shownfor comparison. (b) Scaledreflectance.The IMP spectraareintermediate to the two spectralunits, as expected.

9076

MURCHIE ET AL.' PHOBOS' AND DEIMOS' SPECTRAL PROPERTIES

m

-

HST, leading hemisphere

m

m

011

iLI Viking Grundy and Fink, albedo, IMPleading regr hem. m m m

0.1 0 0.09

•

IMP, 2700-89ø W

--

IMP, 3080-86 ø W

m

m

m

m

m

0.08

m

m m

m

m m

m m

m

0.06

m

m

O.O5

m

m

0 O4

4th order polynomial fits

, "'1''"1""1'"'1""1''"1''"1'"' 500 700 3OO

900

m

11oo

Wavelength in Nanometers Figure6. IMP spectra of Deimos, compared withtelescopic andHSTmeasurements of theleading hemisphere. sets (Figure 5b), eliminating albedo differences,the IMP spectraare intermediatein shapeto the redderand bluer spectral unitsbutarecloserto theredderunit.Apparently, the redder unit dominatesthe observedpart of the sub-Mars hemisphereof Phobosjust as it dominatesthe anti-Mars hemisphere.

Figure6 compares IMP spectraof Deimoswith terrestrial ground-based and HST spectraof the leading hemisphere, whichis comparablylow in albedoto the regionmeasured by

channels. The level of uncertainty in single-channel measurements is exemplifiedby measurements at 670 and 965 nm,which were taken through comparablefilters in different

eyes.The discrepancies are of the samemagnitudeas singlechannel spectral "features."To circumventthis source of uncertainty,we fit both IMP and previous spectrausing a fourth-orderpolynomial,exceptfor the Phobos2 spectrumof Phobos' bluer unit, for which a third-order polynomial was used.For the IMP spectraof Phobos (Figure 5), the fits are highly reproducibleand characterizedby a broad,but very shallowfeaturenear700 nm anda possible shouldernear965 nm.The 965 nm shoulder is highly uncertainbecauseit is stronglyinfluencedby the single 1000 nm channel.The 700nm featureis also marginal,becauseuncertaintiesin the data alsoallow an almoststraight-linespectrumin this wavelength region.Analogousfitting of Phobos2 datacoveringthe same wavelength rangerevealsa similarbroad,very shallowfeature near700 nm,which was not pursuedby Murchie and Erard [1996] becauseof the possibility of mismatchin data

IMP. The differencesbetween the two IMP spectraare only marginalbecause of the low signal-to-noiseratio of the righteye-onlyspectrum.Overall, the IMP spectracompareclosely in shape with previous measurements of the leading hemisphere. However,the matchwith albedopredictedfrom Viking clear-filterdata [Thomaset al., 1996] is relatively poor.Thismaybeaneffectof structure of the IMP CCD, errors in the triaxial ellipsoid model of satellite shape used for photometric correction,or errorsin the absolutescaleof either or bothdatasets.The good agreementwith HST and groundbased telescopic data (also scaled to Viking albedo calibration between instruments.(Phobos 2 data at 770 nm comefromISM or channelof the imager.)The detailsof the IMP mostlikely sourceof error is Viking albedo of the sub-Mars the near-infrared of Deimosaremoreambiguous dueto greaterscatterin hemisphere, whichis more affectedby shadowsthan are other spectra the data,poorreproducibility, andthe lack of two spectrawith partsof the satellite. The relationship between Viking albedo and visible- identicalwavelengthcoverage(Figure 6). wavelengthcolor, fromIMP and previous data, is shown in Figure4. Only the IMP measurements of Deimosare pertinent 4. Discussion to testingthis relationshipbecausethe Phobos measurements The IMP measurements describedhereand by Thomaset al. covermainly the redderunit, which is distinctive spectrally. Addition

of

the

IMP

measurements

of

Deimos

to

the

[this issue]clearlycorroboratethe post-1989view of the

preexistingspectraldata covering Deimosand Phobos' bluer grossspectral properties of Phobosand Deimos,asdescribed unit actuallylessensthe correlationbetweenalbedoand color above.That is, both Phobosand Deimosare actuallyred rather wavelengths,and they to insignificance. The resultsfrom Mars Pathfinderthusdo not thangrayat visible and near-infrared to C-typeasteroids.In fact,they areredder support a consistent relationship of albedo and color arenot analogous than most low-albedo asteroids, and among low-albedo throughoutDeimosand the bluer unit on Phobos. to very red D-types.Also, the IMP The evidencefor Fe mineral absorptionsis contained in asteroidsarecomparable Figures5 and6. The uncertainties in individual measurements resultscorroboratethe compositePhobos2 spectraderivedby leads us to disregard"features"restrictedto single spectral Murchie and Erard [1996] and the HST spectraof the

MURCHIE

ET AL.'

PHOBOS'

AND

satellites [Zellner and Wells, 1994]. By implication, inferencesabout the composition, origin, and evolution of Phobosand Deimos basedon a reportedC-type spectrummay not be valid.

The possible broad, shallow absorption-like 700-nm feature in the spectrum of Phobos is suggestive of an Fecontaining phase. One possible explanation of this feature,if

indeed it isreal,isanFe2+-Fe 3+charge transfer in some typeof

DEIMOS'

SPECTRAL

PROPERTIES

9077

chondrites. If the 700-nm feature is real on Phobos, its

presencewould be counter to the general correlation of the 700-nm and 3000-nm bands among low-albedo asteroids becauseall observationalevidence to date [Bell et al., 1989b; Bibring et al., 1990; Langevin, 1991; Murchie and Erard, 1996] suggeststhat Phobosand Deimos lack the the 3000-nm H20 absorption. Vilas [1994] identified specific asteroids which exhibit a 700-nm featurebut lack an absorptionat 3000 nm, and suggestedthat their ferric-containingsilicates may be desiccated.This explanationmay be particularlytenable in the

ferric-containing silicate. Vilas [1994] presented evidence that this feature is widespread among low-albedo asteroids, and that it is correlated with strength of the 3000-nm case of Phobos and Deimos because of their unusual absorptiondue to chemically bound H20. Vilas interpreted dynamical environment: ejecta which escapethe moons are the phaseresponsibleas a ferric-containinghydrated silicate, mostly gravitationallyboundto Mars, and they reaccreteon a such as is present in low metamorphicgrade carbonaceous short timescale,resulting in a level of impact processing

1.8 • .......

1.6-

m

Phobos from IMP All soil

1

m

m

m

• a

1

1.4-

D class avg C classavg T class avg

m

•

ß

m

m

m

m

m

1.2-

m

m

m

m

1.o-

m

mm

m

m

0.8-

mm

m ,

m

m

0.6 300

500

700

900

11oo

Wavelength in Nanometers 0.12 ----

Phobos from IMP All soil / Halon

a •

T classavg D classavg

.......

0.10-

•o 08-

m

_•

, • =1..

= C.class av• ' •'1'

oE 006o04-'

1

m

0.02

300

b

500

700

900

1100

Wavelength in Nanometers

Figure ?, Comparison the IMP spectrumof Phoboswith the spectraof severalpossiblecompositional analogs to the Martian satellites.(a) Scaledreflectance.(b) Albedo.

9078

MURCHIE

ET AL.: PHOBOS' AND DEIMOS'

unusual for the regoliths of such small bodies [Hamilton, 1996]. Alternatively,the 700-nmbandmayitselfbe part of a

SPECTRAL PROPERTIES

Murchie,K. Lumme,K. Muinonen,J. Peltoniemi, T. Duxbury,B.

Murray, K. Herdenhoff,F. Fanale,W. Irvine, and B. Smith,Results of TV imagingof Phobos(ExperimentVSK-FREGAT), Planet. broad, composite absorption due to ferrous-containing Space Sci.,39, 281-295, 1991. silicatessuchas olivine or chromianpyroxene[cf. Vilas et al., Bell, J., D. Davis, W. Hartmann, and M. Gaffey, Asteroids:The big 1997]. This origin would be consistentwith the apparent picture,in AsteroidsII, editedby R. Binzel, T. Gehrels,and M. Matthews,pp. 921-945,Univ. of Ariz. Press,Tucson,1989a. 1000-nm band in Phobos 2 ISM spectra. Certain ferric oxyhydroxidesand oxides (e.g., hematiteand goethite) also Bell, J., J. Pisciteili,and L. Lebofsky, Deimos: Hydration state from infrared spectroscopy(abstract), Lunar Planet. Sci., XX, 58-59,

exhibit an absorptionat 600-700 nm, but they also exhibit very strongabsorptionsat shorter wavelengthsand at 850900

nm

which

are

absent

fi'om

IMP

and

Phobos

2

measurements.

What do measurements of Phobos and Deimosby IMP add to our knowledge of the moons' composition and their relationship to asteroids?The major contribution of the IMP datais confirmationof very red spectraof the Martian moons. However, this red spectral slope cannot, by itself, be construed as evidence for the satellites' composition. As shown in Figure 7a, a similar red slope occursin putatively compositionallyprimitive assemblages(D-like asteroids)as

1989b.

Bell, J., F. Fanale,andD. Cruikshank,Chemicaland physicalproperties of the Martiansatellites,in Resources of Near-EarthSpace,editedby J. Lewis,M. Matthews,andM. Guerrieri,pp. 887-901,Univ. of Ariz. Press,Tucson,1993. Bibring, J.-P., M. Combes, Y. Langevin, C. Cara, P. Drossart, T. Encrenaz, S. Erard, O. Forni, B. Gondet, L. Ksanfomality,E. Lellouch,P. Masson,V. Moroz, F. Rocard,J. Rosenqvist, C. Sotin, andA. Souffiot,ISM observationsof Mars and Phobos:First results, Proc. Lunar Planet. Sci. Conf., 20th, 461-471, 1990. Burns,J., Contradictoryclues as to the origin of the Martian moons,in Mars, editedby H. Kieffer, B. Jakosky,C. Snyder,and M. Matthews, pp. 1283-1301,Univ. of Ariz. Press,Tucson,1992.

well as in chemically highly evolved assemblages (space-

Duxbury, T., and J. Callahan, Phobos and Deimos cartography (abstract),LunarPlanet.Sci.,XIII, 190-192, 1982.

weatheredmaficmineral assemblages,such as the lunar mare

Duxbury, T., and J. Callahan, Phobosand Deimos control networks,

soils). ?hobos and Deimos are much redder than the asteroid

analogsproposedpreviously,suchas C- and P-type asteroids. The

IMP

results

show

that

?hobos

and

Deimos

are

Icarus, 77, 275-286, 1989.

Gaffey, M., Spectral reflectance characteristicsof the meteorite classes, d. Geophys.Res.,81,905-920, 1976. Grundy,W., andU. Fink,Deimos:A reddish,D-type asteroidspectrum, inAsterolds,Comets,Meteors 1991, pp. 215-218, Lunar and Planet.

comparablyred to D-type asteroids.The affinity is especially Inst., Houston, Tex., 1991. closeto T-type asteroids,a spectraltype relatedto D-types Hamilton, D., The asymmetrictime-variablerings of Mars, Icarus, 119, but with a strongerUV reflectancefalloff and a shoulder near 153-172, 1996. 900 nm.However,the albedosof the satellitesare atypically Klaasen,K., T. Duxbury,and J. Veverka, Photometryof Phobosand Deimosfrom Viking Orbiter images,d. Geophys.Res., 84, 8478high for D-or T-asteroids [Tedesco, 1989], and more 8486, 1979. comparable to the lowest albedo space-weatheredmafic Ksanfomality,L., and V. Moroz, Spectral reflectivity of Phobos's assemblages, such as somelunar maresoils [Pieters, 1993] regolithin the range315-600 nm, Icarus, 117, 383-401, 1995. (Figure 7b).

Langevin,Y., Phobosandothersmallbodiesof the solarsystem,Planet. Space.Sci., 39, 377-394, 1991.

Murchie,S., and S. Erard, The spectralpropertiesand composition of 5. Conclusions Phobos1?ommeasurements by Phobos2, Icarus, 123, 63-86, 1996. Murchie,S., D. Britt,J. Head, S. Pratt,P. Fisher,B. Zhukov, A. Kuzmin, L. Ksantbmality, A. Zharkov,G. Nikitin,F. Fanale,D. Blaney,M. The IMP measurements of Phobos and Deimos corroborate Robinson, and J.F. Bell III, Color heterogeneityof the surfaceof and extendpreviously acquired,well-calibrated measurements Phobos.Relationshipsof geologic features and comparisonto of spatiallyoverlappingregionsof the two moons[Grundy meteoriteanalogs,d. Geophys.Res.,96, 5925-5945, 1991. and œink, 1991; Zellner and Wells, 1994; Murchie and Pang,K., J. Pollack,J. Veverka,A. Lane,andJ. Ajello, The composition

Erard, 1996]. Nearly all of the currently available data show that Phobos and Deimos are actually quite red and not analogousto C-type asteroids or to carbonaceouschondrites. The IMP data also provide tantalizing but inconclusive

evidence for thepresence of sometype of Fe mineralhavingan absorptionnear 700 nm. Observationsof the Martian satellites

of Phobos: Evidence for carbonaceous chondrite surface from

spectralanalysis,Science,199, 64-66, 1978. Pang, K., J. Rhoads, G. Hanover, K. Lumme, and E. Bowell, Interpretationof whole-diskphotometryof Phobosand Deimos,d. Geophys.Res.,88, 2475-2485, 1983. Pieters,C., Compositional diversityand stratigraphyof the lunar crust derived from reflectance spectroscopy,in Remote Geochemical Analysis Elementaland MineralogicalComposition,edited by C. Pietersand P Englert, pp. 309-340, Cambridge Univ. Press,New

fromthe surfaceof their parentplanet haveprovided us with greater confidencein the accuracy of currently available York, 1993 spectral data, but important questions regarding the Pollack,J, J Veverka, K Pang,D. Colburn,A. Lane, and J. Ajello, Multicolorobservations of Phoboswith the Viking Landercameras: composition, evolution,and origin of the enigmaticob,jects remain

Evidencefor a carbonaceous chondritecomposition, Science,199,

unresolved.

66-69, 1978.

Acknowledgments.

We wish to thank P. Thomas for

providing digital image and albedo maps of Phobos and

Simonelli,D., M. Wisz, A. Switala, D. Adinolfi, J. Veverka, and P. Helfenstein,Photometric propertiesof Phobossurfacematerialsfrom Viking images,Icarus,131, 52-77, 1998.

Deimos,and H. McSween,L. Lebofsky,and an anonymous Smith,P., J.F.Bell III, N. Bridges,D. Britt,L. Gaddis,R. Greeley,H. Keller, K. Herkenhoff,R. Jaumann,J. Johnson,R. Kirk, M. Letomort, reviewerfor constructivecommentson the manuscript.This J. Make, M. Malin, S. Murchie, J. Oberst, T. Parker, R. Reid, R. work was completedunder JPL contract960893.

Sablomy,L. Soderblom,C. Stoker, R. Sullivan, N. Thomas, M. Tomasko, W. Ward, and E. Wegryn, Results from the Mars Pathfindercamera,Science,278, 1758-1765, 1997a.

References

Smith,P., M. Tomasko,D. Britt, D. Crowe, R. Reid, H. Keller, N. Thomas,F. Glien,P. Rueffer,R. Sullivan,R. Greeley,J. Knudsen,M. Madsen,H. Gunnlaugsson, S. Hviid, W. Goetz, L. Soderblom,L.

Avanesov,G., B. Zhukov, Ya. Ziman, V. Kostenko,V. Murav'ev, V. Fedotov,B. Bonev,D. Mishev,D. Petkov,A. Krumov,S. Simeonov, Gaddis,andR. Kirk, The imagerfor Mars Pathfinder experiment, d. Geophys.Res., 102, 4003-4026, 1997b. V. Boycheva,Yu. Uzunov,F. Kempe, B. Rebel, G.-G. Weide, D. Halmann, W. Neumann, I. Petuchova,W. Possel,J. Head, S. Thomas, N., D. Britt, K. Herkenhoff, S. Murchie, B. Semenov, H.

MURCHIE ET AL.: PHOBOS' AND DEIMOS' SPECTRAL PROPERTIES Keller, and P. Smith, Observationsof Phobos,Deimos, and bright starswith the Imager for Mars Pathfinder,J. Geophys.Res., this issue.

Thomas,P., J. Veverka, J. Bell, J. Lunine, and D. Cruikshank,Satellites of Mars: Geologichistory,in Mars, editedby H. Kieffer, B. Jakosky, C. Snyder,and M. Matthews,pp. 1257-1282,Univ. of Ariz. Press, Tucson, 1992. Thomas, P., D. Adinolfi, P. Helfenstein, D. Simonelli, and J. Veverka, The surfaceof Deimos:Contributionof materialsand processes to its uniqueappearance, Icarus, 123, 536-556, 1996. Veverka, J., and T. Duxbury, Viking observationsof Phobos and Deimos:Preliminaryresults,J. Geophys.Res.,82, 4213-4223,1977. Veverka, J., P. Thomas,A. Harch, B. Clark, J.F. Bell Ill, B. Carcich, J. Joseph,C. Chapman, W. Merline, M. Robinson,M. Malin, L. McFadden,S. Murchie, S. Hawkins, R. Farquhar,N. Izenberg, and A. Cheng,NEAR's flyby of 253 Mathilde:Images of a C asteroid, Science,278, 2109-2114, 1997. Veverka,J., P. Thomas,A. Harch,B. Clark, J.F. Bell III, B. Carcich, J. Joseph,S. Murchie, N. Izenberg, C. Chapman, W. Merline, M. Malin, L. McFadden, and M. Robinson, NEAR encounter with asteroid253 Mathilde: Overview, Icarus, in press,1998. Vilas, F., A cheaper,faster,better way to detectwater of hydrationon solarsystembodies,Icarus, 111,456-467, 1994.

9079

Vilas, F., E. Jensen,and L. McFadden, Extractingspectralinformation about253 Mathilde usingthe NEAR photometry,Icarus, 129, 440449, 1997.

Zellner, B. and E. Wells, Spectrophotometryof the Martian satellites with the HubbleSpaceTelescope(abstract),Lunar Planet.Sci.,XXV, 1541-1542, 1994.

J.F. Bell III, Centerfor Radiophysicsand SpaceResearch,424 Space SciencesBuilding,CornellUniversity,Ithaca,NY 14853. D. Britt, Lunar and Planetary Laboratory, University of Arizona, Tucson, AZ 85721.

K. Herkenhoff, Jet PropulsionLaboratory, Pasadena, California 91125.

S. Murchie, Room 7-368, Applied PhysicsLaboratory, The Johns HopkinsUniversity,JotmsHopkins Rd., Laurel, Maryland 20723. (email: scott_murchie•jhuapl.edu) N. Thomas,Max Planck Institutefor Aeronomy, Postfach20, 37189, Katlenburg-Lindau,Germany.

(ReceivedFebruary25, 1998;revisedJune23, 1998; acceptedJune30, 1998.)