sign in sign up
<<
Home , Fretted terrain, Hydaspis Chaos, Hydraotes Chaos, Iani Chaos, Ravi Vallis, Shalbatana Vallis

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 99, NO. El, PAGES 1983-1996, JANUARY 25, 1994

Thermal studiesof channelsand valleys using Termoskan data

BruceH. Betts andBruce C. Murray Divisionof Geologicaland PlanetarySciences, California Institute of Technology,Pasadena

The Tennoskaninstrument on boardthe '88 spacecraftacquired the highestspatial resolution thermal infraredemission data ever obtained for . Included in thethermal images are 2 km/pixel,midday observations of severalmajor channel and valley systems including significant portions of Shalbatana,Ravi, A1-Qahira,and Ma'adimValles, the channelconnecting Vailes Marineris with HydraotesChaos, and channelmaterial in Eos .Tennoskan also observed small portions of thesouthern beginnings of Simud,Tiu, andAres Vailes and somechannel material in GangisChasma. Simultaneousbroadband visible reflectance data were obtainedfor all but Ma'adimVallis. We find thatmost of the channelsand valleys have higher thermal inertias than their surroundings,consistent with previousthermal studies. We show for the first time that the thermal inertia boundariesclosely match flat channelfloor boundaries.Also, butteswithin channelshave inertiassimilar to the plainssurrounding the channels, suggesting the buttes are remnants of a contiguousplains surface. Lower bounds ontypical channel thermal inertias range from 8.4 to 12.5(10 -3 cal cm-2 s-1/2 K-I) (352to 523 in SI unitsof J m-2 s-l/2K-l). Lowerbounds on inertia differences with the surrounding heavily cratered plains range from 1.1 to 3.5 (46 to 147 sr). Atmosphericand geometriceffects are not sufficientto causethe observedchannel inertia enhancements.We favornonaeolian explanations of the overall channel inertia enhancements based primarily upon the channelfloors' thermal homogeneity and the strongcorrelation of thermalboundaries with floor boundaries. However,localized, dark regions within some channels are likely aeolian in natureas reported previously. Most channelswith increased inertias have fretted morphologies such as flat floorswith steep walls. EasternRavi and southernAres Vailes, the only major channel sections observed that have obvious catastrophic flood bedforms, do nothave enhanced inertias. Therefore, we favorfretting processes over catastrophic flooding for explainingthe inertiaenhancements. We postulatethat the inertia enhancements were caused either by the original fretting process or by a processinvolving the bondingof finesdue to an increasedavailability of water,either initially or secondarily.

INTRODUCTION missions.The term channelhas beenwidely used for Mars, Enormouschannels and valleys are some of Mars' most althoughit is somewhaterroneous in itsusage [Sharp and Malin, intriguingfeatures. Most, includingthose studied here, are now 1975;Carr, 1981]. For simplicity,we usethe term channel to generallyaccepted to have been cut by water or ice related refer collectivelyto featurespreviously classed as channelsor processes[Carr, 1981;Baker, 1982;Baker et al., 1992]. These valleys. processesprobably included catastrophic flooding and sapping processes.Studies of Martian channelsurface properties and BACKGROUND morphologiesyield importantimplications for Mars' geologic, hydrologic,and climatic history. The Termoskan Instrument and Data The SovietPhobos '88 Termoskaninstrument provided the In Februaryand March 1989 the Termoskaninstrument on highestspatial resolution thermal data obtainedso far for Mars boardthe Phobos'88 spacecraftof the USSR acquireda limited [Selivanovet al., 1989; Murray et al., 1991; Betts, 1993], set of high spatial resolutionsimultaneous observations of includingobservations of severallarge equatorial channels and reflected solar flux and emitted thermal flux from Mars' valleys.Here we presentthe results of thefirst detailed study of equatorialregion. Theseimage panoramas cover a largeportion channelsusing the Termoskan data. We includea descriptionof of the equatorialregion from 30øS to 6øN latitude. Termoskan theinstrument and the observations, a description of thechannels was an optical-mechanicalscanning radiometer with one visible observed,a review of geologicclassifications and previous channel(0.5-1.0 g,m) and one thermal infrared channel (8.5-12.0 thermalstudies, qualitative results and implications, quantitative gm). Theinstrument was fixed to thespacecraft, pointing in the thermal inertia determinationsand implications,critiques of antisolardirection. Thus, all observationsare at 0ø phaseangle possiblehypotheses, and proposedtests using future Mars andonly daytime observations were acquired. More complete descriptionsof the Termoskaninstrument and data appearin Murray et al. [1991] andBetts [1993]. 1Nowat the San Juan Capistrano Research Institute, San Juan Capistrano, California. Termoskan'sbest resolution per pixel was 1.8 km for threeof the panoramasacquired and 300 m for the remainingpanorama Copyright1994 by the AmericanGeophysical Union. [Selivanovet al., 1989;Murray et al., 1991]. Theseresolutions are much better than thoseobtained by the Viking infrared Paper number93JE03173. thermalmapper (IRTM) (approximately5 to 170 km/pixel,with 0148-0227/94/93 JE03173505.00 onlya smallfraction of the datanear 5 km/pixel,and a typical

1983 1984 BETTS AND MURRAY: THERMAL STUDIES OF MARTIAN CHANNELS valueof 30 [Christensen,1986]). Termoskan'sspatial resolution In addition to , several other channels lead either is also better than the 3 km/pixel that was expectedfor Mars into or out of HydraoresChaos (see Figure 1). A large, flat Observer'sthermal emission spectrometer (TES), althoughTES channelenters Hydraores Chaos from to the observationswould have providedglobal 1400 and 0200 local . We will refer to this channelby the unofficial name, time(LT) spectralcoverage. HydraoresChannel. Regionsof chaoticterrain occur both to the Thermal inertia, a bulk measure of the resistance of a unit southand to the northof this channel. Anotherflat, steepwalled surfacearea to changesin temperature,is commonlyused to channel at the northwestcomer of HydraoresChaos begins characterizethe insulatingproperties of planetarysurfaces. It is Simud Vallis. Only approximately75 km of this channelwere definedasI = (kpcp)112,where kis the thermal conductivity, pis observed north of the chaos. At the northeast corner of thedensity, and c3 is thespecific heat. Low-inertia materials Hydraores Chaos, Termoskanobserved about 150 km of a exhibit the largestday-to-night temperature variation and the channel(here called Tiu West) that splitsaround a large butte. smallestthermal skin depths. We use the units for thermal This channelthen meets anotherobserved channel (here called inertia often usedfor the [e.g., Kieffer et al., Tiu East) comingfrom HydaspisChaos to the east. When these 1977]:10 '3 calcm '2 K'1 s'it2. Thermalinertias in SI units(J m'2 setsof channelsmeet north of the Termoskancoverage area, they K'1 s '1/2) can be obtained bymultiplying by41.86. form Tiu Vallis proper. Smallportions of the headwardreaches of AresVallis were alsoobserved. Most of thispart of Aresdoes ChannelDescriptions and Geographic and Geologic Settings not show flat floors, but rather appearsscoured and is locally Termoskanobserved several large channelsnear the eastern anastomofic[Sharp and Malin, 1975]. Simud, Tiu, and Ares, end of Valles Marineris including significantportions of like ShalbatanaVallis, all debouchinto ChrysePlanifia several hundred kilometers downstream. ShalbatanaVallis, Ravi Vallis, the channelconnecting Valles Channel materials were also observed in two of the eastern Marineriswith HydraotesChaos, and channelmaterial in . In the sameregion, Termoskan also observed small Valles MarinerisChasma: Eos andGangis (see Figure 1). Flow portionsof the southernbeginnings of Simud,Tiu, and Ares from theseregions presumably headed to the eastand eventually Valles as well as channelmaterial in the northernportions of northeastin the directionof HydraoresChannel and Chaos. Only GangisChasma. On the otherside of the planet,Termoskan the northernmostpart of Gangis was observed. A separate observedtwo majorvalleys in the AeolisQuadrangle: A1-Qahira Termoskanpanorama shows most of EosChasma. Both chasma Vallis and Ma'adim Vallis (see Table 1). All the channel contain flat, smooth appearingareas classifiedby Scott and sectionsobserved by Termoskancut throughancient cratered Tanaka [1986] as Hesperianchannel materials. The channel terrainof Noachianage [Scottand Tanaka,1986; Greeleyand materialsare situatednext to steepwalls, buttes,and at leastin Guest, 1987]. EosChasma, between regions of chaoticterrain. ShalbatanaVallis (see Figure 1) appearsto emanatefrom a Termoskan also observedMa'adim Vallis and A1-Qahira zone of chaotic terrain at 0øN, 46øW and heads northward.It Vallis, two isolated channels in the Aeolis qua&angle. narrowsto a low-sinuositychannel with a reasonablyuniform Termoskanobserved the northernmost(distal) 350 km of the 700 width of approximately10 km. It eventuallysplits into two km long,gently winding, 15- to 25-km-wideMa'adim Vallis. It distributaries. In all, it extendsover 1000 km. Termoskan headsnorthward until hookingnorthwest after breaching a 30-km observedapproximately the southern400 km of the channel. crater. It debouches into another 30-km crater. Ma'adim is Just to the east of Shalbatanais the 300-km-longRavi Vallis, unusuallyold for a large channel[Baker, 1982; Masurskyet al., whichalso emanates from a regionof chaoticterrain (Aromatum 1980]. Ma'adimhas steep walls and smoothfloors except where Chaos). The channelthins and proceeds east, eventually ending benches exist. in thewestern portion of HydraotesChaos. In contrastwith most Al-QahiraVallis is locatedapproximately 800 km to the west of the channelsdiscussed here, easternRavi has groovedterrain of Ma'adim. Termoskan observed all 300 l•m of this channel. It on its floor. In addition,its walls are not as high or as steepas originatesfrom short tributaries,runs mainly east, then turns those of the other observed channels. northand widens as it takesa very straightcourse. It alsohas a

TABLE 1. ChannelLocations, Seasons Observed, and Types Channel Channelsection observed Ls T•,[•e ,

20øS,199øW to 14øS,195øW 18ø Runoffchannel [M76]; longitudinalvalley [B92];fretted channel [C81]

20øS,183øW to 16øS,184øW 18ø Runoffchannel [SM75]; longitudinalvalley [B92]; frettedchannel [C81]

Shalbatana 2øS,46øW to 5øN, 44øW 6ø Outflow channel[SM75]

Hydraotes Channel, 7øS,36øW to 5øN, 37øW 6ø Simudand Tiu: outflowchannels [SM75]; the observedportion of Simud: Simud, and Tiu Vailes fretted [S73]

Ravi Vailis 2øS,44øW to 0øN, 39øW 6ø Outflow channel[B82]

Ares Vailis løS, 16øW to 5øN, 19øW 6ø Outflow channel[SM75]

Eos Chasma 15øS,44øW to 10øS,37øW 18ø

GangisChasma 8øS,49øW to 7øS,44øW 6ø Classificationsarefrom th• following sources: [S73]: Sharp [1973]; [SM75], Sharp and Malin [1975]; [M76], Mutch etal. [1976]; [C81], Carr [1981]; [B82], Baker [1982]; and [B92], Bakeret al. [1992]. BETrS AND MURRAY: THERMALSTUDIES OF MARTIAN CHANNELS 1985

%*:½7>

Fig. 1. Viking photomosaic(a portionof file ME00N045from U.S. GeologicalSurvey [1991a]) of the channelsstudied in the easternVallis Marinerisregion centered approximately upon 4øS, 33øW. Northis at top in all imagefigures. Note thatvirtually all the channelshave smoothflat floors and steep,scalloped walls, suggestiveof fretting. broad, flat floored channeland heavy cratering[Baker, fluvial flow. Longitudinalvalleys may have begunas small 1982]. Like Ma'adim, its terminationis rather indistinctand valleys,then becomeenlarged by wall retreat as lower courses showsa markedlack of largescale deposits. becamedeeply incised [Baker et al., 1992]. Carr [1981] classes Al-Qahira and Ma'adim as fretted ChannelClassifications channels.Indeed, whatever caused the original valleys, the wide, Martian channelshave been classifiedby several authors. flat floorsand steepwalls with scallopedappearances indicate Table I summarizesprevious geologic classifications for the that a frettingprocess [Sharp, 1973] hasbeen active for Ma'adim observed channels. All of the named channels in the eastern and Al-Qahira. Fretting presumablyinvolved sappingof VallesMarineris region are classifiedat leastin part as outflow groundwateror ice, causingundercutting of the walls. Debris channels[e.g., Sharp and Malin, 1975;Cart, 1981;Baker, 1982; flows,possibly facilitated by ice (as suggestedfor frettedterrain Baker et al., 1992]. These includeShalbatana, Ravi, Tiu, Simud, by $quyres[1978]), may have then movedmaterial away from and Ares. Sharp and Malin [1975] define outflowchannels as the walls, allowing more undercuttingerosion to occur. We mostlylarge featuresthat start full-bom from localized sources. emphasize,however, that fretting is not a well-understood They are broadestand deepestat their head. Someare scoured process,but it is morphologicallywell definedfor Mars. and display features characteristicof catastrophicflooding. Fretting also appearsto have occurredin the portionsof Many originate from chaotic terrain. are channelsobserved by Termoskannear easternValles Marineris. generallyaccepted to have originally formed by catastrophic Againthis is basedupon the steepwalls and fiat, smoothfloors. flooding,in somecases from releaseof waterfrom chaoticterrain In fact,the channel at thenorthwest of HydraoresChaos (leading [Bakeret al., 1992; Cart, 1986]. to Simud)is specificallyshown by Sharp[1973] as an example Sharp and Malin [1975] classedMa'adim Vallis as a runoff of frettedchannels next to chaoticterrains. Thus, many of the channel,and Mutch et al. [1976] classedAI-Qahira Vallis (which channelsobserved show evidenceof fretted morphologies, was not discussedby Sharp and Malin) as a runoff channel. indicatingthat frettingwas at the very least the last major Sharpand Malin [1975]def'me runoff channels as startingsmall, processto influencetheir large scale morphology. Notable increasingin size and depth distally and having tributary exceptionsare eastern Ravi Vallis and southernAres Vallis, branches,and crustal control may be strong.Baker et al. [1992] whichdo not havewalls that are as steepand whichhave rough class Ma'adim and Al-Qahira Vallis as longitudinalvalleys. floor features, including features such as groovesthat are Mars Channel Working Group [1983] distinguishedMartian indicative of catastrophicflooding. Termoskanobserved few valleysfrom channelsby the absenceof bedformsindicative of classic outflow channel morphologies,although it narrowly 1986 BETTSAND MURRAY: THERMALSTUDIES OF MARTIAN CHANNELS missed several north of the area at the eastern end of Valles Implications Marineris. Extensionof this analysisto more classicoutflow We can draw the followingconclusions from the qualitative channelsusing future mission data will be very interesting. observations listed in Table 2: 1. The floors of virtually all the channels observed Previous Thermal Studies consistentlyhave higher inertia (implyingcoarser material, more Several researchers have undertaken thermal studies of bonded material, or more rocks) than their surroundings, Martian channelsand valleysusing Viking IRTM data. These consistentwith previousstudies. analysesand the channelsstudied include Christensenand 2. Boundariesof thermalinertia (which representsthe upper Kieffer [1979], Kasei,Ares, Shalbatana,Simud, and ; few centimetersof the surface) closely match channel floor Zimbelrnan[1986] and Zimbelmanand Leshin [1987] A1-Qahira boundaries,particularly for wide, flat floors. and Ma'adim Valles; and Craddock et al. [1988] and Craddock 3. Dark, presumablyaeolian deposits, do not dominatethe [1987], Dao, Hormakis, Ma'adim, Mangala, and Shalbatana inertia of the channels as a whole. This contrasts with Valles. These studiesconcluded that manychannels and valleys conclusionsdrawn by some previousresearchers [Zimbelman, have higher thermal inertia than their surroundings. In 1986; Craddocket el., 1988], whoseresults were basedupon explainingthe causeof the inertiaenhancements, these studies lower-resolution,nonimaging IRTM data. In Termoskan emphasizedthe presenceof dark, high-inertia,presumably observations,channel inertias are still higher than thoseof the aeoliansaltation deposits within thechannels. surroundingseven outside the localized,dark deposits.The dark Zimbelrnan[1986] and Craddocket al. [1988] concludedthat depositsare very likely saltationtraps for dark sand,similar to severalcentimeter thick aeoliandeposits dominate the inertia of the localizedintracrater deposits seen near somechannels and the channelfloors. At least for the channelsthey studied,they investigatedplanet-wide by Christensen[1983]. We discussthis concludedthat thermal observationsmay not be related to the morefully in the aeolianexplanations section. processesthat producedthe channels.In contrast,we conclude 4. Channelsare examplesof featureswhose inertia correlates that thermal observationsof much of the channelfloors may be well with morphology,which is rare on Mars [Christensenand sampling some material and textures from channel floor Moore, 1992]. formation. IRTM studiesof channelswere limited either by insufficientspatial resolutionto resolve the channelsor by QUANTITATIVETHERMAL INER• DETE•A•ON limited arealcoverage of the highest-resolutiondata. In contrast, Termoskandata provide high-resolutionimages with nearly Method completespatial coverage (i.e., no gapsor gores). We have used the Termoskan data in combination with thermalmodeting and albedoinformation from Viking to derive QUALITATIVEANALYSES thermalinertias for pointswithin channelsand for pointson the surroundingplains. For A1-Qahiraand Ma'adim Valles, we Observations choseapproximately 10 locationsinside and 10 outsideeach Termoskanobtained thermal images of severalchannels in the channel. For the other, shorter, channel segments, one easternVailes Marinerisregion (see Figures 2, 3, and 4) andof representativechannel point and one representative surrounding Al-Qahira Vailis (Figure 5) and Ma'adim Vailis (Figure 6). pointwere chosen. Points inside each channel were selectedto Simultaneous broadband visible channel data were obtained for representthe channel, to avoid large slopes that would all but Ma'adim Vallis. All of the channel systemswere significantlyalter inertia determinations, and for A1-Qahiraand observednear midday,between 9.87 H and 13.00 H, exceptEos Ma'adim, to give goodcoverage over the lengthof the channel. Chasmaat 15.15 H (where 24 H = 1 Martian day). All were For eachlocation inside, a pairedpoint devoid of extremeslopes observedwith an approximateresolution of 1.8 km/pixel and at was chosennearby on the surroundingplains. Areas were nearly 0ø phase angle. Mangala Vailis was also observed,but determinedto be devoidof extremeslopes based upon a lack of we do not discussit in detail. The Mangala data are badly apparentsun-facing (bright and warm) or anti-sun-facing(dark foreshortenedand interpretationis further complicatedby the and cool) slopesbased upon both Termoskandata and Viking early morning (postdawn)and late afternoon(presunset) local Orbiter camera images. For each location, we noted the times of the observations. temperatureand visible signal from single Termoskanpixels. The generalcharacteristics of all the channelobservations are Thesesingle-pixel values generally matched to within I K and summarizedin Table 2. Termoskan'shigh-resolution images oftento within0.5 K of the averageof a 3 x 3 pixel box (if the show for the first time that thermal boundariesvery closely box was entirely within the channel). Latitude and longitude matchchannel floor boundaries,usually to within the resolution were determinedfrom USGS photomosaics.Local time of day of the instrument. For thesemidday observations, the coolerand was calculatedfor eachpoint based upon its longitudeand the darker (or similar albedo)channels must have higher thermal absolute time of the observation. inertiathan their surroundings.The easternend of Ravi Vallis To derive thermal inertias, we used an adaptationof the and southernAres Vallis appeardifferent from mostof the other Clifford et al. [1987] finite difference,homogeneous thermal channelobservations. They do nothave flat floorsor steepwalls, model of the Mars surface. This model numericallysolves the andthey appearthermally similar to their surroundings.Thermal heat diffusionequation using the boundaryconditions of thermal distinctivenesswithin channels is strongly correlated with equilibrium at the surfaceand no heat flow acrossthe lower regionsthat have morphologiesindicative of fretting:flat wide boundary. Physically,this model is identical to the Viking floors and steep, scallopedwalls. Comparisonsare limited, thermal model describedby Kieffer et al. [1977, Appendix1], however,because few other channeltypes were observedby althoughcomputationally it differsslightly. Termoskan except Ravi and Ares, although many occur Ideally, thermal inertia is determined from diurnal elsewhereon the planet. observationsusing temperature alone, as was donefor example BETTSAND MURRAY: THERMAL STUDmS OF MARTIAN CHANNELS 1987

Fig. 2. (Top) Termoskanthermal and (bottom)visible imagescentered approximatelyupon løS, 39øW. North is top. In all thermalimages shown here, darkeris cooler. Shalbatana,Simud, and Tiu Valles all continuefor severalhundred kilometers north of this image. Note the cool and generallyuniform floors of all channelsexcept the eastern(and rough floored) end of Ravi Vallis. Note also that the thermalboundaries closely match the boundariesof the channelfloors and depart significantlyfrom albedoboundaries seen in the visible image. Note also the dark, presumablyaeolian deposits localized within the southern portionsof ShalbatanaVallis andthe southwesternportion of HydraoresChaos and spreadingonto the surroundingplains 'm both cases.Buttes, including the largelabeled one in the northeastof the image,within the channelsappear similar in temperatureand appearanceto the surroundingplains, not the channels. 1988 BETTS AND MURRAY: THERMAL STUDIESOF MARTIAN CHANNBLS

::::: :":"

.;

• ;--;:z:..; :: ...... :;:;.•.::.:: • '-• ::::::::::::::::::::::::::::::::::::::::::: :•.::•:-.:.&::::•:.:::::.:..::

•:.•:::.-:::. :&::. •::•:•

......

.:.:::½

. ,':.:;::•':': ...... ß':. .:.':'::-:•: .•:::?:...... •:½:.:::.:::•.. :-•::

. ::::::::::::::::::::::: ....• ,,,..:::::::::::::::::::::::::::::: '.., :. •--.:-{-.. . ;-,..======'.':::.•...½:•::}.•,,:-'-:•:,'..-,:•:...... •:. ,..":• ../{"½'½3: ...... -...... ;{,.,.:

:L'&......

ß .;::. .. '.:'.. {.;.>..,- ½....: ;:...::,:'-'- < ':.::::: -:.{ ::•. ..

......

.. :...

.

.

, ======

;..-:<:.... ::.... • . ::.... ::'.:

Fig. 3. (Top) Termoskanthermal and (bottom)visible images centered approximately upon løS, 23øW. North is top. Westem portionoverlaps slightly with Figure2. Verticalblack lines have been added where lines were missing in the originaldata. Note that the centralpart of that runsnorth from Iani Chaosis not thermallydistinct from its surroundings.This region showssignificant catastrophic flooding bedforms and lacksfretting morphologies. The smallerchannels that join the centralAres channelfrom the westand from the eastare in someareas cooler. However,as opposedto mostother channels, the coolerareas occurin patches,possibly indicative of aeolianprocesses. by Palluconi and Kieffer [1981] and Kieffer et al. [1977]. and used by Christensen[1983] and others). We solvefor the However, Termoskanacquired only one observationof each of inertia, I, in the expression the channels studied. Thus, we use an alternate method that uses 8Tm a single temperatureobservation combined with bolometric Tob = Tm+ ( • ) albedo (similar to what was describedby Kieffer et al. [1977] 8l(l-lm) BETTSAND MURRAY: THERMAL STUDIFS OF MARTIAN CHANNELS 1989

;:iSi.::::.;;-,:---,..:..•::' '*:*:*E•i.

..::...... :';:¾2 *"':::*'" q.::,.':'...... **:- ...... ?•':*;,:? .:,.,.:::.,.,-- :.-.:.•':*:':: *:*'< **•2: ...... ';:.?,;... .x;:'?'"::?':*' • ....:*::';*"*' ....:?' ":'*:""' ...... :,.:...... ;:,.'- * ... •;•:::::'>-:.:,:. :..'"'4•.,,--.-......

...... :f ';: :?...... :,a.::-c*:.:.::;-::":?½•;':. :..: ... ..-;;..*•;;-';½•':';'*:;::'::;:'...:'•;• ...... ,,'...... ? . '/;*¾.:...??

..

Fig.4. (Top)Termoskan thermal and Coottom) visible images showing Eos Chasma and centered approximately upon 14øS, 41øW. Within Eos Chasma, flat floored channel floor materials [Scott and Tanaka, 1986], for example south of the labeled butte, arecool relative to surroundings. whereTob s is theobserved Tcrmoskan brighiness temperature. model produced temperature outputs for every0.25 H andfor I m is thestandard model inertia of 8.0, whichwas chosen as a every 2ø latitude. The temperatureoutputs were then representativemidpoint inertia for the channels studied. Tm is a interpolatedfor each locations' latitude and time of dayto give standardmodel temperature. Tm wasdetermined by first Tm. To derive•}Tm/•}l, we used an identical process to derive modelingthe surface using an inertia of I m, analbedo of Am modeltemperatures for othervalues of inertia. We usedinertias (discussedbelow), and the correct Martian season (Ls = 6ø or nearI m (within1.5 units) to minimize errors in •}TmfOlcaused Ls = 18ø, whereL s is the areocentricsolar longitude). The by thenonlinearity of temperaturewith inertia. 1990 BETTSAND MURRAY: THERMAL STUDlES OFMARTIAN CHANNELS

of thechannels. Generally, the albedos varied by lessthan 0.01 forall binssurrounding and including the channels. The albedo valueswe usedin our modelfor eachchannel are shownin Table 3. Dueto its large width, Hydraores channel is theonly channel for which we could estimate an albedo separatefrom the surroundings. Mostof thechannels are signif'xcanfly narrower than the 1ø x 1ø bin sizeof Pleskotand Miner [1981]. Thus,the Pleskotand Mineralbedos are likely good estimates for the surroundings, but notnecessarily forthe channels. The Tennoskan data show that

Fig.5. Crop)Termoskan thermal and (bottom) visible images showing AI-QahiraVallis and centered approximately upon 17øS, 197øW. Vertical blacklines have been added where lines were missing in theoriginal data. AI-QahiraVallis shows smooth broad floors, a tributarypattern, and a straight,possibly stmcturally controlled, northern section. Note dark, presumablyaeolian, material localized in thesouthern portions of the valleyand on the surroundingplains, and in the largecrater to the northwestof the valley. The channelfloors appear cooler than the surroundingsboth where the dark deposits are and where they are no[.

Bolometricalbedo (A m) is requiredwhen deriving 'mertias usingsingle observations. Due to instrumentlimitations, atmosphericvariations, and very limited phase angle viewing Fig. 6. Crop)Terrnoskan thermal and (bottom)visible images showing geometry,even approximate estimates of bolometric albedo from Ma'adimVailis and centeredapproximately upon 17øS, 183øW. Visible the Termoskandata have thusfar not yieldedhigh confidence datawere only obtained in thewestern portion of thisregion. Once again, results[Murray et al., 1991; Bens, 1993]. Thus, we use notehow cooltemperatures follow channelfloor. Localizedcooler spots bolometricalbedos from the 1ø x 1ø binned albedosof Pleskot may have aeolian causes,but overall cooler channeltemperatures andMiner [1981]. We averagedadjacent bins along the course probablydo not. BETFSAND MURRAY: THERMAL STUDIES OF MARTIAN CHANNELS 1991

TABLE 2. Stunmaryof TermoskanChannel Observations Visible

Channelfloors consistently cooler than surroundings by3 K to Channelsdarker or similarto surroundings 10K

Thermalboundaries closely follow channel floor boundaries Visibleboundaries do notclosely match channel boundaries

Relativelyuniform on floorsof channels Not nearlyas uniform as thermal

Homogeneousfloors even in serpentineregions around buttes Nothomogeneous in serpentine regions around buttes

Visibleboundaries do not closely match thermal boundaries

Channelfloors cooler than surroundings even outside dark, localizedaeolian deposits. Temperature corresponds better to floorboundaries than to darkdeposits' boundaries

Wherebenches along walls exist, their temperaturesare between those of the flat channel floors and those of the surroundings

Tributariesgenerally appear thermally distinct from their surroundings

Landslidesand ejecta blankets appear thermally similar to surroundingchannel floors, although they are just at thelimit of resolutionof the Termoskandata, so this observationis guarded

thechannel floors are actually all darkeror similarin albedoto ForA1-Qahira and Ma'adim Vallis, where Termoskan obtained theirsurroundings. Therefore, considering the midday local significantlength coverage, there are no systematic differences in timesof the observations,our model results represent lower inertiawith distance along the lengths of eachchannel. Also, boundsfor thechannel inertias. Similarly, we determinelower thereis no correlationbetween channel widths and inertia, as boundsfor the inertia differencesbetween channels and reportedfor someother channels [Craddock et al., 1988; surroundings.For the albedos,inertias, and times of day ChristensenandKieffer, 1979]. involved,we foundthat a decreasein modelalbedo of 0.01 Alsoof interestare buttes ("islands") seen prominently in wouldcause a derivedinertia increase of approximately0.4. HydraotesChannel and in EosChasma. These include the 70 km Due to its largewidth, we wereable to estimatean albedofrom x 140km butteat 3øN,32øW between Hydraores Chaos and Tiu Pleskot and Miner [1981] for HydraotesChannel of Vallis(Figure 2). Althoughsurrounded by channels,the buttes approximately0.17 versus surrounding albedos of approximatelyhave inertias similar to theplains surrounding the channels. 0.20 to 0.21. HydraotesChannel also showedthe largest This is consistentwith the resultsfound by Christensenand differencein visible signal (DN) between channel and Kieffer[1979] for a 20 x 90 km buttewithin Kasei Vallis. Thus, surroundingsof any of thechannels except within dark, localized asthey concluded for theKasei Vallis region, our results imply aeoliansplotches. Thus, an 0.04 decreasefrom the albedosused thatthe buttes were part of a contiguoussurface prior to channel representsan approximatelower bound on channel albedos. formation.The processes that led to the developmentof the Thiscorresponds to an approximateupper bound on possiblebutte and plateausurfaces probably acted prior to channel channelinertia increasesrelative to derived inertias of 1.6 due to formation.Less likely, the buttes and plateaus may be currently albedo uncertainties. modifiedby a similarprocess. However, this process would have to haveaffected them despite the presence of the channelsand Results withoutaffecting the surface properties of thechannels. Ourquantitative results back up the qualitative conclusion that all locationson thechannel floors have higher inertias than the WHY Do CHANNELSAND VALLEYS HAVE HIGHER INERTIA .9 surroundings.Figure 7 andTable 3 showour derived average In General inertiasand results from previous IRTM studies.Lower bounds ontypical channel thermal inertias range from 8.4 to 12.5(352 to TheTermoskan data are consistent with the idea developed for 523 in SI units). The lowerbounds on the averageinertia KaseiVallis by Christensen and Kieffer [1979] that one process, differencebetween the channelfloors and the surroundingspossibly associated with channelformation, increased the inertia variedfrom 1.1 (46 SI) for SimudVallis to 3.5 (147 SI) for throughoutthe channels. A second,probably aeolian, process HydraotesChannel. Our derived inertias for the surroundingsconcentrated a coarse, low-albedo component in certainareas. arein goodagreement with the corresponding2ø x 2ø binned Here we assumethat the low-albedolocalized areas are indeed inertiasof Palluconi and Kieffer [1981], with the average inertia aeolianin nature. We find the more generalinertia difference (Termoskanderived inertia minus Palluconi and enhancementsto be stronglyassociated with fretting. We Kiefferinertia) between paired points being +0.3 with a standard postulatethat the inertia enhancements werecaused either by the deviation of 0.8. originalfretting process or bya processinvolving the bonding of 1992 BETTSAND MURRAY: THERMALSTUDIES OF MARTIAN CHANNELS

AI-Qahira Ma'adm Hydra.Ch. Shalbatana Simud TiuWest 'l'iuEast Gangis Channel Name

Avg.Channel Iner. Avg.Surround. Iner

Fig. 7. Inertiasderived from Termoskandata for thermallydistinct channels and the associatedsurrounding plains. Derived inertiasfor EosChasma and surroundings are not shownbecause they are unreasonably high due to theafternoon cooling effect (seeTable 3 for furtherexplanation). Also not shownare Aresand Ravi Vailes,which are mostlysimilar in inertiato their surroundings. finesdue to an increasedavailability of water. Beforediscussing what their orientation, east-west or north-south. Theoretical frettingin moredetail, we first consideralternate origins of the modellingshows that for the widths (tens of kilometers)and apparentoverall inertia enhancement:atmospheric or geometric depths(few hundredmeters and in rare casesup to 3 km) of the effects,aeolian causes, or catastrophicflooding causes. channels studied, the decreasein overall flux due to this effect is very small becauseof the low sun anglesand small amountsof Atmosphericand GeometricEffects timeinvolved. Specifically, we calculatedthe decreasein energy Two atmosphericeffects will increasethe apparentinertia of a received at the surface to be less than 0.1% for the worst case surface with decreasingelevation [Christensenand Kieffer, scenario,i.e., usingthe largestheight to width ratio observed 1979]: an increase in surface conductivitywith increasing (approximately1/10) anda north-southrunning channel. pressureand an increasein apparentinertia with increasing Observing geometry combined with roughnessof the thermal opacity due to atmosphericdust. Over a range of surroundingplains versus the channels could possibly explain the elevationsrepresentative of one of the most extreme channel- observations,although the explanationis quite strained. It surroundingselevation differences (-2 km to I km in Hydraotes requiresthe surroundingplains to havehigher average slopes and Channel[U.S. GeologicalSurvey, 1991b]), the apparentthermal morelarge scale roughness than the channelfloors. The slopes inertia will changedue to pressurevariation from 6.3 to 6.8 for a facingthe sun, and thus the warmest slopes, were also the slopes surfacewhose actual inertia is 6.5 at 0 km [Kiefferet al., 1973]. that faced the Termoskan instrument. Thus, rougher With a visible opacityof 0.3 at -2 km and a scaleheight of 10 surroundingswould have appearedwarmer. However,IRTM km, the apparentinertia could increase from 6.5 at 1 km to 6.8 at observationsfound channels to have higherinertias, and IRTM -2 km [HaberleandJakosky, 1991]. Thus,although these effects obtained nighttime observationsand multiple phase angle may accentuatetemperature differences, they are too small to observations. Thus, the temperaturevariations observed by explainsolely the observedinertia differences. Termoskanare likely not causedby the "smoothness"of the Increasedmorning and late afternooncooling in the channels channels alone. due to shadowingalso is not a major effect. Becauseof the shadowingeffects of the walls, sunrisewill occur later and sunsetearlier in channelsthan on the surroundingplains. This AeolianIncrease of AverageParticle Size will cool the channelfloors. However,observationally, we infer Visible wind streaksand intracratersplotches in the areas this is probablynot significantfor the channelsstudied because surroundingthe channelsand generalcirculation model (GCM) the channelfloors are cooler than the surroundingsno matter surfacewind predictions[ et al., 1993] indicate that BETTSAND MURRAY: THERMAL $TUDIF.• OF MARTIAN CHANNELS 1993

TABLE 3. DerivedInertias, Summary of ResultsFrom Previous Studies, and Model AlbedosUsed Channel ChannelI SumI E•ltaI PreviousChannel Results From Thermal Studies ChainA SumA

AI-Qahira 8.7 (0.9) 7.3 (0.4) 1.4 (0.8) [Z86]: In combinationwith Ma'adim:3.5-12.5 with mode8. 0.21 0.21

Ma'adim 9.3 (0.9) 7.3 (0.4) 2.1 (0.8) [Z86]: In combinationwith AI-Qahira:3.5-12.5 with mode8. 0.21 0.21

Shalbatana 9.0 6.2 2.8 [CK79]: 7-8+, thermallydistinguishable south of 10øN. 0.24 0.24 [C88]: 9-12, not correlated with dark materials; rock abundanceas high as 14%.

Hydraotes 12.5 9.0 3.5 [CK79]: 12-13near chaos. 0.17 0.20 Channel [PK81]:near 11, althoughmost of the binssample significant portionsof the surroundingplateau as well. [C86]: Relatively high rock abundance,>14% in some places,although not well resolvedfrom surroundings.

SimudVallis 8.4 7.4 1.1 [CK79]: 8-10;thermally distinguishable S. of 10øN. 0.24 0.24

Tiu West 10.6 9.3 1.3 [CK79]: 11-12;thermally distinguishable S. of looN. 0.22 0.22

Tiu East 11.0 8.6 2.4 [CK79]: 10-11;thermally distinguishable S. of 10øN. 0.24 0.24

Gangis 10.4 7.8 2.6 0.19 0.19

Eos Chasma 19.9 15.6 4.4 0.17 0.17 'Abbreviatedcolumn headings are'as follows: Channel I, derived channel inertias (10-3cal cxn-2 K -1 s -1/2, multiply by41.86 for SI units); Sum I, derivedinertias for the surroundings;Delta I, averageinertia difference between paired points in channeland outsidechannel; Chan. A, albedoused for channelin thermalmodel; Surf. A, albedoused for surroundingsin thermalmodel, from Pleskotand Miner [ 1981]. Numbersin parenthesesrepresent standard deviations, presented to givean ideaof the rangein inertias.These do not representerrors, since the data werecollected for severalpoints within the channelsand on the surroundings that actually have different inertias. Previousresults are from the following sources: [CK79], Christensen and Kieffer [1979](note that inertia numbers were read off theircontour plots basedupon 1/2 ø x 1/2ø bins,whereas descriptions are derived from their text); [PK81], Palluconi and Kieffer [ 1981] (datafrom 2ø x 2ø bins);[C86], Christensen[1986]; [Z86], Zimbelman[1986], includesportions of thechannels that were not observedby Termoskan;and [C88], Craddocket al. [1988], includesportions of the channelsthat were not observedby Termoskan. ChannelI andDelta I areprobably lower bounds (see text). Our derivedinertias for thesurroundings are in goodagreement with the corresponding 2ø x 2ø binned inertias of Palluconiand Kieffer [1981], with the averageinertia difference(Termoskan derived inertia minus Palluconi and Kieffer inertia)between paired points being +0.3 with a standard deviation of 0.8. Derivedinertias for EosChasma are probably much too high due to the so-calledaftemoon cooling effect, in whichthe surfaceof Mars is observed to coolmuch faster in theafternoon than predicted by thermalmodels of thetype used here [Jakosky, 1979; Ditteon, 1982]. EosChasma, observed at 15.2 H, wasthe onlychannel observed after 13 H, andhence the only channel for whichthis was a majorfactor. For reference,the Palluconiand Kieffer[1981] inertia for thearea modeled as the Eos Chasma surroundings was 9.2, versus the 15.6derived here. Inertiaswere not derivedfor Ravi or AresValles because their temperatures generally appear similar to theirsurroundings. aeolianprocesses have been and are probably still activein the channels.Based upon Earth analogs,most of the saltating channelregions studied here. In addition,the channelsmay materialwill keepmoving until it reachesand piles up in lower focus winds or createthem preferentiallydue to differential wind velocity/ adverseslope traps such as the observeddark heating of walls versus the floors [Craddocket al., 1988], deposits.The exactcorrelation of inertiawith fiat floor bottoms althoughthis is lesslikely for the widestchannels. We agree is alsoinconsistent with an aeolianexplanation. One would with previousstudies that localized dark splotches and streaks expectan aeolian process to spreadsome of its thermalsignature within the channelsare probablyaeolian sand depositsonto terraces or againstwalls. Thatis exactlywhat is observed [Christensenand Kieffer, 1979; Zirnbelman, 1986; Craddock et in thevisible with somelocalized dark deposits that do spread al., 1988]. The questionremains, however, whether aeolian out onto the surroundingplains (e.g., see southernShalbatana processesare responsiblefor the enhancedinertia of the rest of Vallis and also HydraotesChannel in Figure 2). Aeolian the channels. Possible aeolian causesinclude (1) small-scale explanations,whether depositional or deflational, are also deposition,i.e., betweenrocks, but not burying all rocks; (2) inconsistentwith the thermal homogeneitiesin serpentine large-scaledepositional blankets; and (3) deflation, i.e., wind- regions of the channels,such as around buttes in northern induced removal of fine material, exposing higher inertia HydraotesChaos. material. Becausea blanketof suspension(dust) sizedparticles A several centimeterthick sand blanket within the channels, within a channel would cause a lower inertia, not the observed althoughperhaps consistent with the thermal homogeneity,is higherinertia, we consideronly saltationsized particles. inconsistentwith other observations. One would expect a The thermalhomogeneity of the channelfloors argues against complete sand blanketing to spread somewhat to the anytype of aeolianprocess, short of a uniformsand sea, causing surroundings.Also, the visible heterogeneityseen in some the channelinertia enhancement.Aeolian processeson Earth placesmay be inconsistentwith a sandsea. Dune featuresare and as observedon Mars inevitablycause spatial heterogeneity. also not obviousthroughout the channelsin Viking images, This is consistentwith having localized, dark depositswithin althoughresolution is a problem. 1994 BETrs AND MURRAY: THERMAL STUDmSOF MARTIAN CHANNELS

Preferential aeolian deflation within the channels also seems Frettingas a generalcause of channelinertia enhancements inadequateto explainthe overallinertia enhancement,although maybe consistentwith IRTM thermalstudies of channelregions it may play somerole. As with aeoliandeposition, one would not observedby Termoskan. Christensenand Kieffer [1979] expectaeolian deflation to causegreater thermal heterogeneity, found Simud, Tiu, and Shalbatana Valles to have inertia particularlyin serpentinechannel regions. Any topographic enhancementssouth of 10øN, but not north of there. To the obstacleor channelbend would presumably affect the amountof north, these channels show increasedcatastrophic flooding deflationand eventualdeposition of particles. An even greater bedformson their floors such as grooves,and they do not difficultywith aeoliandeflation as a sole explanationis that to commonlyhave steep, scalloped walls. Christensenand Kieffer matchthe observations,it mustuniformly strip flat floors,but not found that higherinertia on the floor of Ares Vallis was most buttes,benches, or the surroundings.Then, even if aeolian apparentnorth of 10ø N andin a regionnear 7øN. Someof these deflationhas taken place, what remainson the surfacesof the portionsof Aresare not obviouslyfretted, but mostdo not show floors?Deflation could not havebeen too effectivestripping the obviouscatastrophic flooding floor featuresas do the least channels down to rock. The observed inertias are far below the thermallydistinct portions south of about6øN [Sharpand Malin, inertias of at least 30 or 40 expectedfor bare rock on Mars. 1975]. Althoughrock abundancesfor some channelsare higher than KaseiValles, found by Christensenand Kieffer [1979] to have averageMars, they are still probablyless than 20% (basedupon enhancedinertia, was classedas a modified fretted channelby Christensen[1986] and Craddock et al. [1988]). Thus, even Sharp and Malin [1975]. Althoughit may show significant deflationwould have to leave significantfines behind. This catastrophicflooding features in certainregions, it waslikely last couldbe accomplishedby bondingof the fines, somekind or modified in most regionsby fretting type processes[Baker, armoringof the surfaceby rocks, or a self-limitingstripping 1982]. Craddoclcet al. [1988]reported that Mangala Vallis did processwhere a natural limit is reached on the amount of not appearthermally distinct from its surroundings.It shows saltatingfine materialthat can be strippedaway (M. C. Malin, significantcatastrophic flooding floor featuresover muchof its personal communication,1993). However, one still has length[Sharp and Malin, 1975;Baker, 1982]. Thus,although it difficultyexplaining thermal homogeneity and flat floor thermal is speculativeto extenda frettingexplanation to a widerrange of correlation,including in serpentineregions. channelswithout higher resolution thermal data, IRTM data do ChannelFormation Processes: Fretting Versus Flooding seemgenerally consistent with a fretting explanation. Future missions'increased global coverage will allow a more thorough Two categoriesof channel formation processesmay have testingof the generalityof thefretting hypothesis. resultedin channelinertia enhancement:catastrophic flooding or Fretting could have increasedchannel inertias either by fretting. Most of the thermally distinct portionsof observed channelshave flat, wide floors devoid of large-scalebedforms. increasingthe averagerock abundanceversus the surroundings, Steep, sometimesscalloped, walls are also associatedwith most or by preferentiallyincreasing the bondingof fine particles.We considerthese two possibilitiesin turn. thermallydistinct channels. These morphologiesare indicative of fretting [Sharp, 1973]. EasternRavi (Figure2) and southern Increasedrock abundance. An increasedareal percentage of Ares(Figure 3) arethe only majorchannel sections observed that rocks(in the formof boulders,cobbles, gravel, or evenpebbles) are not clearly thermallydistinct. They also are the only major could be the cause of the channel inertia enhancement. Rocks channelsections observed that have bedformsclearly indicative may have been emplacedas debris derived from fretting, of catastrophicflooding, but not fretting [Sharp and Malin, 1975; althoughit would have been challengingto both transportthe rocks several lcilometers to several tens of kilometers and still Baker, 1982]. Thus, we favorfretting over catastrophic flooding as the cause of the inertia enhancements that are observed. preservea relatively uniform thermal inertia floor signature. Frettinghere refers to wet or dry sapping,mass wasting, and However, particularly given the uncertaintyof the fretting possibledebris flow [Sharp, 1973], althoughwe emphasizethat process,this may havebeen possible. IRTM data indicatethat it is not a well-understoodprocess. For example,Baker and rock abundancesare higher for some channelsincluding Kochel [1979] identified a whole range of mass movement, [Craddock et al., 1988] than for the slope, and periglacial featuresassociated with scallopedand surroundingterrains. The percentagesof rocksfound, although fretted channel margins. Significanfiy,these features did highfor Mars, are still nowherenear a completecovering of the contrastwith the suite of cataclysmicflood bedformsfound on surface. Thus, a finer clasfic componentstill must play a the floorsof somechannels [Baker and Milton, 1974]. Fretting significantrole in determininginertia. is morphologicallywell defined for Mars, and it does contrast Bondingoffine materials. Variationsin maturityof a duricrust with channels showing well-defined catastrophicflooding (i.e., degreeof bondingof a case-hardenedcrust) were suggested bedforms. Thus, whateverthe exact frettingprocesses, fretted by Jakoskyand Christensen[1986] to explainmost of the low morphologiesdo appear to be associatedwith the channels resolution thermal inertia variations on Mars. Areas where fines showingenhanced inertias in this study. Althoughcatastrophic have been more efficienfiy bondedwill have higher thermal floodingundoubtedly occurred in someof thesechannels, fretting inertiasdue to increasedthermal conductivity. Duffcrusts were likely followed. Only the last significantprocess to affectthe observedat the Viking Landersites [Binder et al., 1977;Mutch channelswill affectthe upperfew centimetersthat are sensedby et al., 1977]. Water and/orsalts were proposed as the agentsof diurnal thermal measurements. duricrustformation [Jalcosky and Christensen,1986]. Fretting Chaoticterrain is often associatedwith nearbyfretted areas, mayhave increased bonding of fine materialswithin the channels suchas in the HydraoresChaos region. Chaoticterrain may due to increasedpresence of water and possiblybrines either representan intermediatestage that in somecases was eventually initially or secondarily. smoothedto form fretted areas [Sharp, 1973]. It often has Whetherof primaryor secondaryorigin, water for the bonding enhancedinertia (e.g., in HydraoresChaos) as observedboth in of materialswas likely moreaccessible on the low, flat channel the Termoskandata and in IRTM data [e.g., Christensenand bottomsthan on the surroundingplateaus or on the intermediate Kieffer, 1979]. These inertia enhancementscould be related to inertiabenches. Bonding is alsolargely consistent with thermal early stagesof fretting. variations strictly following the channel bottoms, even in BETTSAND MURRAY: THERMAL STUDmS OF MARTIAN CHANNELS 1995 serpentineregions. The absoluteinertia values of the channels matchflat channelfloor boundaries.Atmospheric and geomclxic are consistentwith valuesthat couldbe obtainedby bondingfine effects are not sufficient to cause the observed inertia materials in combination with some rocks on the surface. The enhancement.We agreewith previousresearchers that localized, channelinertias are similar to somewhathigher inertias in the dark, high-inertiaareas within channelsare likely aeolianin areasof the Viking Landers(approximately 9 and 8 [Kieffer, nature. We disagreewith someresearchers that aeolian deposits 1976]). The lander sites showedrelatively thin duricrustsas fill the channelsor areresponsible for the overallthermal inertia well as relativelyhigh percentagesof rocks[Binder et al., 1977; enhancement. Small-scale aeolian deposition or aeolian Mutch et al., 1977]. deflationmay play rolesin the inertia enhancement.However, In additionto being consistentwith the idea of Jakoskyand largelybecause of the thermalhomogeneity of the channelfloors, Christensen[1986] that most inertia variations on Mars are due we favor alternateexplanations. to variationsin duricrustmaturity, increased bonding in channels Fretting or catastrophicflooding may have emplacedmore is also consistentwith ChristenseWs[1986] finding that most rockson channelfloors or causedincreased bonding of finesdue thermal inertia variations on Mars are due to variations in the to the presenceof water. We favor fretting processesover fine componentinertia, not thepercentage of blocks. This theory floodingfor thecause of thehigher thermal inertia because of the doesnot explainthe increasedrock abundancesfound in some flat floorsand steepscalloped walls in mostregions that show channels,but it is consistentwith findingsthat regionswith inertiaenhancements. Alternatively, at some time after channel higherrock abundancesgenerally have higher frae-component formation,water that was preferentiallypresent due to the low, inertias[Christensen, 1986]. flat frettedfloors may have enhanced bonding of originalfines or Increased water or brines on channel floors may have come dust fallout. Future missions should be able to distinguish from the initial frettingdue to sappingwater flow or ice flow. between competing theories of inertia enhancement. The This would require that the fines alreadybe presenton the possibilitythat the flat channelfloors owe their high inertia to surface at the time of channel floor formation and that the water-relatedprocessing (bonding of fines) arguesfor assigning bonded material survive since the time of channel formation, highpriority to thesesites in futureexploration. which is difficult consideringthe long time sinceformation and The loss of Mars Observer(MO) occurredwhile this paper the relative activity of the surface. In particular,dust storm was beingrevised. We still havechosen to includea discussion fallout mustbe dealt with. One possiblescenario is that aeolian of MO instrumentsand their potentialcontribution to the study deflationstrips the new dust fallout off over time. This of channelsand valleysbecause the commentswill be generally combinedtheory of aeoliandeflation is favoredover deflation applicableto whateveranalogous instruments may fly on future alonebecause the surroundingscan be strippedsimultaneously, Mars missions.The unprecedentedhigh resolutionof the Mars but the resultreexposes the bondedmaterial. Thus, the channel ObserverCamera (MOC) (up to 1.4 m/pixel[Malin et al., 1992]) alonedoes not haveto be preferentiallystripped and a relatively would have enabled channel floor surface morphologies uniform surfacewill be exposed. There is an alternativethat indicative of aeolian, flooding, or fretting processesto be doesnot requirethe originalsurface to be preservedand that recognized.These morphologies may include:dunes, water flow keepsmany of the attractivefeatures of a bondingtheory. morphologies,large boulders,and somemass wasting features. Water may havebeen preferentially present on frettedchannel The Mars Observer laser altimeter (MOLA) profiles, with a floors after initial channel floor formation. Fretting may have verticalprecision of about2 m andhorizontal resolution of about emplacedwater or ice nearthe surface,or the uniformfloor level 300 m [Zuberet al., 1992], would have complementedMOC by of fretted channelsmay representthe original depth of frozen providingdetailed topographic information and someroughness ground[Sharp, 1973]. Processesacting over long time scales information that would show how flat and smooth the floors such as evaporation,adsorption and diffusion,or some other really are and how steepthe walls are, allowing slope versus processmay then haveprovided water at the very surfacethat angleof reposecomparisons for the walls. The high resolution accentuatedthe bondingof fines. The actualmaterial bonded stereocamera (HRSC) on Mars '94 (M94) is designedto obtain could have been original or could have been dust that was bothhigh-resolution imaging and topographic information. depositedover time in the channelsas a result of dust storm Also on MO, TES [Christensenet al., 1992] and the pressure fallout. Post floor formationbonding would be consistentwith modulatorinfrared radiometer (PMIRR) [McCleeseet al., 1992], the apparentTermoskan observation that landslidesand ejecta via globalthermal inertia and albedocoverage, would have given blanketsappear thermally similar to the surroundingchannel insightinto the originof the channelinertia enhancements. The floors. However, these features are just at the limit of correlation of channel inertia enhancements with fretted Termoskan'sresolution, so this observationwill have to be morphologiesversus purely catastrophicflood morphologies confirmedwith futurehigher resolution missions. would have been tested globally. TES also would have Thus, post floor formation bonding avoids some of the contributedsignificantly to understandingthe small-scalecause difficulties involvedwith preservingan original surface. The of the enhancementsvia rock abundancesand frae-component theoryis somewhatspeculative, however, given uncertainties on inertias in 3 km/pixel maps derived using multiwavelength how this mechanism would act and at what rates. Whatever the methodssimilar to Christensen[1983, 1986]. Thus, increased- actualprocess, the associationof waterwith the formationof the rock theories could have been evaluated versus increased-fine- channels and the location of channel floors closer to water and componentprocesses, whether larger particles or bonding. TES ice tables arguefor the hydrologicplausibility of a preferential spectralmapping would have indicated compositional differences bondingexplanation. betweenchannels and surroundings. Termoskan2 onM94 is expectedto increasespatial resolution SUMMARY AND CONCLUSIONS anotherorder of magnitudefrom mostof the Termoskan1 data. Utilizing the Termoskandata, we concludethat channelson The high spatialresolution targeted upon channelswill enable Mars generally have higher thermal inertia than their tests of whetherthe thermal signal remainsuniform at those surroundings,consistent with IRTM studies. For the first time, resolutions.In addition,Termoskan 2 can observemore channels we observethat the thermal inertia boundariesvery closely of varyingmorphologies elsewhere on Mars. Observationsof 1996 Bt!TTSAND MURRAY: THERMAL STUDIES OF MARTIAN CHANNBLS small cratersand their ejecta and landslideswill test inertia deposits:Comparisons with general circulation model results, J. Geophys. enhancementtheories and timescales. Res., 98, 3183-3196, 1993. Itaberle, R. M., and B. M. Jakosky,Atmospheric effects on the remote The Omegaimaging spectrometer on M94 is designedto give determinationof thermalinertia on Mars, Icarus, 90, 187-204, 1991. importantnear-IR spectral information about the channels.In Jakosky,B. M., The effectsof nonidealsurfaces on the derviedthermal particular,Omega is designedto maptrace amounts of hydrated propertiesof Mars,J. Geephys.Res., 84, 8252-8262,1979. minerals as was done at lower resolutionfor other Mars regions Jakosky,B. M., andP, R. Christensen,Global duricmst on Mars: Analysis of with the Phobos'88 imagingspectrometer for Mars (ISM) remotesensing data, J. Geephys.Res., 91, 3547-3560,1986. Kieffer, H. H., Soil and surfacetemperatures at the Viking landersites, instrument[Erard et al., 1991]. We would expect enhanced Science, 194, 1344-1346, 1976. hydrafionsignatures for thechannel floors if significantbonding Kieffer,H. H., S.C. Chase,Jr., E. Miner, G. Mttnch,and G. Neugebauer, and duricmstformation has occurred. Thermally distinctive Preliminaryreport on infrared radiometric measurements from the channelfloors also representinteresting locations for future 9 spacecraft,J. Geophys. Res., 78, 4291-4312,1973. landersdue to their unique history and the probable surface Kieffer,H.H., T.Z. Martin, A.R. Peteffxeund,B.M. Jakosky,ED. Miner, and F.D. Palluconi,Thermal and albedo mapping of Mars duringthe Viking presenceof material from variousstratigraphic layers and primarymission, J. Geephys.Res., 82, 4249-4291,1977. locations. Malin, M. C., G. E. ,A. P. Ingersoll,H. Masursky,J. Veverka,M. A. Ravine,and T. A. Soulmile, MarsObserver camera, J. Geephys.Res., Acknowledgments. We thank Phil Cl•stensenand Vic Baker for 97, 7699-7718, 1992. thoughtfulreviews of the submittedmanuscript, Michael Malin for helpful MarsChannel Working Group, Channels and valleys on Mars,Geol. Sec. discussions,and Doug Nash,Dewey Muhleman,and Andy Ingersollfor Am. Bull., 94, 1035-1054, 1983. helpfulcomments on the manuscript. Also, thanks to JanYoshimizu for help Masursky,H., A. L. Dial, Jr., andM. E. Strobell,Martian channels-A late in producingprints of the images. Funding for this research was provided by Vikingview, NASA Tech. Memo., 82385, 184-187, 1980. NASA grantsNAGW-1426 and NAGW-2491. Divisionof Geologicaland McCleese,D. l., R. D. Haskins,J. T. Schofield,R. W. Zurek, C. B. Leery, PlanetarySciences, Califomia Institute of Technologycontribution 5287. D. A. Paige,and F. W. Taylor,Atmosphere and climate studies of Mars usingthe Mars Observerpressure modulator infrared radiometer, J. Geophys.Res., 97, 7735-7758,1992. Murray,B.C., et al., Preliminaryassessment of termoskan observations of Baker,V. R., The Channelsof Mars, Universityof TexasPress, Austin, Mars, PlanetSpace Sci.,39(112), 237-265, 1991. 1982. Mutch,T. A., R. E. Arvidson,J. W. Head,K. L. ,and R. S. Saunders, Baker,V. R., M. H. Carr,V. C. Gulick,C. R. Williams,and M. S. Marley, TheGeology of Mars,Princeton University Press, Princeton, N.J., 1976. Channelsand valley networks, in Mars,edited by H. Kiefferet al., pp. Mutch, T. A., R. E. Arvidson,A. B. Binder, E. A. Guinness,and E. C. 493-522,University of ArizonaPress, Tucson, 1992. Morris,The geologyof theViking Lander 2 site,J. Geophys.Res., 82, Baker, V. R., and R. C. Kochel,Martian channelmorphology: Maja and 4452-4467, 1977. KaseiValles, J, Geephys.Res., 84,7961-7983, 1979. Palluconi,F. D., andH. H. Kieffer,Thermal inertia mapping of Mars from Baker, V. R., and D. J. Milton, Erosionby catastrophicfloods on Mars and 60øSto 60øN,Icarus, 45, 415-426, 1981. Earth,Icarus, 23, 27-41, 1974. Pieskor,L.K., and E.D. Miner, Time variability of Martian bolometric Bells,B. H., Thermaland visible studies of Mars usingthe Termoskandata albedo,lcarus, 45, 179-201, 1981. set,Ph.D. thesis,Calif. Inst. of Technol.,Pasadena, 1993. Scott,D. H., and K. L. Tanaka, Geologicmap of the westemequatorial Binder,A. B., R. E. Arvidson,E. A. Guinness,K. L. Jones,E. C. Morris,T. regionof Mars,U.S. Geol. Surv. Misc. Invest. Ser. Map, 1-1802A, scale A. Mutch,D.C. Pieri,and C. ,The geologyof theViking Lander 1 1:15,000,000, 1986. site,J. Geophys.Res., 82, 4439-4451,1977. Selivanov,A. S., M. K. Naraeva,A. S. Panfilov,Yu. M. Gektin, V. D. Cart, M. H., The Surfaceof Mars, Yale UniversityPress, New Haven, Kharlamov,A. V. Romanov,D. A. Fomin,and Ya. Ya. Miroshnichenko, Conn., 1981. Thermalimaging of thesurface of Mars,Nature, 341,593-595, 1989. Carr,M. H., Mars:A water-richplanet, Icarus, 68, 187-216,1986. Sharp,R. P., Mars:Fretted and chaotic terrains, J. Geophys.Res., 78, 4073- Christensen,P. R-, Eolian intracraterdeposits on Mars: Physicalpwperties 4083, 1973. andglobal distribution, Icarus, 56, 496-518,1983. Sharp,R- P., andM. C. Malin,Channels on Mars, Geol. Soc. Am. Bull., 86, Christensen,P. R., The spatialdistribution of rockson Mars,Icarus, 68, 593-609, 1975. 217-238, 1986. Squyres,S. W., Martianfretled terrain: Flow of erosionaldebris, Icarus, 34, Christensen,P. R., and H. H. Kieffer,Moderate resolution thermal mapping 600-613, 1978. of Mars:The channelterrain around the ChryseBasin, J. Geephys.Res., U.S. GeologicalSurvey, Mosaicked digital image model OdDIM), in 84, 8233-8238, 1979. Missionto Mars: Digital Image Map CD-ROM, vol. 1, Flagstaff,AZ, Chfist•en, P. R., and H. J. Moore,The Martiansurface layer, in Mars 1991a. editedby H. Kiefferet al., pp. 686-729,University of ArizonaPress, U.S.Geological Survey, Topographic map of Mars,U.S. Geol. Surv. Misc. Tucson, 1992. Geol.Inv. Map, 1-2179,1991 b. Christensen,P. R., et al., Thermalemission spe•rometer experiment: Mars Zimbelman,J.R., Thermal properties of channelsin the aeolisquadrangle: Observermission, J. Geephys.Res., 97, 7719-7734,1992. Topographictraps for aeolianmaterials, in Symposiumon MECA,pp. Clifford,S. M., C. J. Bartels,and E. P. Rubenstein,The Mars thermalmodel 112-114,Lunar and Planetary Institute, Houston, Tex., 1986. (MARSh: A FORTRAN77 finite-differenceprogram designed for Zimbelman,J.R., and L.A. Leshin,A geologicevaluation of thermal generaldistribution, Lunar and Planetary Institute, Houston, Tex., 1987. propertiesfor theElysium and Aeolis quadrangles of Mars,Proc. Lunar Craddock,R- A., High resolutionthermal infrared mapping of Martian Planet.Sci. Conf. 17th, Part 2, J. Geophys.Res., 92, suppl.,E588-E596, outflowand fretted channels, M.S. thesis,Dep. of Geol.,Ariz. StateUniv., 1987. Tempe, 1987. Zuber,M. T., D. E. ,S.C. Solomon,D. O. Muttleman,J. W. Head, J. Craddock,R. A., R. Greeley,P. R. Christensen,and F. T. Aldrich,Martian B. Garyin, J. B. Abshire, and J. L. , The Mars Observerlaser channelmaterials and the formationof channelwinds (abstract), Lunar altimeterinvestigation, J. Geophys. Res., 97, 7781-7798,1992. Planet. Sci., XIX, 215-216, 1988. Ditteon,R., Daily temperaturevariations on Mars,J. Geephys.Res., 87, B. H. Betts,San Juan CapistranoResearch Institute, 31872 Camino 10,197-10,214, 1982. Capistrano,San Juan Capistrano, CA 92675. Erard,S., J.-P.Bibring, J. F. Mustard,O. Fomi,J. W. Head,S. Hurtrez,Y. B.C. Murray, PlanetaryScience 170-25, CaliforniaInstitute of Langevin,C. M. Pieters,J. Rosenqvist,and C. Sotin,Spatial variations in Technology,Pasadena, CA 911-25. compositionof the VailesMarinefts and hidis Planirisregions derived fromISM data,Proc. Lunar Planet. Sci. Conf.,21st, 437-455, 1991. Greeley,R., andJ. E. Guest,l:15M Geologicmap of theeastem equatorial regionof Mars,U.S. Geol. Surv. Misc. Invest. Map, I-1802-B, 1987. (ReceivedJune 28, 1993;revised November 5, 1993; Greeley,1L, A. Skypeck,and J. B. ,Martian aeolian features and acceptedNovember 9, 1993.)