Edinburgh Research Explorer

Terrestrial analogs and thermal models for Martian flood

Citation for published version: Keszthelyi, L, McEwen, AS & Thordarson, T 2000, 'Terrestrial analogs and thermal models for Martian flood lavas', Journal of Geophysical Research, vol. 105, no. E6, pp. 15027-15049. https://doi.org/10.1029/1999JE001191

Digital Object Identifier (DOI): 10.1029/1999JE001191

Link: Link to publication record in Research Explorer

Document Version: Publisher's PDF, also known as Version of record

Published In: Journal of Geophysical Research

Publisher Rights Statement: Published in Journal of Geophysical Research: Planets by the American Geophysical Union (2000)

General rights Copyright for the publications made accessible via the Edinburgh Research Explorer is retained by the author(s) and / or other copyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associated with these rights.

Take down policy The University of Edinburgh has made every reasonable effort to ensure that Edinburgh Research Explorer content complies with UK legislation. If you believe that the public display of this file breaches copyright please contact [email protected] providing details, and we will remove access to the work immediately and investigate your claim.

Download date: 09. Oct. 2021 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 105, NO. E6, PAGES 15,027-15,049, JUNE 25, 2000

Terrestrial analogs and thermal models for Martian flood lavas

L. Keszthelyiand A. S. McEwen Lunar andPlanetary Laboratory, University of Arizona,Tucson

T. Thordarson Divisionof Explorationand Mining, CSIRO MagmaticOre DepositsGroup, Floreat, Western Australia

Abstract. The recentflood lavason appearto havea characteristic"platy-ridged" surfacemorphology different from that inferredfor mostterrestrial continental flood flows. The closestanalog we havefound is a portionof the 1783-1784Laki flow in Icelandthat hasa surfacethat was brokenup andtransported on top of movinglava during major surgesin the eruptionrate. We suggestthat a similarprocess formed the Martian flood lava surfacesand attemptto placeconstraints on the eruptionparameters using thermal modeling. Our conclusionsfrom this modelingare (1) in orderto produceflows > 1000 km longwith flow thicknessesof a few tensof meters,the thermophysicalproperties of the lava shouldbe similarto fluid basalt,and (2) the averageeruption rates were probably of the order of 104m3/s, with the flood-like surges having flow rates of the order of 10s - 106m3/s.We alsosuggest that thesehigh eruptionrates should have formed huge volumes of pyroclastic depositswhich may be preservedin the MedusaeFossae Formation, the radar"stealth" region,or eventhe polar layeredterrains.

1. Introduction Some of the most recent and best preservedflood lavas lie The objective of this paper is to provide initial qualitative in eastern Elysium Planitia (the Cerberus Formation of and quantitativeinterpretations of new data on Martian flood Plescia [1990]) and in a portion of Amazonis Planitia lavas. The Mars Orbital Camera (MOC) on board the Mars centeredat 30ø N, 160ø W [McEwen et al., 1999b]. These Global Surveyor (MGS) spacecrafthas returnedspectacular two regions are connectedby Marte Vailis, which is also new images of Mars [Malin et al., 1998]. We were partially filled by flood lavas. The surfaceof the Cerberus particularlyinterested in imagesof the Martian flood lavasin plainsrecords a complexhistory of volcanic,fluvial, impact, order to comparethe emplacementof Martian and terrestrial tectonic,aeolian, and possiblylacustrine and glacialprocesses flood . [e.g., Mouginis-Mark et al., 1984; Tanaka and Scott, 1986; Flood volcanism is one of the most significant - Plescia, 1990; Scott and Chapman, 1991; McEwen et al., forming processesidentified on Mars [e.g., Greeley et al., 1998, 1999c]. In this paper we focus solely on the best 2000; Greeleyand Spudis,1981 ], and the extremepaucity of preserved(and presumablyyoungest) volcanic features. cratersin someof the imagesof the flood lavassuggests that Well-preservedflood lavasof similarappearance (although this style of volcanismmight plausiblypersist to this day on probably older) are also presenthigh on the Bulge, Mars [Hartmann, 1999; Hartmann et al., 1999; Hartmann near Tharsis Tholus. Layers exposedin the walls of Valles and Berman, this issue]. Understandingthe eruptionsthat Marineris are also likely to be yet older flood lavas [McEwen formedthese flood lavasis a critical stepin understandingthe et al., 1999a]. Figure 1 showsthe locationof theseflood lava evolutionof the surface,interior, and atmosphereof Mars. exposureswith respect to other major features on Mars. Before proceeding,we shouldnote that we use the term Crater countingand modelssuggest ages of about 10 Ma for "flood lava" to denotea packageof sheet-likelava flows that the youngestof the flows in the CerberusPlains [Hartmann inundatesa large region, producinga smoothplain without and Berman, this issue]. However, MOC imagesalso reveal •-100 m tall constructs. This use of the word is in line with that there are locationswhere the lava is eroding out from that presentedby Greeleyand King [1977] and its application beneath the Medusae Fossae Formation, which may have to terrestrial continental flood [Bates and d•ackson, protectedthe lava from small impact craters. While many 1984] and is essentiallyidentical to the useof the term "mare" processescan complicate the age estimates from crater on the (but withoutthe requiredlow albedo). However, counts, the crisp, undegradedmorphology of the flows someother workershave used "flood lava" to simply refer to supportsthe idea that theseflood lavas are geologicallyvery long lava flows with no clear channels or tubes [e.g., young. Together,the recent flood lavas appearto cover an Cartermole,1990]. areaof about107 km 2, roughly the size of [McEwen et al., 1998]. The minimum thicknessof the Cerberusdeposits Copyright2000 by theAmerican Geophysical Union is estimated to be in the hundreds of meters, and the new Papernumber 1999JE001191. MOC images support the estimate by Plescia [1990] that 0148-0227/00/1999JE001191509.00 individualflow frontsare only of the orderof 10 m thick.


These flood lavas have a distinctive surfacemorphology dominated by plates and ridges [McEwen et al., 1999b] (Figures 2-9). The platy-ridgedmaterial is interpretedto be lava becauseit is boundedby lobatemargins characteristic of lava flows (Figure 4). At high spatial resolution(2 to 20 m/pixel), these flows exhibit a complexsurface morphology interpreted to include rare channels (Figures 3 and 4), common rafted crustal slabs (Figures 2, 3, 5, 6, 7, and 9), ubiquitous pressure ridges (Figures 2-9), some ponded surfaces(Figures 5-7), and possiblesqueeze-ups (Figures 2 .. .,.,,. and 5-7). Coherentslabs of crustappear to rangein size from a few hundred meters to a few kilometers.' Ridges are typically only tens of meters wide but up to kilometersin length. Most ridges are interpretedto have formed at the edges of plates as they are rammed into each other during :i:; 5• emplacement(i.e., pressureridges). Other ridges may have ., .,%• formedby liquid lava oozingup throughextensional cracks in ß.:.• ..,.•. the plates(i.e., squeeze-ups). ... These flows are remarkably flat, with about 400 m of vertical elevation over 1500 km of lateral distance(0.027% •:• ...... slope) [Smith et al., 1999]. However, the region is radar ß•.•(½•..:•:**..:.•,•;•. . •:•.:..:s,:•:•:.•..:.•½::.•;•:•::s•....,-.• ...::...... •:•::.:•,:,•a. ..•:•:•...e•:•½:.-.,. bright, implying a rough surface on the decimeter scale [Harmon et al., 1992, 1999]. .'• ....::-,•: •::• •;:•...... Vents in the flood lava regions are difficult to identify, probably as a result of burial by the flows, their small size, and/or rapid removal of the relatively fragile vent structures. However, the CerberusRupes fracturesand nine low shields on the western side of the Cerberus Plains were identified as • •;•:.•:::•::::::::::::::::::::::::::::::::::::::: ..>:•:.:•:•;:•,•::•',-•:' .....-..'..,.: .• likely ventsfrom Viking data[Plescia, 1990; Edgettand Rice, 1995]. The most recentMOC imagesshow small lava flows • .:%;.•:•t•:;• ..... •:• ...... ' emanating from both sides of some of the CerberusRupes .:: ;"::•*•:...... ,..?:.,•*'*$;:•;:• :• fissures,confirming that they indeedserved as volcanicvents [McEwen et al., 1999c]. However, there is still no conclusive :::. proof that most of the large flood lavas were fed from CerberusRupes. Also, the CerberusRupes fractures cut all the lava flows, indicatingthat it has been tectonicallyactive more recentlythan it hasbeen a volcanicvent. One of the ongoinglimitations in workingwith the Martian ,,• •,-,.•::e•.::.: [-/• ...... •, ...... flood lavas is our inability to trace individual flows. The

Viking data do not have the resolutionto allow flow margins ..:...... •..-. -, ...... :...e?. :-:;,:.•;•;:•.:.:.,;•.• :;:.½,•.:,: .... to be confidently traced. The MOC data cover only a tiny fraction of the Cerberus Plains, allowing only extremely discontinuoussampling. Our bestestimate for the extentof a singleflow is shownin the lightestshading in Figure 1. This area correspondsto the brightestradar returns reportedby Harmon et al. [1999]. It also follows the local topographic gradient away from the presumed vents along Cerberus Rupes. It is relatively dark in the Viking images, and the freshestlooking lavas in MOC imagesfall within this area. Also, the radar bright area terminatesat the samearea as the flow terminusseen in Figure4. On a smallerscale the flows appearto have 100- to 1000- m-size lobesat the margins,but the largersheets are generally not capturedin any singleMOC frame(3-10 km wide). From the small areas sampledby MOC and the subtlemottling in the Viking images,we tentativelyinfer that both the widths and lengthsof the individualsheets are of the orderof tensof kilometers. Only a few sheet-likelobes can be seento be less than a kilometer wide (e.g., Figure 4). However, if previous suggestionsfor vent locationsare correct,the lavasmust have generallyspread for >500 km and may have exceeded2000 km in somecases. This impliesthat eachflow is composedof a greatnumber of smallersheet-like lobes. KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS 15,029

Figure 2. MOC imageSP1-21905 and Viking contextimage from ElysiumPlanitia centered at 5.7øN,209.0øW. In this andthe followingMOC images,north is up, andthe Viking contextis froma basemapmosaic [Davies, et al., 1992]. Note raftedplates of lava thatcan be jigsaw fit in a patternconsistent with flow to theNW or SE. The wakebehind an obstaclein the NE cornerof the image(arrow) showsthat the flow directionwas actuallyto the NW. Larger,curving ridges in the platesare probablypressure ridgesformed during the collisionof plates. Thin, involuteridges within the brightermaterial between the platesare probably squeeze-upsof lavaformed after the lavasurface stagnated.

2. Terrestrial Analogs lava flow in may be the bestavailable analog to the Martian flood lavas. Without well-studied terrestrial analogsto provide a qualitativemodel for the Martianflood lavas,it is impossible to pin downthe physicalprocesses to includein quantitative 2.1. Continental Flood Basalts lava flow models. Currentlava flow modelsrequire some The Martian flood lavas derive their name from their basicassumptions about how the flow advances(e.g., tube- obvious similarities to terrestrial continental flood basalts fed versus channel-fed). Even empirical relationships (CFBs). Continentalflood basalt provincestypically cover between eruption parametersand flow dimensions[e.g., morethan 106 km 2 witha lavapile kilometers thick [e.g., Walker,1973; Pieri and Baloga, 1986;Kilburn and Lopes, Macdougall, 1988]. When not deeply eroded,they are also 1991] are criticallyaffected by the style of emplacement.In characterizedby flat, plateau-like,topography [e.g., Greeley this sectionwe will showthat portionsof the 1783-1784Laki and King, 1977]. The best preservedand most extensively 15,030 KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS studiedCFB provinceis the ColumbiaRiver Basalt Group lavas have radiometricages between 17 and 6 Ma, but the (CRBG). The CRBGconsists of 174,300ñ31,000 km 3 of majorityof the lava was eruptedbetween 16.5 and 15.3 Ma dominantly basaltic-andesitelava covering 163,700 ñ5000 [Tolan et al., 1989]. Comparisonto other CFB provinces km2 of area.These lavas are divided into approximately 40 showsthat most CRBG lava flows are similar to the majority stratigraphicunits that contain>300 individual"flows". The of lava flows found in other provinces [e.g., Hooper,1982; Self et al., 1998]. Although continentalflood basalt provinceshave been studiedextensively, their flows are not sufficientlywell preservedor exposedfor the originalsurface morphology to be directlyobserved. Instead, the surfacemorphology must be inferred from observationsmade in cross sectionsalong natural and artificial cliffs. Such examination of the CRBG lava flows indicatesthat they were primarily emplacedas inflated pahoehoe lava flows [Self et al., 1997, 1998]. However, it is importantto note that only one unit, the Roza Member of the Wanapum Formation,has the combinationof exposureand preservationthat has allowed the detailsof an entire flow field to be described[e.g., Shaw and Swanson, 1970; Swanson et aL, 1975; Martin, 1989; Thordarson, 1995; Thordarsonand Self, 1998]. Most other CRBG flow fields are either too poorly exposedor too extensivelyeroded for suchdetailed study. Thus there is a continuingdebate about the role of inflation in the emplacementof other CRBG lava flows [e.g., Reidel, 1998]. Other terrestrial flood basalt provinces (including the Deccan, Entendeka, Kerguelen, Parana,and Keewena)also appear to be dominatedby inflated pahoehoe [Self et al., 1997, 1998; Jerram et al., 1998; Keszthelyi et al., 1999; Frey et al., 2000]. However, a substantial fraction of flows in the CRBG and other flood basaltprovinces have brecciatedflow tops that do not fit in the aa versuspahoehoe categorization [Macdonald, 1953; Self et al., 1998;Frey et aL, 2000]. Based on the surface morphology of large inflated pahoehoe flows in , New Mexico, Australia, and elsewhere [cf., Theilig and Greeley, 1986; Keszthelyiand Pieri, 1993; Hon et al., 1994; Stephensonet al., 1998;Self et al., 1998], we infer that a substantialportion of the surfaceof CFB lavas consistsof a convoluted set of lobes, tumuli, and plateaus,all inflated to aboutthe samelevel. Sucha surface has a distinctive mottled appearance at 1-10 m/pixel, especiallywhen partially infilled by aeoliandeposits. Figure 10 shows this surface morphology of terrestrial inflated pahoehoe lava flows in a variety of settings. This morphology is diagnostic of inflation; no other style of eraplacementproduces this morphology. Do we expect inflated flows to be common on Mars? Inflated flows requirethe presenceof a coherentupper crust

Figure 3. MOC imageSP2-38804 and Viking contextfrom Matte Vailis centeredat 6.9øN, 183.0øW. Note platesin the lava flow consistentwith the centerof the a channelwinding North. The parallel ridgesin the top of the imageappear more similarto compressionalpressure ridges formed by the collisionof solid platesthan to leveesformed by overbank depositsof liquid lava. The streamlinedfeature SE of the center of the image appearsto be sheetsof crust piling up againsta topographicobstacle, leaving a wake in the crust. This type of wake is seenon a much smallerscale in Hawaii whenchannelized lava piles againsta tree. The groovesseen in the flow field in Iceland are more similar in scale(see Figures12 and 13). KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS 15,031

Figure 4. Cutoutwindows from MOC imageSP2-40703 and Viking context. Imagecentered at 19.2øN, 174.6øWinside Marte Vailis. Lower (left) cutoutwindow showsthe terminusof a flow lobe that was confinedto preexistingchannels presumed to be fluvial in origin. The marginsof the flow are typicalfor basalticlava flows. The albedomarkings within the flow are suggestive of multiple channelsthat did not have the opportunityto drain. The flow marginsand surfaceappear to be only minimally modifiedby postemplacementgeologic processes. The upper(right) cutoutwindow shows the marginof a flood lava flow with smoothsheet-like channel overbank deposits and disruptedchannel surface consistent with flow to the NE. Also of interestare the small fluvial channelsthat the lava embays. The marginof the flow is partiallycovered by fluvial and/oraeolian sediments. Numberson contextimage are latitudeand longitude.

and conditions under which it is easier for the lava to raise the the flow and the force holding the crust down. Thus the crustthan to advanceforward. The major differencebetween reducedgravity shouldsignificantly enhance the tendencyof the and Mars is the lower gravitationalacceleration on lava flows to inflate [Keszthelyiand Self, 1998a]. The other Mars. This decreasesboth the force driving the advanceof major difference between the Earth and Mars is the •100 :...... •.,

.:•. .•:.• -- ..:•,.. ,t, .....** .•-• :4::.•..: .:...'. •:•;•.• -•:;** .... **:,:,:,' ' ":a•.•?'.:'*,.' •::.. .::'.:.''"•i• :.'"'• .'•, ;• ,..:,. •:.•'::.....* .:.: '" ......

:•?' .....:•:..• :.....:...... •;•:....•::...... :...... ** ...,...... :,....•... ,..** ::..:.:**.::•:.

"* --.:, ' .:-.-..,, ..:,:, :::,.'::•:• .½:*':'"::::. ::...,• .....,..";:::;**•:.. :*';'" ..a:'.• ....' ', ...... *''": '"'•::• :.:..•..;•,.::...;.,...:•:,- .•,¾ '*

':...... •:. .:-:?. :., ':'•::*:: ,.,,,:;:.::"•:;:$:.::.,"':':•,...::.;:•:-'":: ::-:½::**.'- ...... ;..: .....,:. ;: .: ..:..• :..• . :•:'•":',-:':•.....-.:...'::::: •:::....•:::•,...... :.::• :.-:,..-.:..:::•.• ...... ß ::::::

....ß.::;•,:.•:" ..... , '", :•:- .a:.:.:.',•':..:::•:-'•:::? ,.,...... ,,- .,.-.... •..: ..... •-•::'" ,.,...... :.•:a, ß...... :• .... ,...... -.•' ..:..,.:.'." 4..::""-:-'.--: ..... "::X: ":"::.'-_*:., .:

":.::;.:::.:•a...... *.* :-:',;:'."'..:':*•:::'"•:..... •.-...,...... :-:;.': :..:: .... ,- :.,...... •.:?..,.....-::<;:..,?:;,:..:..,:., :'";•;.,-.;.-. :';;j : :.•'.:•...*::-:::: '%, •'•.*.:;•:• -'"" ., 4:...•;•½..::•;::.,:•*.".....::'":-•*•,• ,..... q:.*'.'•:.::. :::.:, ...... •-::'::•--a ': :..,•:::::::::'•...... :'--; "%,:.': •.- ':':,'G•:: :* :.:'•"4-, "''...... :...•:•:a :;:'•i..:',;:..., ::::.:•:.•::,.::::':•:'::•"::::.::'"' ::•: .:;.;•'::e•:::' ,?..'•;':--*"'":':•::•:::':•:';::::-•*::': •:"':-,"::'::'-"::::P/•:• •-•;:*,...... ;F'•*:*;:" ':'.;::::"::'/:'""*::;:::½'"':....:½•.:,:'• :•:" S:• ½:.'"::::' ':.:::.....: :::...... :. -...., ß.::--:..:::-:'.:':" '-" :.'.'...Z•;::%':'"'"':4:•...... :::'...... 2' :-.. ' ':;."-'• -.•::'".... •;,:.:.'-½..:• :--½'•...... :• r •* ß

. ...:-:.-',-:'• .. ß **:...., '";....•,:.-;:•:':;'•--.....'..:. :--:, • ..:. ::' • .::.:..;. , .,. -.. .;::., '.?:

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

::?•:..'...:, , "-.... -:.:....,.-.:<•r'..:...... j ::..:::'.:::...... :::::::::::::::::::::::::::::

Figure 5. Cutoutwindows from MOC imageSP1-24203 and Viking contextfrom AmazonisPlanitia. Imagecentered at 24.6øN, 163.6øW. In the upper (right) cutoutwindow, note the ridgeson the plateson the lava surfaceat the top of the image. Our interpretationof the formationof this platy-ridgedsurface is the following:First, the ridgesformed as compressionalhummocks perpendicularto flow direction. Theseridges do not appearto be inflationfeatures (see Figure 10) and insteadare similarto compressionalpressure ridges formed in brecciatedflow tops(slab pahoehoe, aa, etc.). We suggestthat belowthe brecciatedflow top, the lava solidifiedinto a coherentslab. A later surgein lava flux then raftedpieces of this crust,translating and rotating blockswith ridgeson their surface(see Figure 15). In the lower (left) cutoutwindow the sameflow exhibitsa much smoother appearance.The randomorientation of the web of thin convolutedridges on the smoothlava surfacesuggests that they are squeeze-upsof moltenlava comingup betweencracks in the cruston a pondedarea. Pressureridges would be expectedto showa clearpreferred orientation perpendicular to flow. KESZTHELYI ET AL.' MARTIAN FLOOD LAVAS 15,033




Figure 6. MOC imageSP2-42604 and Viking contextfrom Amazonis Planitia centered at 25.7øN,167.0øW. Groups of parallel ridgesare probablycompressional pressure ridges of disruptedcrust, and more sinuous,isolated ridges are probablylarge squeeze-upsin the flat pondedsurfaces. Local flow directionis ambiguous,but motionto the SW is mostconsistent with the ridgeorientations. times lower atmosphericpressure on Mars. This shouldlead the resolutionneeded (<<10 cm/pixel) to identify diagnostic to reducedconvective cooling of the lava surfaceand slower features (tumuli, inflation clefts, lobe boundaries,vesicle crustgrowth on a lava flow. However,the initial formationof distribution,etc.) in the Martian cliffs. Instead,we makethe a crust is governedby thermal radiation,which will be very assumptionthat mostof the Martianflood lavasthrough time similar on Earth and Mars [e.g., Wilson and Head, 1994]. have had surfaces like the recent flood lavas. Our Overall, the conditionson Mars appearto be very conducive examinationof the MOC imagesof the exposedsurfaces of for the formation of inflated flows. What remains to be seen the well-preservedMartian flood lavas suggeststhat only is if the eruptionson Mars were also in the regimethat forms isolated segments of their margins might be inflated inflated flows. pahoehoe. The rafted platesthat characterizethe Martian Locally high strainrates (relative to the lava ) flood lavas are not seenon inflatedterrestrial pahoehoe flows to tearingand disruptionof the crust[Peterson and Tilling, and insteadsuggest a disruptedcrust that translateslaterally 1980]. In Hawaii, aa flows with disruptedcrusts begin to during emplacement. Thus, while the scale and gross format effusionrates above about 15 m3/s[Rowland and physiographyof the terrestrialand Martian flood lavas appear Walker, 1990]. However, the Roza Member of the CRBG is similar, the more detailed flow morphologyrevealed by the an inflated pahoehoe flow that was emplaced at MOC images suggeststhat they formed by significantly approximately2000 m3/s [Thordarson, 1995; Thordarson and differentprocesses. Therefore we look for other terrestrial Self, 1998]. In the caseof the Roza eruption,the lava front analogsto providethe neededinsight into the emplacement was tens of kilometerswide, so the much higher volumetric processesof Martianflood lavas. effusion rate did not translateinto rapid flow rates or high strain rates. Thus high volumetric effusion rates, by 2.2. The 1783-1784 Laki Eruption themselves,do not rule out emplacementas inflatedpahoehoe sheet flows. The eruption in 1783-1784 at Laki, Iceland, produceda The layereddeposits seen in the walls of Vailes Marineris largeflow field that hasmany similarities to CFB lava flows, appearsimilar to the layersof lava flows commonlyseen in but alsohas areaswith a morphologythat is strikinglysimilar CFB provinces[McEwen et al., 1999a]. However, we lack to the Martian flood lavas. The Laki eruption is the largest 15,034 KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS


•;;,::...,•'•.•:;*.'..':.'?":' ...... :.;,......


ß.:..-; :"¾....:, ..... -,::,.? ,::::,....

...,...... :':.,'..'.... ;...S•,'

•:.. '.• . ....

Figure7. MOC imageSP2-53504 and Viking context from Amazonis Planitia, centered at 23.9øN,164.3øW. Rough ridged areasand the most prominent linear ridges appear to bepressure ridges of brecciatedmaterial. Smooth areas appear to beponded lava. In general,the rough areas match the dark mottles in theViking context images. The margin of theflow, where it abutsthe hillsto thenorth, is not morphologicallydistinct from the rest of the lava. Dunesalong the flow marginand the sparsecraters suggestthat this flow might be slightly older than some of theElysium Planitia flows. Flow direction is ambiguous,with no clear interaction between the lava flow and the older hills. KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS 15,035

Figure 8. MOC image SP2-54304 and Viking contextfrom lava plains near TharsisTholus, centeredat 17.0øN, 84.1øW, showingolder lava surfacewith ridgesand grooves,similar to thoseseen on the recentflood lavasin Elysiumand Amazonis. Despite significantaeolian cover, including>100-m-long dunes (arrows), these ridges are readily recognized. However, the marginsof platesare generallyobscured.

historicaleruption and the detailed written recordsof the effusionrates appear to have reachedvalues as high as 8700 eyewitness observationsof this eruption (e.g., J. m3/sfor shortperiods during the early part of theeruption Steingrimsson,written communications, July 4, 1783 and [ Thordarsonand Self, 1993]. November24, 1788;O. Stephensen,written communication, Ten eruptiveepisodes fed fissuresegments 1.6-5.1 km in August15, 1783)are particularly helpful in understandingthe length along a fissure system with a 27 km total length eruptivehistory and behavior of thelava. (Writteneyewitness [Thordarsonand Self, 1993]. Generally, only one fissure reportsare publishedin Icelandicby Einarssonet al., 1984 segmentwas active at any given time, but for a shortperiod and the accountof Jon Steingrimssonis translatedinto sectionsof three fissure segmentswere active over a total English[Steingrirnsson, 1998]). Also, while the Laki flow distance of-10 km [Thordarson and Self 1993]. The has been carefully examinedby severalIcelandic workers eruptionreleased 350 Mt of CO2, 240 Mt of H20, 120 Mt of [e.g.,Jakobsson, 1979; Oskarsson et al., 1982; Thordarson, SO:, 15 Mt of HF, and7 Mt of HCI [ Thordarsonet al., 1996]. 1990; Thordarsonand Self, 1993],little is publishedon the The fluorine in particularwas deadly to the local livestock flow morphology.Much of thefollowing description is based (primarily sheep),which led to a massivefamine that killed onpreviously unpublished field work by T. Thordarson. approximately 20% of the population of Iceland The eruption started on June 8, 1783, and lasted until [Thorarinsson, 1979]. The tops of the buoyant eruption February7, 1784,erupting 15.1 km 3 of basalt(dense columnspenetrated the tropopause,and SO2 was distributed equivalent (DRE)) that advanced>60 km from the vent acrossthe Northern Hemisphere[Woods, 1993]. The Laki [Thordarsonand Self, 1993; Thordarson et al., 1996](Figure eruptionwas associatedwith an unusuallycold winter, with 11). About 10%of the mass(and >50% of the volume)was the Danube River freezing over at Vienna and rafted ice depositedas pyroclastics. Of the lavaflows, 60% (by mass filling the Mississippi River at New Orleans [Thordarson, and volume)was eruptedin the first 1.5 monthsand 90% in 1995; Thordarsonet al., 1996]. thefirst 5 months[Thordarson and Self, 1993]. Theaverage The lavasflowed away from the fissuresystem and passed effusionrate for the entire eruption was 730 m3/s [Thordarson through two river gorgesbefore fanning out on the coastal and Self, 1993], not muchhigher than historicalMauna Loa plains(Figure 11). Field mappingand examinationof aerial eruptions[Rowland and Walker,1990]. However,based on photographsshow that the majority of the lava formed theeyewitness accounts of therate of advanceof lavasurges, inflated pahoehoe. However, one sectionof the distal end of 15,036 KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS

Figure 9. MOC image SP2-54303 and Viking context from lava plains near TharsisTholus, centeredat 14.2øN, 83.7øW, showingolder, degradedexample of flood lava with similar ridged morphologyas the more recent Elysium and Amazonis examples. Groovesin the crustsuggest flow to the NW during the initial emplacement.However, the crackindicates opening to the NE, suggestingthat the flood-like breakoutfrom this lobe movedin that direction. Note that in theseimages, there is no correlationbetween the lava morphologyand the albedomottling in the Viking image. the flow field is composedof arcuatepressure ridges, rafted breaksup during springfloods ("undanhlaups"in Icelandic). crustalplates, and longitudinalgrooves (Figures 12 and 13). Recordedobservations of the flows advancingthrough the The lava in this areais coveredby a few centimetersof , areasshown in Figures 12 and 13 includereports of the lava but the underlying flow top can be seen to be brecciated, surgingforward at 2-5 timesthe averagerate for severalhours forming a lava surfaceunlike any commonexamples in [Steingrirnsson,1998]. Hawaii. The closestHawaiian analog to thisdisrupted crust is Based on field work and examination of the aerial slabpahoehoe, a form of lava transitionalbetween pahoehoe photographs(Figures 12 and 13), we conclude that the and aa [Macdonald,1953]. groovesin the flow surfaceare wakes left in the brecciated We infer that the raftedcrustal plates initially formedon a crust as it translated past isolated obstacles. The arcuate relativelystagnant flow andwere disrupted and transported by ridges appearto be locationswhere the brecciatedcrust was a later surgein the flow rate. This is directlyanalogous to the compressedas it ponded before overtoppingtopographic manner in which the ice cover on a partially frozen river ridges. This part of the Laki flow field is underlain by KESZTHELYI ET AL.' MARTIAN FLOOD LAVAS 15,037


'":•:..;--:,::•::.i:;;•;;:•,:•.:'i;;;•;•:;::i;"•:,•';•;•...:,:...... •:;::•;;;:;• .:;i":' •;::•:,,..;•::;.;•;:.½•;;;;,•½.::,-?.. -:• ...... •: .....%..'::;•'.o. ;:•::::-:j•':;;•Z :•:•---*-'•---•:•'•'•-4•:,:• ...... '-:","- ...... ':'"•:•:::'?:•;•;•:•; ;•;-•::•: ½ :%'-: •:•.:•:-:•,•..;•:•..;::,• •,;:½•;;:•;;:(•...:½*:-•I'2Z'•;;:•'• ""'•""•------•-••.•-;-:•-••: ...... •-•..•••:•.•.

. .

;::;•;;::::-•::;:E•:,:%•:•:•:':'•½'•½•J;&.,-:'•':•;-•::::,.::'• .... •:•'-•-:' :,:½;'.;•½'*- •4::..":'•'.;"•':•:• ?%{ :::::.--•½:• •:::...:,.-;;::•<.:•.• ,=•:•:•.:--.-:.•, ..:•:-•...½:;:.,::•.• .• :,:• :½. ::-:-•,---.•.--•,:•;•:•::::,...-•-.:..•:•;::•:..--.:.,.4::.•......

...... ½...½--::,•.½::.::;•. •:..•.:• -, .;::•,•s.•,•...,.:.•,.,..•::•:.-.-:•:•,•:..•.. :.....• •.:.::•,,,:..,...... • -.'•:•.'"<•½:": :;•::½•:•:'•,:.: -•.':>'•::.(;•-' ...... :•-½•':';': .•:..::•.•:•:•.:::..:. -,½':'•: ,::;...:;•,•!:•..... ""-•;.._•.:..':%;•:( ...... ,•;•...:½•...:;'.•.•:•;•-----•;•:•.•:,:•..-,z•;'.;;-":'•:'•-'•.•..•;•: "•:•;½•;:;•T*••..•** ...... ;•::•,:•:•:½"•"f•:••4:•...... • ...... •"•-:.. ,.. '"';½•;::•.:L.½•:•:-•;;;;::;•;.;:½.:.•.•...... •:.....•:::::,:..•.: .:•.•:•:.j;:•:.:....•.>•..:.•:•;•4•(.•::•%•:4:..•:•;;•:J-• ' ::"";"•

...... '.'.:•;---:-½.;.-'•:?;..-.- . :.•..• ._•:•-•..:•..•.½•"

-.:-•,:-•-*:% ...:;.;.;-.•-.' •



Figure 10. Examples of terrestrial inflated pahoehoeflow surfacemorphology. (a) Aerial photographof SE corner of the Macarthy'sflow field, New Mexico. Note largeinflation plateau in centerof imagewith inflationpits. Smallerinflation features on hummockypahoehoe surfaces appear as mottling at this resolution. (b) Aerial photographof SE portion of Carrizozo flow field, New Mexico. Note the largeinflation plateau along the flow marginwith inflationpits and the moremottled appearance of the hummockyportions. Dark delta-shapedfeatures are smallpatches of slabpahoehoe breaking out from the inflatedpahoehoe lobes. (c) Aerial photographof SE portionof the Laki flow field, Iceland. Note the large inflationplateau in the centerof this lobe and the more hummockytextures. Highway 1 crossesthe lava in this image. (d) Obliqueaerial photographlooking over the west end of the Pisgahflow field, California. Note mottled patternof the hummockypahoehoe surface made more visibleby partialfilling with aeoliansands. Trucks on Interstate40 areshown for scale.Photo courtesy of BruceC. Murray.

rootlesscones in the older (934 A.D.) Eldgja flow field. raised edges and streamlinedshape of the groovessuggest These cones occur both in lines and in isolated clusters of that the brecciatedtop of the lava had a significantyield cones, making them the only likely candidate for the strength, while pahoehoebreakouts show that the interior topographicobstacles that the Laki lava encountered. The remained fluid. 15,038 KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS






. -.--.:....-:>..m.,- .::..: -...:-:-.--.-....::...- ...... - .:.- -. :-- .-

Figure 10. (continued)

Thissurging and autobrecciating mode of emplacementhas beforenarrow channelsor tubesevolved within the flow). not beendescribed in Hawaii, but may applyto the -•20% of Thesenear-vent flows are characterized by highlocal effusion the flows in the Columbia River flood basalt provincethat ratesand hot, gas-rich lavas with relatively low . haverubbly flow topsand smooth pahoehoe-like flow bases Where the flows have ponded,their surt•tcesare also [Selfet al., 1998]. It mayalso help explain the enigmatic characterizedby veryshallow slopes (<<1ø). brecciatedflows seenon the KerguelenPlateau [Frey et al. 20001. 3. Modeling In Hawaii, longitudinalgrooves and pressureridges are largelyabsent within lava flows. However,we havefound The observationsfrom Iceland,and to a lesserextent, from somepieces of raftedcrust that are tens of metersin scale. Hawaii and continentalflood basalt provinces,strongly Raftedpieces of crustare visible along the SouthwestRift suggestthat the Martian flood lavas were emplaced assheet Zone of Kilauea and in the summit of flowswith intermittent,surging advances. In thissection we (Figure14). The raftsare mostly confined to thenear-vent describethis qualitative emplacement model in moredetail faciesof lavaflows (i.e., thebroad sheets of highlyvesicular andthen attempt to placequantitative constraints oneruption lava within a few hundred meters of the vent that formed parameters.Surging emplacement is inherently difficult to KESZTHELYIET AL.' MARTIAN FLOOD LAVAS 15,039

Figure 10. (continued)

modeland we lack critical measurements suchas the length crust.The second involved open sheet flow with a disrupted andwidth of individual outpourings. Therefore it is currently and mobile crust. The thermal models from Keszthelyi and impossibleto produce a quantitative model replicating the Self [1998b] are readilymodified to Martianconditions and eraplacementof a specific Martian flood lava flow. However, candescribe each of thesetwo distinct flow regimes. it is possibleto showhow the flows were not emplaced. 3.2.1. Insulated sheet flow. The thermal model for Eliminationofthe physically impossible provides surprisingly insulated sheet flows presented by Keszthelyiand Serf[ 1998b] usefulconstraints on the formationof the flows. is a simplemodification of the thermalmodel for lava tubes presentedby Keszthelyi[ 1995]. Bothmodels involve steady 3.1. Qualitative Model stateconductive heat loss through the uppercrust and heat The brokenrafts of crustand the observationsof the Laki generationvia steadystate viscousdissipation. Flow flowfield suggest that the Martian flood lavas have disrupted velocities within the sheetflow are calculatedusing the flowtops. The observed rafts of crustare highly suggestive equation for flow betweentwo parallel platesand the of surgestraveling through the flow, inflating it to the point assumptionsof laminarflow, a Newtonianrheology, and thata suddenbreakout carries away large pieces ofpreviously shallow slopes. The resultingset of equationsis givenby solidifiedcrust (Figure 15). Thecrust riding the flood is Keszthelyiand Self[ 1998b]as brokenup as it issheared along the margins of theflow and as theflow moves between and around topographic obstacles. = pg0 H 2 / 8 q, (1) Asthe surge wanes, the base of the disrupted upper crust may Qvi• = p g H 0, (2) becomecoherent via solidification of the liquid lava , Qcond= k (T-Ta)/ Hc, (3) allowinginflation to resume. While thereis someevidence •r/•x = (Q,•- Q•o•d)/ ( p Cp*H), (4) forthe formation of marginallevees, based on the images availableto date,the flow doesnot seemto be restrictedto where is the averageflow velocity,p is the bulklava ,g is thegravitational acceleration, 0 is slope, H isthe narrowchannels. The disrupted crust on theseopen sheet flowsshould have experienced a steady process ofproduction thickness of theliquid lava, q is viscosity,Qv•,½ is theheat anddestruction as it is churnedby the flow pastbends, generatedby viscousdissipation, Q½o,a is the heatloss by constrictions,and other types of obstacles. conduction,k is thermal conductivity, T is thetemperature of theliquid lava, T• is the temperature of flow surface (assumed 3.2. Quantitative Thermal Model to beambient), Hc is thethickness of thesolid crust, OT/Ox is thecooling per unit distance of flow, and Cp* is heat capacity The surgingflow model implies that the lava was includingthe effectof latentheat (dominated by the transportedin two very differentregimes. The first flow crystallizationof ). Thismodel is verycrude, and regimeinvolved flow under a thick,stationary, insulating results should be consideredaccurate to no betterthan a factor 15,040 KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS


...."Laki fissures


7 August

,. .,

.. •:-•' -...... 18June 9 August

20 J,uly


Figure11. Mapof Lakiflow field, Iceland. Positions of theflow front at knowndates are shown. Large, flood-like breakout withridged surface and rafted plates (Figure 13) is indicatedby location1. Smallerflood-like breakout with rafted plates (Figure 12)is shownby location2. Classicinflated pahoehoe structures (Figure 10c) are shown by location3. KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS 15,041

:.•.;...... ::.::...'"•;:'½?.:'.';'?..-";.:?,•.•.:.;:•-- • ..'• ..•. ••:• - -'"'•'"?-::;.:-:;.:?.-..•½•;..:. **•:,'•:•-, •.'• ...... •':;:; .,.-'..;:.•'"'. .;,(';.:- •.,..•.'-.•:•,• :.' :..-.::.•:,;:.;•;..•.:..;•;:•.• '" .•...' ':•...-. .::. ;•--..... •.'•.'-. :..'.&: .... .: •: ..-,.• :..:; ...•:,• ..... ' .... ;• •".:'•i•?½:...' -:..

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

::;-.•..;•-...... ::.;-.-...;..;:•.::•:•:..,.:;:...:...:....::; ...... •;..• ...:::...... •:. . •,..;•,..;:.;;.,...:...:;.:½...... -: .: .•...•...... :. •:•;:::...•,.:.:.?•.•½:••:•--:; '--'-•:•••'•'•-•:•..•;:•:::•:::•t•-•;:----: ..•: •-- .•:•e•:.:f½•½:•'?..:..-,•-;:• . .-••• .-.•...... ,..•-• ...... •.::---:..-.-::; •:.,,..•...-:*.•:::.:..-•:.:::•(•;..•..• ...... -.:• •:::..•:•:,?.,ß...... :..:...... '•...... - .:..:..•... •-•-.•- .•..-•,.-:::;:-g•.• • -. ,.•:..:,-..;..j•..... ,:-:•.•..•,•..•,-.:,-•-.-.:...... • ,•.....z. ...:.:....,..'...... ½•'::;?:.....•..,.,..:...:...... •:":':".':.:':•...,%• *..,..•....• ...... •'•'- ...... • ...... :•;.-...... 4..,':½ •...... •'"•/ •.• '•,'½•;-..•<:,.½.<..;..•...... •:....•....."•"• '" ...... '•••••:•s:•':•'-'•.'-:-*;:'.:½: ..,,,...:-., •,...,: .:•...•...-...-.. .:::*';':::...... :...... :','•.•.•, :.:*•.-,":'.•:..::..: ...... :::•:•...... ,'/*•,• %...... ,•:....: ..-'.;•-;•:'...... ,..•::.:...•,...... ;;'½.*.-. :..:,:•.;.:?,:.-½,½.. ::**:.•::..•: ...... •.,...... •:.*?•.'•.--...:.,$(•:,;.•,; .. •::a,.:.,:..,..::;,.;.;., ...... :.$..:.:,...•.-•..-:::',•::':-•,,•...... ;.•;.,:,;,.:..:;•?..•: :•:a.:...:,:..;•:•., -'-•?.;.•.•,?:.:....::,?:, .. -**:'**•.",.:":*:•,.***-'•.,..:--.• ,....•. :.,.:...... :.•..•: ...... ::.?a;:.,.,;'.:; ..... ::,,,.... •:•:',::.•...... :..••-..-;"...... •".-,?;a.:•:, -:.. •...

'::;• ...... ;•;•':::•::;•2 -•,.½::'""•?'•':"•: :•¾•::•.; " .. ';&:•:•'• '•'•'•;•'••;:e• ':'":'•'" ' •"•::; .... .-•.. .:,.. ..•..:5.:....;½..::...... •.,.,:..:.•..½.½, ...... : ..... •, .;• •. ::•.•.-.:.• ...... :..... •½. •.....

•:•, ...... :;::•.•:.•.•:; .,

•:;• •.•;• "•-•:...•.• .•:;• • :.::

.. •..:.:;•:•::•::::..:...... :.::::::•:•.....:::::::::::::::::::::::.;.:..::•..;.....:...:..: .::..::•.:::::....::;..:.:•:: :.....:..•:::.•.:•:..::.:..•;:. :.:..::...... :..:...... :..::.:::...,:...... ::•.:::: ...... :.•...... :...... :.... :::::. :...... :.;.:..:.. :.•:.::..:.:....:•.•:•.:•..:•::: :.....:•:....•.....:::..:•.....:...: ::... .. :.:...... :...... :...... : . ...:%.½:?.

z lkm

Figure 12. Aerial photographof portionof the Laki flow field, Iceland(labeled 2 in Figure11). (a) Comextshowing the Skafia River, Highway 1, and farmsas well as the locationsof Figures12b and 12c. (b) Close-upof the northernchannel which was activeon July 17, 1783,showing 100-m-scale rafts of lavacrust transported in an openchannel. This channelbroke out from a lava frontthat had stagnatedaround July 12, 1783. (c) Close-upof this earlierlava, showingthat it formeda broadsheet with pressureridges, plates, and grooves. The ridgesare piles of ,the plates are pieces of moreiraact crust, and the groovesare "wakes"left in the brecciatedtop as it movedpast topographic obstacles. 15,042 KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS

Figure 13. Aerial photographof portionof the Laki flow field, Iceland(labeled 1 in Figure 11), showingpart of the largeflood- like breakoutthat formedin mid-June,1783. Note the largerafts of ridgedcrust tom from the earlierformed flow top, as well as long groovesin the ridged lava. The ridged lava surfaceconsists of pressureridges in the brecciatedflow .top,not tumuli as suggestedby Thedigand Greeley[ 1986]. The groovesformed as this brecciated flow toptranslated over topographic obstacles.

of 2. However, more sophisticatednumerical models [e.g., roughly correspondingto typical values for ultramafic, Sakimoto and Zuber, 1998] do not produce significantly basaltic,and basaltic-andesiteflows. differentresults, only moreprecise ones. Severalpoints in Table I require somediscussion. First, The major modifications required to adjust the model irrespectiveof the chosenviscosity, a flow that can extend presentedby Keszthelyiand Se/f[1998b] to the Martian flood 1000 km under a l-m-thick crust will be turbulent because of lavas are the reductionin both slopeand gravity. Ambient the high flow ratesrequired. Because it is difficultto envision temperatureis also reducedto-30øC, but this parameterhas a stationaryinsulating crust surviving over a turbulentflow, very little effect on the model results. Also importantis that we regard this scenarioas unlikely. In the case of the the Martian flood lavas appearto have advancedas much as ultramafic (10 Pa s) lavas, the crust must be •20 m thick 2000 km from the presumedvent areas,while Keszthelyiand beforeflow velocitiesdrop into the laminarregime. While Self [ 1998b] only examinedthe requirementsfor a flow to the •25 m total thicknessrequired by this scenariomay be extend •100 km. compatiblewith someobserved flow thicknesses,it suggests In examining the Martian flood lavas, we chose to that it mightnot be possibleto produceinsulating sheet flows investigatethe effectof lava viscosityand the thicknessof the with extremely fluid lavas, even with low gravity and insulating crust on the minimum flow thicknessand flow remarkablyshallow slopes. This arguesagainst ultramafic velocities needed for a flow to travel 1000 km without compositionsfor the Martianflood lavas. However,we note freezing. As in the work of Keszthelyiand Self[ 1998b] we that recent studiessuggest that mildly turbulentterrestrial assume that a temperature drop of 50øC within the lava lava flows may have been emplacedas insulated transportsystem would lead to significantcrystallization and pahoehoesheet flows [Hill et al., 1995]. the flow stopping.Thus the maximumcooling rate allowable For a viscosityof 1000 Pa s, the minimum model flow for a 1000-km-longflow would be 0.05øC/km. Table 1 lists thicknessesare of the orderof 60 m. This alreadyexceeds the the model runs and results for the Mars Observer Laser measuredflow thicknesses,so we can confidently rule out Altimeter (MOLA) derived slope of 0.025%. Viscositiesof more evolved and viscouscompositions (e.g., )for 10, 100, and 1000 Pa s for the fluid lava are considered, these flood lavas. We conclude that physical properties KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS 15,043


Figure 14. Aerial photographsof platy lava in Hawaii. (a) Pondedflows at the northernend of Mauna Loa's summitcaldera (Mokuaweoweo)and in North Pit. In this case,there is no clear directionof flow recordedin the orientationsand shapesof the crustalrafts. (b) The 1974 Kilauea SouthwestRift Zone openchannel flow. Lava pondedagainst the face and formeda solid crustbefore the flow surgedonward. Boththese areas are characterizedby high flow ratesand shallowslopes.

similarto terrestrialbasalts are probablythe mostappropriate least1-5 m2/s for the lava to be transported > 1000 km without for furthermodeling of the Martianflood lavas. freezing. The available images suggestthat the sheetsare Since the model constrainsboth flow velocity and flow generally on the order of 10 km wide, requiring average thickness,we can also make quite robustconclusions about effusionrates on the order of l-5xl04 m3/s or greater. This is minimum flow ratesper unit width of flow. This mustbe at approximatelyI orderof magnitudehigher than estimated for 15,044 KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS


Figure 15. Cartoonof formationof rafted plateson Martian flood lavas. (a) Lava entersthe area, most likely as a rapid flood. (b) The flow front stagnatesand a largevolume of liquid lava is storedwithin the confinesof a thickening,insulating crust. Lava is fed into the areaunder a thick insulatingcrust. (c) The flow front fails and lava breaksout in a runawayflood, rafting away piecesof crust. The sequenceis repeatedas the floodwanes and slows, and an insulatingcrust begins to growagain.

Table 1. Results From Insulated Sheet Flow Model terrestrialflood basalt eruptions[Thordarson, 1995;Self et al., 1997, 1998; Thordarsonand Self, 1998]. Viscosity Crust TotalFlow Average (Composition) Thickness Thickness FlowVelocity As an aside,the Keszthelyi [ 1995] thermal model for lava Pa s m m m/s tubesindicates that 1000 km long flows could be formedon 10 1 (turbulent) theMartian plains with an effusion rate of only33 m3/s and a (ultramafic) 5 (turbulent) tube diameter of 22.7 m. However, since we have not found 10 (turbulent) any evidencefor narrow lava tubesin the Martian flood lavas, 20 24 0.25 and becausesuch lava tubesare not expectedon theseshallow slopes[Keszthelyi and Self, 1998b], we find it very unlikely 100 I (turbulent) that the Martian flood lavasseen in the recentMOC images (basaltic) 5 27 0.75 were emplacedat low effusionrates. 10 26 0.39 20 32 0.21 3.2.2. Open sheet flow. In order to quantify the more rapid, surging,stage of our hypothesizedemplacement model 1000 1 (turbulent) for the Martian flood lavas, we require a different set of (basalticandesite) 5 70 thermalmodels. Crisp and Baloga [1994] developeda model 10 56 0.33 for the open channel aa flows during the 1984 Mauna Loa 20 53 0.18 eruptionthat is generallyapplicable to the open sheetflows Insulatedsheet flow not consideredphysically possible with we postulate. The Crisp and Baloga [1994] modelcalculates turbulent flow. the coolingwithin the well-mixed core of the flow, including KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS 15,045

the effects of reductionin radiative coolingby a disrupted While laminar, averageflow velocity for Newtoniansheet crust,mixing of cold crustback into the flow, and the latent flow is givenby [e.g.,Bird et al., 1960] heat of crystallization. Keszthelyi and Self [1998a,b] presenteda minor modification of the Crisp and Baloga = pg 0 H2 / 3 fl. (10) [1994] model to convertthe inputs to parametersthat are somewhat easier to constrain for flows that are not observed There are many different formulationsfor turbulentflow, but while active. In particular, the increasein crystallinity is Keszthelyiand Self[ 1998b] explainedwhy the one presented calculatedfrom the coolingrather than using petrological data by Goncharov [1964] appearsto be the most applicableto from samplescollected from the activechannel. Atmospheric lava flows. Thus for moderatelyturbulent flow we use convectivecooling and viscousheating were alsoadded, with flow ratescalculated assuming a Newtonianrheology while 2= gH0 / C• (11a) the flow is laminar and using the formulas presentedby Cf = •,/2, (1 lb) Goncharov[1964] whenthe flow is turbulent.Equations (5)- ),,= {4 1og•o[6.15((Re'+800)/41)0.92]}-2, (11c) (10) describethe Keszthelyiand Self [ 1998a,b] modification Re' = 2Re, (11d) of the Crisp and Baloga [ 1994] model: whereCf is a frictionfactor and •, is an empiricallyderived fit c3r/c3x= { Qvisc-Qrad-Qatm-Qentr} / { 19Cp, S}, (5) to experimentsof fluid flow that are moderatelyturbulent [Goncharov,1964]. where T is the core temperature,x is the down flow direction, Combining these equations, the Keszthelyi and Self Qviscis the viscousheating, Qrad is radiativeheat losses, Qatm is [1998a,b] model calculatesthe flow rate and coolingrate for atmosphericheat loss, Qentris the heat lost from the core of the core of a sheetor wide channelflow with a disrupted the flow via entrainmentof the crust, is the averageflow crust. However, it must be noted that the model does have velocity,p is lava density,Cp* is heat capacityincluding several critical shortcomingswhen applied to the Martian latent heat, and H is the flow depth. Viscous heating is flood lavas. First,the pulsatingnature of the proposedmodel describedby equation(2), and radiative heat lossescan be is not incorporatedinto this simple model. However, by expressedas examiningthe two flow regimesseparately, we are able to make some reasonable inferences about the overall flow Qrad= •;*f(74- ra4), (6) emplacement. Second, even if the liquid lava is approximatelyNewtonJan, a thick brecciatedcrust can impart where e is emissivity (-0.95), , is the Stefan-Boltzmann the bulk flow with a significantyield strength[e.g., Booth and factor,f is the fractionof coreexposed, and T, is the ambient Self, 1973; Hulme, 1974; Fink and Zimbelman, 1986]. temperature[Crisp and Baloga, 1990]. The atmosphericheat However, inputtinga reasonableviscosity of 500 Pa s for the losshas a similarexpression, Laki lava, our flow modelproduces flow velocities(1-4 m/s) anderuption rates (1500-9000 m3/s) that are consistent with Qatm-'h f(T- Ta), (7) the eyewitnessobservations. This suggeststhat the crust riding on this type of flood might have a relatively minor where h is the atmosphericheat transfercoefficient, which effect on the bulk flow rheology. Third, the modelcannot be hasan approximatevalue of 70 W m'2 K 'l on the Earth used to examine the temporal and spatial evolution of the [Keszthelyi and Denlinger, 1996]. This value shouldbe flow. Still, the successof the original Crisp and Baloga directlyproportional to atmosphericdensity and heat capacity [1994] model suggeststhat this modified version shouldbe [e.g., Arya, 1988], suggestinga value of-1 for Mars. Heat more than adequate to provide broad constraintson the transfer due to entrainment of the crust is eruption parameters. More rigorous modeling of the flow behavior,as in the work by Glaze and Baloga [1998], will Qentr= 19re, Hc (T- T0 / z, (8) becomeapplicable only when individualflows and breakouts can be mappedout. where Hc is the thicknessof the crust, T• is the average The key inputs into this model are the thermophysical temperatureof the crust,and z is the averagetime a pieceof propertiesof the lava (initial temperature,bulk viscosity,heat crustsurvives before being entrained back into the coreof the capacityincluding latent heat and bulk density)and the crust flow. (average temperature, survival timescale, and thickness). The sheet flow is assumed to be laminar while the Slopeis the singlemost important factor [Keszthelyiand Self, Reynolds number (Re) is less than 500. We note that the 1998a], but is well constrainedfor the mostrecent flood lavas valueof Re markingthe transitionto turbulentflow depends by the new MOLA data at about0.025% [Smithet al., 1999]. on both the exact versionof the Reynoldsnumber used and The thermo-physicalproperties are related to the assumed the geometryof the flow. The variouspermutations of Re and compositionof the lavas. We have determinedthat the open the critical Re that permeatethe publishedliterature were sheet flow model does not help further constrain the discussedby Keszthelyiand Self[1998b]. In this paperwe thermophysicalproperties of the lava and thus only discuss use runsusing the basalticcomposition suggested by the insulated sheet model. Re = p rh / rl, (9) The propertiesof the crust are poorly constrained. The only published observational constraints on the crust whererh is the lengthscale of the largestphysical perturbation parametersare from Crisp and Baloga [1994] for the 1984 that canpropagate down the lengthof the flow (i.e., the flow Mauna Loa openchannel aa flow. After examininga rangeof thicknessfor a sheet), is the averagevelocity, and rl is possiblecrustal parameters,we have chosento report just viscosity. three examples which illustrate the main results. We call 15,046 KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS

Table2. CrustalProperties for OpenSheet Flow Model extend more than a few tens of kilometersbefore cooling CrustType 'c,s f, % Hc, m T-Tc, øC >50øC, the flow must be >20 m thick. At such a flow Thin 100 50 0.05 100 thickness the calculated flow velocities will be of the order of Medium 1000 10 1 500 a few metersper second,and the localflow rateswill be >40 Thick 100,000 1 10 700 m2/sper unit flow width. For sheets-10 km widethis translatedto minimumlocal volumetric flow rates of-4 x 10s m3/s. If individualflood-like breakouts traveled closer to 100 these three cases"thin" crust, "medium"crust, and "thick" km,the volumetric flow rates are likely to be-106 m3/s and crust. The parametersused for eachcase are listed in Table 2 flow thicknesseshad to be >40 m. However, it is likely that but requiresome justification. In general,we assumethat a the final flow thicknessdropped as the flood stageof the thinnercrust will be bothhotter and more rapidly destroyed. emplacementwaned, explaining why >40-m-tallflow margins In the "thin" crustcase, we examinethe possibilitythat the havenot beenobserved in this region. crust is only centimetersthick. Such a thin crustwould be expectedto be disruptedand reincorporated into the flow in a 4. Conclusions and Discussion matterof minutes. The extensivelydisrupted crust would also Based on comparisonwith the morphology of large exposea relatively large portionof the interior of the flow. terrestriallava flows and thermalmodeling, we concludethat The values listed in Table 2 for the thin crust case are for a the recent Martian flood lavas are broad sheet flows that were crustmarginally thinner, hotter, and more disrupted than that emplacedby alternatingbetween two flow regimes:insulated reportedby Crisp and Baloga [1994] for the earlypart of the sheet flow and flood-like breakouts. The closest terrestrial 1984 Mauna Loa eruption. The "medium" crust case analogwe have identifiedis part of the 1783-1784Laki lava examinesparameters related to a crustapproximately a meter flow in Iceland that was subjectedto large surgesin lava thick. Crustalblocks might survivefor tens or hundredsof influx. minutesand would be substantiallycooler than the liquid Modeling results suggest that the flows should have interior of the flow. The "thick" crust is taken to be about 10 thermo-physicalproperties close to basalts. More evolved m thick with blockssurviving for roughlya day. The thick compositions(i.e., andesites)would require flow thicknesses crustcase is likely to be the bestrepresentation of the rafting greaterthan thoseobserved. Ultramafic lavasare not likely, lava on Mars,where pieces of crustare strongenough to form because they would require the formation of a stable, rafts kilometers across. While these three cases do not cover insulatingcrust over a turbulentflow. the full rangeof plausiblecrust properties, they do provide Average eruption rates for basaltic Martian flood lavas importantinsight into the dynamics of the surgingflood lavas. wereprobably--•10 4 m3/s, about 1 orderof magnitudehigher Despite the shortcomingsof the model noted earlier, than those estimated for typical terrestrial flood basalt severalpoints are clearly demonstratedby the modelresults eruptions.However, the large surgesthat raftedaway pieces shownin Table3. First,we canrule out certaintypes of flow of crust are likely to have involved short-livedvolumetric behaviorwith a high degreeof confidence.If the disrupted flowrates 1 to 2 ordersof magnitudehigher (10 s - 106m3/s). crustis vigorouslybroken up and mixed into the interiorof This style of emplacementis significantlydifferent from that the flow, the flow will be remarkablythermally inefficient. In inferredfor the majorityof terrestrialflood basalts. However, fact, the "medium"thickness crust, with a survival timescale if typicalMartian flood lava eruptions produce 103-104 km 3 of of 15 minutes,cools almost an orderof magnitudefaster than lavaat -104 m3/s, the eruption durations are expected tobe of the "thin" crust that losesheat via essentiallyuninhibited the orderof a year to a decade,very similarto thoseestimated thermal radiation. This counterintuitiveresult is actually for some terrestrialflood basalt eruptions[e.g., Thordarson, easily explained. The crustriding the flood-likebreakout is 1995; Thordarsonand Self, 1998]. the productof daysor weeksof cooling. When this material Theseconclusions naturally raise a hostof new questions. is mixed into the interior of the flow, the cumulative heat loss Perhapsthe mostobvious is, Why are the stylesof eruptionof from the many daysof coolingis suddenlytransported into flood lavas different on the Earth and Mars? We speculate the previously insulatedflow interior. Not even wholesale radiative heat loss from the surface of the flow can match this rateof cooling. Table3. ResultsFrom Open Sheet Flow Model The modeling suggeststhat in this most thermally Flow Flow Flow Crust Cooling inefficient mode, the flood should not have been able to Thickness Velocity Rate Type Rate advancemore than 1-12 km before losing >50øC from the m m/s m2/s øC/km interior of the flow and being brought to a halt by the "thin" 71 resultingcrystallization. The fact that we seethe sheetlobes 5 0.19 9.5 "medium" 520 with rafted piecesof crust extendingfor tens of kilometers "thick" 73 strongly indicatesthat this thermally inefficient style of "thin" 8.8 emplacementwas not the norm for the Martian flood lavas. 10 0.78 7.8 "medium" 65 This rapid mixing of the disruptedcrust and freezingof the "thick" 9.1 lava would be most likely if the flow was turbulent,again "thin" 1.8 arguingagainst very fluid ultramaficlavas. 20 1.8 36 "medium" 14 The model runs with the "thick" crust are our current best "thick" 1.9 attemptto mimic the undanhlaup-likephase of the Martian flood lavas. Even thoughthe thick crustcase has a crustthat "thin" 0.57 is stableon the timescaleof a day, the occasionalentrainment 40 3.0 120 "medium" 4.2 of blocks of crust leadsto substantialcooling. In fact, to "thick" 0.58 KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS 15,047

thatthis may be theresult of thethicker lithosphere on Mars. Another location where the predictedrecent pyroclastic Wilsonand Head [1983, 1994] have shownthat larger depositsmight be preservedis in the polar regions. Again, volumesof magmamust accumulate to breakthrough the detailed in situ observationswill probably be needed to thick lithosphereon Mars. The resultingdikes should be distinguishash fall layers formed by an eruptionfrom dust- widerand produce higher eruption rates. The gas phase in the rich layers causedby dust storms. Such observationscoult propagatingdike is expectedto be concentratedin the upper enable us to observe the interaction between volcanic partof thedike [e.g.,Spence et al., 1987],leading to an initial eruptionsand climate, in the samemanner as the terrestrialice vigorousfountaining phase followed quickly by a waning cores[Zielinski et al., 1994]. phaseas the volatilesare rapidlyexpended. Each phase of extendingthe fissuresystem could lead to a pulsein the flux Acknowledgments. This researchwas partially funded by grant of lavatraveling through the flow, as in the earlypart of the NAG5-8308 from the NASA Mars Data Analysis Program. We Laki eruption. Suchcycles are commonfor terrestrialbasaltic thank David Pieri (JPL) and Bruce Murray (Caltech) for aerial photographsof the Carrizozoand Pisgahflow fields. We alsowould eruptions[Wadge, 1981], but might have been even more like to thankTracy Greggand David Crownfor very helpfulreviews. dramatic on Mars. We thank Mike Malin, Ken Edgett,Elsa Jensenand othersat Malin The hypothesisthat the Martianflood lavas were fed by SpaceScience Systems for the excellentimages. fissureeruptions with repeatedvery vigorous phases to predictionsthat may be testable with current and future References missionsto Mars. The hypothesizedstyle of eruptionshould producevery little in the form of vent structuresbut should Arya, S. P.S., Introductionto Micrometeorology,307 pp., Academic, result in extensivepyroclastic deposits. The lava flows SanDiego, Calif., 1988. Bates, R. L., and J. A. Jackson(œds.), Dictionary of Geological withoutextensive vent structures seen around Cerberus Rupes Terms.,5th ed., 571 pp., AnchorBooks, New York, 1984. [McEwenet al., 1999c]are consistentwith ourhypothesized Bird, R. B., W. E. Stewart, and E. N. Lightfoot, Transport styleof eruption. Phenomena,780 pp., JohnWiley, New York, 1960. The predictionof large volumesof pyroclasticsis also Booth, B., and S. Self, Rheologicalfeatures of the 1971 Mount Etna testable.If we assumethat the productsof the lava fountains lavas,Philos. Trans.R. Soc.London, Ser. A, 274, 99-106, 1973. Butler, B. J., 3.5-cm radar investigationof Mars and Mercury: on Marsare similarto thoseof highbasaltic lava fountains on Planetologicalimplications, Ph.D. thesis,Calif. Inst. of Techno!., the Earth,we estimatethat 10%of the massof the erupted Pasadena,1994. lavawill bein theform of pyroclastics[Thordarson and Self, Cattermole,P., Volcanic flow developmenton Alba Patera, Mars, 1993].The pyroclasts are expected to includea widerange of Icarus, 83, 453-493, 1990. Crisp, J., and S. Baloga, A model for lava flows with two thermal materialsincluding larger droplets(Pele's tears), open components,J. Geophys.Res., 95, 1255-1270, 1990. frameworkfoam (reticulite), and fine ash [e.g., Mangan and Crisp, J., and S. M. Baloga, Influence of crystallization and Cashman,1996]. Thesematerials are typically 10-100 times entrainmentof coolermaterial on the emplacementof basalticaa lessdense than the lava flows. This suggeststhat the total flows,J. Geophys.Res., 99, 11,819-11,831,1994. volumeof pyroclasticsshould be approximately1-10 times Davies, M. œ., R. M. Batson, and S.S. C. Wu, Geodesyand Cartography,in Mars, editedby H. H. Kieffer, B. M. Jakosky,C. thevolume of thelava, implying about 106-107 km 3 of recent W. Snyder,M. S. Matthews,pp. 321-342, Universityof Arizona, pyroclasticdeposits. Tucson, Ariz., 1992. It is likely that much of this pyroclasticmaterial is Edgett,K. S., and J. W. Rice, Very youngvolcanic, lacustrine, and preservedon the surface of Mars. The denser fraction, fluvial featuresof the Cerberusand Elysium Basin region, Mars: Whereto sendthe 1999 Mars SurveyorLander, Lunar Planet.Sci. includingthe Pele'stears and piecesof spatter,should be Conf Abstr., 26, 357-358, 1995. concentratednearest to the Elysium and Amazonis Planitia Edgett, K. S., B. J. Butler, J. R. Zimbelman, and V. E. Hamilton, regions,while the extremelyporous (>>99% void space) Geologic context of the Mars radar "Stealth" region in reticulitewould have traveled farther, perhaps reaching the southwesternTharsis, d. Geophys.Res., 102, 21,545-21,567, baseof the Tharsisvolcanoes. The fine-grainedash might 1997. Einarsson,T., G. M. Gudbergsson,G. A. Gunnlaugsson,S. Rafnsson, havebeen transported globally. and S. Thorarinsson(Eds.) Skafidrledar 1783-1784: Ritgerdir og The MedusaeFossae Formation (MFF) is characterizedby Heimildir,279 pp.,Mill of Menning,Reykjavik, 1984. a weaklylithified material that is readilycarved by the wind Fink, J. H., and J. R. Zimbelman,Rheology of the 1983 Royal [Ward, 1979;Scott and Tanaka,1982, 1986]. The MFF is Gardensbasalt flows, Kilauea , Hawaii, Bull. Volcanol.. mostlylocated between the radar bright freshflows of the 48, 87-96, 1986. Frey, F., et al., KerguelenPlateau-: A large igneous CerberusPlains and the "stealth"region [Butler, 1994; Edgett provinceSites 1135-1142, Proc. OceanDrilling ProgramInitial et al., 1997](see Figure 1). A smalldegree of weldingor Reports,[CD-ROM], Volume183, 2000. alterationof basalticpyroclastic deposits may providethe Glaze, L. S., and S. M. Baloga,Dimensions of Pu'u 'O'o lava flows appropriatedegree of lithificationfor the MFF. on Mars,J. Geophys.Res., 103, 13,659-13,666,1998. Goncharov,V. N., Dynamicsof ChannelFlow, 317 pp., translated It is intriguingthat our estimated volume of pyroclasticsis from Russianby Israel ProgramSci. Transl., U.S. Dep. of similar to the estimated volume of the Medusae Fossae Commer.,Off. of Tech.Serv., , D.C., 1964. Formation [Sakirnotoet al., 1999]. The MFF partially Greeley,R., and J. S. King, Volcanismof the EasternSnake River overlapsthe stealthregion, but the stealthregion is generally Plain, Idaho:A ComparativePlanetary Geology Guidebook, 308 farther to the east [Muhlernanet al., 1991;Butler, 1994; pp., NASA, Washington,D.C., 1977. Greeley,R., and P. D. Spudis,, Rev. Geophys., Harmonet al., 1999].The stealth region is of theorder of 106 19, 13-41, 1981. km2 and has a radar-absorbing cover at leastseveral meters Greeley, R., S. A. Fagents,N. A. Bridges,D. A. Crown, and L. S. thick [Butler, 1994]. Extremelyporous reticulite is a prime Crumpier,Volcanism on the red planet:Mars, in Environmental candidatefor this surfacecover, but in situ samplingwill Effectson VolcanicEruptions: From Deep to DeepSpace, editedby J. R. Zimbelmanand T. K. P. Gregg,Kluwer Acad., probablybe necessaryto confirmthis hypothesis[Butler, Norwell, Mass.,in press,2000. 1994;Edgett et al., 1997]. Harmon, J. K., M.P. Sulzer, P. J. Perillat, and J. F. Chandler, Mars 15,048 KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS

radar mapping:Strong backscatter from the Elysium Basin and McEwen, A. S., M. Malin, L. Keszthelyi, P. Lanagan,R. Beyer, W. outflow channel,Icarus, 95, 153-156, 1992. Hartmann,Recent and ancientflood lavason Mars (abstract),Lun. Harmon,J. K., R. E. Arvidson,E. A. Guinness,B. A. ,and Planet.Sci. Conf., [CD-ROM], 30, abstract1829, 1999b. M. A. Slade, Mars mapping with delay-Doppler radar, d. McEwen, A., R. Beyer, D. Burr, L. Keszthelyi, and P. Lanagan, Geophys.Res., 104, 14,065-14,089,1999. Well-preservedvolcanic and fluvial featureson Mars (abstract), Hartmann,W. K., Martian cratering:VI, Cratercount isochrons and Eos Trans.AGU, 80 (46), Fall Meet. Suppl.,609-610, 1999c. evidence for recent volcanism from Mars Global Surveyor, Mouginis-Mark,P. J., L. Wilson, J. W. Head, S. H. Brown,J. L. Meteorit. Planet. Sci., 341, 167-177, 1999. Hall, and K. D. Sullivan, Elysium Planitia, Mars: Regional Hartmann,W. K., and D.C. Berman,Elysium Planitia lava flows: geology, volcanology,and evidence for volcano-groundice Cratercount chronology and geologicalimplications, J. Geophys. interactions,Earth Moon Planets, 30, 149-173, 1984. Res., this issue. Muhleman, D. O., B. J. Butler, A. W. Grossman,and M. A. Slade, Hartmann, W. K, M Malin, A. McEwen, M. Carr, L. Soderblom,P. Radarimages of Mars, Science,253, 1508-1513,1991. Thomas, E. Danielson, P. James, and J. Veverka, Evidence for Oskarsson,N., G. E. Sigvaldason,and S. Steinthorsson,A dynamic recent volcanism on Mars from crater counts,Nature, 397, 586- modelof rift zone petrogenesisand regionalpetrology of Iceland, 589, 1999. J. Petrol., 23, 28-74, 1982. Hill, R. E. T., S. J. Barnes,M. J. Gole, and S. E. Dowling, The Peterson,D. W., and R. I. Tilling, Transitionof basalticlava from volcanologyof komatiitesas deducedfrom field relationshipsin pahoehoeto aa, KilaueaVolcano, Hawaii: Field observationsand the Norseman-Wilunagreenstone belt, WesternAustralia, Lithos, key factors.J. Volcanol.Geotherm. Res., 7, 271-293, 1980. 34, 159-188, 1995. Pieri, D.C., and S. M. Baloga, Eruption rate, area, and length Hon, K., J. Kauahikaua,R. Denlinger,and K. Mackay,Emplacement relationships for some Hawaiian lava flows, J. Volcanol. and inflation of pahoehoe sheet flows: Observations and Geotherm. Res., 7, 271-293, 1986. measurementsof active lava flows on Kilauea Volcano, Hawaii, Plescia, J. B., Recent flood lavas in the Elysium region of Mars, Geol. Soc. Am. Bull., 106, 351-370, 1994. Icarus, 88, 465-490, 1990. Hooper,P. R., The ColumbiaRiver basalts,Science, 215, 1463-1468, Reidel, S. P., Emplacementof Columbia River flood basalt,J. 1982. Geophys.Res., 103, 27,393-27,410,1998. Hulme, G., Interpretationof lava flow morphology,Geophys. d. R. Rowland, S. K., and G. P. L. Walker, Pahoehoe and aa in Hawaii: Astron. Soc., 39, 361-383, 1974. Volumetric flow rate controlsthe lava structure,Bull. Volcanol., Jakobsson,S. P., Petrologyof recentbasalts of the EasternVolcanic 52, 615-628, 1990. Zone, Iceland, Acta Naturalia Islandica, 2, 1-103, 1979. Sakimoto,S. E. H., and M. T. Zuber, Flow and convectivecooling in Jerram, D. A., H. Stollhfen, V. Lorenz, S.C. Milner, and A. R. lavatubes, J. Geophys.Res., ! 03, 27,465-27,487,1998. Duncan, Plume related pahoehoe lava flows of the basal Sakimoto, S. E. H., H. V. Frey, J. B. Garvin, and J. H. Roark, Entendeka flood basalts,NW Namibia, Int. Volcanol. Congr. Topography,roughness, layering, and slope propertiesof the MeetingAbstr., p. 28, 1998. Medusae Fossae Formation from Mars Orbiter Laser Altimeter Keszthelyi,L., A preliminarythermal budget for lava tubeson the (MOLA) andMars OrbiterCamera (MOC) data,J. Geophys.Res., Earthand planets, ,/. Geophys.Res. 100, 20,411-20,420,1995. 104, 24,141-24,154, 1999. Keszthelyi, L., and R. Denlinger, The initial cooling of pahoehoe Scott, D. H., and M. G. Chapman, Mars Elysium Basin: lava flow lobes,Bull. Volcanol., 58, 5-18, 1996. Geologic/volumetricanalysis of a young lake and exobiologic Keszthelyi, L. P., and D.C. Pieri, Emplacementof the 75-km-long implications,Proc. Lun. Planet.Sci. Conf 21st,669-677, 1991. Carrizozo lava flow field, south-centralNew Mexico, J. Volcanol. Scott, D. H., and K. L. Tanaka, Ignimbritesof AmazonisPlanitia Geotherm.Res., 59, 59-75, 1993. regionof Mars,J. Geophys.Res., 87, 1179-1190,1982. Keszthelyi,L., and S. Self, Why are therelong channelized flows on Scott, D. H., and K. L. Tanaka, Geologic map of the western and Mars, but not on the Earth? (abstract),Lun. Planet. equatorialregion of Mars, scale 1:15,000,000, Map 1-1802-A, Sci. Conf. [CD-ROM], 29, abstract1529, 1998a. U.S. Geol. Surv., Reston, Va., 1986. Keszthelyi, L., and S. Self, Some physicalrequirements for the Self, S., T. Thordarson, and L. Keszthelyi, Emplacement of emplacementof long basalticlava flows, ,/. Geophys.Res., 103, continentalflood basaltlava flows, in Large IgneousProvinces, 27,447-27,464, 1998b. editedby J..J. Mahoneyand M. Coffin, Geophys.Monogr. Set., Keszthelyi, L., S. Self, and T. Thordarson,Application of recent vol. 100,pp. 381-410,AGU, Washington,D.C., 1997. studieson the emplacementof basalticlava flows to the Deccan Self, S., L. Keszthelyi, and T. Thordarson,The importanceof Traps,Mem. Geol. Soc., 43, 485-520, 1999. pahoehoe,Annu. Rev. Earth Planet.Sci., 26, 81-110, 1998. Kilburn, C. R. J., and R. M. C. Lopes,General patterns of flow field Shaw,H., and D. Swanson,Eruption and flow ratesof flood basalts, growth: Aa and blocky lavas, J. Geophys.Res., 96, 19,721- in Proceedingsof the SecondColumbia River BasaltSymposium, 19,732, 1991. editedby E. Gilmour and D. Stradling,pp. 271-299, East. Wash. Macdonald, G. A., Pahoehoe, aa, and block lava, Am. J. Sci., 251, StateCollege Press, Cheney, 1970. 169-191, 1953. Smith,D. E., et al., The globaltopography of Mars and implications Macdougall,J. D., ContinentalFlood Basalts, 341 pp., KluwerAcad., for surfaceevolution. Science,284, 1495-1503, 1999. Norwell, Mass., 1988. Spence,D. A., P. W. Sharp, and D. L. Turcotte, Buoyancy-driven Malin, M. C., et al., Early views of the Martian surfacefrom the crack propagation:A mechanismfor magmamigration, d. Fluid Mars Orbiter Camera on Mars Global Surveyor, Science,279, Mech., 174, 135-153, 1987. 1681-1692, 1998. Steingrimsson,J., Fires of the Earth: The Laki Eruption1783-1784, Mangan, M. T., and K. V. Cashman,The structureof basalticscoria translated by K. Kunz, pp. 1-95, Univ. of Iceland Press, and reticulite and inferencesfor vesiculation,foam formation, and Reykjavik, 1998. fragmentationin lava fountains,J. Volcanol.Geotherm. Res., 73, Stephenson,P. J., A. T. Burch-Johnston,D. Stanton,and P. W. 1-18, 1996. Whitehead, Three long lava flows in north Queensland,J. Martin, B. S., The Roza Member, Columbia River Basalt Group: Geophys.Res., 103, 27,359-27,382, 1998. Chemical stratigraphyand flow distribution,in Volcanismand Swanson,D., T. Wright, and R. Helz, Linear vent systemsand Tectonism in the Columbia River Flood-Basalt Province, edited estimatedrates of magmaproduction and eruption for the Yakima by S. P. Reidel and P. R. Hooper,Spec. Pap. Geol. Soc.Am., 239, basalton the ,Am. J. Sci.. 275, 877-905, 1975. 85-104, 1989. Tanaka,K. L., and D. H. Scott,Channels of Mars' Elysiumregion: McEwen, A. S., K. S. Edgett, M. C. Malin, L. Keszthelyi,and P. Extent,sources, and stratigraphy,NASA Tech.Memo, 88383, 403- Lanagan,Mars Global Surveyorcamera tests the ElysiumBasin 405, 1986. controversy:It's lava not lake sediments,Geol. Soc. Amer. Abstr. Theilig, E., and R. Greeley, Lava flows on Mars: Analysisof small Programs,abstract 50268, 1998. surfacefeatures and comparisonswith terrestrialanalogs, Proc. McEwen, A. S., M. C. Malin, M. H. Carr, and W. K. Hartmann, Lunar Planet.Sci. Conf. 17th,Part 1, d. Geophys.Res., 91, suppl., Voluminous volcanism on early Mars revealed in Vailes E 193-E206, 1986. Marineris, Nature, 397, 584-586, 1999a. ThorarinssonS., On the damagecaused by volcaniceruptions with KESZTHELYI ET AL.: MARTIAN FLOOD LAVAS 15,049

special referenceto tephra and gases,in VolcanicActivity and Walker, G. P. L., Lengths of lava flows, Philos. Tran., R. Soc. Human Geology,edited by P. D. Sheetsand D. K. Grayson,pp. London, $er./t, 274, 107-118, 1973. 125-159,Academic, San Diego, Calif., 1979. Ward, A. W., Yardangson Mars: Evidenceof recentwind erosion,J. Thordarson,T., The eruptionsequence and eruptionbehavior of the Geophys.Res., 84, 8147-8166, 1979. Laki 1783-1785 eruption,Iceland: Characteristicsand distribution Wilson, L., and J. W. Head, A comparisonof volcanic eruption of eruption products,M.S. thesis, 239 pp., Univ. of Texas at processeson Earth, Moon, Mars, , and Venus,Nature, 302, 663- Arlington, 1990. 669, 1983. Thordarson,T., Volatile releaseand atmosphericeffects of basaltic Wilson, L., and J. W. Head, Mars: Review and analysisof volcanic fissureeruptions, Ph.D. thesis,446 pp., Univ. of Hawaii at Manoa, eruption theory and relationshipsto observedlandforms, Rev. Honolulu, 1995. Geophys.,32, 221-264, 1994. Thordarson,T., and S. Self, The Laki (SkaftfirFires) and Grimsv6tn Woods, A. W., A model of the plumes above basaltic fissure eruptionsin 1783-1785,Bull. Volcanol.,55, 233-263, 1993. eruptions,Geophys. Res. Lett., 20, 1115-1118,1993. Thordarson, T., and S. Self, The Roza Member, Columbia River Zielinski, G. A., P. A. Mayewski,L. D. Meeker, S. Whitlow, M. S. Basalt Group: A gigantic pahoehoelava flow field formed by Twickler, M. Morrison, D. A. Meese, A. J. Gow, and R. B. Alley, endogenousprocesses?, J. Geophys.Res., 103, 27,411-27,445, Record of volcanism since 7000-BC from the GISP2 1998. ice coreand implicationsfor the volcano-climatesystem, Science, Thordarson,T., S. Self,N. (Sskarsson,and T. Hulsebosch,Sulfur, 264, 948-952, 1994. chlorine,and fluorinedegassing and atmosphericloading by the 1783-84 AD Laki (Skaffftr Fires) eruption in Iceland, Bull. Volcanol., 58, 205-225, 1996. L. Keszthelyiand A. S. McEwen, PlanetaryImage Research Tolan, T. L., S. P. Reidel, M. H. Beeson,J. L. Anderson,K. R. Fecht, Laboratory,Lunar and PlanetaryLaboratory, University of Arizona, and D. A. Swanson, Revisions to the estimatesof the areal extent 1629E. UniversityBoulevard, Tucson, AZ 85721-0092. and volumeof the ColumbiaRiver BasaltGroup, in Volcanism ([email protected]) and Tectonism in the Columbia River Flood-Basalt Province, Th. Thordarson,CSIRO, Divisionof Explorationand Mining, edited by S. P. Reidel and P. R. Hooper,Spec. Pap. Geol. $oc. PrivateMail Bag,P.O. Wembley,WA 6014, Australia. Am., 239, 1-20, 1989. Wadge, G., The variation of discharge during basaltic (ReceivedSeptember 28, 1999;revised February 14, 2000; eruptions,J. Volcanol.Geotherm. Res., 11, 139-168, 1981. acceptedFebruary 29, 2000.)