JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 97, NO. B13, PAGES 19,729-19,737, DECEMBER 10, 1992 SolidificationandMorphologyof SubmarineLavas' A Dependenceon ExtrusionRate Ross W. GRIFFrms ResearchSchoolof Earth Sciences, AustralianNational University,Canberra JONATHAN H. Fn, rK GeologyDepartment,ArizonaStateUniversity,Tempe Theresultsof recentlaboratoryexperiments withwax extrudedbeneathrelativelycoldwatermaybe extrapolated to predictthesurfacemorphology of submarine lavasasa functionof the extrusion rateandmelt viscosity.The experiments withsolidifying waxindicated thatthesurface morphology wascontrolled byasingleparameter, theratio of thetimetakenfor thesurfaceto solidify,anda timescalefor lateralflow. For submarine basaltsa solutionof the coolingproblem(whichisdominated byconduction in thelavabutconvective heattransferin thewater)andestimates of lava viscosities placethisparameterwithintheempiricallydetermined "pillowing"regimeovera widerangeof extrusionrates.This resultis consistent with the observation thatpillow basaltsare the mostcommonproductsof submarine eruptions. Smoothersurfaces corresponding to thevarioustypesof submarine sheetflowsarepredicted for sufficientlyrapidextrusions of basalticmagma.Stillhighereruptionratesin regionsof low topographic relief may producesubmarine lavalakes.Minimumemplacement timescanbecalculated for submarine volcanicconstructs of a singlelava flow type. INTRODUCTION Submarine basalt flows are the most common volcanic rocks on solidifyingcruston the advanceof lava flowsanddomes,submitted to Journalof Fluid Mechanics,1992].Thusthesurfacecoolingand development of thesolidcrustis a criticalprocess in determiningthe surfacemorphologyof lavasand,in somecases,may leadto crustal earth.However, becausethe eruptionof submarinelavashasnever beendirectlyobserved,ourabilityto inferemplacement conditions control of the flow's advance. from their morphologyis quitelimited.Despiterecentconcerted In a seriesof laboratoryexperiments withpolyethyleneglycolwax effortsto catchdeep-seaeruptionsin action[e.g.,Chadwicket al., extruded onto the base of a tank filled with cold water, Fink and 1991;Haytnoneta/., 1991],the likelihoodof beingableto make Griffiths [ 1990, 1992] showedthat formationof solidcrustled to a accurate measurements withpresentlyavailabletechnology issmall. numberof typesof surfacemorphologyanalogousto thoseseenon This situationcontrastswith that for subaerialflows, whosemorlava flows (Figure 1). Theseincludedleveesaroundflow margins, phologyhasbeenrelatedquantitatively to bothlavarheologyand surfacefolds similar in appearanceto thosefoundon ropy basalts, effusionratein severalstudies[e.g.,Hulme,1974;FinkandFletcher, striatedsurfacesseparatinglargesmoothplatesalongrifting zones, 1978; Borgia et al., 1983; Baloga and Pieri, 1986; Rowlandand andsmallbulbousstructures similartopillow basalts(Figure2). Fink Walker, 1990]. One way to overcomethis limitationto our underand Griffiths also found that the type of surfacemorphologydomistandingis to usetheoreticaland laboratorymodelsto relatelava nantin eachexperimentdependedin a systematicmanneron therate morphology to effusionrate andrheology.Thisis theapproach of of extrusionandtherateof formationof crust:themostrapidcooling thispaper. and solidificationled to "pillow" morphology,whereas"rifting," When lava is extrudedontothe Earth's surfaceor the seafloor,it "folding,"and"levees"occurredfor progressively slowersolidificaspreadslaterallyunderthe influenceof gravity.It alsobeginsto tion ratesor more rapid extrusionrates.At the lowest coolingrates, solidify as a result of radiative or convectiveheat loss from its no crustat all formed within the durationof an experiment. surface. By observation weknowthatathinlayeratthesurface ofthe A theoreticalpredictionof the kind of surfacemorphologyto be lava can become a solid crust within minutes in the case of subaerial expectedunderprescribedlaboratoryor field conditions is currently eruptionsor within a few secondswhenthelava is underwater.The beyondour capabilities;it wouldrequirecompletesolutionsfor the timefor formationof thefirstsolidis determined by theeruption coupledtemperatureand flow fields taking into accounttemperatemperatureand rate of heat loss from the surface,and thesesame ture-dependentrheology,solidification,and the mechanicsof the parameters determinethesubsequent rateof thickeningof thesolid crust.Various attemptshave been made to predict aspectsof the crust.The propertiesof the crust,in turn,controlits deformationin overallspreading of lavaflows[Walker,1973;Malin, 1980;Huppert response to stresses appliedby the motionof the underlyingmelt eta/., 1982; Blake, 1990], some consideringonly the thermal [e.g.,Fink and Fletcher,1978].Undersomeconditions (namely problemin whichcoolingandsolidificationthroughoutmuchof the rapidsolidification, smallextrusion rates,or highmeltviscosities) depth of the lava column is assumedto bring the flow to a halt thecrustcanalsoexercisesomedegreeof controloverthespeedof [Pinkerton,1987; Crisp and Baloga, 1990]. However, thesepapers advanceof the flow front [R. W. Griffiths andJ. H. Fink, Effectsof have not addressedthe dynamicaleffectsof the cruston the overall flow nor the relationshipbetweendeformationof the crustand the Copyright1992by the AmericanGeophysical Union. underlyingflow. In conjunctionwith theirlaboratoryexperiments, F inkandGriffiths Papernumber92JB01594. 0148-0227/92/92JB-01594505.00 [1990] employedan alternativeapproach.On the basisof a dimen19,729 19,730 GRIFFrrHS ANDFINK:SUBMARINE LAVAMORPHOLOGY flow would have in the absenceof solid crust.In the experimentsthe flowof wax withoutcrustwouldhavebeengoverned by a balance betweengravity and viscousstresses. Equivalently,• is the lateral distancefrom the vent,measuredin flow depths,at which solidcrust is predictedto appear.Evaluationof this parameterfor each wax experimentusing a theoreticalpredictionfor the surfacecooling, alongwith the observationsof morphology,enabledempiricalvaluesof ß to be found at the variousmorphologicaltransitions:when •>50, the wax flow reached the walls of the box without crust appearing; at25<•<50, solidformedneartheflowfromandcreated levees which had to be broken or overridden before the flow could advance;at 10<•<25 therewere cross-stream-oriented folds similar Fig.la. Transverse folding onflowfroma pointsource (widthoftankis 30 • ..... :- .; ........ . .... . ......... >½;•.• ........ ...:..•...: .............. .,..,,•j•-;;:; .....;. ....::•:•. •..;,. •. tothoseonropypahoehoeflows;at3<•<10 thewaxcrustdeveloped platesandrift-like fractures,withmoltenwaxexposedalongtherifts ultimately freezing and addingto the rigid surfaceplatesthat were movingapart;and at W<3, crustdevelopedrapidlyat the vent and spreadingproceededthroughlargenumbersof bulbousoutgrowths, or "pillows." The transitionsbetweendominantmorphologicaltypesoccurred atapproximately thesamevaluesof •Pforbothradialspreading from a smallhole[Fink andGriffiths,1990]andtwo-directional spreading from a narrow slit [Fink and Gri•ths, 1992] (with spreadingrate appropriately scaledfor eachcase).In mostrespects themorphology was also independentof the roughnessand slope of the base, althoughfor a longnarrowslit(a line source)the"pillow"regimewas observedonlywith a roughbase.With a smoothbase,"pillows"were replacedin linear (two-directional)spreadingby an extensionof the "rifting" regime, presumablybecausethe crust did not grip the smoothbasenearthe flow front andthe two largeplatescouldslide away on eachsideof the line sourcewithoutdecelerating.A gently slopingbase(uptosixdegrees)gaveapreferreddirectionof flow, but the dominanttype of surfacemorphologywasotherwiselittle affected.Thusthelaboratoryexperimentsshoweda reasonablyrobust sequenceof transitionsbetweensurfacemorphologies, transitions whicharedescribedby valuesof a singleparameterrelatingcooling rateto spreadingrate.The dependence of morphologyon theparameter •P alone(at least for a givenmaterial)showsthat the surface morphology,and thekind of behavioroccurringnearthe flow front, Fig. lb. "•ng" or Cate-lib spr½•ingon an ½x•ion •om a •n½source can be described in terms of the sourceconditions, extrusion rate, and wh•e •sifion is in•c• by •ows (t•k is 20 cmfromtopto surfaceheat flux, with some influenceof basaltopography. The laboratory observationsof surfacemorphologyprovide a .................... ?:;• •:-•?• ;;•:•,• ...... ::...::,.. .•.• .......... :.•:•...: ..• ............ .......... ½:.L:•..-...•.:.•.•:;:•:%. .............. s•:..:.s ........... •?•::•:::,•½;::•. -:•:;::½:½•--•:,.•:.•. .;•,•. .....';.•-•---•------------•--•...•..,•:..," -.-........•..: ....... ,.:... .... --------------•• .......•..........•.,:•: .....•--<• Fig.lc.Sideviewofspr½•i•gfroma•½ soure•rough"½low"formation ½-cm-•ck flow movedtoward•e camera). Fig. 1. •oto•aphs of t•½½mo•hologic ½n½ glycolwax uMcr coldwater,from•½ [•90, •9•]. potentially powerful tool for interpretationof the eraplacement conditionsfor lava flows. The experimentshave been carriedout with convectivecoolingof wax beneathcold water, and the results shouldbe of direct relevanceto submarinelava eruptions.In this paperwe evaluatetheparameter•Pandmaketentativepredictions of flowmorphologyfor submarinebasalticlavas.Thesepredictions are functionsof the volumeflux, eruptiontemperature,andlava viscos- ityatthe vent. However, ifweconsider only mid-ocean ridge basalts, therangeof venttemperatures andviscosities is likelyto bequite limited,andtheextrusionrateis left asthemaindeterminant of flow morphology. Thisopensthepossibility thatobserved morphology maybeusedtoplaceconstraints onvariations of extrusion ratealong a ridge. sionalanalysisandanassumption of dynamicsimilarity,theyargued SPREADING OF VISCOUS LAVA thatthesurfacemorphology depends primarilyontherateof formation of crust relative to the rate of lateral flow. The behavior can be characterizedfor eachexperimentby a singledimensionless parameter, •, which they definedas the ratio of the time takenfor the surfaceof the wax to solidify and a time scale for horizontal advection,the latterbeinga measurementof thetime requiredfor an elementof extrudedwax to travel a distanceequal to the depththe The depthandlateralvelocityof lava neara vent,with or withore cooling,are determinedby the sourcevolumeflux Q, sourcegeometry, lava theology,gravitationalforce,andbasetopography.Cooling as a result of heat loss from the surfaceor baseof the lava influencesthespreadingby alteringtherheologyof partsof theflow. If thereis noverticaladvectivetransportof heatwithinthelava (i.e., GRIFF1THS ANDFINK:SUBI•IARINE LAVAMOm'HOLOO¾ Fig.2a. ALVIN viewof basaltpillowswith spreading cracks(iqdividualpillowsapproximately 1m across)fromtheEastPacificRise. Photocourtesyof RobertEmbley,NOAA. Fig.2b. Archaean pillowsfromsouthern Ontarioshowing darkglassyrinds.Pennearbottomof largestpillowfor scale.Photocourtesy of StephenJ. Reynolds,ArizonaStateUniversity. 19,731 19,732 GmmT•S ANDFINK:SUBMARINE L^v^ MORPnOLOG¾ no thermal convectionor shear-generated turbulence),coolingis confined to thin conductive boundary layers: a thin solid crust develops.In this paperwe consideronly laminar(or near-laminar) flows: thosehaving viscositiessmall enoughand velocitieslarge enoughfor a Reynoldsnumberto be greaterthan about100 are generallyunstableor turbulent, and conductionis not the sole mechanismfor heat transportto their surfaces.Becausesolidification commencesmuchmore slowly at the basethanat the exposed surfaceof a lava,we alsoneglectanyeffectsof basalcoolingonthe overallflow or on developmentof the surfacecrustanditsmorphology. Thermalconvectionin the lava cannotcommence withinthe extremelyshorttimesrequiredfor a solid surfaceskinto form in SURFACECOOLING AND SOLIDIFICATION SurfaceFluxes An analysisof the coolingof lava, assumingthat verticalheat transferwithin the lava is by conductionandthatthe surfaceflux is determinedby turbulentnaturalconvectionin the environment,is summarizedby Fink and Griffiths[1990]. Convectivetransportis expected tobethedominantmechanism forheatlossfromsubmarine eruptions.However,thereis alsoa smallradiativetransfer,andfor completeness we includeestimates of thisflux.Thusthesurfaceflux F is givenby thesumof theconvective flux Fcandtheradiativeflux Ft. The magma motion is assumed to be laminar (at leastnear its contact with water. surface),andwe showlaterthatthereis noneedto considerthermal Assumingthatthe isothermalmelt beneaththesurfaceboundary convectionwithin the magma. layercanbe satisfactorilydescribedby a uniformviscosity,therate of formationof surfacecrustcanbe comparedto therateof spreading that would prevail in the absenceof cooling. For a flow with Consideranelementof lavaextrudedat a timet=0 andtemperature T1undera deeplayer of seawaterat temperature Ta. At extremely small times after any given element of melt comes into contactwith Newtonianrheologyand a very smallReynoldsnumberextruded the seawater, a conductive thermal boundary layer buildsup in the ontoa horizontalplane,scalesfor thedepthH andvelocityU of the water adjacent to the lava surface until it becomes unstableandis flow [Fink and Griffiths,1990] are,for radialspreading froma point sweptaway by convection.Using a boundarylayer analysis(and source, assumingsingle-phaseflow) it is straightforward to showthat this H- (pvQ/gAp) TM, U- (gApQ/pv) •/2 (la) and, for a line (slit) source, H- (pvq/gAp)•r3, surface heatflux. However, attimesmuchgreater than10-2s,only U- (gApq2/pv) it3. (lb) The motionis drivenby the gravitationalaccelerationg actingon the densitydifferenceAp=p-pa, wherep is the densityof thelava and Pa is density of the environment.Flow is retardedby stresses associatedwith the kinematic viscosityv (v--q/p, where •1 is the dynamicviscosity).In (la), Q is the volumeflux from the vent, whereasin (1b), q is thevolumeflux perunitlengthof thelinesource. Constantsof proportionalityarenot shownherebutwill be of order one; their precise values are not needed,as the dimensionless parameterwhichwe will formusingtheabovescales mustin anycase be evaluatedempiricallyat the variousmorphological transitions, usinglaboratoryanalogexperimentsor field data. The time for advectionof lava througha distanceequalto one depthscale,ta=H/U, is foundfrom (1). Theratioof solidificationand advectivetime scales•=ts/ta (whichcanbe thoughtof astheinverse of a "solidification Peclet number," instead of the usual "heat flow • = (gAp/pv)3/4Q1/4ts a temporallysmoothed flux isrequired.Increased surfaceslopeand roughness shouldreduceboththeboundarylayertimescaleandthe magnitudeof the fluctuations in heatflux. The convectiveflux Fcestablished for convectionatlargeRayleigh numbersin water,with smalltemperaturedifferencesandnoboiling, is given by Fc= J(Tc-Ta)4/3, J = paCaT (gOtalCa2/Va) 1/3, (3) whereTc(t) is thelava temperatureat thecontactsurface,Ota,ra, Va, Pa, and Caare the thermal expansioncoefficient,diffusivity,kinematic viscosity,density,and specificheat(at constantpressure)of the convectingfluid, and), is a constantwhosevalueis closeto 0.1 [Turner,1973;HuppertandSparks,1988].Themolecularproperties of water are almostconstantin all existingtestsof (3). However,at hightemperatures, seawaterexhibitsa strongdependence of thermal expansioncoefficienton temperature[Bischoffand Rosenbauer, 1985]:for the seawaterat mid-oceanridges(3.2% to 3.5% salinity and100-300bar),otvariesfrom1.5xl0--4attheambient temperature (0ø-10øC) [Sverdrup, 1945]to 1.0x10 -2at350ø-420 ø(depending on Peclet number") then becomes Radialspreading boundary layertimescale isof order10-2s(taking seawater propertiesin Table 1). This is alsothe timescalefor the boundarylayer to reform. Boundarylayer intermittence resultsin fluctuationsof the (2a) pressure).At conditionsclose to the critical point of seawater (approximately 400øC,300 bar)theexpansion coefficientbecomes even larger. It is therefore important to make allowance for a nonlineardensity-temperature relation. In orderto estimateW for a givenflow it is necessary to evaluate Convectiveheattransferfrom submarinelavasiscontrolledby the the time tsrequiredfor the surfaceof thelavato coolfrom thevent dynamicsof a thin unstableboundarylayerin thewater(aswell as temperatureto thesolidificationtemperature. Thetimetsisusedhere conductionthrougha thin boundarylayer in the lava). As far aswe astheprimaryindicatorof thedevelopment of crustduringtheearly are aware, the effects of stronglynonlineardensity-temperature stagesof solidificationof flow from thosepositionswherenewmelt relationson this form of convectionhave not beenparameterized. is exposedto theenvironment.At muchgreatertimeswhenthe lava However,we arguethattheconvectiveheatflux dependson thenet surfaceis farther from the vent and its temperaturehasapproached buoyancyin the boundarylayer and thereforeon an "effective" the ambienttemperature,the crustthicknessbecomesinsensitiveto averagevalue for Ota,which might be definedsuchthat Ota(Tc-Ta)= thesurfacesolidificationtimeandisinsteadcontrolledlargelyby the &llo•Ot(T)(T-Ta)dz, where z=0atthecontact surface and15 isthe thicknessof thethermalboundarylayer.However,we proceedunder thicknessof the unstableboundarylayer. The bestjustifiable aptheassumption thatcrustmorphology,suchaspillowsandtransverse proximation for the effective Otais simply Ota=Ot(T*),where folds,is determinedby the early stagesof developmentof thecrust T*=(Tc+Ta)/2. For the early stagesof coolingconsideredin this and that the structureof any part of the flow is setlong beforethat paperwe will seethatT* variesfrom 350øCto 550øC,a rangeover portionof the surfacehastime to coolto the ambienttemperature. whichtheminimum expansion coefficient is Ota=3X10 -3,andthe Linesource • = (gAp/pv)2/3ql/3ts . (2b) GRIF2rlTHS ANDFINK:SUBMARINE L^v^ MOReHOt. Oa¾ 19,733 TABLE 1. Values of Parametersfor Seawaterand BasalticLavasUsedin the Computationof CoolingandTime Scalesfor LateralFlow of theLava Parameter Description Value Reference Seawaterat Depths1000-4000m Xla,Pas Va,m2s'l, Ca,Jkg'lK'1 r•a,m2s'1 Pt,kgm-3 dynamic viscosity kinematic viscosity specific heat thermal diffusivity density lx10-3 lx10-6 4.0x103 lx10-7 1.00xl03 otat ambientT, K-1 ova at -400øC,K-1 thermalexpansion 1.5x10 -4 -10-2 Ta, K temperature 280 [Fofonoff,1962] [Weastet al., 1989] [Fofonoff,1962] [Sverdrup,1945] [BischoffandRosenbauer,1985] [Diet•ch et al., 1980] xl,Pas 102-104 [Hall, 1987] v, m2s-1 0.04-4 Balsaltic [Dietrich et al., 1980] Lava c,Jkg-1K-1 1.2x103 • m2 s-1 5x10-7 p,kgm-3 2.6x103 Ti, K Ts, K • -1423 1003 0.9 emissivity [Turcotteand Schubert,1982] [WilliamsandMcBirney,1979] [Hall, 1987] (estimatefromMacdonald[1972]) [Ryanand Sammis,1981] value assumed The solidificationterrq•eratureTs is assumedto be the glasstransition. maximumisof order10-1,for whicha morerepresentative valueis tta=0.01[BischoffandRosenbauer,1985]. An estimateof the radiative flux Fr which will be sufficientfor our purposesis Fr = ec•(Tc 4- Ta4), (4) where c• is the Boltzmann constantand •; is the emissivityof the surface.Stefan'slaw (4) may tendto overestimatetheradiativeflux, where somefraction of the radiationat long infraredwavelengths will be absorbedwithin the thin conductivethermalboundarylayer in the water and thereforeleft to be carried away as a part of the convectiveflux Fc. However, most of the radiationpenetratesbe- The problemof findingthetemperaturein thelavacanbereduced, usingstandard techniques, tosolutionof anintegralequation[Carslaw and Jaeger, 1959, p. 76]. The equationcanbe transformed into .._ -1_•.•l0tF(3*) e-Z2/4•c(t-3') d3*' (5a) T(z,t) -T1 pc0c•)l/2 t•-3* wherez is thedistancefrom thecontactsurface,and3.is a dummy variable.In our casethe time-dependentsurfaceflux is F(t) = eo[Tc(t)n- Ta4] + J[Tc(t)- Ta]n/3, (5b) wherep, c, and•carethedensity,specificheat,andthermaldiffusivity ofthelava(values areshown inTable1),andTc=T(z=0). Thedepth yondtheboundary layer(athickness of-3xl0 -2mm)andaddstothe ofisothermscanbe tracedasfunctionsof time. The contacttemperatotal flux. ture is found by solvingwith z=0. Griffithsand Fink [1992] presentexplicitvaluesfor the surface The computeddepthsof severalisothermsare shownin Figure• fluxescalculatedfrom (3) and(4) andcomparethemwith thefluxes for thecaseof a basaltextrudedat 1423 K (1150øC)into seawaterat from lava surfacesin subaerialenvironments. Theyrangeup to 5000 280 K (7øC),assuming 0ta--0.01.The isothermat 1373K(1100øC)is kW m -2. chosenasanestimateof thesolidustemperaturefor basalts;1003+15 K (730øC)is thebestavailablevaluefor theglasstransition temperaThe ContactTemperature ture[RyanandSaramis,1981];and1200Kis simplyanintermediate temperature.When 1 s haselapsed,the solidusandglasstransition The idealizedcoolingproblemto be solvedis thatin whichmelt temperaturehave propagated2 mm and0.5 mm, respectively,into of uniform temperatureTl is suddenlybroughtinto contactwith the lava. After this time the depth of theseisothermscontinuesto overlyingwaterat temperature Ta.Conductionin thelavaismatched increase as t1/2.In Figure4 we plot the evolutionof thecontact to the total surfaceflux, which in turn dependson the surface temperaturefor the above extrusion. The influence of radiative temperatureandambientconditions.No allowancefor latentheatis transportissmall,contributingonlyapproximately 5 K tothesurfacd requiredover the small time scalesof interest,since the rapid coolingor decreasingby 4% the time takenfor Tc to reachthe glass quenchingby waterleadsto vitrificationof thelava surfacewith no transition. The resultsare not radically differentif 0tais variedby a crystallization (seebelow).A solutionfor the contacttemperature factor of 5. Tc(t),assuming onlyconvective transferfromthesurface,wasgiven in dimensionless form by Fink and Griffiths[ 1990].Here we calcuSolidificationTimes late explicitly both the contacttemperatureandthicknessof solidified crustunderconditionsappropriatefor submarineextrusionsof Our model is basedon the hypothesisthat flow morphologyis basalt.Given this limited range of conditions,the resultsare more largelydependenton therelativeratesof lateralspreadingandcrust accessiblewhen expressedin dimensionalform. formation.We have presentedevidence[Fink and Griffiths, 1990, 19,734 GRwFrms^NI)FzNs:: SUBM^RtS•. L^v^ Mo•P•IOI•OG¾ solidustemperature andbeforeit reachestheglasstransition. Work onthecrystallization kineticsofigneousrocksandlunarbasalts[e.g., Uhlmannet al., 1979;Dowry,1980;Kirkpatrick,1981]indicatesthat homogeneous nucleationandgrowthof crystalsto significantvol- umefractions requires minimum timesof order103-104 safterthe melt is cooledfrom its liquidus.Heterogeneous nucleationdue to preexistingcrystalsmay decreasethe time to reachlargecrystal fractionsbutnotby morethananorderof magnitude. Thussolidificationnear the quenchedsurfaceinvolvesonly the formationof 10.3 glass,a prediction thatis confirmed bytheglassysurfaces foundon freshsubmarine basalts[e.g.,BallardandMoore,1977;Chadwicket al., 1991]. Carryingthe calculations of isothermdepthsto greater timesthanshownin Figure3, we findthatlargevolumefractions of crystalsareunlikelytoformbeforethelavasolidifiestoa glass,until the solidificationfront hasmigratedto depthsof 50-100 mm (or at 0-4 aroundI hourafterthatpartof thesurfacewasexposedtothewater). This prediction,too, is consistent with the thickness (50-60 mm [MooreandLockwood,1978]) of observedglasslayersontheouter 10-5 0.01 0.001 0.1 surfaceof pillow basalts. 1 We conclude that the best estimate of the solidification time, rs, Time (s) appropriate for usein extrapolating laboratory observations of surto submarine extrusions of basaltis thetimetaken Fig. 3. Numericalsolutions for the depthof threeisotherms beneatha facemorphology horizontalsurfaceof lavainitiallyatuniformtemperature (T1=1423K) after for thelavasurfacetocoolto theglasstransition temperature. Foran it isbroughtintocontactwithseawater atTa=280K. Valuesusedforphysical extrusion temperature of 1423K, rs-0.07s.Thiscalculated solidifiquantifies aregivenin Table1. A slopeof 1/2is shownfor comparison. cationtime alsoappearsto be consistent with observations of the 1992] thatan importantmeasureof therateof crustdevelopment is thetime takenfor the contactsurfaceto beginto solidify.This time visible radiation from submarinelavas. The red glow from an elementof lavapersists for approximately 1-3s afterit isexposed to thewaterat theneckof a newpillow outgrowth[TepleyandMoore, canbe usedto evaluatethedimensionless parameter•, butwedonot 1974]. From independent measurements the last visiblered glow imply thatflow morphologyis in someway established at thetime fromlavacorresponds to a temperature of around900 K [Williams solidfirstbeginsto formin theearlystages of anextrusion. Timesat andMcBirney,1979].Althoughwepredictthesurface tocooltothis whichthe submarinelava surfacepassesthroughits solidusandits temperature within0.2s,thevisibleradiation isexpected topenetrate glasstransition temperature havebothbeendetermined, andcorrefrom the hot interior througha coolerglassycrustfor sometime. sponding solutions fordifferenteruption temperatures areplottedin Hence, the two observationsand the presentcooling model are Figure5. Thetimerequiredfor coolingto theglasstransition varies consistent if theglassis upto 1 mmthickbeforeit becomes opaque. from 0.14 s at T1=1600K to 0.04 s at Tl=1300 K. Over thisrangethe surface always coolsbelowthelavasolidus inlessthan10-2s. PREDICTED SUBMARINE MORPHOLOGY The calculatedcoolingtimesaremuchtoosmallfor anycrystallization to occurin the thin surfacelayer after it coolsbelow the Having an estimatefor the surfacesolidification time, ts, the parameter ß canbeevaluatedusingequations (2a) and(2b).Comparisonof these values of ß with thosefor the laboratorywax 1400 '• •• 1 0ø - solidus ß ß ß ß , .... , .... 1300 10'1 1200 transitio• glass 1100 glass 1000 10'2 N\ solidus. 900 i i ii•1111 , ßiIllIll I .......I ß• •kl ii 800 10'4 0.001 0.01 0.1 1 Time (s) Fig. 4. Solutionsfor thetemperature of thecontactsurfacebetweenlavaand seawaterfor the caseT1=1423K, Ta=280 K, and0ta=0.01with bothradiative and convectivetransport(solid curve),0ta=0.01and only convectiveflux (dottedcurve), and 0ta=0.002with bothradiationand convectionfrom the lava surface(dashedcurve). 10-3 10-4 1300 . . , t, , , I 1500 i ß ß 1600 Eruptiontemperature(K) Fig.5. Thetimeelapsed forthesubmarine lavasurface tocooltothesolidus andglasstransitiontemperature, asfunctions of theextrusion temperature. GRIFFITHS ANDFINK:SUBMARINE LAVAMORPHOLOGY 19,735 experiments may thenhelpusto understand thesurfacemorphology totalflux from thevent,sincebranchingof theflow will effectively of the submarine lavas. reducethelocalflow rate.Filmsandvideosof submarine portionsof In Figure 6 we have plotted the calculatedvalues of ß for Hawaiianbasaltflows thateruptedsubaerially[Tepleyand Moore, submarinebasaltsas a functionof the melt viscosity,q, and the 1974;Moore, 1975;Sansome, 1990]showthatpillowformationand extrusionrate,Q andq, for bothradialspreadingfrom a pointsource growthis associated withlocalflowratesof between 0.1 to2.0m3s-1. (Figure6a) andspreadingin twodirectionsfromaline source(Figure Theseestimatesfor viscosityand flow rate imply that W<3 is 6b), respectively.Estimatesof the viscositiesof basaltsfall in the probablefor mostsubmarineeruptions(Figure6). That is, extraporange102to 104Pas (103-105poises) [Macdonald, 1972;Hall, lation of the empirical morphologicalregimes for the wax analog 1987].Flowratesforindividualeruptions atmid-ocean ridgesarenot placessubmarinelava eruptionslargelyin the"pillow"regime.This wellknown butmaywellbeupto103m3s-1m-1(102km3d-1kin-l). result is consistentwith the observationthat pillow basaltsare By way of comparison,theextrusionof basalticandesitemagmain commonat bothoceanridgesandintraplatevolcanoes[e.g.,Moore, the 1991 eruptionof Hekla Volcanoin Iceland(a volcanicsystem 1975]. If the wax resultscan be carded over in sucha directmanner, attributedto a mantle hotspotat the mid-oceanridge) rangedfrom wealsopredict thateruptions withlavaviscosities lessthan103Pas 2000m3 s-1duringinitialfissure-fed activityto 1-12m3s-1during andflowramsgreater than10m3s-1mayformintobroadplate-like the following 2 months when lava issuedfrom a single crater sheetflows (the "rifting" regime)which wouldbe eithersmoothor coveredwithmillimeter-scale, flow-parallelstriations. At evenhigher (another hotspot volcano) pouredoutupto 780m3 s-1of tholeiitic flow ratesor lower viscosities,transverse foldsmay develop.Any basalt[LipmanandBanks,1987].Magmaflow ratesrelevantto the reductionof flow rate due to branchingof theselow-temperature formationof surfacemorphologymayoftenbemuchsmallerthanthe basalticflows will accentuatethetendencyto formpillows. Submarine lava lakes,describedin detailfrom sitesalongtheGalapagosRift [Gudmundsson et al., 1992]. The 1984 eruptionof Mauna Loa 1 02 ß " ß "' .... I " " ...... I ........ [ ........ I ........ I ' [Ballard et al., 1979] and East Pacific Rise [Lonsdale, 1977a; Ballard et al., 1981], and extensive flood basaltson the Hawaiian ' ''' arch[Holcombet al., 1988] may representflowsthat advancewith relatively little crustformation.These would be analogousto the "levee" or "no crust" regimesidentified in our experimentsand levees or no crust folding 1 01 wouklrequireextremelyhigheffusionrates(greater than2000m3s-1 m-1or3000m3s-1forlavawithaviscosity of 102Pasextruding from rifting W 11=10 2Pas 1 0ø pillows 10-1 ()•••• 1 '•"104 • / 10-2 10'3 . I I IIIIII I , . II1.11 I I ..,,.,1 , . , ,..,,1 10'1 I . , ,11,.1 ' ' , i . .. 101 Q (m3/s) 102 ........ i ........ i ß . ...... i ........ i ', .... i . • i ß ß levees or no crust folding 1 01 rifting 10ø rl=102Pas • riftingor pillows 10'1 104 10'2 -" .'•' ...... • 10 3 / / ........ b • ........ 10-1 ' ........ q (m2/s) • ........ 101 . . , 111 1o3 a linear or point sourcevent,respectively). If magmaissuesfromalongsubmarine fissure,theflow rate,q,per unit lengthof the fissureis smallerthanthetotal volumeflux, Q, and Figure 6b indicatesa shift to smallerW values,placingsubmarine basaltsstill farther into the "pillow" regime than indicatedby the point sourcemodel. On the other hand, the experimentswith line sourcesrevealedno "pillow" regime when a smoothbasewas used. Thusif the surfacesof previoussubmarineflows aresmoothenough sothatnew lobescan slideacrossthem,thenelongateventsat ridges may initiallyproducesheetflowsat theexpenseof pillows.As was concludedfor point sources,sufficientlyrapidsubmarine eruptions from longfissurescouldproduceflowswhosesurfacesaresmooth, apart from centimeter-scaletransversefolds and millimeter-scale striations.The folds may be confinedto a region near the flow marginsif extrusionfrom the ventwasrapidbutof shortduration. Subaerialdike-fedfissureeruptionsnormaflyare thermallyunstable,constrictingto one or a few point-sourceventswithin a few hours[Delaney and Pollard, 1981]. This tendencywould be enhancedfor eruptionsinto cold seawater.Thus the experimentsin whichwax extrudesfrom a fissureareprobablyonlyrelevantto the early stagesof most submarineeruptions. The resultsshownin Figure 6 can help explain distributionsof flow typesdescribed fromavarietyof submarine environments [e.g., Lonsdale, 1977a, b; van Andel and Ballard, 1979; Ballard et al., 1979, 1981;BonattiandHarrison,1988].Forexample,Bonattiand Harrison [1988] measuredthe areascoveredby sheetflows and pillowsonmapsfrommanydifferentmid-oceanridgesegments and founda positivecorrelationbetweenthe proportionof sheetflow coverageand the spreadingrate of the ridge.They suggested that sheetflow formationis favoredin lavasthathavehighereruption temperaturesand extrusionrates and lower crystallinitiesand viscosities.Althoughit seemslikely thatsomeoftheseconditions might be associatedwith fast spreadingridges,Bonatti and Harrison Fig. 6. Values of the dimensionless solidificationtime W for submarine basaltseruptedat T]=1423K asafunctionof extrusion rateandmeltviscosity for (a) radialspreading fromapointsourceand(b) spreading in twodirections from a line source.Morphologicregimesindicatedby thelaboratoryexperi- [1988] could not determine which would most influence flow mormentsof Fink and Griffiths [1990, 1992] are shown. phology. In contrast,the dimensionlessvariable, W, allows us to 19,736 Gram'ms ANDFINK:SUBMARINE LAVAMO•I•OI. OG¾ calculatethe relative importanceof thesefour factors.For example, if we assumethateruptiontemperatureandcrystallinitycontrollava viscosity(v), as describedby variousexperimentalstudies[e.g., Shaw, 1969; Pinkerton and Stevenson,1992], then from Fink and Griffiths[1990, equations(3) and (12)] we find thatthe distanceat which surfacesolidificationfirst occurs(and hencethe flow mor- phology) variesas(Q3/v)I/8.Thusvolumeflowrateshould havea Acknowledgments. We thankDerek CorriganandTony Beasleyfor constmctionof equipmentand assistance with the experiments and Robert EmbleyandWilliamChadwickforhelpfuldiscussions aboutsubmarine flow morphology. Thepaperwasimprovedsignificantly byreviewsfromWilliam Chadwickand an anonymous referee.Researchwassupported by NASA grantNAGW 529 from the PlanetaryGeologyandGeophysics Program, NationalScienceFoundation grantEAR 9018216,anda visitingfellowship from the AustralianNationalUniversity. strongerinfluenceonmorphologythancrystallinityorlavaviscosity. REFERENCES Some of the most detailedmaps of submarinelava flows yet compiledarefrom the summitregionof Axial Volcanoon theJuan Ballard,R.D., andJ.G.Moore,Photographic Atlasof theMid-AtlanticRidge Riff Valley, 114p., Springer-Verlag, New York, 1977. de FucaRidge[Embleyet al., 1990]. Thesemapsdistinguish pillows from lobate,ropy, andjumbled sheetflows. Althoughthereare no Ballard,R.D., R.T. Holcomb,and Tj.H. van Andel, The GalapagosRift at 86øW, 3, Sheetflows, collapsepits, andlava lakesof the rift valley,J. observationsof the latterthreetypesduringformation,their appearGeophys.Res.,84, 5407-5422, 1979. ancesuggests thattheymaycorrespond tothemorphologic sequence Ballard,R.D., J. Francheteau,T. Juteau,C. Rangan,andW. Normark,East seenin our experiments(with jumbled sheetflows equivalentto PacificRiseat 21øN:The volcanic,tectonic,andhydrothermal processes leveedflows).Thus,if we assume a lavaviscosity of 102Pas,Figure 6a lets us infer that the flow typesrequirethe following extrusion rates: pillows < 1 m3s-1 lobate sheet flows < 1 m 3 s-1 < 100m3s-1 3000m3 s-1 ( ropysheetflows < 3000m3s-1 ( jumbledsheetflows 100 m 3 s-1 of the centralaxis, Earth Planet. Sci. Lett., 55, 1-10, 1981. Baloga, S.M., and D. Pieri, Time-dependent profilesof lava flows. J. Geophys.Res.,91, 9543-9552, 1986. Bischoff, J.L., and R.J. Rosenbauer,An empirical equationof statefor hydrothermal seawater(3.2 percentNacl), AM J. Sci., 285, 725-763, 1985. Blake, S., Viscoplasticmodelsof lava domes,in Lava Flowsand Domes: Emplacement Mechanismsand Hazard Implications,IAVCEI Proc. Volcanol.,vol. 2, editedby J.H. Fink, pp. 88-128,SpringerVerlag,New Theseestimatesconstitutethefirstattemptstoestablish thephysical York, 1990. conditionsneededtoform specifictypesof submarine lavamorphol- Bonatti, E., and C.G.A. Harrison,Eruptionstylesof basaltin oceanic spreadingridgesandseamounts: Effectof magmatemperature andviscosogy. However, they are basedon a directextrapolationfrom wax ity, J. Geophys. Res., 93, 2967-2980, 1988. experiments,and their validity for lava, with its vastly different mechanicalpropertiesand generallymore complexemplacement Borgia,A., S.R. Linneman,D. Spencer,L.D. Morales,and J.A. Brenes, Dynamics of lavaflow fronts,ArenalVolcano,CostaRica,J. Volcanol. histories, must be further tested in future studiesthat include careful Geotherm•Res., 19, 303-329, 1983. examinationof lava flow structurefrom a variety of submarine Carslaw,H.S.,andJ.C.Jaeger, Conduction ofHeatinSolids, 2ndEd.,510pp., OxfordUniversityPress,New York, 1959. settings. Finally, for submarinevolcanicconstructs madeof a singlelava Chadwick,W.W., R.W. Embley, and C.G. Fox, Evidencefor volcanic eruptionon the southern Juande FucaRidgebetween1981and 1987, flow type, like pillow mounds,our modelingallowsus to calculate Nature, 350, 416-418, 1991. minimumemplacementtimes.For example,Figure6a indicatesthat Crisp,J.A., and S.M. Baloga,A modelfor lava flows with two thermal J. Geophys. Res.,95, 1255-1270,1990. a subcircular pillowmoundwitha volumeof 106m3andaviscosity components, Delaney,P.T., and D.D. Pollard,Deformationof hostrocksandflow of magmaduringgrowthof minettedikesandbreccia-bearing intrusions near days (it could take longer if effusion were not continuous).These ShipRock,New Mexico.U.S.Geol.Surv.Prof.Pap. 1202,61 pp., 1981. typesof estimatesare neededto developstrategiesfor observing Dietrich,G., K. Kalle, W. Krauss,andG. Siedler,GeneralOceanography, submarineeruptionswhoselocationsaredetectedby thepresence of 625 pp., JohnWiley, New York, 1980. hydrothermalmegaplumes[Embleyetal., 1991] or othersignsof Dowty, E., Crystalgrowthandnucleationtheoryandthe numericalsimulation of igneouscrystallization, in PhysicsofMagmaticProcesses, edited activity. of 102Pa s couldnothaveformedin lessthan106s, or roughly12 by R.B. Hargraves,pp. 419-485, PrincetonUniversityPress,Princeton, N.J., 1980. CONCLUSIONS A sequence of morphologictransitionsobservedon thesurfaceof a wax analogto lava flows can be comparedto the structureof submarinelavas by using a simple viscousgravity-drivenflow scaling,alongwith estimatesfor coolingof the lava surfacedueto turbulent natural convection in the seawater. The main task involved is a calculationof the time requiredfor the lava surfaceto solidify after it comesinto contactwith seawater.A directapplicationof the laboratoryresultsfor morphologictransitionsto submarineconditionspredictsthatbasaltsaremostlikely to form intopillows,while lobate, ropy, and jumbled sheet flows and lava lakes form from progressively higher flow rates and/or lower viscosities (crystallinities).Extrusionfrom a fissurealsofavorsdevelopment of pillows unlessthe substrateis smoothenoughto allow the new lava to slideoverit, in whichcasethevarioustypesof sheetflow will be favored.In mostcasessuchfissureeruptionswill evolveto oneor a few isolatedsubcircular ventsquiterapidly.Themorphologic similarities betweenlaboratorywax flows and submarinelavasare an encouragingindication that these types of simulationsmay help answerimportantquestionsaboutthe emplacementof lavason the sea floor. Embley,R.W., K.M. Murphy,andC.G. Fox,High-resolution studiesof the summitof Axial Volcano,J. Geophys.Res.,95, 12,785-12,812,1990. Embley,R.W.,W.W. Chadwick,M.R. Perfit,andE.T. Baker,Geologyofthe northernCleft segment,JuandeFucaRidge:Recentlavaflows,sea-floor spreading, andtheformation ofmegaplumes, Geology, 19,771-775,1991. Fink, J.H., andR.C. Fletcher,Ropypahoehoe: Surfacefoldingof a viscous fluid, J. Volcanol.Geotherm.Res.,4, 151-170, 1978. Fink,J.H., andR.W. Griffiths,Radialspreading of viscous-gravity currents withsolidifyingcrust,J. FluidMech.,221,485-509,1990. Fink, J.H., andR.W. Griffiths,A laboratoryanalogstudyof themorphology of lavaflowsextrudedfrompointandlinesources, J. Volcanol.Geotherm. Res.,in press,1992. Fofonoff,N.P., Physicalproperties of seawater,in TheSea,vol. 1,Physical Oceanography, editedby N.B. Hill, JohnWiley, New York, 1962. Grdfiths, R.W., and J.H. Fink, The morphologyof lava flows in planetary environments: Predictions from analogexperiments, J. Geophys. Res.,in press,1992. Gudmundsson, A., N. Oskarsson, K. Gronvold, K. Saemundsson,O. Sigurdsson, R. Stefansson, S.R.Gislason,P. Einarsson, B. Brandsdottir, G. Larsen,H. Johannesson, and T. Thordarson,The 1991 eruptionof Hekla, Iceland, Bull. Volcanol., 54, 238-246, 1992. Hall, A., IgneousPetrology,573 pp., LongmanScientificandTechnical, Harlow, England,1987. Haymon,R.M., andthe ADVENTURE Team,Activeeruptionseenon East PacificRise, Eos, Trans.A GU, 72 (46), 505-507, 1991. Holcomb,R.T., J.G.Moore,P.W. Lipman,andR.H. Belderson, Voluminous GRIFFITHS ANDFINK: SUBMARINE LAVAMORPHOLOOY submarinelava flows from Hawaiian volcanoes,Geology,16, 400-404, 1988. Hulme,G., The interpretation of lava flow morphology, Geophys. J. R. Astron. Soc., 39, 361-383, 1974 19,737 Sansome,F., Pele meetsthe sea, 27 minute video, LavaVideo Productions, Honolulu, Hawaii, 1990. Shaw,H.R., Rheologyofbasaltin themeltingrange,J. Petrol.,10, 510-535, 1969. Huppert,H.E., andR.S.J.Sparks,Meltingtheroof of a chambercontaining Sverdrup,H.U., Oceanographyfor Meteorologists,246 pp., Allen and Unwin, London, 1945. a hotturbulentlyconvecting fluid,J. FluidMech.,188, 107-131,1988. Huppert,H.E., J.B. Shepherd,H. Sigurdsson, and R.S.J.Sparks,On lava Tepley,L., andJ.G. Moore,Fire UndertheSea:The Originof Pillow Lava, domegrowth,with applicationto the 1979lavaextrusionof theSoufriere 16mm soundmotionpicture,MoonlightProductions, MountainView, of St Vincent, J. Volcanol.Geotherm.Res., 14, 199-222, 1982. Calif., 1974. Kirkpatrick,R.J.,Kineticsof crystallization of igneousrock,Rev.Mineral., Turcotte,D.L., andG. Schubert,Geodynamics, 450 pp.,JohnWiley, New 8, 321-398, 1981. Lipman, P.W., and N.G. Banks,Aa flow dynamics,Mauna Loa 1984, in Volcanismin Hawaii, editedby R.W. Decker,T.L. Wright, and P.H. Stauffer,pp. 1527-1567,U.S.Geol.Surv.Prof.Pap. 1350, 1987. Lonsdale,P.,Structuralgeomorphology of afast-spreading risecrest,theEast PacificRisenear3ø25'S,Mar. Geophys. Res.3, 251-293,1977a. Lonsdale, P.,Abyssalpahoehoe withlavacoilsattheGalapagos rift, Geology, 5, 147-152, 1977b. Macdonald,G.A., Volcanoes, 510pp.,Prentice-Hall,EnglewoodCliffs,N.J., 1972. Malin, M.C., Lengthsof Hawaiianlava flows, Geology,8, 306-308, 1980. Moore,J.G.,Mechanismof formationof pillowlava,Am.Sci.,63, 269-277, 1975. Moore,J.G.,andJ.P.Lockwood,Spreading cracksonpillowlava,J. Geol., York, 1982. Turner,J.S., BuoyancyEffectsin Fluids, 368 pp., CambridgeUniversity Press,New York, 1973. Uhlmann,D.R., P.I.K. Onorato,and G.W. Scherer,A simplifiedmodelfor glassformation,Proc. Lunar Planet. Sci Conf., lOth, 375-381, 1979. vanAndel,Tj.H., andR.D. Ballard,TheGalapagos Rift at 86øW,2, Volcanism,structure, andevolutionoftherift valley,J. Geophys. Res.,84, 53905406, 1979. Walker,G.P.L., Lengthsoflava flows,Philos.Trans.R. Soc.London,Ser.A, 274, 107-118, 1973. Weast,R.C., D.R. Lide, M.J. Astle,andW.H. Beyer(Eds.),Handbookof Chemistryand Physics,70th ed., CRC Press,BocaRaton,Fla., 1989. Williams,H., andA.R. McBirney,Volcanology, 397pp.,Freeman,Cooper, San Francisco,Calif., 1979. 86, 661-671, 1978. Pinkerton,H., Factorsaffectingthe morphology of lavaflows,Endeavour, 11, 73-79, 1987. J.H. Fink, GeologyDepartment, ArizonaStateUniversity,Tempe,AZ 85287-1404. R.W. Griffiths, ResearchSchool of Earth Sciences,Australian National Pinkerton,H., and R.J. Stevenson, Methodsof determiningtherheological propertiesof lavasfrom theirphysico-chemical properties, J. Volcanol. University,GPO Box 4, CanberraACT 2601, Australia. Geotherm.Res.,in press,1992. Rowland, S.K., and G.P.L. Walker, Pahoehoeand aa in Hawaii: Volumetric flow rate controlsthe lava structure,Bull. Volcanol., 52, 615-628, 1990. Ryan,M.P., andC.G. Sammis,Theglasstransition inbasalt,J. Geophys. Res., 86, 9516-9539, 1981. (ReceivedMarch 2, 1992; revised June29, 1992; acceptedJuly 8, 1992.)
© Copyright 2026 Paperzz