sign in sign up
<<
Home , Dacite, El Tatio, Paniri, Pastos Grandes, Tata Sabaya, Toconce (volcano)

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 99, NO. B9, PAGES 17,805-17,825,SEPTEMBER 10, 1994

Effusive silicic volcanism in the Central : The Chao and other younglavas of the -PunaVolcanic Complex S. L. de Silva,1 S. Self,e P. W. Francis,3 R. E. Drake,4 andCarlos Ramirez R. 5

Abstract. The largestknown Quaternarysilicic body in theworld is CerroChao in northChile, a 14-km-longcou16e with a volumeof at least26 km3. It is thelargest of a groupof severalclosely similar daciticlavas erupted during a recent(< 100,000year old) magmaticepisode in theAltiplano-Puna VolcanicComplex (APVC; 21-24øS)of theCentral Andean Volcanic Zone. The eruptionof Chao proceededin threephases. Phase 1 wasexplosive and produced -• 1 km3 of coarse,nonwelded dacitic pumicedeposits and later block and ash flows that form an apronin frontof themain lava body. Phase2 wasdominantly effusive and erupted -22.5 km• of magmain theform of a compositecoule6 covering -•53km 2 with a 400-m-highflow frontand a smallcone of poorlyexpanded around the vent. The lavais homogeneouswith rare flow bandingand vesicular tops and selvages. Ogives (flow ridges) reachingheights of 30 rn formprominent features on its surface.Phase 3 produceda 6-km-long,3-km- wideflow thatemanated from a collapseddome. Ogives are subdued,and the lava is glassierthan that producedin previousphases. All theChao products are crystal-rich high-K and with phenocrystsof ,, , , sphene, rare , and oxides. contentsreach 40-60 vol % (vesiclefree) in themain phase 2 lavasbut arelower in thephase 1 (20- 25%)and phase 3 (-40%) .Ovoid andesitic inclusions with vesicular interiors and chilled margins upto 10cm are found in thelater stages of phase2 andcompose up to 5% of thephase 3 lava.There is littleevidence for preeruptivezonation of themagma body in composition,temperature (-•840øC), œ07 (10'•), or watercontent, so we proposethat eruption of theChao complex was driven by intrusionof fresh,hot andesitic into a crystallizingand largely homogeneous body of daciticmagma. Morphologicalmeasurments suggest that the Chao lavas had internal plastic viscosities of 10•ø to 10•2 Pa s, apparentviscosities of 109Pa s, surfaceviscosities of 10•5 to 10• Pas, and a yieldstrength of 8 x 105 Pa. Theseestimates indicate that Chao would have exhibited largely similar rheological properties to othersilicic lava extrusions, notwithstanding its highphenocryst content. We suggestthat Chao's anomaloussize is a functionof boththe relatively steep local slope (20' to 3') andthe available volume of magma.The eruptionduration for Chao'semplacement is thoughtto havebeen about 100 to 150 years, with maximumeffusion rates of about25 m• s-• for shortperiods. Four other lavas in thevicinity with volumesof ~ 5 km3 closelyresemble Chao and are probably comagmatic. The suiteas a wholeshares a petrologicand chemical similarity with the voluminous regional to of the APVC andmay be derivedfrom a zoneof silicicmagmatism that is thoughtto havebeen active since the lateTertiary. Chao and the other young lavas may represent either the waning of thissystem or a new episodefueled by intrusionsof maficmagma.

Introduction youngerthan most of the large late Pleistoceneandesitic volcanoesthat dominatethe region. What makesthese lavas We describehere a group of large solitary, well-preserved exceptionalis their largesize. By far the mostspectacular is silicic lava bodies from the Central Volcanic Zone of the Andes CerroChao, a 14-lorn-longcou16e with a flow front•400 m high (CVZ; Figure1). All are highlyporphyritic, high-K dacites and (Figure3). (Figure 2) andmost have a steep-sided,circular plan and Located in northern at 22ø07'S, 68ø09'W, Cerro Chao would be characterizedas "low domes"[cf. Blake, 1990], while (alsoknown as Volean Chao or Chaolava) is a prominentfeature the othersare cou16es[cf. Walker, 1973]. These lava bodies are of the Quaternaryvolcanic front, where its unusualsquat characterand morphologyare in markedcontrast to the older •Departmentof Geographyand Geology,Indiana State andesiticcomposite cones of and Leon which flank it University, Terre Haute. (Figure3). It wasfirst describedby Guestand Sanchez [1969], 2HawaiiCenter for Voleanology,Department of Geology who estimateda minimumvolume of 24 km3. Suchlarge and Geophysics,SOEST, University of Hawaii, Honolulu. volumesof silicicmagma are usually erupted explosively to form 3Departmentof Earth Sciences,Open University, Milton ignimbrites,but large volume silicic lavas are known from several Keynes, England. regions[e.g., Henry et al., 1988;Bonnischen and Kauffman, 4BerkeleyGeochronology Center, Berkeley, California. 1987]. Unlike Chao,these are sheetlike,poorly exposed bodies, 5ServieioNaeional de Geologiay Mineria,, Chile. andin manycases their eruptive style is stilldebated [e.g., Henry and Wolff, 1992]. Copyright1994 by theAmerican Geophysical Union. Apart from its sheersize, Chao posesseveral intriguing questionsregarding its eruption,eraplacement, and regional Papernumber 94JB00652. siginificance.Here we addressthese questions and present field 0148-0227/94/94JB-0065•05.00 observations that 1end further insight into the eruption

17,805 17,806 DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM

68o30,W 67o30' 66 ø30' t t"' / 10 ø \ NVZ 21 ø ,',• o•/ 30' s //

10 ø

20 ø * / 22 ø •0 ø

40 ø

/

•o Plate •" ,...,)#

22 ø 30'

ß i 0 50 100 km

• collapse scarp Individual stratovolcanoes (-10 Ma to present) ...... Major normalfaults Alignmentofstratovolcanoes - - Faults and major structurallineaments Silicicdomes (>1 km 3) ..... Outline of resurgentcenters ':'::•'" Salars _ .- International fronhers

Figure1.Altiplano-Puna Volcanic Complex ofthe Central Volcanic Zone (CVZ) of the Andes showing themajor -structuralfeatures. The region of young silicic lavas and domes described here is outlined and shown in Figure2.Inset map shows distribution ofthe Cenozoic toRecent volcanic zones ofthe Andes and their relation to theplate tectonic framework ofthe western margin ofSouth America. NVZ is Northern Volcanic Zone; SVZ is Southern Volcanic Zone.

mechanismsof the Chao crystal-rich dacite , petrological volcanicfront built upon the basement. Alignment of data bearingon the origin of the magma,and the first thevolcanic front is approximatelyN-S, but the segment between geochronologicdata. We estimaterheologic properties and about22 ø and23øS trends NW-SE. The locationof thevolcanic effusionrate of thelava, and we presentan integrated model for centersinthis portion appears tobe controlled bya seriesof NW- eruptionand emplacement of the lava. We thencompare Chao SEtrending faults possibly related to the southern margin of the with otherassociated domes and volcanic rocks in the regionand PastosGrandes caldera complex (Figures 1 and2). Chaois discusstheir significance in the evolution of theAltiplano-Puna located on theeasternmost arm of theNW trendingportion of the VolcanicComplex (APVC). volcanicfront on a col betweenthe volcanoes of Paniriand Leon (Figure2). GeologicSetting Geochronology Chao and the associateddomes are locatedwithin the APVC, a largezone of silicicvolcanism occupying the 21 ø to 24øS segment Chaois clearly younger than the flanking of Paniri of the Central VolcanicZone of the Andes,which is one of three andLeon, on whoseflanks it is constructed.A glacial activevolcanic zones in the Andes [de Silva, 1989a, and b; de fromVolcan Leon which abuts against the eastern flank of Chao Silvaand Francis, 1991] (Figure 1). In theAPVC, eruption of formspart of a complexofmoraines that terminate atabout 4500 regionallyextensive ignimbrites from several caldera complexes m. Thesemoraines are thought to havebeen deposited during dominatedthe volcanicevolution from the late to the recessionfrom the last glacialmaximum [Hollingworth and Pleistocene.Since then volcanism has been dominantly andesitic, Guest,1967; de Silvaand Francis,1991]. On thesegrounds, forminglarge volcanic cones along a diffuse50-km-wide Chaomust therefore be older than ~11,000 years. DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM 17,807

68ø30'W 67o30'W

21 30' S Pastos

V. San Pedro

Chao

V. Michina

;•),oalama V.

,,?Lr•"4JPre-ignimbrite basement Volcanoactive in Pleistocene I I UpperTertiary ignimbrites VolcanoactiveinHolocene

...... •...... Quatemaryvolcanics T - , S - Sol de Mar3ana

•• Siliciclavaflows anddomes ,/ Intemationalfrontier r..•' (Chile/) B Town

Figure2. Mapof thestudy area highlighting the volcanic geology. Locations of the Chao dacite and other young domesin relationto othermajor volcaniccenters are shown.Centers numbered are 1, Cerro Chanka;2, Chascon/RuntuJarita; 3, Chillahuita;and 4, Tocopuri.P is PiedrasGrandes a 7.5 Ma domecomplex associated with the ignimbrites.

Threebiotite separates from Chao I lavahave been analyzed Geology byconventional K-Ar andhigh precision 4øAr/39Ar single crystal laserfusion (SCLF) techniques (Table 1). An averageage of PyroclasticDeposits 423,000+ 100,000years is suggestedby theconventional method The most extensive erupted products from Chao are a andmay be a realisticage for Chao. However, individual crystals previouslyunrecognized pyroclastic sequence that forms an apron of biotitefrom each separate yielded anomalously old ageson 24 km wide and34 km in extentbeyond the southernflow front SCLFdating. Thus, the K-Ar agesfrom the bulk separates could of Chao lava and is poorly exposedbeneath the lava on the be considerablyolder than the true age. Anomalously old ages eastemside (Figures3a and 3b). An unknownbut potentially from singlecrystals may be the resultof severalcauses, such as largevolume of thesedeposits is thoughtto underliethe lava. We xenocrysticcontamination, excess Ar, andK leaching.During conservativelyestimate the total denserock equivalent(DRE) microprobeinvestigations, several with lower than volumeof thesedeposits to be approximately1 km 3 . normaltotals and low K werefound, suggesting that K leaching Two groupsof depositsare separatedby an erosionsurface andalteration may have occurred in theChao rocks. As a result, andfluvial gravels(Figure 4). The olderChao pyroclastic flows theage of Chaois not well constrained, but the SCLF data permit are 10 to 15 m thick and nonwelded,and rest on fluvial sandsand us to suggestthat the ageof Chaois lessthan the K-Ar ageof gravels and a 5-m-thick debris flow depositcomposed of 423,000 + 100,000; how much lessis unknown. andesiticclasts, presumably derived from Paniri volcano. Large, •7,8o8 DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM

"""•-• ':'" .:•"-i :.'q?'•::::-• :;½:•;:•::- ½.-,:-.::.•;!b:....,.? ..•:? .•: .•. ::.::½C.•:...... :. ::• . :-.:?.':.:;".....::;• ......

'.qi.• ....•:: ....:i:•-:•;: ß •::: :-•' ;!•.• ...... ß-:..if! .: .... :"'•.".,,. ::::½'...... • .... '.*** ...... :•?•-;:. ....'. "•"::' :"'.--*•:½:. :'4:...... -.;...... , -...... '. -.":::,.:.;}; ..-:,. - .•: -.- '• ..... •.. ;:.:"::'::".: ".'--•:.:' :c.' ': -:-:-- ..• .• .. •. .;:.... ::..':.• . .;..-::;:..ß :..-.:.• W-:..- . ,,.....:;.... ;;.:.•.•:: ::.:-:.:. •... $;.:.•.:::'::"•;•. ..

•:•• -' :...::;•:•}.•.. :{. '-'t.... ,:--:?'•::;:: ...... •½ ',. ½.....:.;;;:!•/;•':.;:',.:: ...... ß ..::-•.:;...... :::..::.....:.:•.•.• ;½•:":.'-':",...... ,..c:.::.;e;...:::•;;; •...... :'-:'".;...' .... -. '•' ...... -.... "'"' •:%•:::;**"':•;'::,..:....•;:.•:: ..-v '?:;;..-..., , .:; •.-&': '•::½ •,.':;'":'""!:' 5-::.;...... ß{'½•-':t .;:.:.::.:•½::' t '..... d:-' "."?:':.::::;'•:'- ;•;•:::½••:::'...... '" '-':

...... :.....?:c::½ •;•:•...• •½:.•;';;'"'"'•:"•:•'•'•::':::''.:.:;:...;;•:½::';•;•;•;'.;(::.•..:.:•. :•f':•:'•:..::""•,.•:•:,,-:,... . : ;•..•;..??'½:...::...7';•:::.;•:...•:X:•.•½.-;;;? ...... :;':::-r'";;: :':".::•;:'•%%•:..:..:•:•'•.: ':"'"':'•:•;"-:-::;':----•:::::•:-•-:•:"":':•::•½•..•P;**;½':½"•*:•4½ :::'::.:'•;...... ;½:::;'•5; :•::::::':"•;'•;:-Z:-:".•½;:)'./%:;f:,;::•;(;'% :::•½• .':::;q;•:•&::..;•;.....:;;:•:94½-2; ½.:-...:.;:•;:•:-...;....•,.•:-:'•4;'•-•:•,:.;:;•, .,:;*'*'"::'•'•(•'.. '•;;4;8.;":::::•.":":':.'...... ':'•: .•::•:7.%" ..r:• .... -... ?.&•5;•:•:.:.:

Figure3a. LandsatThematic Mapper (TM) image(Landsat 50522-14001 TM FalseColor Composite of bands7,4, and 2 in Red, Green and Blue channelsrespectively) of 30 x 30 km portionof the AltiplanoPuna Volcanic Complexshowing Chao volcano and neighboring centers Paniri (P), Leon (L), and SierraNegra (N). Note the ribbedsurface of the 14-km-longChao lava due to well developedogives. Lava to the eastis CerroChillahuita. Theselarge silicic lavas contrastmarkedly with the~6000-m-high, -capped, radially gullied and glaciated, centralvolcanoes of theCentral Volcanic Zone. Note the approximate NW-SE trendof thefront and the faults (F) that coincide with this. poorly vesiculardacitic pumice blocks (maximum pumice (MP) phenocrystsof quartz, plagioclase,biotite, and hornblende ~1 m) and denselava clastssimilar in characterand mineralogy accountfor up to 30 vol % of the pumiceand are set in a poorly to the juvenile material of the upperpyroclastic flows and Chao to moderatelyvesiculated glassy matrix. Block (densejuvenile lava are abundant.The uppergroup of pyroclasticflow deposits clasts with breadcrustedexterior) and ash flow units are more are a seriesof coarse,fines-poor, nonwelded, pyroclastic flows prominentin the upperpart of the sequence.These dense clasts (ignimbrite and block and ash flows) of limited extent with have identicalpetrology to the pumiceblocks and the Chaolava thicknesses of about 4-5 m where observed. Remnants of coarse itself. pumiceclast-rich flow ridgesand flow frontsare present. Pumice Two overlappingpyroclastic cones located on the northern clasts are dacitic, large (MP 2-3 m), and crystal rich. Coarse flank of Chao (Figure3), constituteother evidence of explosive DE SILVA ET AL.- EFFUSIVESILICIC VOLCANISM 17,809

V. Paniri 6000m •L..L•/• Sierra Negra,,," // Pumice/ Cone •

III Collapseddome v. Leon ß-.%'% ..... •..,

,.:Phase 2... Phase 3 ofCha• 5500m

i5:;i !r, ", .•.. )?,"'-. ....II :::.:'•,'"•"': 1: "• •!.•,"-... '...... ;i:'.'...... ß. I•.?..?/',•! ...... ½:..;i\'•,.-.,..: .....-...... ß ...... :...... ;..-., '"".: .; ;:..'..':Ji I[1•=" • "":"'":2';...... ' ...... '"' ...... '"'"'"..'" ? ,,,*ø ..,'.,,,.'., o 1 ß ,,.. L I ,.,,,'•.,.' km • Boundarybetween eruptive phases Phase 1 ?...... Ogives(Row ddges) ,Earlypyroclastic '•""'• Talusslopes • SIo•s Rowdirection

Figure3b. Geologic sketch map of Cerro Chao, northern Chile. Main geologic features and neighbouring centers referredto in text are shown. Eruptive phases 1 through 3 are explained in text. Roman numerals I, II, andIII refer to flow unitsof Guestand Sanchez [ 1969]. activityat Chao and mark the site of thevent for the eruption. A of 2 km, theDRE volumeof thepumice cone is estimatedto be shallowdepression filled with pumice blocks is found at the top 0.5 km 3. of thecone. Its northern flank is steep and rises about 100 m from In stratigraphicterms, eruption of thepyroclastic rocks marked its baseon the col betweenLeon and Paniri,and its location thefirst, explosive phase of theeruption. Moderately explosive suggestsa relativelyhigh elevation for thevent; fromhere Chao vulcanianto subplinianactivity produced small, dense pyroclastic lavasflowed into lower elevations tothe west and south, filling in flowswhose deposits accumulated to the south of thepresent thetopography. Although the middle Chao lava unit mmcates the lava.Erosional horizons within the sequence indicate that activity southernpart of thepumice cone, pumice is foundon top of the wassemicontinuous. While most of thepyroclastic deposits are lavasuggesting that the mildly explosive activity forming the probablyproducts of thisearly explosive phase, their similarity pumicecone occurred throughout effusion of themain part of the in compositionand density to theChao lava suggests that some of Chao lava. the uppermostblock and ash flows may have been generated Thecone is consreactedofpoorly vesicular pumice and duringpart of themain extrusive event that produced the Chao blocksup to 30 cm.Plagioclase, quartz, biotite, and hornblende lava,avalanching away from the front of thegrowing mass of phenocrystsmake up about30 vol % of therock. The matrixis a lava. Thusthe transition from phase 1 to 2 wasgradual and was poorlyexpanded devitrified glass. The pumiceis similarto the markedby a declinein pyroclasticflow emplacement.About 1 vesicularportions of thelava from Chao, although its density is km3 ofmagma was erupted during this phase. somewhathigher (Table 2). Assuminga 3/4 conewith a diameter A sparselayer of smallpumice fallout lapilli is scatteredover 17,810 DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM

Table la. Resultsof K-At Age Determinationsof Biotite Separatesfrom ChaoI

Sample K% Radiogenic AtmosphericAge, Ma + 2 c• Ar•0 n18-1 Ar•%

SO 985 6.82 0.08440 97.30 0.310 + 0.263

SO 986 6.96 0.14430 95.60 0.533 + 0.254

SO987 7.21 0.13160 93.00 0.470 + 0.135

Sampleswere analyzedby S. de Silva at BritishGeological Survey Isotope Laboratory, London. the surfaceof Chao,most prominently on the northernside. We lobe diametersranging from 1.8 km on the distalmargin to about originallythought this was associatedwith the pyroclasticcone 0.5 km on the northern side. Most of the flow moved southward, on Chao [de Silva et al., 1988]. However the pumice is following preexistingtopography, but lobate flow frontson the mineralogicallyand chemically distinct from other Chao products northwestmargin indicate that a componentof the flow spilled and we now interpretthis to be derivedfrom the nearbyVolcan over in that direction. Chao II directly overliesChao I and is San Pedro. distinguishedfrom it by a subduedscarp on its westernside. The

Chao Lava combinedvolume of ChaoI andII is ~22km 3, if a flat,uniformly slopingbase is assumed.However, as indicatedin Figure 4, the We retain the subdivisionsChao I, II, and1II establishedby preexisting topography between Paniri and Leon volcanoes Guestand Sanchez(1969) for the lower, middleand upper units probablyhad a concaveupward profile with a relatively steep of Chao lava (Figure 3b). Volume estimatesgiven below are headwall. In this region, the lower flanks of nonglaciated basedon theprofiles in Figure5. volcanoestypically have slopesof ~15-20ø, gradingdownward The lowermostcou16e, Chao I, is the mostextensive part of into talus apronswith slopesof ~3 - 6ø [de Silva and Francis, the Chao dacite.Major featuresof this 14-km-longlobe are its 1991]. Preemptiontopography with thesecharacteristics would ~400-m-highflow front andthe prominentflow ridgesor ogives therefore impart a considerablyhigher volume to the Chao (seebelow and Figures6a and 6b). The flow front is lobate,with complex.

Table lb. Resultsof Ar4ø/Ar39 single crystal laser fusion Technique for BiotiteSeparates from Chao I

Ar4ø/• 9 Ar37/• 9 Ar36/Ar39 Ar4ø*/• 9 Percent Age(Ma) +_2 o Arno Radiol•enic SampleSO 985

22.86918 0.02656 0.07364 1.10861 4.8 0.387 + 0.120 7.265

18.68870 0.01692 0.06172 0.44848 2.4 0.157 + 0.108 4.863

18.18412 0.38870 0.05948 0.60772 3.3 0.212 _+0.080 6.393

22.06160 0.33690 0.07453 0.03901 0.2 0.014 + 0.087 3.660 19.76031 0.02711 0.07338 -1.92386 -9.7 -0.672 + 0.097 3.234

21.51208 0.10204 0.07666 -1.13400 -5.3 -0.396 + 0.116 2.237

SampleSO 986 221.42560 0.13654 0.74034 2.66501 1.2 0.931 -+ 1.131 15.692 23.87207 0.00746 0.07898 0.53098 2.2 0.186 + 0.225 2.414 33.19360 0.02524 0.11550 -0.93756 -2.8 -0.328 + 0.263 4.600 13.69843 0.01875 0.04771 -0.40211 -2.9 -0.141 + 0.100 1.306

26.86425 0.02481 0.09261 -0.50237 -1.9 -0.176 -+ 0.083 10.020 20.98473 0.02398 0.07252 -0.44563 -2.1 -0.560 + 0.064 14.086

SampleSO 987 6.86844 0.00427 0.02140 0.54232 7.9 0.189 +0.105 1.342

32.54490 0.01652 0.06926 12.07863 37.1 4.216 + 0.312 1.728

17.03600 0.02781 0.06182 -1.23131 -7.2 -0.430 + 0.126 1.441

51.32864 0.00796 0.16614 2.23198 4.3 0.780 + 0.191 9.302 68.86053 0.03082 0.20169 9.26097 13.4 3.233 + 0.669 18.614

Sampleswere analyzedby R.E. Drake at the BerkeleyGeochronology Center, Berkeley, California. DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM 17,811

Phase 3

Chao III Underlying Volcaniclastic Units clusions Phase 2 3-4m YoungerChao pumiceflows - possiblysyneruptive with Chao 13hase2 lava. MP -2 - 3 Chao II 5m Andesiticdebds flows and fluvialgravels; variable thickness. No Chao clasts Soil / / c Chaol 15m Older Chao pumice flows. Several flow units; o / thickest 8 m, MP -1 m ß / / 5m Lithified debris flows; andesitic clasts a. / / 10+m Fluvialgravel and sands

Phase 1 Debris flow deposit

Base not exposed

Figure4 Compositeeruption stratigraphy of the Chaodacite, built fromtraverses up the northern,eastern and southernflanks of Chao.MP is maximumpumice diameter. Columns are not to scale.

Macroscopically,Chao II andI areidentical. Flow ridges up appearsdistinct from ChaoI andII, sinceit directlyoverlies the to 30 m highare prominent features on aerialand satellite images pumicecone in the north and Chao II on the easternside. Its vent of Chao(Figure 3). Spacingbetween the ridges varies from 50 m is markedby a collapseddome that forms a steep-sided,crescent- near the flow front to as much as 100 m nearer the vent, where shapedaccumulation of blocky lava. Successivefailures of this theyare not sopronounced. Flow ridgeson the westernflank are domeare indicatedby severalarcuate collapse scars on the inner stretchedout parallel to theflow margin.The lava is massive,and wall of the crescent(Figure 3). The flow front of Chao III is a remnantsof v esicularflow topsand brecciazones are common singleconvex lobe of blocky dacite,about 150 m high, with a but volumetricallyinsignificant. The uppervesicular surfaces of smatteringof theaeolian "sand" derived from weathering of Chao thelava are weathered and altered, causing it to disintegrate.On I and 1I. Ogives in Chao III are less obviousthan on the lower the northern flank of Chao II several 10 to 20-m mounds of lobes of Chao. lithifiedbreccia are found on thetops of flow ridgesclose to the Chaom hasa volumeof about3 km3 andis typicallydense vent.These may representfossil mounds. blockylava, with pumiceouszones common in its upperparts. Petrographically,both Chao I andII are monotonouscrystal- Mineralogically,Chao III is similar to the rest of Chao but has a richdacites with prominent of quartz,plagioclase, lower crystal content and a higher concentration of andesitic biotite,and large up to 2 cm.The texture inclusions;up to 5 vol % of therock in places.This is particularly resultsfrom high but variable crystal contents (about 40 vol % on evidentabout half way up the flow front. Many of the inclusions averagebut in somehand samplesan almostholocrystalline are larger (up to 20 crn) and more vesicularthan thosein Chao II, textureis evident).Flow banding is ubiquitous.Toward the upper but are otherwisesimilar. Effusion of Chao III marked the third, andnorthern (proximal) end of Chao11, small andesiticinclusions final phaseof the Chaoeruption. are common.These are typicallysmall (up to 8 cm), ovoid or lobate,fine grained, and nonvesicular (Figure 6d). Some display glassymargins interpreted as chilledmargins. These inclusions Table 2. Densitiesof Chao Lava and Pyroclastics arenot presentin the areasof ChaoI that were studied. The distinctionbetween Chao I andII madeby Guestand Sample Density Sanchez[1969] was largelymorphological. There are several kg/m3 largeramparts on ChaoI andII that couldbe interpretedas distinctflow lobes, but because of thesimilarity in lithologyand Pumicefrom upperpyroclastic flows (#88069B) 845 petrology of the two lobes, rather than seeing these as Dense breadcrust clasts from representinghiatuses in activity,we suggestthat they are merely upperpyroclastic flows (#88069A) 2030 manifestationsof changesin the rate of lava extrusionand that ProniceCone (#88080) 2340 ChaoII vesicularlava (#88069) 1710 ChaoI and II representpulses of themain eruptive phase of ChaoIll lava (#88067A) 2750 Chao(phase 2). At thevent, mildly explosive activity continued ChaoII lava (slightlyvesicular) (#88081) 2030 throughouteffusion of thelava and built the pumice cone. About 22.5km 3 of magmawas erupted during this phase. Densitieswere determinedby cuttingcubes and weighing By contrast,the 6-km-long, 3-km-wide Chao III coule6 them(estimated accuracy is _+30 kg/m3). 17,812 DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM

•soo ReducedMapshowing

, '.e._•..,_..:..•_...-...... '-• -'"'"' "' "' "'•"•.• .•.';•-.f..".f..';•,,'" section of 1 0• - _. • ß SECTION A A' 1øøø,oo_tI SECTION B B' lOOO• $oo

o SECTION C C'

SECTION D D'

••ooosoo-• .•.;.•'•'"-'.•;. --, ConePumice soo '"'"'"'.'r..;:..,;'..:•

o SECTION E E'

Figure 5. Profilesof the Chaodacite (modified from Guestand Sanchez[ 1969]). Heightsare in metersabove the basalflow front.Horizontal and vertical scales are equal.

Petrography, Mineral Chemistry, and Magmatic coresas high as An63 are found. Quartz grains in ChaoI andII are Conditions large with minor resorption;in Chao III quartz occursas small embayedremnants of larger crystals.Phenocrysts of biotite are Four representativesamples from each major unit at Chao generally fresh in the massive lava but more altered in the (pyroclastics,Chao I, 1I and III, and the inclusions)were chosen vesicular samples. Large twinned hornblendes are common for detailed chemical and petrologicalstudies. Samplingwas throughout Chao. Biotite and hornblende compositionsshow conductedduring traverses up the northern,southern, and eastern similar restricted ranges. Biotite phenocrystsand microlites flanks of the volcano.The Chao rocks are all dacitic (dacitesand throughoutChao arehomogeneous with Mg numberranging from rhyodacites), with phenocrysts of plagioclase > biotite > 58 to 60, while hornblendesare commonlyzoned with coresand hornblende> quartz>> sphene> sanidine> Fe-Ti oxides.Apatite rims showingranges of Mg numbers from 56-61 and 61-65 is themost common accessory mineral, both as microphenocrysts respectively. Hypersthene rimmed to varying extent by and as inclusions in the phenocrysts,together with zircon. hornblendeis foundscattered throughout most thin sectionsof the Allanite was rarely detected.Phenocryst contents are variable, Chao II lava. The hypersthenehas Mg numbers of 80-83, rangingfrom ~ 25 vol % (vesiclefree) in thepumice to ashigh as identical to thosein the andesiticinclusions (see below), and is 50 vol % in the lava. A phenocrystcontent of 40 to 45 vol % is thoughtto be xenocrystic.Glomerocrysts of hornblendeand Fe- typicalfor Chao (Table 3 andFigure 7a). Mineral chemistrywas Ti oxidesare probably resorbed . analyzedon a Cameca SX 50 electronmicroprobe at Purdue In most of the denselava samples,the matrix is a colorless University. Data and operating parametersare available on high-silicarhyolite glass(Figure 7a) which in most of the lava requestfrom the first author. samplesis fresh. However, patchesof spheruliticdevitrification Plagioclasephenocrysts throughout Chao exhibit a varietyof andvariable K20 and SiO2suggest alteration. Some samples also formsranging from simpletwins to complexreverse and normal contain microlites of the phenocrystminerals that make up to zoning. Compositionsthroughout Chao are restrictedwith 90% about50% of thematrix. The combinationof thehigh phenocryst of the rangeis betweenAn2• and An55(n = 100). Rarely, some contentand the partly crystallizedgroundmass leads to a very DE SILVA ET AL.- EFFUSIVE SILICIC VOLCANISM 17,813

.. %_..'.;,'.ß

.. ß 17,814 DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM

Table 3. Modal phenocrystdata for samplesfrom Chao

Pyroclastics PumiceCone ChaoI ChaoII Chao I!I 84063 88069C 88080 88073 84024 88068A 88068C 88081 88067

Rock Type dense pumice pumice lava lava lava vesicular lava lava juvenile lava Glass/Matrix* 60 65 60 55 50 47 57 57 55 Phenocrysts Plagioclase 18 17 17 19 21 22 21 18 21 Quartz 6 5 8 8 10 6 6 7 6 Biotite 7 7 5 9 7 9 7 8 7 Hornblende 4 2 8 5 7 10 6 7 7 Fe-Ti Oxide 1 2 1 1 2 1 1 1 2 Sanidine 2 1 1 2 nd 3 2 1 nd Sphene 2 1 nd 2 2 2 tr 1 Apatite•' tr tr tr tr tr tr tr tr 1 Zircon • tr tr tr tr tr tr tr tr Allanitc •' tr tr tr nd nd tr nd nd tr Percentphenocrysts 40 35 40 45 50 53 43 43 45 All data are volume percenton the basisof 1000 pointscounted (analyzed by S. de Silva). Abundancesfor vesicular samplesare calculatedon a vesicle-freebasis. Tr is traceand nd is not detected. *Glass/matrix is treated as noncrystallinematerial in this data set •' These mineralswere detectedonly as discretemicrophenocrysts or as inclusions high overall crystalcontent for Chao (>60%). In a few samples variation from phase 1 through3 for major or trace elements, the groundmassis devitrified to a fine-grainedaggregate. In the indicatinga homogeneousmagma batch (Figure8). SiO2 varies vesicular samples, the glass is largely devitrified and dusty, from 66.97 to 69.77 wt %, K20 from 3.73 to 3.94 wt %, Rb from althoughrare freshpatches are found. 162 to 170 ppm, Sr from 282 to 327 ppm, andTh from 28 to 29 Andesiteinclusions in ChaoII andIII are sparselyporphyritic ppm.The andesitcinclusions are more with lowerRb, Th, with a few large crystals of plagioclase,hypersthene, quartz, and Nb and significantlyhigher Cr, St, Y, and Zr; Ba is biotite, andhornblende set in a groundmassof randomlyoriented marginallyhigher. A singlerepresentative sample of ChaoI lava acicular hornblende, plagioclase, and euhedral hypersthene (sample84054) was found to be LREE enriched,with a Ce/Ybof surroundedby a small proportionof glassymesostasis. Of these 48, an 87Sr/86Srratio of 0.70805, and a •43Nd/•nnNdratio of groundmassminerals, plagioclase shows a rangeof An65to An74, 0.512247. hornblendes have Mg numbers ranging from 65-68, and The scaReredpumice lapilli airfall foundon the surfaceof the hyperstheneshave Mg numbersof 80-82. The glassymesostasis Chaolava is anomalousin petrologicand chemical character. It is is high silicarhyolite. Of the largecrystals in the inclusionsonly sparselyporphyritic with plagioclase,augitc, and hypersthene hyperstheneappears to be in equilibrium. Quartz is highly phenocrystsset in a clearvesicular glass matrix. are resorbedand has reactionrims of augitc(Figure 7c), plagioclase more calcic (An75to An80) and the glassmore mafic (-70% is commonlyhighly fretted and showsreaction on its rims, and SiO2)than any of thevolcanic products demonstrably of the Chao both biotite and hornblendephenocrysts exhibit breakdownto an volcano.It is more mafic than the dacitesof Chao, and although aggregate of oxides and smaller crystals. A large (1 cm) its major elementcomposition may be generatedby mixing hornblende lath (Figure 7d) in the inclusion has a core between the andesitic inclusions and dacitic lava of Chao, its compositiontypical of the Chaodacite (Mg numberof 60.5) and traceelement characteristics are inconsistentwith suchan origin. a rim typicalof the andesitc(Mg numberof 67.4). It is likely that High Ba and Sr and low Th contents,in particular,are more theselarge crystalsare actually xenocrystsinherited from the typicalof the pumicefalls eruptedfrom the nearbySan Pedro Chao dacitemagma, as at Cerro Purico [Davidsonet al., 1990] volcano[O'Callaghan and Francis, 1986], whichwas the most andChaos Crags [Heiken and Eichelberger, 1980]. probablesource. Compositions of coexisting Fe-Ti oxides indicate little variationin temperatures(--840 øC) andfO 2 (10'n) throughout Physicaland RheologicalParameters the magma(Table 4). Total pressureestimated using the method of Johnsonand Rutherford[1989] indicatesa Ptotof '"0.2 GPa (2 Our observationsindicate levels of crystallinity(40 to 60%) kbars)indicating that the magmaequilibrated at about 7 - 8 km. closeto the critical limit for silicic magmas[cf. Marsh, 1981]. Naney (1983) suggeststhat at thesepressures, the presenceof This raises interestingquestions about the influence of high both hornblendeand biotite indicatesH20 contentsof at least4 - crystalcontent on the rheologicaland physicalproperties of 5 wt% in magma. For a crystal-rich magma such as Chao, silicic lavas.We have estimatedvarious physical and rheological preemptionH20 contentsin the magma (liquid plus crystals) parametersfor Chao(Table 6). Chaohas an aspectratio of -20, would have been-2 wt %. which falls at the extremeend of the knownrange for othersilicic lavas[Henry et al., 1988], a yield strengthof 8 x 105Pa s, Whole Rock Composition internalplastic viscosity in therange 101ø-1012 Pa s, andsurface viscosityin therange 10•5-102• Pa s.These are not exceptional for Whole rock analysesof the the Chao productsshow that the silicic lavas. Further, using the method of Pinkerton and magmawas calc-alkaline,high-K dacite(Table 5). Thereis little Stevenson[1992], we obtainan estimateof 109 Pa s for the DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM 17,815

...

... 17,816 DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM

Table 4. Temperatureand fid 2 EstimatesBased on Fe-Ti Oxide Compositions

Phase 1 Phase 2 Phase 3

Mag Ilm Mag Ilm Mag Ilm SiO2 0.06 0.02 0.00 0.00 0.00 0.00 TiO2 4.30 34.19 4.88 35.72 4.56 34.67 A1203 1.11 0.23 1.56 0.41 1.40 0.14 FeO 84.47 60.81 85.35 58.01 85.89 57.52 M_nO 0.54 0.41 0.56 0.85 0.52 0.71 MgO 0.60 0.58 0.69 1.29 0.86 1.34 CaO 0.10 0.15 0.07 0.05 0.04 0.04 NaO 0.12 Subtotal 91.30 96.39 93.11 96.33 93.27 94.42

FeO* 32.78 29.32 33.84 28.96 33.38 28.07 Fe203* 57.44 35.00 57.25 32.28 58.35 32.73 Total 97.06 99.90 98.85 99.56 99.12 97.70

log (Mg/Mn) 0.29 0.40 0.34 0.43 0.46 0.52

X(usp)'l' 0.13 0.15 0.14 X(ilm)1' 0.66 0.68 0.67

TøC :!: 853.80 864.92 855.23 log fO 2 -10.33 -10.30 -10.37

TøC ô 826.20 832.11 827.15 logfO 2 - 11.22 - 11.24 - 11.25

Mag is magnetite(spinel), and Ilm is ilmenite (rhombohedral).Oxides are in weight percent. Usp is ulvospinel. *Calculated assumingcharge-balanced stoichiometric formula. •Calculatedfollowing Stormer [1983]. •: Calculatedfollowing Spencerand Lindsley [1981]. ô Calculatedfollowing Anderson and Lindsley[ 1985].

apparent(bulk) viscosity, similar to that calculatedfor the Mount resulted.Blake [ 1990] alsopredicts that the materialfirst pilesup St. Helensdome, another crystal-rich silicic lava body.The yield near the vent to form a symmetricallava dome;when the dome strength calculated in Table 6 is approximately an order of collapses,a shortcou16e forms. This appearsto be the sequence magnitudehigher than expectedfor magmasof this composition of eventsshown by ChaollI, whichwas emplacedonto the upper at magmatietemperatures [cf. McBirneyand Murase, 1984] but is surfaceof Chao II on a slopeof about2'. The diameterof the similar to values calculated for other domes and coule6s. domeappears to havebeen about 1.5 km beforeit yieldedto form Formation of a viscous carapace, as indicated by the well- a cou16e,indicating a yieldstrength of 5 x l0 s Pa accordingto developed ogives, was probably an important factor in this the methodof Blake [1990, Figure 14]; this value is consistent elevatedyield strengthestimate [e.g., Blake, 1990]. Griffithsand with our morphologicalestimate. Fink [1993] proposethat the strengthof the ,where well During the furstphase of the Chao eruption,lava probably developed,is more significantthan the physicalproperties of the eruptedonto a much steeperslope, possibly as muchas 15 to internal part of the lava body in controllingmorphology and 20 ø. On theseslopes, any initial domewould have achievedonly distance flowed. very small size (100-200 m in radius) before collapsingand Chaowas eruptedonto an averageslope of about 4ø, although would not readily be dete,etable. Some of the upperpyroelastic the slopemay locally havebeen as greatas 20ø(see below). This flowsprobably formed as a resultof collapsefrom sucha nascent slopestrongly influenced the growthand final morphologyof the dome. Chao lava. For magma with a yield strengtherupted onto a slight Our estimatesshow that althoughChao magma was extremely slope, Blake [1990] suggeststhat for coul6es to form, the crystalrich, its rheologicalproperties are typical of siliciclavas in condition general.This runscounter to the conventionalview thatincreased crystal!inityresults in increasedbulk viscosity, leading to the R>ho/sin 2 0 systemlocking up when a critical crystallinityof -40 to 60% is reached[cf. Marsh, 1981]. The fact that many lavaswith crystal must be satisfied,where R is the dome radius, ho is the natural contentsclose to or greaterthan the critical limit existimplies that lengthscale, and 0 is theslope. At Chaowhere R --4 andho/sin 2 factorsother than the crystalcontent are important.It is possible 0 = 1.26, the condition above was satisfied and coul6e formation that while highercrystal content leads to increasedbulk viscosity, DE SILVA ET AL.. EFFUSIVE SILICIC VOLCANISM 17,817 •7,8•8 DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM

I

66 69 3.5 4.0 600 650 ?8 30 25 55 58 65 800 900 SiO2 K• Be Th An Mg# (Hbl)

Figure8. Variationof petrologicaland chemical indicators with eruption stratigraphy (see Figure 4). Only the juvenilecompositions are plotted. Data shown are from 16 samplesof Chao(inclusions not included), four samples fromeach unit. Horizontal bars indicate range of analysesfor eachunit, and dashed vertical lines join meanvalues. theresulting increase in yield strengthcoupled with the growthof independentrelationships in the literature (most of theothers deal a crustreduces heat loss,which in turn reducesthe bulk viscosity with exclusivelymafic flows) andit allowsan estimationof the [e.g. Dragoni and Tallarico, 1994]. The massiveogives on the time requiredfor a lava field to achievespecified dimensions. surfaceof the Chao lavasindicate that the carapacewas thick and For Chao I and II, the measuredmorphological parameters are strong, so it is possible that the yield strength-temperature W== 7 x 103m; L== 14x 103m; H = 500m, andaverage ct of relationshipsmay haveoutweighed the effectof crystalcontent in 4 ø (seeabove); these parameters yield T, the eruptionduration, determiningthe overalltheology of the Chaolavas. asabout 250 years. This implies growth rates of 2 to 3 ms s -• for the main phaseof Chao. Eruption Duration and Growth Rate of Chao Our calculatedgrowth rate is typical ot historicallyobserved domesworldwide like Mount St. Helens [Swansonet al. 1987] Although Chao was erupted in three stages,each stage and Santiaguito[Rose, 1987]. Further,Manley [1992] suggests appearsto have been continuous.For instance,the Chao I /II that durationsof hundredsof years are not unreasonablefor cou16eshows little evidenceof breaksin activity, althoughthere cooling(and perhapsflow) of sheetlikesilicic lava bodies. may have been variationsin the effusion rate as suggestedby Nevertheless,we considerthe estimateof theeruption duration to ramparts.Intuitively, a steadysupply of magmawould have been be an overestimatefor Chao. First, there is considerablescatter in neededto keep the flow moving.This leadsus to questionsabout the Kilbum and Lopesdata, and the errorsassociated with the the rate of magma supply, eruptionduration, and growth rate. analysisare ~ 70% for the left handside and 10% for thefight Unfortunately, such questions are not easily answered for hand side of equation(1). The errorsare of the order of the prehistoricflows becauseof the errorsin translatingmorphologic uncertaintiesin measurementsof someof theparameters at Chao, characteristicsinto kinetic processvariables and becauseof the particularlydepth and slope.Further errors are introducedfor subtlebut importantdifference between effusion rate (flux at the large-volumeeruptions like Chaobecause their relationship needs ventregardless ofdensity, Vf) ofmagma at the vent and eruption to be extrapolatedbeyond the boundsof the data. Second, rate (DRE or massequivalent per second,V e ) estimatedfrom degassedand solidified lava. Nevertheless,we attemptto provide Table 6. Estimatesof RheologicalProperties of Chao Lava someinsight into thesequestions below. We use morphometricparameters to estimategrowth rates Value Method becausethese are the leastproblematic for our purposes.Kilburn Intemalplastic viscosity 10]ø - 10]2 Pa s Mooreet al. [1978] and Lopes [ 1991] useddata from historiceruptions to proposed the following empirical relationshipbetween growth rate and Surfaceviscosity during 10]5 - 10Z•Pas Fink[1980] morphometricparameters of lava fields: folding Yield Strength 8 x 10• Pas B/ake[1990] (W=/L,.)H2 sin ct = 0.1T (1) Aspectratio (H/V) 20 Walker [ 1973] whereW,, is themaximum width, L= is themaximum length, H is the mean depthof the flow, ct is the slope,and T is the time H is the diameter of circle with the basal area of the lava, (days).This approachis one of the few apparentlycomposition- and V is the thickness of the lava. DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM 17,819 althoughthe growth rates are appropriateby analogy with the eruptionof lava has continuedfor 71 years, albeit at a lower rate historic domes,these grow exogenouslyin a seriesof eruptive thanwe envisagefor Chao. episodesfollowed by periodsof stasis,whereas the main phaseof In conclusion,the estimated peak eruption rate (~25 mz s-1) for Chao(24 kmz) appearsto havegrown endogenously during a Chao is low in comparisonto eruptionrates of historicbasaltic singleeruptive episode. During an eruptiveepisode the effusion lava flows like the eruption of Laki, in 1783 rate peaks soon after the beginning of the eruption and then [Thordarsson and Self, 1993] but is probably realistic in graduallydrops to lower values as the eruptionproceeds [e.g., comparisonwith observederuptions of silicic lavas.Low eruption Wadge, 1981; Stasiuket al., 1993]. The main volume of a lava rates are also consistentwith the fact that no caldera collapse may thereforehave been effusedin a very shortperiod of time, occurred;rapid eruptionof 24 kmz of magmais likely to have but the last few percentmay have taken a considerablylonger resulted in caldera collapse. We also conclude that the time. theologicalproperties of Chao were similar to other silicic lavas. To obtaina more realisticestimate of the eruptionduration at Chao is exceptional,then, only for its volume; few otherdomes Chao an estimateof the effusionrate is required.This may be or cou16es have volumes close to the minimum 26 km z estimate done by using one of the available fluid dynamic and thermal for Chao. All elsebeing equal, slopemay have been the crucial models [e.g., Huppert et al., 1982; Fink and Griffiths, 1992; factorin determiningthis. Stasiuket al., 1993]. However, theseare more appropriatefor Assuminga large reservoirof magmais available,the volume interpretingflows and domes with detailed recordsof effusion of a eruptedonto a flat surfacewill be limited by the and morphologicevolution, and they involve several variables relation between the overpressuredriving the eruption and the that are poorly constrainedin prehistoricflows like Chao. A less massof the lava above the vent. Figure 9 showsthat the critical problematicapproach is to use the dimensionlessGraetz number control for effusion of lava is that is related to the cooling of a warm fluid moving througha cold pipe. This is commonlyused to calculateeffusion rates for P• > Pa + pgh . flows where only morphometricparameters are available [e.g., Hulme and Fielder, 1977; Wilson and Head, 1983] and therefore Where P, is the overpressurein the ,P a is may be appropriatefor our purposes.Essentially, atmosphericpressure, p is the density, g is accelarationdue to gravity, and h is the height of the dome. P, can be assumedto Vf =300rW L /H have been similar for each of the lavas, as all came from a where,W, L, andH are as in equation(1), and •: is the thermal coherentmagma zone (seebelow) and there is little differencein vent elevation;Chao has the highest vent elevationof 4950 m, diffusivity~ 5 x 10'7 m2/sec(•: = K/Po cv ;valuesfor K Chillahuitais at ~4750 m, and and Chasonare -4900 (thermalconductivity), Po (nonvesicular lava density),and c v m. Once a critical value of h is reached,no furthermagma can be (specificheat capacity) are 1.26J m4 s4 K'l, 2750 kg m'z, and erupted. In the case of Chao a large volume of magma was 1150J kg4 K4 respectively).This yields effective eruption rates available (see below) and was eruptedonto a slope sufficiently of about25 mz s4. Thistranslates to aneruption duration of about steepto allow lava to flow away from the vent area, preventing 20 years for Chao. We believe this duration to be an the critical height from being reached.Observations on the other underestimate,because the Graetz number approachassumes a lava domesin this regionsuggest the criticalheight is of the order continuoussteady state eruption and thusignores variables such of 200 m. aschanges in conduitdiameter and local slope and any changes in effusionrate that may have occurredduring the eruption.We Eruption Mechanism suggestthat the Graetz number approachyields eruptionrates closeto the peak value for Chao. This is not unreasonablewhen Although there was some explosive activity during the first comparedto Mount St. Helens, where in individual eruptive phase of the Chao eruption, this was volumetrically episodesfrom June1980 to August1982, eruptionrates ranged inconsequential.An outstandingquestion therefore,is why Chao from2 to 26 mz s4 [Swansonet al., 1987,Table 1]. shouldhave eruptedeffusively when all the ignimbritesin the Our analysisillustrates the difficultiesin estimatingeruption region,which it resemblesin termsof composition,were erupted parametersfor pre-historicflows like Chao. We have used what explosively.A magmawith a moderatewater contentsimilar to we considerto be the most appropriatemodels and obtained Chao (-2.5%) shouldbegin to vesiculaterapidly and fragment resultswhich are an orderof magnitudedifferent. The difference explosivelyon ascentas confiningpressures decline below 0.1 is rooted in the different approachesof the two models; the GPa (1 kbar or ~4 km). Kilburnand Lopes [1991] approachyields an eruptionduration We dismissthe possibility that tephra ejectionfollowed by that includesperiods of stasisand is thereforean overestimate, lava extrusionreflects eruption from a magmazoned in volatiles whilethe Graetz number approach assumes a continuous eruption [Eichelbergerand Westrich,1981; Fink, 1983], becausethere is andtherefore underestimates the eruption duration. It is likely that little evidenceof zonationin the Chao eruptiveproducts. Another the eruptionrate for phase2 of the eruptionprobably peaked at possibilityis that an initially volatile-richsilicic magma somehow about25 mz s4 soonafter onset and decayed rapidly during the releasedits gas as it ascended.A theoreticalmechanism for loss courseof the eruptionto a more nearly steadystate effusion, of volatiles from a silicic magma during eruption, termed the duringwhich the bulk of thevolume was erupted. Short pulses of "permeablefoam" model [Eichelbergeret al., 1986], has gained higheruption rate are suggestedby themorphology of ChaoI/•. widespreadsupport despite objections by Friedman [1989] and, Latereffusion rates were probably much lower but probablystill more recently, Fink et al. [1992]. In this model, an ascending higherthan the growthrate predictedby the Kilburn and Lopes water-richsilicic magma vesiculates to porositiesas high as 55- [1991] approach.On this basis we suggestthat an eruption 75 vol %, at whichpoint it becomeshighly permeable to gasflow durationof about100 to 150 yearsfor the main phaseof Chao is due to interconnectingbubbles. Jaupart and Allegre [1991] more appropriate.We note that at Santiaguitoin further developedthe permeablefoam idea, and suggestedthat 17,820 DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM

^) pgh + pgh +

Figure 9. Schematicillustrating pressures involved in eruptionof lava onto(a) flat and (b) inclined topography. Eruptionis drivenby an overpressureP, andopposed by the pressureof the growingmass of lava abovethe vent (pgh+ P,,; wherep is density,g is accelerationdue to gravity,h is heightof domeand P,,is atmosphericpressure). Lengthof arrowsdenotes relative magnitudes of pressure.Note in Figure9a thatonce a criticalheight h is reached, P, is exceededand no furtherlava canbe erupted.In Figure9b, h neverreaches a criticalvalue because the lava is continuallymoving away from the vent area, and P, is neverexceeded.

explosivity of a magma may dependon eruptionrate; a slow Silva et al., 1993; Davidson et al., 1990; Bacon, 1986]. We eruption rate allowing gas to escape from a highly inflated interpretthese inclusions as evidencethat the dacitemagma body magma throughpermeable conduit walls. Jaupart and Allegre of Chao was intruded by a hotter, more mafic magma. The [1991] strengthenedtheir case by demonstratingthat the time appearanceof theseinclusions toward the middle of the eruption scales involved in their model are consistent with the slow (in Chaoll) suggeststhat the andesitewas forciblyinjected into, eruptionrates observedfor silicic lavas like that of the Mount St. and dispersed throughout, a large part of the chamber as Helens dome. disaggregatedblobs. These chilled rapidly in the host dacite, An implication of the permeablefoam model is that silicic producingthe randomlyoriented network of acicularcrystals in lavasshould be eruptedentirely as highly inflatedpumice (55-75 the groundmass(Figure 7d). Xenocrysts in the dacite derived vol % vesicles),which then collapsesdue to its own weight to from the mafic magma suggestthat completedisaggregation of form a lava with porositiesof less than 5%. At Chao, the initial someblobs occurred. Some phenocrysts from the dacitewere also explosive activity may have served to create a relatively incorporatedinto the andesiteblobs. The record of more mafic permeable conduit environment in wall rock (probably rim compositions on the hornblendes and reaction rims of ignimbrite)where degassingof the magmacould take place.A clinopyroxene on quartz is consistentwith an increase in low ascentrate is certainlyconsistent with our calculationsabove. temperatureeither locally or throughoutthe magmachamber and However, there are two major objectionsto the permeablefoam supportsan intrusive event. An alternative argumentis that the model for Chao. andesiticinclusions represent a hotter,more mafic layer to the First, it may be difficult to attain the required levels of Chaomagma chamber, but thereis little evidencefor a thermally inflation in crystal-richmagmas. Even if crystallizationof the andchemically stratified magma body. matrix took place after or during emplacement,the phenocryst We suggestthat the intrusionevent may have triggeredthe contentof the Chaomagma was still very high (50 vol % in some eruptionof Chao:magma mixing/mingling is a commoneruption samples).This objectionis supportedby the poor vesicularity triggerfor volcaniceruptions [cf. Eichelberger,1978; Sparkset (<30 vol % vesicles)of thepumice associated with Chao.Second, al., 1977]. Since the inclusionsand host containsimilar glass completepreemption degassing did not occur at Chao, because compositions,thermal equilibrium between host and inclusions syneruptiveexplosive activity proceededthroughout phase 2; was reachedduring the considerabletime it took for the Chao thusthe magma was not completelydegassed, as requiredby the eruptionto occur. permeablefoam model. We thereforebelieve it is unlikely that If the intrusionof the andesitewas the eruptiontrigger, two thismodel is applicableto Chaoor, possibly,to othercrystal-rich scenarioscan be envisaged. magmas.Further, as we explainbelow, the apparenthigh water 1. Vesiculationof the inclusionsmay haveresulted in a release contentof the Chao magma may be a "red herring" and not a of volatiles sufficient to trigger the eruption of Chao and factor in its eruption. accountingfor the first explosivestage of the eruption. We proposethat the andesiticinclusions played an important 2. Intrusioncaused an overpressurein the chamberdue to its part of the Chao eruption.They are a prominentfeature of the excessvolume [cf. Blake, 1981], and the Chao magma would laterproducts, appearing in the middleof the sequence(Chao 11) have been forced out. andbecoming progressively more commonin the laststage (Chao A combination of the two scenarios, in which volatile release I11).Their petrologiccharacter, ovoid shapes, vesicular interiors, fueledthe initial explosivephase while the later effusionswhere andchilled margins (Figure 6d) aretypical of inclusionsresulting simply forcedout slowly, may be near the truth. In addition,heat from magmamixing throughoutthe CVZ andelsewhere [e.g., de addedto the systemfrom the andesiteitself and from the latent DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM 17,821 heatof crystallizationmay have lowered the bulk viscosity of the diameter, while the larger is a compositeflow with a quasi- Chao magma sufficiently to erupt it. We considerthat the circularplan with maximumdiameter of 3 km. Eachhas the flow- intrusionof mafic magmawas criticalto the eruptionof Chao, ridgedsurface typical of siliciclavas. whichotherwise may have stayedbelow and crystallizedas an Like Chao, the ventsof all theseyoung silicic centersappear intrusion;it appearsto havebeen well on the way to complete to be controlled by a series of NW-SE trending faults which crystallization. characterizethe tectonicfabric of this part of the APVC (Figure 1). Chillahuitais locatedjust to thenorth of theline of centersof Other Silicic Lava Bodies in the APVC whichChao forms a part, but it appearsto be controlledby the same group of faults. Tocopuri lies at the SE extensionof this Several other beautifully exposed crystal-rich dacite or alignmentof volcanoes.Cerro Chasconand the associatedCerro lavas closely similar similar to Chao in age and Runtu Jarira domes appearto be the surfacemanifestation of a petrologicalcharacteristics occur within 100 km (Figure10). NW-SE trendingdyke, whereasCerro Chankais at the northern CerroChillahuita (22'10'S, 68'02'W; Figures 3a and10a), 20 extent of the NW-SE trendingInacaliri graben(Figure 2). All km east of Chao, and Cerro Tocopuri or La Torta (22'26'S, thesefeatures are locatedwithin, or dose to, the southernmargin 67'$8'W; Figure 10b), $0 km to the southeast(Figure 2), are of the late Mioceneto PliocenePastos Grandes caldera complex. circular bodies with areas of ~11 km 2 and volumes of about 4 The two active geothermalareas of E1 Tatio (Chile) and Sol de km• each.They havesteep, scree-covered flanks about ~200 m Mafiana(Bolivia) lie on the sameNW-SE trend(Figure 2). high and flat tops and are locally referredto as tortas(cakes). The PiedrasGrandes dome complex is anotherexample of a Flow folds rib their uppersurfaces. The vent for Chillahuitais large silicic lava body in this region. However, this complexis clearly visible on the southernside of the lava, which flowed mucholder, about -7.5 Ma [Marinorig and Lahsen,1984], and north and eastwardspreading down the 3-4* slope(Figure 3a). is an extrusionrelated to eruptionof the Toconceignimbrite Tocopuri on the other hand was eruptedonto a flat terrain formation [de Silva, 1989b]. It is not part of the episodeof betweenVolcan Michina (Tocopuri)to the eastand Volcan Tatio activity discussedabove but is included in the following to the west and is symmetricallyarranged about a centrally discussionfor comparativepurposes. locatedvent (Figure 10b). Marinovif andLahsen [ 1984]report a K-Ar ageof •1 Ma for Tocopuri,which overlieslate Pleistocene RelationshipsBetween Lavas and Ignirnbrites moraineson its southeasternflank andis thusprobably Examinationof Table 7 reveals that with the exceptionof (•10,000 years)in age. Chanka,the youngautonomous lava bodiesare closelysimilar in Both Chillahuitaand Tocopuriappear to be the productof composition.Chao and Chillahuita are particularly striking in the single extrusiveevents. No evidencefor explosiveactivity is similarity of their trace element and rare earth element (REE) observed;a scatteredpumice fallout depositaround Chillahuita compositions,but all the domesare LREE enriched,with closely has the same chemical compositionas that on Chao and is similar abundances and therefore similar Ce/Yb and La/Sm; probablyalso from Volcan San Pedroto the west. 87Sr/S6Srvaries from 0.70801 to 0.70805 and •43Nd/•44Ndfrom Cerro Chascon(21'$3'S, 67'$4'W; Figure 10c) is located 0.512244 to 0.512247. These chemical characteristicsclosely within the caldera in Bolivia some 60 km to the resemble those of the older ignimbrites and associatedsilicic northof Chao(2 in Figure2). It is the largestbody in a chainof lavas (Table 7), but differ from the silicic lavas eruptedfrom small silicic bodiesknown as Cerro Runtu Jarita(Figure 10d) compositecones throughout the CVZ, which have higher large whichis •obably the surfacemanifestation of a NW-SE trending ion lithophileand high field strengthelement concen•ations and dyke similar to that describedfrom the Inyo domes,California lower Sr and higher Nd isotope ratios. Chanka shares the [Fink, 1985]. Although located within the Pastos Grandes characteristicsof the lavas from the compositecones, and this, Caldera,these domes are not thoughtto form part of this caldera coupledwith its apparentolder age,indicates that it doesnot form complexsensu strictu, however they may be relatedin a broader partof therecent episode of silicicvolcanism we aredescribing. Their traceelement and isotopicchemistry suggest that the Chasconis a torta with a ventlocated roughly in its geometric CVZ ignimbrites were produced dominantly by large-scale center,from which lava lobes extendradially like petalsof a crustalmelting, possiblytriggered by mafic magmasfrom an flower.Beneath the lithified basal dome of the lavabody asthenosphericsource. Similar mafic magmasmay alsorise into is a thin (1 m) rhyoliticpumice fall containingobsidian clasts. the upper crust,experiencing contamination on the way, to feed Clearly, some minor pyroclasticactivity precededthe lava the magma chambersof the central volcanoes,where further extrusion. Cerro Runtu Jarita consists of seven domes. The two assimilationand fractional crystallization yield smallvolumes of northern domes are extremely crystal-richdacites, while the silicicmagma [de Silva, 1989a,Figure 5]. southerndomes are andesitic.Mixing of the two magmatypes is The close resemblancein compositionof Chao and other ubiquitousthroughout (R. Wattsand S. L. de Silva unpublished young domes to the older ignimbritesdemonstrates that silicic data, 1993). magmasof largelysimilar composition have been erupting in the The other young silicic lava bodiesin the region are Cerro APVC since the late Tertiary. This combinedwith the domes' Chanka(Pabellon) and two smalldomes nearby (1 in Figure2). similarity in age, proximity to each other, and their location Chankais a steep-sideddome consisting of threelobes each with within the zone of silicicmagmatism, allow us to speculatethat a diameter of-1.5 km located at 21'48'S, 68'15'W on the the domes may represent a discrete new period of silicic westernflank of Volcan Azufre. Chanka is not as youthful magmatism.de Silva [1989a] has argued that the continuing morphologically as the other domes described. Roobol et al. geothermalactivity and the presence of theyoung silicic volcanic [1974] reportedan ageof 1.5 + 0.1 Ma from thiscenter. The two productsindicate that the silicic magma systembeneath the neighboringdomes are locatedon the easternside of Azufre and APVC may still be active. Chao and the other domesmay be are morphologicallymore youthful than Chanka.The smalleris leaks from this large silicic magma systemand thus shouldbe CerroApacheta, a simpledome with a circularplan about 1 km in regardedas the most recent expression of continuingactivity. 17,822 DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM 17,823

i i i i i i i i i

(",1 0

i

o(•1

o [...,.. •.• 17,824 DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM

RegionalSignificance of the Young SilicicLavas the olderregional ignimbrites suggests that all form partof a long lived silicic magmaticsystem below the APVC. Chao and the Given that the other young silicic lavas all share a general other young domes may have been derived from a waning family resemblanceand may represent the same magmatic magmaticsystem containing "old" volatile-poorsilicic magma episode, is it possible that intruding mafic magma was as left after eruptionof the voluminousignimbrites. Alternatively, importantin their eruptionsas at Chao?Recent work at Chascon they may representa new crustal melting episodefueled by and Runtu Jarita where magma mixing is abundant,certainly intrusionof fresh hot mafic magma.The locationof the young implicatesthis as an eruptiontrigger (R. Watts and S. L de Silva, silicic lavasprobably defines the extentof the presentlyactive unpublished data, 1993). All the lavas appear to be in an magmatic system and may indicate the location of a future advancedstage of evolution,and thereseems little likelihoodthey caldera. would have eruptedwithout sometrigger. We note that periodic intrusionsof mariemagma are thoughtto be an importantfeature Acknowledgements.An early versionof this manuscriptbenefited from in the long term evolutionof large continentalsilicic magmatic thoroughand insightfid reviews by Jon Fink, Don Swanson,and Mike provinces [cf. Hildreth, 1981]. We concludethat intrusion of Sheridan.These, discussions with Bill Rose,Jon Fink, CurtisManley, and marie magmawas likely to havebeen the eruptiontrigger. StephenBlake as well as official reviewsby CurtisManley and John If the youngsilicic lavasdescribed here do indeedrepresent a Eichelbergerhave improved this work immeasurably.Fieldwork was conductedin severalstages but largelyunder the auspicesof NASA grant discretenew episodeof leakage from a regional silicic magma NAGW-1167 while S. de S. and P.W.F. were at the Lunar and Planetary reservoir, two interestingpossibilities emerge. On the one hand, Institute(LPI) in Houston.The LPI is operatedby the UniversitySpace the episodemay constitutethe waning stagesof the magmatic ResearchAssociation under NASA contractNASW-4066. Lindsey O' system.In tiffscase, the dacitelavas represent"residual" volatile- Callaghan,Jon Davidson(UCLA), and PeterMouginis-Mark (University Ix)or magmaleft behindafter the massiveexplosive eruptions that of Hawaii) are thanked for field supportduring various campaigns producedthe ignimbritesand then refueled by intrudingmafic between1985 and 1989.This paper is SOESTcontribution #3549. magma. In this case, the moderatewater contentof the dacites References inferred from their mineralogy would be an artefact of the originalmagma, and thereforea "red herring"that is not relevant Anderson,D.J., and D.H. Lindsley, New (and final!) models for Ti- to the eruptionmechanism of the lavas. magnetite/ilmenitegeothermometer and oxygenbarometer (abstract), On the otherhand, the youngdacitic lavas may heraldbirth of Eos Trans. AGU, 66, 416, 1985. a "new" growing silicic magma system beneath the APVC, Bacon, C.R., Magmatic inclusionsin silicic and intermediatevolcanic producedin responseto the melting of upper crustalmaterial rocks,J. Geophys.Res., 91, 6091-6112, 1986. (probablythe plutonicequivalents to the ignimbrites)by mantle- Blake, S., Volcanism and the dynamics of open magma chambers, Nature, 289, 783-785, 1981. derived mafic magmas.Either of these scenarioswould explain Blake, S., Viscoplasticmodels of lava domes,in Lava Flows and Domes: why essentiallysimilar magmaserupted explosively over a period EraplacementMechanism and Hazard Implications, IAVCEI Proc.in of 10 m.y. ashuge volumes of ignimbriteduring the early history Volcanol.vol. 2, editedby J. Fink,.pp. 88-126, Springer-Verlag, New of the magmasystem, followed by small, quiet effusionsin this York, 1990. last episodewe are describing. Bonnichsen,B., and D.F., Kauffman, Physicalfeatures of rhyolitelava It is not clear which of these is more likely. The interval flows of the SnakeRiver Plainvolcanic province, southwestern Idaho, between the last major ignimbrite eruption and the eruptionof inThe Eraplacementof Domes and Silicic Lava Flows, edited by J. Chao is about 1 m.y., and volatile-depleted magma would Fink, Spec.Pap. Geol. Soc.Am. 212, 119-145, 1987. probably have solidified during this time. In this light, the Davidson,J.P., S.L. de Silva, P. Holden,and A.N. Halliday, Small-scale remelting scenariobecomes more appealingbut is problematic disequilibriumin a magmaticinclusion and its more silicic host. J. Geophys.Res., 95, 17,661-17,675.1990. becauseit is difficult to seehow phenocrystssuch as hornblende de Silva, S.L., Large-VolumeExplosive Silicic Volcanismin the Central are preserved without resorption. Further work is needed to Andes,Unpub. Ph.D. thesis,304 pp. OpenUniversity, Milton Keynes, resolve these issues,but in either case, Chao and its neighbors England,1987 may define the extent of the presentlyactive magmaticsystem de Silva, S.L., Altiplano-Punavolcanic complex of the CentralAndes, and, in the latter case, may indicate the location of a future Geology, 17, 1102-1106, 1989a. caldera. de Silva, S.L., Geochronologyand stratigraphyof the ignimbritesfrom the 21ø30'Sto the 23ø30'Sportion of theCentral Andes of N. Chile,J. Summary and Conclusions Volcanol. Geotherm. Res., 37, 93-131, 1989b. de Silva, S.L., andP.W. Francis,Volcanoes ofthe CentralAndes, 213 pp. Chao is an exceptionally large volume, compositecoulre Springer-Verlag,New York, 1991. composedof homogeneous,crystal-rich, dacite. Its eruption de Silva , S.L., S. Self, and P.W. Francis, The Chao dacite revisited proceededin threemain phases;a shortinitial explosivephase (abstract)Eos Trans. AGU, 69, 1487-1488, 1988. was followed by two dominantlyeffusive phases during which de Silva, S.L., J.P. Davidson, I.C. Croudace, and A. Escobar, over 90% of the magmawas erupted.Chao is similarto other Volcanologicand petrologicevolution of volcanTata Sabaya,S.W. silicic lava bodies in rheologic and eruptive parameters. Bolivia, J. Volcanol. Geotherm.Res., 53, 305-335, 1993. Inclusionsin thelater erupted portions of thelava suggest that its Dragoni, M., and A. Tallarico, Effect of crystallizationon the rheology eruptionwas triggered by intrusionof mafic magmainto a and dynamicsof lava flows, J. Volcanol. Geotherm.Res., 59, 441- 452, 1994. homogeneouscrystal-rich dacite magma body. Chao is thoughtto Eichelberger, J.C., Vesiculation of mafic magma during the havebeen emplaced over a periodof about100 to 150 years,with replenishmentof silicicmagma reservoirs, Nature, 288, 446-450, maximum effusion rates of about 25 m 3 s4. The local 1978. topographicslope led to the formationof large coulresrather Eichelberger,J.C., and H.R. Westrich, Magmaticvolatiles in explosive than small domes.Petrologic and physicalsimilarities between rhyoliticeruptions. Geophys. Res. Lett., 8, 757-760, 1981. Chao andneighboring young domes suggest they form part of the Eichelberger,J.C., C.R. Cardgan, H.R. Westrich,and R.H. Price,Non- samemagmatic episode. Their generalfamily resemblancewith explosivesilicic volcanism, Nature, 323, 598-60, 1986. DE SILVA ET AL.: EFFUSIVE SILICIC VOLCANISM 17,825

Fink,J.H., Surfacefolding and viscosity of rhyoliteflows, Geology, 8, ' Naney,M.T., Phaseequilibria of rock-formingferromagnesian silicates 250-254, 1980. in graniticsystems, Am. J. Sci.,283, 993-1033,1983. Fink, J.H., Structureand eraplacementof a rhyolificobsidian flow: Litfie O'Callaghan,L. J., and P.W. Francis,Volcanological and petrological GlassMountain, Medicine Lake Highland,Northern California, Geol. evolution of San Pedro volcano, ProvinciaE1 Loa, north Chile, J. Soc. Am. Bull., 94, 362-380, 1983. Geol. Soc. London., 143, 275-286, 1986. Fink, J.H., Geometryof silicic dikesbeneath the Inyo domes,California, Pinkenon, H., and R.J. Stevenson, Methods of determining the J. Geophys.Res., 90, 11127-11133,1985. theologicalproperties of magmasat sub-liquidustemperatures, J. Fink, J.H., and R.W. Griffiths, A laboratory analog study of the Volcanol. Geotherm.Res., 53, 47-66, 1992. morphologyof lava flows extrudedfrom line and point sources.J. Roobol,M.J., P.W. Francis,W.I. Ridley, M. Rhodes,and G.P.L. Walker, Volcanol. Geotherm. Res., 54, 19-32, 1992. Physico-chemicalcharacters of the Andeanvolcanic chain, between Fink, J.H., S.W. Anderson,and C.R. Manley, Textural constraintson latitudes 21ø and 22 ø south, paper presented at Symposium silicic volcanism:beyond the permeablefoam model. J. Geophys. Intemacionalde Volcanologia,Int. Assoc.of Volcanol.and Chem,of Res., 97, 9073-9083, 1992. Earth'sInter., Santiago, Chile, Sept. 9-14th, 1974. Friedman,I., Are extrusiverhyolites produced from permeablefoam Rose, W.I., Volcanic activity at Santiaguito, 1976-1984, in The eruptions?Bull. Volcanol., 51, 69-71, 1989. Emplacementof Domesand Silicic Lava Flows, editedby J. Fink, Griffiths,R., andJ. Fink, Effectsof surfacecooling on the spreadingof Spec.Pap. Geol.Soc. Am., 212, 17-28, 1987. lava flows and domes.J. Fluid Mech..,252,667-702, 1993. Sparks,R.S.J., H. Sigurdsson,and L. Wilson., Magma mixing as a GuestJ.E., and J. Sanchez, A large daciticlava flow in NorthernChile, mechanismfor triggeringacid explosive volcanic eruptions, Nature, Bull. Volcanol., 33,778-790, 1969. 267,315-318, 1977. Heiken, G., and J.C. Eichelberger,Eruptions at Chao Crags, Lassen Spencer,K.J., andD.H. Lindsley, A solutionmodel for coexistingiron- Volcanic National Park, California, J. Volcanol. Geotherm. Res., 7, titanium oxides, Am. Mineral., 66, 1189-1201, 1981. 443-481, 1980. Stasiuk,M.V., C. Jaupan,and R.S.J. Sparks, On thevariation of flow rate Henry,C.D., andJ.A. Worff, Distinguishingstrongly rheomorphic tuffs in non-explosivelava eruptions, Earth Planet. Sci. Lett. 114, 505-516, from extensive silicic lavas, Bull. Volcanol., 54, 171-186, 1992. 1993. Henry, C.D., J.G. Price, J.N. Rubin, D.F. Parker, J.A. Wolff, S. Self, Stormer,J.C., The effects of recalculationon estimatesof temperature R.R. Franklin,and D.S. Barker, Widespreadlava4ike volcanicrocks andfugacity from analysesof multicomponentiron-titanium oxides, of theTrans-Pecos Texas, Geology, 16, 509-512, 1988. Am. Mineral., 68, 586-595, 1983. Hildreth, E.W. Gradientsin silicic magmachambers: Implications for Swanson, D.A., D. Dzurisin, R.T. Holcomb, E.Y. Iwatsubo, J.J. lithosphericmagmafism, J. Geophys.Res., 86, 10,153-10,192,1981. Chadwick,Jr., T.J. Casaderail, J.W. Ewen, and C.C. Heliker, Growth Hollingworth,S.E., andJ.E. Guest, Pleistoceneglaciafion in the Atacama of the lava domeat Mount St Helens,Washington, (USA), 1981-1983. desert,Northern Chile, J. Glaciol., 6, 749-751, 1967. in The Eraplacementof Domesand Silicic Lava Flows, editedby J. Hulme, G., and G. Fielder, Effusionrates and theologyof lunar lavas. Fink, Spec.Pap. Geol. Soc.Am. 212, 119-145,1987. Philos.Trans. R. Soc.London., A, 285, p. 227-234, 1977. Thordarsson,T., and S. Serf, The Laki (Skaftgr Fires) and Grimsvotn Huppen,H.E., J.B. Shepherd,H. Sigurdsson,and R.S.J. Sparks, On lava eruptionsin 1783-85,Bull. Volcanol.,55,233-264, 1993. dome growth with application to the 1979 lava extrusion of the Wadge,G., Thevariation of magmadsicharge during basalticc eruptions, Soufriere of St. Vincent, J. Volcanol. Geotherm. Res., 14, 199-222, J. Volcanol. Geotherm.Res., 11,139-168, 1981. 1982. Walker, G.P.L., Lengthsof lava flows, Philos.Trans. R. Soc.London, A, Jaupan,C., andC. Allegre,Gas content, eruption rate, and instabilities of 274,484-495, 1973. eruptionregime in silicic volcanoes,Earth Planet. Sci. Lett., 102, Wilson, L., andJ.W. Head, A comparisonof volcaniceruption processes 413-429, 1991. on Earth, Moon, , Io and Venus. Nature, 302, 663-669, 1983. Johnson,M., andM. Rutherford,Experimentally determined conditions W6mer, G., S. Moorbath,J. Entemann,L. Lopez-Escobar,S. Horn, R.S. in the Fish CanyonTuff, Colorado,magma chamber, J. Petrol., 30, Harmon, and J.P. Davidson, VariacionesGeoquimicas, locales y 711 - 737, 1989. regionales,en el frentevolcanico cuatemario de los AndesCentrales Kilbum, C.J., andR.M.C. Lopes, Generalpatterns of flow field growth: (17ø 30'-22ø00'S),none de Chile,Rev. Geol. Chile, 19, 37-59, 1992. aa andblocky lavas, J. Geophys.Res., 96, 19,712-19,732,1991. Manley, C.R., Extendedcooling and viscousflow of large, hot rhyolite S.L. de Silva, Departmentof Geographyand Geology, Indiana State lavas: Implications of numericalmodelling results.J. Volcanol. University,Terre Haute,IN 47089. (e-mail:gesilva@scifac. indstate. edu) Geotherrn.Res., 53, 27-46, 1992. R.E. Drake, BerkeleyGeochronology Center, Berkeley, CA 94720 Marinoviq, N., and A. Lahsen, Hoja Calama. Region de , P.W. Francis,Department of EarthSciences, Open University, Milton scale 1:250,000 cana geologicade Chile 58, Serv.Nac. Geol. Min., Keynes,MK7 6AA, England.(e-mail: [email protected]) Santiago,1984. C. RamirezR, ServicioNacional de Geologiay Mineria, Casilla2032, Marsh,B.D., On the crystallinity,probability of occurrenceand rheology Santiago,Chile. of lava andmagma, Contrib. Mineral. Petrol., 78, 85-98, 1981. S. Self, Hawaii Centerfor Volcanology,Department of Geologyand McBimey, A.R., and T. Murase, Rheologicalproperties of magmas, Geophysics,SOEST, University of Hawaii,Honolulu, HI 96822.(e-mail: Annu. Rev. Earth Planet. Sci., 12, 337-357, 1984. [email protected]) Moore, H.J., D.W.G. Arthur, and G.G. Schaber,Yield strengthsof flows on the Earth, Mars, and Moon, Proc. Lunar Planet. Sci. Conf.,9th, (ReceivedJune 14, 1993;revised February 25, 1994; 3351-3378, 1978. acceptedMarch 7, 1994.)