Lava and Ice Interaction at Stratovolcanoes: Use of Characteristic Features to Determine Past Glacial Extents and Future Volcanic Hazards

Portland State University PDXScholar

Geology Faculty Publications and Presentations Geology

10-10-2000 Lava and Ice Interaction at Stratovolcanoes: Use of Characteristic Features to Determine Past Glacial Extents and Future Volcanic Hazards

David T. Lescinsky Arizona State University

Jonathan H. Fink Portland State University, [email protected]

Follow this and additional works at: https://pdxscholar.library.pdx.edu/geology_fac

Part of the Geology Commons, and the Volcanology Commons Let us know how access to this document benefits ou.y

Citation Details Lescinsky, D. T., & Fink, J. H. (2000). Lava and ice interaction at stratovolcanoes: use of characteristic features to determine past glacial extents and future volcanic hazards. Journal of Geophysical Research: Solid Earth (1978–2012), 105(B10), 23711-23726.

This Article is brought to you for free and open access. It has been accepted for inclusion in Geology Faculty Publications and Presentations by an authorized administrator of PDXScholar. Please contact us if we can make this document more accessible: [email protected]. JOURNAL OF GEOPHYSICALRESEARCH, VOL. 105,NO. B10,PAGES 23,711-23,726, OCTOBER 10, 2000

Lava and ice interaction at stratovolcanoes: Use of characteristic features to determine past glacialextents and future volcanichazards DavidT. LescinskyI and Jonathan H. Fink Departmentof Geology,Arizona State University, Tempe

Abstract. Structuresresulting from lava and ice interactionare common at glaciated stratovolcanoes.During summit eruptions at stratovolcanoes,meltwater is producedand travelsfreely down steep slopes and thin permeable valley glaciers, eroding the ice and enlargingpreexisting glacial drainages. As a result,eruptions in thisenvironment have producedfew catastrophicfloods. Lava flowing into the open channels and voids in the glaciersbecomes confined and grows thicker, filling the availablespace and producing steep- sidedbodies with smooth,bulbous contact surfaces. Quenching of lava againstice or by waterforms small-scale features such as tensional fractures and glass. As the amountof meltwaterin contactwith the lava increases,the type and abundance of smaller-scalefeatures becomesimilar to thoseproduced during subglacial eruptions into meltwaterlakes. Identificationof large-and small-scale lava-ice contact features in the field canbe usedto reconstructpaleoglacial extent and, combined with geochronologyof lavas,to determinepast paleoclimate.An understandingof lava-iceinteraction allows us to betterassess the hazards posedby futureeruptions at glaciatedvolcanoes.

1. Introduction In this studywe examinelava-ice interaction as it relatesto Suddenmelting of glaciersduring volcanic eruptionscan stratovolcanoeslike Mount Rainierthat havealpine glaciers producelarge debrisflows and floodsthat placeneighboring (ice thicknesses<150 m). First,we synthesizeobservations of populationsat risk. Mount Rainier,Washington is a glaciated historicaleruptions to identify the dominantprocesses and hazards of lava-ice interaction on stratovolcanoes. We then stratovolcanothat is closeto the greaterTacoma-Seattle area (2,600,000 residents)and has a high probabilityof a future presentfield studiesof older lava flows, characterizing eruption [National Research Council, 1994]. Due to the fractureand landformmorphologies produced during lava-ice hazards it poses, this volcano was designateda "Decade interaction.These features are then used,along with insight Volcano" by InternationalAssociation of Volcanologyand gained from eruption observations,to reconstruct two Chemistryof the Earth's Interior(IAVCEI) duringthe 1990s examplesof eruptionswith lava-iceinteraction and to predict -the United Nations' International Decade for Natural theprobable course and hazards of a futurelava flow eruption at Mount Rainier. Hazard Reduction [National Research Council, 1994]. Mount Rainier has producedlava flows and small explosive events[Sisson, 1995; Sissonand Lanphere,1997]. Thus it is 2. Eruption Observations likely that lava will erupt onto one of the summit's many Much of our knowledgeof lava-iceinteraction comes from glaciersin the future. observationsof historicaleruptions and thermal events at Surficial lava flows at glaciatedvolcanoes, unlike other glaciatedvolcanoes. Below we synthesizeobservations from styles of eruption, rarely initiate debris flows or floods 18 eruptionsto highlight the important processesand because the rate of meltwater production is low and a recurring phenomenaduring lava-ice interaction(detailed mechanismfor meltwater ponding is often lacking [Major summariescan be foundin AppendicesA andB of Lescinsky and Newhall, 1989]. At stratovolcanoes,meltwater can drain [1999]). Theseevents occurred primarily at stratovolcanoes throughrelatively thin alpineglaciers (up to 100-150 m thick) that were dominantlymafic. becausethe ice is permeable[Smellie, 2000]. In contrast, Most lava flows onto glaciatedvolcanoes follow a brief largevolumes of water may be trappedby thick ice capsand period of increasedheat output and/or explosiveactivity. ice sheets(>>150 m thick) that have an impermeablebasal Conductiveheat flow and fumarolicactivity may cause layer. As a result, lava eruptionsunder thick layers of ice gradualmelting and glacialsubsidence over periods of weeks oftenproduce outburst floods (jOkulhlaups). to years (1960s-1970s Mount Wrangell, Alaska, and 1975 Mount Baker, Washington[Easterbrook, 1975; Bensonand lNow at Departmentof EarthSciences, University of Western Motyka, 1978]). With the start of explosiveactivity, hot Ontario, London, Canada pyroclasticdebris and spattermay quicklymelt snowand ice (both referredto as "ice" below) [e.g., Walder, 1999] or Copyright2000 by the American Geophysical Union. shatterice and increasepermeability. Floods have been Papernumber 2000JB900214. producedduring this earlyphase of eruption(1971 Villarrica, 0148-0227/00/2000JB900214509.00 Chile, and 1991 Mount Hudson, Chile [Gonzdlez-Ferran,

23,711 23,712 LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES

associatedwith thick ice (1991 Hudson and 1983-1984 and 1993-1994 Veniaminof [Yountet al., 1985; McClelland et al., 1989; Naranjo et al., 1993; SmithsonianInstitution, 1993, 1994b]). Lavas erupted into this pondedmeltwater likely behaveas submarinelavas [e.g., Batiza and White,2000]. When lava is eruptedperiglacially (in proximityto, but not in contactwith, ice) and intersectsa glacier,it will eitherflow onto the glacial surface(1983, 1993, and 1994 Kliuchevskoi [Khrenov et al., 1988; Vinogradovand Murav'ev, 1988; Smithsonian Institution, 1994a; Belousov, 1996; Zharinov et al., 1996] or be deflectedand flow alongthe glaciermargins (1945 Okmok, Alaska [Byerset al., 1947]). Lava traveling over ice causeslimited and gradual melting that resultsin localized subsidence of the flow (1983 Kliuchevskoi [Vinogradov and Murav'ev, 1988]). Such lava flows are proneto instabilityand collapse,particularly on steepslopes FigureI. Lavaflow visible in a •50-m-deeptrench melted and during periods of increased effusion rate and basal throughsummit glacier on L!aima volcano, Chile, during its melting (1984-1985 Villarrica and 1994 Kliuchevskoi [ May1994 eruption. Main summit of volcanois outsideof McClelland et al., 1989; Smithsonian Institution, 1994a; phototo the left. Belousov,1996]). Lava flow collapsescan producevolcanic "mixed" avalanchesof broken rock and ice, which scour the glacierand can form floodsand/or debris flows [Piersonand danda, 1994]. 1973; Naranjo et al., 1993]). Pyroclasticflows that travel over glaciers may abrade and melt ice and/or trigger avalanches,resulting in largedebris flows [e.g., Waitt, 1989; 3. Lava Flow Structure and Morphology Pierson et al., 1990; Trabant et al., 1994; Meyer and Lava flow featuresformed during historicaleruptions of Trabant, !995]. glaciatedvolcanoes are difficult to study due to their rapid As ice melts, water not convertedto steamtravels freely burial by ice. Therefore studiesof lava flow structureand down slopethrough the existingdrainage system. Flowing morphologywere conductedin the North AmericanCascades water erodesice mechanicallyby scouringand thermallyby Range and at Mount Redoubt, Alaska, where glacial retreat friction. Frictional erosion is accentuatedon steep slopes has exposed lava flows that have interacted with ice. where flow velocitiesare high and may accountfor a large Mathews [1951] first proposedthat glassylava marginswith portion of channelerosion into glaciers(50 to 75% of the total erosionfor water near 0øC [Marston, 1983]). Ratesof subhorizontalpolygonal joints or "columns"indicate lava quenching against glacial ice. We have used these basic erosion are greatly increasedfor flows of warm meltwater criteria and knowledge of previous glacial distributionsto associatedwith volcanicactivity. I/inogradovand Murav'ev [1988]observed glacial incision rates of 275-550x10-6 m/s identify a wide varietyof lava structuresthat can be attributed to lava-ice interaction. (1-2 m/h) by the 50øC meltwater runoff during the 1983 K!iuchevskoieruption. In sum,the flow of warm meltwater 3.1. Glassy Margins and Fracture Types on steepstratovolcanoes may rapidly enlargeglacial tunnels and/orCUt trenches through a glacierto bedrock. Lava flows that have cooled next to ice typically have a Lava flows follow the newly modified glacial topography glassyouter shell around a crystallineflow interior(Figure 2) and are frequentlyconstrained by ice, either within trenches [Lescinskyand Sisson,1998]. Fracturemorphologies found (1983-1984 Veniaminof, Alaska [Yount et al., 1985; Reeder within the glassyouter zone are classifiedin this paperusing and Doukas, 1994]) or subglacialtunnels (1991 Hudsonand the following terms: polygonal,sheet-like, pseudopillow, 1994 Llaima eruptions[Naranjo et al., 1993; Moreno and hackly,and shard-forming(Table I and Figures3 and4). The Fuentealba, 1994]). Undercutting and collapse of the fracture morphologygroups are somewhatgradational and confiningice walls widen trenches,open tunnels upward, and increasein complexity. As the spacingbetween fractures may temporarily dam meltwater (1983 K!iuchevskoi; 1991 becomessmaller, the fractures become less planar and the Westdah!, Alaska; 1991 Hudson and 1994 L!aima, Chile; propagationdirections become more variable. These small- Figure 1 [l/inogradov and Murav'ev, 1988; Smithsonian scale fracture features, together with larger-scalelandform Institution, 1991; Naranjo et al., 1993; Moreno and morphology,may be usedto identifylava-ice interaction; thus Fuentealba, 1994; Reederand Doukas, 1994]). This process we describe each in detail below. may resultin suddenmeltwater production and accompanying Polygonal fractures (columnar joints) intersectto form phreatic explosions (1983 Kliuchevskoi, Kamchatka polygons(Figure 3a). They occur at the flow margins, [ l/inogradovand Murav'ev, 1988; I/inogradovet al., 1990]). surface, and base of lavas that have interacted with ice Intensesteaming occurs when moltenlava comesinto contact (Figures 2, 4a, 4b, and 5b). Along the flow marginsand with ice and substantialrunoff is produced,particularly when surface,the fracturesform narrow columns(4-8 cm diameter effusion rates are high (1983 K!iuchevskoiand 1984-1985 for rhyolitic-rhyodacitelavas, 5-10 cm diameterfor andesitic- Vi!!arrica [Khrenovet al., 1988; McClelland et al., 1989]). dacitic lavas, and up to 20 cm diameter for basaltic lavas). Substantial accumulations of meltwater have occurred Fractureorientations are perpendicularto the coolingsurface. within the topographicconfines of large cratersor calderas In some cases, column diameters increase into the flow as LESCINSKYAND FINK: LAVA AND ICE INTERACTIONAT STRATOVOLCANOES 23,713

.o,ß ...... &•-...:•'... ß .::.:...... •?,..•...';;•'-,:':...... •::.•.;;*-'!:-';::'•'•:'""?-:':'. '.'.".;....i'; ;;..;,:i-.;.;.';;;.i;.:..'•;;i•?•. ?•:!-.. •"'• ß •"•'•'•••••.•••4•--•:.•:.:,,..•..... ß...... :...... •-.- ..... "•-: .... -'• ...... " .•'--,';.:•:.:::':;..•.:'::.'...... '...? ......

. : ......

..... ;;;;;•';::,:•,(*:'"'":*';"...... '?"" . , . , ., ...... ::-:;::i...... --W".. .:. :;;2';" • "' ' .....-,....:•:.. '.::::-.., ::-'-';'-:.*':'...... "-."..... '"'ß '..... -, ...... :' '" . • • •. o.r...... :..:-"'"' --- . .::.::,:;.•;;..:.;..:;;•,:.. _ ...... • :.. ?:*.::•..:7•*::.7"::" ß--- ii ...... ' ...... ¾.

'•• •:-.•:..... •.,,. . ... - •...... , .....ß • ,.•• ...... •'z?'.L,•.y*.•...... •...• .•'.:•..•.•,..,'..•.•.:,. • ...... '..,•.,,.'.,...•.• ...... ,.•.:'•' •..,•-.• .... , '...... • 2'...... • ....•.- ... ß ,...... :.'•,.•.,:5'.. ,....:.,.-•..-.....:.,•.•:.,:.•:.-.. •'-*• . ':.." ...,.'...... • ' ..•...... • 'broad.i.n:ieriOr...... '"'7,'...... :...... •':.'"" .,...-:•?•.::::•::"?:':.•*"•'

...... "' --:-' ...... ':"...."...... ß...... :...... •...,.:..:.•:.:•-•.::::•.•..:...."" ...... •??'7•' ...... 7%:.:-:•:- Figure 2. Cross-sectionalview of Mazama Ridge flow below the old Paradiseice caves, Mount Rainier, Washington.Polygonal joints (columns) are visible in the glassyflow marginat rightand along the flow base. In the crystallineflow interior,platy fractures are presentin a zoneof broad,poorlydefined columns and in a zoneof moremassive lava above this. The lavaflow surfacehas been glacially eroded with theexception of a portionof the rightside of the flow wherea glassyflow surfaceis preserved.The flow is-200 m acrossand 35 m highin this photo. much as 3 times over distancesof <50 cm by a processof changeover a shortdistance (-50 cm). Polygonalfracture fracturetermination (Figure 4b). Verticalcolumns at theflow patternscan be quite complicated,including a varietyof fan- baseare commonly broader (30-50 cm diametersfor andesite like and curving arrangementsin the resultant columns. to rhyoliteand 40-60 cm diametersfor basalt). At the Bandsfound on the fracturesurfaces (Figure 3a) areproduced intersectionof marginaland basal zones, fracture orientations by incrementalfracture advance[e.g., Ryan and Sammis,

Table 1. Descriptionof FracturesFound in LavasThat HaveInteracted With Ice

Fracture Type Description Inferred Origin

GlassyFlow Marg#7s Shards Small fragments of glassy lava, commonly Rapid cooling and fracturing (granulation)of altered to palagonite;found disaggregatedas lava in contact with water, in some cases hyaloclastitesin associationwith lava pillows followedby spallation. or pillow fragments. Hackly fractures Chaotic, arcuate fractures with a range of Rapid cooling and fracture sometimes orientations and crosscuttingrelationships; "explosively" steam migration disrupting may havepalagonite along fractures. isotherms. Pseudopillowfractures and Arcuate fractures with associated small, Cooling by water/steampenetration into rock secondaryfractures perpendicular,secondary fractures; palagonite with relativelywell establishedisotherms that alongfracture. become locally disrupted resulting in secondaryfractures. Sheet-like fractures Parallel,planar fractures extend beyond the sides Cooling with well-establishedisotherms with of an individual column, in some places extension of wide fractures with well- reachinglengths of more than 5 m; may be developedorthogonal cross cutting fractures. palagonitealong fracture. Polygonalfractures, Adjacent fracturesintersect to define regular Cooling with well-establishedisotherms with columnarjoints polygons. Normally parallel, but may fan or isotropicstress field; little or no fluid flow. bend.

CrystallineFlow h•terior Platyfractures, sheeting Thesefractures form overlapping plates that are Since these interior fracturesparallel flow joints oftenclosely spaced, 0.5-4 cm apart,have an direction and occurred after the lava had average breadth of about 80 cm, and break substantiallycrystallized, they are probably around phenocrysts. Fracturestend to be related to late stage shear of the lava flow parallel to the flow base, exceptnear the [Bonnichsenand Kauffmann,1987] and/or boundarywith the carapace,where they are microliteorientation [Walker, 1993]. orientedparallel to theflow margin. Broadpolygonal fractures, Broad(>1 m diameter)polygonal (four or five- Late stage cooling and contraction,somewhat megacolumns sided)joint setscutting through flow interior. irregulardue to preexistingplaty fractures;in Fracturesbreak around phenocrysts. some cases,one face will have formed earlier, priorto developmentof platyfractures. 23,714 LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES

primary fracture bases. Fracture spacingranges significantly (from tens of propagationdirection(a-d) I centimetersto <1 cm) over short distances(<1 m). Bands on a Polygonal fractures b Sheet-like fractures fracture surfaces are arcuate, and their widths also vary greatly (Figure 3d). Mud and/or palagoniteare commonly found along hackly fracturesurfaces. ..::..,.'• Small shardsof glass(millimeters in scale) are formed by ,::½•!{'"•i:•: primaryfractures'--'•• the intersection of large numbersof closely spacedfractures (Figure 4h). The glass shards are commonly altered to fr ter 'rnlnations'• fractureband palagonite and are found as disaggregatedsediments (hyaloclastites)in close associationwith lava pillows. Glass secondaryfractures primaryfractures shardsand hyaloclastitesare rare at the lava-ice interaction localities that we studied. Two additional types of fractures (platy and broad polygonalfractures) occur in the crystallineflow interiorand, in some cases,extend 10-20 cm into the glassyflow margin. a•' secondary fracture P!aty fracturestend to be orientedparallel to the flow base, propagation direction (c only) exceptwhere they approachand becomeoriented parallel to c Pseudopillow fractures d Hackly fractures the flow margins. These fracturesform overlappingplates that are often closelyspaced, 0.5-4 cm apart, have an average Figure3. Sketchesof fracturemorphologies. (a) Polygonal fractures. (b) Sheet-like fractures. (c) Pseudopillow width of about 80 cm, and break aroundphenocrysts, unlike fractures.(d) Hacklyfractures. White surfaces correspond to fractures in the glassy zone that propagate through cooling fracture faces, with gray lines representing phenocrysts.Broad polygonal fractures (>50 cm in diameter) incremental fracture terminations. The surface between two form poorly defined megacolumnsthat are commonin flow of these fracture terminations is called a "band" or "striae". interiorsand may crosscutplaty fractures (Figure 2). Heavyblack lines represent cooling fractures. Lighter black lines representincipient cooling fractures: in (a) theseare 3.2. Large-Scale Lava Flow Morphology poorlydeveloped primary fractures; in (c) theseare secondary fractures.Propagation direction for all primaryfractures (a-d) Landforms produced by lava-ice interaction at is towardthe bottomof the figure. Propagationdirections for stratovolcanoesvary significantlyin their morphologyand secondaryfractures in (c) areinto and out of thefigure. size. They can be separatedinto three groupsof landforms: (group 1) glassymargins with polygonalfractures; (group 2) glassymargins and continuousfractures; and (group3) glassy marginswith pseudopillowfractures (Figure 5). 1978; DeGraff and Aydin, 1987; Aydin and DeGraff, 1988; 3.2.1. Group I landforms: Glassy margins with DeGraffand Aydin, 1993] and rangefrom severalmillimeters polygonal fractures. Steep-walledridge flows, smallerlava to severalcentimeters in width. In general,band width varies flows and one varietyof domehave steepglassy flow margins with column diameter; columns 4 cm in diameter have band with numerouspolygonal fractures on the mm'gins(Figures widths of 0.3-1 cm and columns 30 cm in diameter have band 5a-5c). Lava flows formingsteep-walled ridges (Table 2) can widths of 2.5-3 cm. be up to 450 m thick and extend up to 25 km from the Sheet-likefractures have parallel setsof long fractures(55 volcano's summit. Lessvoluminous, thinner lava flows (-20- m long) that are separatedby up to l0 cm. Perpendicular 30 m thick) include severalthinner lobes of the ridge flows. fractures connect adjacent long fractures producing Domes are even smaller (down to 20 m in diameter) and rectangular columns (Figures 3b and 4c). Bands extend include discrete bulges (= 4 m high and • 8 m wide) continuouslyalong the length of long fractures and have connectedby thinner fingers of lava. These flows ranged widths similar to those observedin polygonalfractures. In from andesitcto rhyolitein composition. some cases, bands on the shorter connectingfractures are The width of the glassymargins ranges up to 5 m in a few wider and more arcuate than those on the adjacent long thick lava flows (>100 m thick) and to <2 m in thinner flows fractures(Figure 3c). and domes(<50 m thick). Flow frontsof severalof the ridge Pseudopillowfractures are arcuatefractures that extendfor flows have glassy zones that extend up to 12 m inward several meters and commonly occur in roughly subparallel [Lescinskyand Sisson, 1998, Figure 2]. As flow thickness setswith some intersections(Figures 3c, 4d and 4e). Short decreases, glassy flow surfaces become more common; fractures extend perpendicularto, but rarely connect, the almostall flows or flow lobesof_<10 m thicknesshave glassy pseudopillowfractures. Bands on the shorterfractures are flow surfaces. Where preserved,the upper portions and orientedparallel to the pseudopillowfractures, indicating that surfacesof lava flows >30 m thick are crystalline,oxidized, theseare secondaryfractures that propagatedaway from the and vesicular. Flow basestend to be somewhatless glassy pseudopillowfractures (Figure 3c) [Watanabe and Katsui, than flow margins, and basal breccias are either thin or 1976]. Mud and/or alteredglass (palagonite) is commonly absent. found in pseudopillowfractures but not in the shorter Lava flows forming steep-walledridges have 2-5 m thick secondaryfractures. basal zones of glassy, broad (30-40 cm in diameter)and Hackly fracturesintersect chaotically and form irregular vertically oriented columns (Figures 2 and 5a). Basal angularblocks or "splinters"of rock (Figures3d, 4f, and4g). columns in smaller flows and domes are not significantly The angularblocks and splintersgenerally trend away from broader(15-20 cm in diameter)than the associatedmarginal the flow margin or radiatefrom small voids at the lava flow columns(-10 cm in diameter). Hackly, pseudopillow,and LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES 23,715

...... :

:!•.i.•.:.:;-•...: ...... ?. :%:

Figure 4. Photographsof fracturetypes produced during lava-ice interaction.(a) and (b) Polygonalfractures near Yakima Point, BurroughsMountain flow, Mount Rainier, Washington. The columnsformed by the fracturesin Figure4a are quite regularand have an averagediameter of•l 0 cm. Figure4b was taken severalmeters away and representsa point 30-50 cm farther inward from the outer flow margin. Note that the columnsare much broader (20-30 cm in diameter),are less regular,and show the terminationof fracturesthat form narrowercolumns. (c) Sheet-likefractures trending diagonally across the picturefrom KokostickButte, South Sister,Oregon. Smaller cross-cuttingfractures are alsovisible. (d) Pseudopillowfracture at the centerof the photowith an irregulararcuate trend and secondaryfractures extending perpendicular from it. Figure 4d is from the glassyflow surfaceof the Mazama Ridge flow, Mount Rainier,Washington, visible in Figure2. (e) Pseudopillowfractures, left of center,in a pillow lobe in SprayPark, Mount Rainier,Washington. Note the darker,rough outer surface of the pillow lobe on the right side. (f) Hackly fracturesradiating from the right sideof a caveat the baseof a lava flow next to Nisqually Glacier, Mount Rainier, Washington.Presumably, the cavewas producedby the meltingof an ice block that had beencovered by the lava flow. (g) Hackly, intersectingfractures in a GrandeRonde Basalt flow alongthe Tieton River, Washington.(h) Gradationalcontact between disaggregated hyaloclastite on left and shardsin coherentlava on right at Blfihnfikur,Torfaj/3kull, Iceland. Fracturesin Figures4a-4e are viewed perpendicularto the cooling surface,propagation was into the photo. Fracturesin Figures4f and 4g are viewedoblique to the coolingsurface; propagationwas toward the right of the photo and inward. Lens cap in Figures4a, 4d, 4g and 4h is 5.2 cm in diameter. Pen in Figures4b and 4c is 14 cm long. Ice axe in Figure4f is •90 cm long. Field of view in Figure4e is •150 cm. 23,716 LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES

GROUP 1' glassy margins with polygonal fractures GROUP 2: glassy margins & continuous fractures

100- 450 rn l/ttll,,,,l I I

_ 30mI I I I 250-300 m 100-400 m

C 2O m

30 m I I I I 50-100 m 100-200 rn

GROUP 3: glassy margins with pseudopillow fractures

lOO m 4- 6rn I I I I 400-500rn 6-12rn

KEY polygonalfractures_-- withmegacolumns & pseudopillow subaerial • glassymarginwith/ / crystallineplaty fractures flowinterior fractures• flow"'"-'•""•'ßsurface lavae•pillows

Figure5. Idealizedsketches showing the structural cross-sections andfracture distributions indifferent landforms associatedwith lavaand ice interactionat stratovolcanoes.Group 1 landforms:(a) ridge-forminglava flow; (b) smallerlava flow; and (c) polygonally jointed subglacial dome. Group 2 landforms:(d) flat-toppedmountain [after A4athews,1951]; and (f) esker-likelava flow. Group3 landforms:(f) subglacial dome with pseudopillow fractures; and (g) pillow lobe flow.

sheet-likefractures occur in someplaces on the flow surfaces margins(Figures 5a and5f). Polygonalfractures in the lava of thinner lava flows, flow lobes, and small domes and are flows form narrow columnsin an upper"entablature" zone alsofound associated with voidsat flow bases(Figures 4f and and broad columns in a lower "colonnade" zone. In the 6). Hackly fracturesoccur closestto the flow margin or esker-like lava flows the entablature zones are •-80% of the surface,and pseudopillow,sheet-like, and polygonalfractures entire flow thickness and have columns 15-20 cm in diameter; occurprogressively inward toward the flow interior(Figure columns in the colonnade are 40-60 cm in diameter. 6). Also in the smaller flows and domes,there is a zone of Alongthe marginsof theselandforms, column orientations massive glass (roughly 1 m wide) with broad fractures changefrom vertical to subhorizontal,and the flows are between the marginal columns and the crystallineflow locallyglassy (Figures 5d and5e). Severalof the esker-like interior(Figures 5b and 5c). lava flows havediscrete lava pillowsor flattenedellipsoidal The flow interiorsof all three landformtypes are massive, pillow lobes(6-15 m acrossand about a meterthick) along are crystalline,and commonlyhave platy fracturesand poorly theirmargins and scoriaceous flow bases.Hyaloclastites are formed,broad columns (Figures 2 and 5a-5c). Platyfractures absentexcept for thin layersof interstitialhyaloclastites in the flow interiorare orientedconcentric to the shapeof the associatedwith pillows. lava flow at its margins(i.e., nearvertical, Figure 5b) andare 3.2.3. Group 3 !andforms: Glassy margins with absentor poorlydeveloped at the centerof the flow. pseudopillowfractures. Theselandforms have subparallel 3.2.2. Group 2 !andforms: Glassy margins and pseudopillowfractures in the glassymarginal zone and continuous fractures. This group is characterizedby includelava domes(Figure 5f) andpillow lobeflows (Figure polygonal(and rarely hackly) fracturesthat extendthrough 5g andTable 2). The pseudopillowfractures are filled with the entireflow thickness.The groupincludes flat-topped lava brownmud/palagonite and extend inward 3-4 m beforegiving mountains(composed of multiplelava flows; Figure5d) and way to a roughly2 m thick zone of massive,irregularly sinuous"esker-like" lava flows (Figure5e). Theselandforms jointedglass (Figure 5f). In the pillow lobes,pseudopillow are composedof relativelythick lava flows with steepflow fracturesextend up to I m andcommonly have altered bread- LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES 23,717

4. Discussion

4.1. Lava Flow Eraplacementand Morphology The majority of lava flows and landformsstudied are consistent with observations of lava flows in ice trenches. Group I !andformshave glassymargins with polygonal fractures,are commonlythick, and have steepmargins with subhorizontal columns due to flow confinement and quenchingagainst ice walls[Mathews, 1952a; Lescinsky and Sisson,1998]. Flat-toppedmountains (group 2) alsoform by ice confinementwith successivelava flows filling sustained glacialvoids [Mathews, 1951 ]. In addition to evidence of flow confinement, several of the

~lm landformsshow evidenceof substantialamounts of ponded water. Esker-likeflows (group2) havepillows and localized Figure6. Idealizedsketch showing progression of fracture areasof vesiculationat their bases,indicating that theselavas typesfrom a void at the baseof lavaflow to the crystalline traveledthrough shallow water and/or saturatedsediments. flow interior.Different fracture types are separated by dashed Group 3 domeshave abundantpseudopillow fractures that line. Labelsare a, hacklyfractures; b, pseudopillowfractures; form due to water infiltration [Watanabeand Katsui, 1976; c, sheet-likefractures; and d, polygonalfractures. The transitionbetween fracturetypes is gradational. Wider Yamagishi,1991 a, 1991b]. The final !andforms,pillow lobe polygonalfractures are found at the flow baseand irregular, flows (group3), also have pseudopillowfractures but show broadpolygonal fratures and platy fractures are found in the no evidence of confinement. flow interior. We can use the evidence of flow confinement, abundant water, and landformsize to graphicallyseparate the features studied(Figure 7). End-membersshow evidenceof either crust like surfaces that form rinds several centimeters thick lava flow confinementor of emplacementin water. The (Figure4e). Both domesand pillow lobeshave crystalline confined flow end-memberincludes (in order of decreasing and platy jointed flow interiors,although this is less size): flat-toppedmountains (group 2), steep-walledridge pronouncedin the thinnerpillow lobes. flows, smaller lava flows, and domes(group 1). In between

Table 2. Typesand Locationsof Lava-IceInteraction Landforms Studied in the Field

Landform type Examples

Group 1: GlassyMargins WithPolygonal Fractures

Steep-walledridge flows Numerousexamples at MountRainier, Washington, most notably: Emerald Ridge flow (46ø48.23'N,121ø52.08'W); Mazama Ridge flow (46ø47.98'N,121ø42.78'W, area visiblein Figure2; 46ø45.43'N,121ø39.68'W, terminus of flow); and Burroughs Mountainflow (46ø54.43'N,121ø34.62'W); Kulshan Ridge flow Mount Baker, Washington(48ø49.66'N, 121ø41.21 'W); flowat piedmontlobe of Drift RiverGlacier Mount Redoubt,Alaska (60ø32.40'N, 152ø45.90'W); flows at Mount Rainierare andesitesand dacites [Sisson and Lanphere,1997]. Smaller lava flows MoraineLake flow (44ø03.45'N,121ø46.03'W) and KokostickButte (44ø01.38'N, 121ø47.88'W),South Sister, Oregon; Whitewater Glacier lower dacite (44ø42.43'N, 121ø48.04'W),Mount Jefferson, Oregon; several thinner lobes of theridge flows listed above,most notably above Heather Meadows (48ø50.51'N, 121ø41.32'W), Mount Baker,Washington; flows are andesites and dacites [ Taylor et al., 1987;Conrey, 1991]. Domes WikiupPlain (44ø03.09'N, 121ø47.46'W) and west side (44ø03.45'N, 121ø46.03'W), SouthSister, Oregon; Glacier Creek (44ø05.13'N, 121ø49.37'W), North Sister, Oregon; flowsare rhyodacites and rhyolites.

Group2: GlassyMargins and ContinuousFractures

Flat-toppedlava mountains TheTable (49ø53.69'N, 123ø00.83'W), near Garibaldi Lake, British Columbia; flows are andesites[Mathews, 1951 ]. Esker-like lava flows Severalexamples (50ø03.90'N, 123ø07.93'W) in CheakamusRiver valley near Garibaldi Lake, BritishColumbia; flows are basalts [Mathews, 1958].

Group3: GlassyMargins •ith PseudopillowFractures Domes Three small buttes(40ø30.65'N, 121ø13.06'W;40ø30.13'N, 121ø13.27'W;40ø29.13'N, 121ø13.04'W), CaribouWilderness east of Lassen,California; domes are andesitesand basalticandesites [Guffanti et al., 1996]. Pillow-lobe flows SprayPark (46ø55.39'N, 121ø48.67'W), Mount Rainier, Washington; NE of Park Butte (44ø43.96'N, 121ø47.93'W),Mount Jefferson,Oregon; flows are basaltic-andesiteto andesite[Conrey, 1991; Sisson and Lanphere,1997]. 23,718 LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES

Feature Size Scott, 1987]. These are not includedin Figure 7 for this reasonbut may provide evidenceof lava-ice interactionin combination with other observations.

4.2. Fracture Production and Cooling Rate Fracturesin the glassyflow marginsreflect the cooling history of the lava during flow advance and lava-ice ..•oXflow:'/Esker-like\ interaction,rather than posternplacement.These fractures formed incrementallyby thermal contraction,producing the 7 (Gr.1:;'øws(Gr'2' ! •"'%'• bands on their faces (Figure 3) and propagatedinward as the • / Ridge-forming• flows cooled [e.g., Ryan and Sammis,1978; Long and Wood, 1986; DeGraff and Aydin, 1987; Aydin and DeGraff, 1988; Flow /"77 Flat-toppedfiø•s(Gr'l)• sømetuyas?/ tindars • Water DeGraff and Aydin, 1993]. Since fracturing followed the Confinemen/møuntains(Gr.2)/ • ',•Present solidification front, the fractures tend to be oriented / confinedflowt largelyconfined • lotsof ponded water• perpendicularto the closestflow surfaceor margin[e.g., Long littleor noponded water someponded water noconfinement of lava and Wood, 1986; Aydin and DeGraff, 1988]. The abrupt Figure7. Qualitativediagram separating landforms produced transitionin fracturetype and increasein crystallinityat the duringlava and ice interaction,based on landformsize and boundarywith the flow interiorsuggest that the flow margins the presenceof evidencefor flow confinementor subaqueous formed a rigid shell around the still moving flow interior emplacement. End-membersthat show evidenceof flow [Lescinskyand Sisson,1998]. Examplesof fracturesformed confinement(great flow thickness,steep lava flow margins with horizontalfractures) but no evidenceof pondedwater after flow emplacementinclude those in columnarbasalts that (pseudopillowfractures, shards, hyaloclastites,or lava penetratethe majority of the flow thickness,experience only pillows)were likely emplacedin trenchesand voidsin thin gradual change in spacing, and have predominantly glacierson the flanksof stratovolcanoes.End-members that subverticalorientations [e.g., Longand Wood, 1986; DeGraff showevidence of pondedwater but no evidenceof lava flow and Aydin, 1993]. confinement were likely emplaced in englacial lakes. The fractures observed in lava flows that have interacted Intermediatelandforms showing evidence of bothwere likely with ice show an inward progressionin form (Figure 6). At emplacedin small water-filledglacial voids or voidswith flow margins, fractures are irregular and closely spaced shallow water. Landform groups listed (groups 1-3) are (shard-forming and hackly fractures), and toward flow discussed in the text. interiors,fractures are regularand morewidely spaced(sheet- like and polygonalfractures). This progressionin fracture spacingand regularitycorresponds to decreasingcooling rates the two end-members,esker-like flows (group2) and domes inward from the flow margins(Figure 6) [Long and Wood, (group 3) show evidence of both flow confinement and 1986; Reiter et al., 1987; DeGraff et al., 1989; DeGraff and emplacementin water. Aydin, 1993; Budkewitschand Robin, 1994]. Substantial Although examples of lava flows emplaced in water changesin cooling rate result from variations in cooling (englaciallakes) are rarein the Cascades(include only pillow mechanisms. Cooling by radiation and atmospheric lobe flows, group3, andthe tindarsof LoneButte and Crazy convection at a lava flow surface is slower than convective Hills, southernWashington [Hammond, 1987]), they are cooling by water and steam penetrating the flow commonin Iceland,Antarctica, and the Tuya-Teslinregion of [Saemundsson,1970; Hardee, 1980; Long and Wood, 1986; Canada and are included in Figure 7. These landforms, DeGraff and Aydin, 1993]. This differencein cooling rate composedof large quantitiesof hyaloclastites,lava pillows produceszones of regular, polygonal fractures (colonnade and massivelava, reflectthe shapeof the water-filledvoids in zone) and zones of irregular, narrower hackly fractures which they formed due to confinementof the hyaloclastite (entablature) in columnar basalts. Convecting water in beds [e.g., Smellie, 2000], rather than confinementof lava subaqueousenvironments cools lava even more rapidly and flows. At the larger end of the size spectrum,tuyas and causesgranulation and the formationof glass shards[e.g., tindarsform mountainsup to 5 km acrossand 600 m high Batiza and White, 2000]. The relationshipbetween cooling [Mathews, 1947; Bemmelenand Rutten, 1955]. Smaller, rate and fracturespacing/type is presentedgraphically (Figure mound-formingor esker-likebodies may reachmore than 1 8), with the lowest cooling rates associatedwith subaerial km acrossand 200 m high [Walker and Blake, 1966; Smellie emplacement(no glacial ice or meltwater)and the highest and Skilling, 1994]. coolingrates associated with subaqueousemplacement (such Since the developmentof englacial lakes is limited to as within an englaciallake). topographicdepressions or thick, impermeableglaciers, we Fracturespacing is controlledlargely by the fracturedepth can use Figure 7 as a crude discriminatorof ice thickness. (striation width) [Lachenbruch, 1961; Nemat-Nasserand Landformscomposed of confined lavas correspondto thin Oranratnachai,1979], which is a functionof thermalgradient glaciers(<150 m thick), and landformscomposed of lavas [Ryan and Sammis, 1978; DeGraff and Aydin, 1993; emplacedin water correspondto thick glaciers(>>150 m Grossenbacherand McDuffie, 1995]. Large thermal thick). gradients result in narrow striations and closer fractures. Lava flows emplaced on top of glaciers are not well Repeated narrow striations and sustainedhigh thermal represented in the geologic record and appear as gradientsindicate a constantcooling rate at the solidification monolithologicbreccias and abruptlyterminated lava flows front. If the cooling rate at the solidificationfront were to [Mathews,1952b; Bemmelen and Rutten,1955; Wright, 1980; decrease,we would expectto seea changeto smallerthermal LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES 23,7!9

Fracture Spacing Wide ß Narrow Hull andCaddock, 1999]. However,these approaches rely on materialproperties that are poorlycharacterized (i.e., glass transitiontemperature, fracture toughness, Young's modulus, RadiativeCooling&Convective by Atmosphere Poisson'sratio). A promisingapproach would be to use relaxationgeospedometry to measurethe coolingrates of naturalglassy samples from different fracture types [Wilding • •• / convective et al., 1996,2000] andto developan empiricalrelationship. • •i• . Cooling rates for different fracturetypes could also be E• 0-o constrainedusing the relationshipbetween cooling rate and • . • ... lava flow surfacemorphology (Figure 8). Experimental studieshave demonstrated that different surface morphologies (levees,folds, rifts, andpillows) are producedas a functionof / coolingrate, eruptionrate, and lava flow viscosityand/or / yield strength[Fink and Griffiths,1990; Griffithsand Fink,

Low • High 1997; Fink and Griffiths, 1998]. Also, studiesof natural Cooling Rate lavashave found that the sizeof individualpillows increases with increasingviscosity [Fumes et al., 1980;Fridlei•son et Figure8. Qualitativemodel relating fracture type to cooling al., 1982; Walker, 1992]. rate, coolingmechanism, and lava composition.Fracture It is apparentthat lava flows that interactwith ice have spacingincreases, and fracturemorphologies become more regularas flow coolingrates decrease [e.g., Long and Wood, complexand heterogeneouscooling histories. In contrast, 1986; DeGraff and •4ydin,1993], resultingin the sequence modelsof lava flow coolingonly considerthe averageheat fromshards at highcooling rates to polygonalfractures at low flux from an entireflow [e.g., daeger, 1968; Hardee, 1980; coolingrqtes. Coolingrates are sensitiveto bothcooling Dragoni, 1989;Manley, 1992;DeGraff and Aydin, 1993;Hon mechanismand flow composition.Cooling by radiationand et al., 1994; Keszthelyiand Denlinger, 1996]. Relating atmosphericconvection at a lavaflow surface (no shading) is cooling rate and fracturetype would allow constructionof slowerthan convective cooling by steam(light shading) or by more accurate cooling histories for these lava flows. water(dark shading) penetrating the flow [e.g.,Hardee, 1980; Improved cooling histories would make it possible to Longand Wood, 1986]. Coolingof lavais largelya function determineheat fluxes and meltwater productionrates over of the lava'sthermal diffusivity. Sincesilicic magmas have time. This would enablepredictions of the potentialextent of largerthermal diffusivities, they can cool more rapidly mafic magmas[Gregg and Fink, 1996]. Fracturespacing is also floodingduring lava-iceinteraction, as highlightedby Smellie controlledby compositionallydependant material properties [2000]. (Young'smodulus, Poisson's ratio, and thermal expansivity) [Nemat-Nasserand Oranratnachai, 1979] that result in more 4.3. ReconstructingEruption History and closelyspaced fractures in rhyolitesthan in basalts.Analog Paleoenvironments studieshave demonstratedthat differentsurface morphologies We can use the characteristic fractures and landforms (levees,folds, rifts, and pillows) are produced as a functionof producedduring lava and ice interactionas indicatorsof coolingrate, eruptionrate, and lava flow viscosityand/or yieldstrength [Fink and Griffiths,1990; Griffiths and Fink, glacial thickness and location during the eruption. The 1997]. Studiesof naturallavas have found that the sizeof presenceof glassyor crystallineflow surfacescan be usedto individual pillows increaseswith increasingviscosity map out subglacialflow and fracture morphologiescan be [Walker, 1992]. used to indicate where water may have accumulatedand whether it accumulatedin significantamounts (i.e., forming ponds or lakes). By reconstructingpast cases of lava-ice interaction we can determine situations where meltwater gradients,wider striations,and more widely spaced fractures productionand accumulationare most commonand therefore (Figures4a and 4b). better assesswhere it may occur during future eruptionsat Lava compositionalso plays an importantrole. Higher glaciatedvolcanoes. Examplesof such reconstructionsare thermal diffusivities for silicic magmas(relative to mafic presentedbelow. magmas) enable more rapid cooling and result in a 4.3.1. Reconstruction of Mazarna Ridge lava eruption, compositionalvariation between cooling rate and cooling Mount Rainier, Washington. The Mazama Ridge lava at mechanism(Figure 8) [Gregg and Fink, 1996]. Silicic Mount Rainier is a high silica andesite-daciteflow that was magmasalso have lower valuesof Young's modulusand erupted-90 kyr ago [Sissonand Lanphere,1998, availableat Poisson'sratio and higher thermal expansivity, which all tend http://www.geosociety.org/pubs/ft1998.htm]. It extends to enable closer fracture spacing [Nemat-Nasser and down the southernflank of the volcanoto the TatooshRange Oranratnachai,1979]. This relationshipis observedin the and then eastalong the southwall of StevensCanyon, where field with narrowerspacing of polygonaljoints in side it forms several lava benchesthat coincide with valley marginsof rhyolitic lavas (2-6 cm) than in andesiticor confluences(Figure 9a). The abundanceof glassand well- basalticlavas (5-10 cm and>10 cm, respectively). developedfractures along the lava flow marginsand terminus Quantitatively, the relationships between fracture makesthis flow a goodcandidate for reconstructingthe extent spacing/type,cooling rate, and compositionremain poorly of the syneruptiveglaciers. constrained. Analog models of cyclic fracturing have The majority of lava and ice contactfeatures crop out employed desiccation contraction rather than thermal along the Mazama Ridge flow margin exposedin Stevens contraction[Corte and Higashi, 1964;Miiller, 1998a, 1998b; Canyon,indicating that a largeglacier was presentat the time 23,720 LESCINSKYAND FINK: LAVA AND ICE INTERACTIONAT STRATOVOLCANOES

Figure9. Reconstructionof Mazama Ridge flow (stippledarea) and syneruptive glaciers (shaded areas) on the southflank of MountRainier. Contours are in feet. Dashedlines in Figures9e and9f indicatethe extension of lava beneaththe ice (SGL, subglaciallava). (a) Presenttopography [based on U.S.Geological Survey, 1971] and extent of lavaflow. Lavabenches are marked by "I," "II," and"III." An areaof slopinglava near the headof Stevens Canyonis markedby "X." (b) PreeruptiveStevens Canyon glacier as reconstructedusing evidence of lava-ice interactiondiscussed in text. Successivestages in flow advanceand bench formation as lavaintersects confluence with tributaryglaciers from the (c) SunbeamCreek, (d) UnicomCreek, and (e) MapleCreek drainages. (f) Extent of the MazamaRidge flow andStevens Canyon glacier at the endof the eruption.

of the eruption. The preservedglassy and fracturedsloping Combiningthe differentmeasures of the ice thickness(up surface of the lava flow at the head of Stevens Canyon to 300 m thick), we can extrapolatethe distributionof the (Figure9a) is consistentwith subglacialflow andcan be used preeruptionglaciers within Stevens Canyon and the adjacent to estimate the minimum ice thickness. The lava flow drainages(Figure 9b). Priorto andduring the eruption, voids benchesconsist of thickerportions of the flow that have flat and channelswere erodedalong the marginsof the Stevens tops and steepscarps at their downflowboundaries. These Canyonglacier allowing the lava to flow forward.Pauses in benchesprobably represent sites where lava pondedbefore lava advance and bench formation occurred when these erosionalchannels developed and allowed continueddown- channelsslowly erodedacross intersecting tributary glaciers valley flow. At the end of the Mazama Ridge lava a (Figures 9c-9f). This process probably involved a spectacularwall of lava columns indicatesthat the flow combination of surficial and subglacial water flow, ponded against ice in a way analogousto the bench undercuttingof the glacierwalls and limitedsubglacial flow formationsup slope. We can thereforeuse the heightof the (Figure9e). Thetotal meltwater produced during the eruption lava benches and the flow terminus wall as additional wasestimated to be-•0.35km 3 by determiningthe volume of evidenceof the glacialice thickness. ice removedduring the eruption;this calculationwas made LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES 23,721

Figure 10. Reconstructionof Pleistoceneandesite flow (stippledarea) and syneruptive glacier (shaded area) on the southflank of SouthSister, Oregon. Contoursare in feet. (a) Currenttopography [based on U.S. Geological Survey,1988] and extentof flow. Dashedlines indicate the inferredboundary of flow. Crosses,quench features (glassand fractures)associated with lava-iceinteraction; circles, oxidized, vesicular flow surfaceindicative of subaerialemplacement; c, caveat lavaflow base(such as in Figures4f and6), presumablycaused by meltingof ice thathad beencovered by lava. (b) Preeruptiveglacier as reconstructedusing field evidencein Figure10a. Devils Hill and KaleetanButte formed topographic highs that likely roseabove the glacier. (c) As the lava flow reached thesetwo topographicbarriers, it was deflectedto the eastand noah. Severalsmall offshoots of the lava flow extendeda shortdistance under the ice (SGL, subglaciallava). The box showsthe locationof Figure10a.

using an assumedpreeruption subglacial topography (not Abundant lava and ice contactfeatures are exposedaround shown; based on field estimates of lava flow thickness and the marginsof this flow (Figure 10a). moderntopography in Figure 9a), the estimatedpreeruption The South Sister lava flow is considerablythinner (30-40 glacialtopography (Figure 9b), andthe estimatedposteruption m thick) than the Mazama Ridge flow. Hackly fractures glacial topography(Figure 9f). This volume is a minimum radiateoutward from severalcaves at the flow base(Figure estimatesince not all glacial melting will have occurredin 10a), suggestingthat the lava overrode blocks of ice. contactwith the lava flow. Melting was probablyepisodic, Pseudopillowand sheet-likefractures occur along the baseof with pauses during bench formation and rapid melting severalsmall flow lobes (5-6 m across)that extenddownhill, associated with renewed lava advance. This behavior would away from the main lava body (Figure 10a). The flow lobes haveresulted in suddenpulses of waterdownstream that may have columnarand sheet-likefractures on their uppersurfaces have causedfloods or debris flows, such as thosedocumented indicativeof subglacialemplacement. The thicknessof these for the morerecent geologic history [Scott et al., 1995]. subglacial lobes was used to estimate the minimum ice 4.3.2. Reconstructionat South Sister Volcano, Oregon. thicknessadjacent to the flow. The second reconstruction is for a Pleistocene andesite lava At severalpoints, the lava flow surfaceappears to have flow [Tayloret al., 1987] at SouthSister, Oregon. This flow cooledsubaerially (Figure 10a). It is scoriaceous,crystalline extends down the south flank of the volcano in the Moraine and oxidized, similar in appearanceto other nearbyandesite Lake region and bends eastward at the preeruptive flows that show no evidence of lava and ice interaction. topographic highs of Kaleetan Butte and Devils Hill. Below theseapparently subaerial lavas, the flow marginsare 23,722 LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES steepand exhibit ice contactfeatures. We can use this zone flow initiation near the vent. Warm meltwater will travel of transition from subaerial to subglacial emplacementto down the Nisqually drainage,thermally and mechanically locate the upper surface of the glacier at the time of the eroding the ice along the topographicallylower western eruption. marginof the glacier(Figure l lb). At the confluencewith The field evidencelets us reconstructthe glacierthat filled Wilson Glacier, meltwater flow will likely enter subglacial the depression between the flank of South Sister and tunnels,but melting and collapseof overlyingice should KaleentanButte/Devils Hill dr/ringthe eruption(Figure 10b). eventuallyresult in the formationof an opentrench with the Lava flowed down a channeleroded through this glacier and NisquallyGlacier on oneside and Wilson Glacier on theother slowed when it reachedthe gentler slope of the depression. (Figure l lb). Lava flowing down the Nisqua!lydrainage As the meltwaterchannel became more poorlydefined, small would then be funneledalong the erodedglacier margin. lava lobes undercutthe glacier and flowed short distances Near where the lava flow would cross the 12,000-foot underthe ice (Figure 10c). Some meltwaterflowed over and (3658 m) contour,the subglacialtopography slopes steeply accumulatedat the base of these of the subglacial lobes, away from the glaciermargin. As a result,continued erosion resulting in water infiltration and the production of of the glacialmargin will likelyresult in someundercutting of pseudopillow fractures. Estimates of the total meltwater the NisquallyGlacier (Figure l lb). Collapseof the undercut produced(--0.07 km3; determined in thesame manner as for ice walls would causechannel widening and might result in the MazamaRidge flow) are muchlower than for the Mazama suddenmeltwater productionor temporaryice dams. At Ridgecase (--0.35 km3), and unlike the larger example, the lower elevations,glacial undercuttingshould be reduceddue presenceof pseudopillowfractures at South Sisterindicates to the flatter subglacialfloor of the valley. smallamounts of meltwaterponding. Lava ponding and bench formation may occur at the confluence of the Nisqually and Wilson Glaciers if a 4.4. Evaluating Future Hazards meltwater channel does not exist. Once this channel is formed, the lava flow would become constrainedalong a The glacierspresent on modernstratovolcanoes are much currentlysubglacial ridge that separatesthe glacialdrainages. smallerthan when the Cascadelava flows were emplacedin Undercuttingby subglacialmeltwater flow downthe sidesof contact with ice. As a consequence,potential hazardsare this ridge would promotelocal water accumulationsand the lower today. Today, many previouslyice-covered volcanoes formationof small lava lobesaway from the main lava flow are now unglaciatedor have glaciersof insignificantvolume. (Figure l lb). Beyondthe Wilson Glacier, the lava flow Lava eruptedperiglacially commonly diverts around glaciers should continuealong the westernmargin of the Nisqually ratherthan intrudingor flowing over them. In addition,thin Glacier, potentiallyto its terminus. glaciers(<100 m thick) are less likely to constrainlarge lava The primary hazardsassociated with such an eruption flows (on the scaleof ridge-forminglavas) and lesslikely to would be from flooding. By estimatingthe differencein ice trap largevolumes of meltwater. volume before and after the hypotheticaleruption (Figures Hazards still exist at volcanoes, such as Mount Rainier, 11aand I lb, respectively),wefind that at least0.03 km 3 of that still have significant glacial cover. On the basis of meltwatercould be produced. The greatestmelt discharge historic observationsof lava-ice interaction,we can predict rate would likely be during initial establishmentof the that hazards will be greatest early in an eruption when channel. Possible sudden surgesof meltwater might be extrusion rates and heat flow are highest (especiallywhen caused by temporary damming of the channel during accompaniedby lava fountainingonto ice) and when drainage subglacial flow or by collapse of glacial walls due to channelsand trenchesare beingeroded in the ice. After such undercutting. channels are established, both the rate of meltwater Probably the greatest hazards during this hypothetical productionand the potential hazardsare reduced. Hazards eruptionwould comefrom suddenincreases in lava effusion would increaseagain if the channelwere to becomeblocked, rate. While suchincreases are hardto predict,we know they allowing meltwaterto accumulateor if there were a sudden can have devastatingconsequences. During eruptionsat increase in effusion rate. Viiiafrica in 1971 and 1984, suddenincreases in effusion rate resultedin dramatic increasesin meltwaterproduction and 4.5. Hypothetical Flow Down Nisqua!!yGlacier, Mount also collapseof the leadingedge of the lava flow [Gonz•Jlez- Rainier Ferrari, 1973; McC!elland eta!., 1989]. Lava flow collapses We havecombined historical eruption observations and the can produceblock and ash flows capableof scouringthe insight gained through field studiesof lava-ice interaction glacial surfacesand producinglarge debrisflows. Work by featuresto predictthe behaviorof a hypotheticalfuture lava t/allance [1995] and T.W. Sisson(personal communication, flow down the Nisquallydrainage at Mount Rainier (Figure 1997) suggeststhat lava flow collapseand resultantdebris l la). The Nisquailydrainage is a likely flow path for lava flow productionhave occurred on numerousoccasions during eruptedfrom Mount Rainier's most recentlyactive summit the Quaternary at Mount Rainier and therefore must be crater and has existing maps of subglacialtopography considered seriously during an evaluation of potential [Driedger and Kennard, 1986]. We have chosenan initial hazards. meltwaterchannel width of 150 m, usingthe samechannel Mount Rainieris currentlyat a low level of activityand the width as the Mazama Ridge lava flow describedabove and pathsand effusion rates of future lava flows are uncertain. comparableto that observedduring the 1994 Llaima, Chile, The hypotheticaleruption presentedabove is for only one eruption (from 50 m to 150 m) [Moreno and Fuentealba, drainage. A morethorough evaluation of potentialhazards at 1994]. Mount Rainier, or at other volcanoes, could be made by At the onset of eruption, glacial runoff will increase constructing flow projections for all of the voicano's dramaticallydue to increasedheat flow, explosions,and lava drainagesassuming a variety of vent locationsand eruption LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES 23,723

%% I I I I

1 km 0.5 km

Figure 11. Predictedpath that a lava flow (stippledarea) would take down the NisquallyGlacier, Mount Rainier, Washington. Shadedareas correspond to glaciers;contours are in feet. (a) Currenttopography and glacial extent [basedon Driedget and Kennard, 1986, Plate 1]. Box showslocation of Figure 11b. (b) Glacial runoff downthe NisquailyGlacier would likely form a trenchalong the westside of the glacierthat would subsequently be occupied by the lava flow. Owing to the slopingsubglacial topography, some undercutting of the ice is expected(indicated by dashedline and SGL, subglaciallava). At the confluenceof the Nisquallyand WilsonGlaciers (indicated by X), meltwaterwould pond and/orerode a trench(as shown)between the two glaciers. At this point the lava would be flowing on alongthe crestof a currentlysubglacial ridge and might have small offshoots to eitherside (SGL). The lavawould continuedown the trenchand might reach the glacialterminus depending on the durationof the eruption.

rates. Alternatively, lava flow projectionscould be made dome) or isolated lava pillows (esker-like lava flows), followingthe onsetof eruptionand after the flow has begun indicativeof the presenceof water, and yet have the steep travelingdown a specificdrainage. glassymargins indicative of flow confinementagainst ice. A second landform with pseudopillow fractures, pillow lobe flows, showsno evidenceof flow confinementand was likely 5. Conclusions eraplacedin an englaciallake. When lava erupts in a glacial environment,two main Fractures formed during lava-ice interaction can be classesof landforms result, dependingon whether or not morphologically characterized as polygonal, sheet-like, meltwateris able to escaperapidly. On stratovolcanoeswhere pseudopillow,hackly, and shard-formingand correspondto underlyingslopes are steepor if ice layersare thin (<150 m increasinglava cooling rates. Pseudopillow,hackly and thick) and permeable,meltwater travels away from the vent shard-formingfractures form duringcooling by a mixtureof area more freely, erodingcavities in the ice by thermal and steam and water penetratingthe lava flow. Polygonaland kinetic processes[cfi Smellie et aL, 1993; Smellie and sheet-like fractures may represent cooling by convecting Skilling, 1994]. These voids, tunnels, and trenchesact as steam. confiningchannels for lava that flows into them, producing Landforms produced by lava-ice interaction at unbranchedflows with steep margins, which may have stratovolcanoescan be divided into three groups:(group 1) unusuallygreat thickness,flat upper surfaces,and esker-like, glassy margins with polygonal fractures;(group 2) glassy or cylindrica,l shapesthat mimic the void geometry. The margins and continuousfractures; and (group 3) glassy !andformsproduced have glassy margins with polygonal marginswith pseudopillowfractures. Theselandform groups fracturesand includeflat-topped lava mountains,steep-walled correspondto different eruption environments;group 1 ridgeflows, smallerlava flows, andone variety of dome. !andformsdevelop during interaction with thin glaciers. Several other types of landformsare producedwhen the The presenceof the differentfracture types can be usedto glacial voids have small volumesof pondedmeltwater. These locate and infer the former presenceof ice and trapped !andformshave pseudopillowfractures (one type of lava meltwaterat the time of eruption. By combininglarge- and 23,724 LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES

small-scale field evidence with historical observations of DeGrafi}J.M., and A. Aydin, Effect of thermalregime on growth glacial modification, it is possible to reconstructpast incrementand spacingof contractionjoints in basalticlava, d. environmentalconditions and describevarious aspects of the Geophys.Res., 98, 6411-6430, 1993. DeGraff, J.M., P.E. Long, and A. Aydin, Use of joint-growth eruption. directionsand rock texturesto infer thermalregimes during Insightgained by reconstructingpast conditions allows us solidificationof basalticlava flows,d. Volcanol.Geotherm. Res., to anticipatethe hazard implicationsof future lava flow 38, 309-324, 1989. eruptionsat glaciated volcanoes. Historic observations Dragoni,M., A dynamicalmodel of lavaflows cooling by radiation, Bull. l,'olcanol.,51, 88-95, 1989. indicatethat primaryconcerns are the rateof ice melt andthe Driedget, C.L., and P.M. Kennard, Ice volumes on Cascade pondingand suddenrelease of meltwater. By considering volcanoes:Mount Rainier, Mount Hood, Three Sisters, and Mount glacialthickness and subglacialtopography we can identify Shasta,U.S. Geol. Surv. Prof. Pap., 1365,28 pp., 1986. whereglacial undercutting and subglaciallava flow are most Easterbrook,D.J., Mount Baker eruptions,Geology, 3, 679-682, 1975. likely, both of which can generateincreased volumes of Fink, J.H., and R.W. Griffiths, Radial spreadingof viscous-gravity meltwater. We can also identify where meltwaterentrapment currentswith solidit3/ingcrust, d. Fluid Mech., 221, 485-509, is mostlikely, within subglacial tunnels, beneath undercut ice 1990. cliffs, and at the confluenceof glaciers. By estimatingthe ice Fink, J.H., and R.W. Griffiths, Morphology,eruption rates, and available for melting and monitoring lava effusion and theology of lava domes: Insights from laboratorymodels, d. Geoph.vs.Res., 103, 527-545, 1998. advance,we can makepredictions regarding increased runoff Fridleifsson,I.B., H. Fumes,and F.B. Atkins, Subglacialvolcanics.-- andpossible floods and debris flows. Suddencollapse of the On the control of magma chemistryon pillow dimensions,d. lava flow front, commonly associatedwith increasesin [•9lcanol. Geotherm. Res., 13, 103-118, 1982. effusionrate, can producemixed avalanchesand largedebris Fumes,H., I.B. Fridleifsson,and F.B. Atkins, Subglacialvolcanics w flows, and relevantprecautions should be taken. On the formationof acid hyaloclastites,d. Volcanol.Geotherm. Res., 8, 95-110, 1980. Gonzfilez-Ferran,O., Descriptionof volcaniceruptions--Villarrica, Acknowledgments. Work in the field benefitedfrom the help Bull. Volcanol.Erupt., 11, 41-42, 1973. and insight of Tom Sisson,Cathie Hickson,Charlie Bacon, Pat Muffler, Phil Long, Wes Hildreth, Cynthia Gardner, Stanley Gregg,T.K.P., and J.H. Fink, Quantificationof extraterrestriallava Williams, Hugo Moreno, Gustavo Fuentealba,Tina Neal, Tom flow effusion rates through laboratorysimulation, d. Geophys. Miller, Tom Pierson,and Terry Keith. We thank John Smellie, Res., 101, 16,891-16,900, 1996. ChristopherWaythomas, Jason Lore, and PenelopeKing whose Griffiths, R.W., and J.H. Fink, SolidifyingBingham extrusions: A commentsgreatly improved this manuscript. This work was modelfor the growthof silicic lava domes,d. Fluid Mech.,347, supportedby NSF grantEAR 9614330and NASA grantNAGW 529 13-36, 1997. to J.H.F., GSA grant 5192-93 to D.T.L., and NSF grant EAR Grossenbacher,K.A., and S.M. McDuffie, Conductivecooling of 9421455 to StanleyWilliams. lava: columnarjoint diameterand stria width as functionof coolingrate and thermalgradient, d. Volcanol.Geotherm. Res., 69, 95-103, 1995. References Gufihnti, M., M.A. Clynne, and L.J.P. Muffler, Thermaland mass implicationsof magmaticevolution in the Lassenvolcanic region, Aydin, A., and J.M. DeGraff, Evolution of polygonal fracture Cali[brnia, and minimum constraints on basalt influx to the lower patternsin lava flows,Science, 239, 471-476, 1988. crust,d. Geopkvs.Res., 101, 3003-3013, 1996. Batiza, R., and J.D.L. White, Submarinelavas and hyaloclastite,in Encyclopediaof Volcanoes,edited by H. Sigurdsson,pp. 361- Hammond,P.E., Lone Butte and Crazy Hills: Subglacialvolcanic 381, Academic,San Diego, Calif., 2000. complexes,Cascade Range, Washington, in CordilleranSection, Belousov, A.B., Block-and-ash pyroclastic flows of basaltic Centen.FieM Guide vol. 1, edited by M.L. Hi!l, pp. 339-344, compositiongenerated by secondaryphreatomagmatic explosions Geol. Soc.of ,4m., Boulder,Colo., 1987. on lavaflows: A newtype of volcanichazard demonstrated by the Hardee, H.C., Solidification in Kilauea Iki lava lake, d. Volcanol. 1994 eruption of Kluchevskoyvolcano (Kamchatka), in Pan Geotherm. Res., 7, 211-223, 1980. Pacific Hazards '96 Conferenceand TradeShow, p. 29, Disaster Hon, K., J. Kauahikaua,R. Denlinger,and K. Mackay,Eraplacement PreparednessResour. Cent., Univ. of B.C., Vancouver,1996. and inflation of pahoehoe sheet flows: Observationsand Bemmelen,R.W.v., and M.G. Rutten, Tablemountainsof Northern measurementsof active lava flows on Kilauea volcano, Hawaii, Iceland,217 pp., E. J. Brill USA, Kinderhook,N.Y., 1955. Geol. Soc. Am. Bull., 106, 351-370, 1994. Benson,C.S., and R.J. Motyka, Glacier-volcanointeractions on Mt. Hull, D., and B.D. Caddock,Simulation of prismaticcracking of Wrangell,Alaska, Geophys.Inst. Univ. AlaskaFairbanks Annu. coolingbasalt lava flows by thedrying of sol-gels,d. Mater.Sci., Rep., 1977-78, 1-25, 1978. 34, 5707-5720, 1999. Bonnichsen,B., and D.F. Kauffmann,Physical features of rhyolite Jaeger,J.C., Cooling and solidification of igneousrocks, in Basalts: lava flows in the Snake River Plain volcanic province, The Poldervaart Treatise on Rocks of Basaltic Composition, southwesternIdaho, in The Eraplacementof Silicic Domesand editedby H.H. Hessand A. Poldervaart,pp. 503-536,Wiley- Lava Flows,edited by J.H. Fink, Spec.Pap. Geol. Soc.Am., 212, Interscience,New York, 1968. !19-145, 1987. Keszthelyi,L., and R. Denlinger,The initialcooling of pahoehoe Budkewitsch, P., and P.-Y. Robin, Modeling the evolution of flow lobes,Bull. I blcanol., 58, 5-18, 1996. columnarjoints, d. Volcanol.Geotherm. Res., 59, 219-239, 1994. Khrenov, A.P., A.Y. Ozerov, N.E. Litasov, Y.B. Slezin, Y.D. Byers, F.M., D.M. Hopkins,K.L. Wier, and B. Fisher,Volcano Murav'ev,and N.A. Zharinov,Parasitic eruption of Klyuchevskoi investigationson Umnak Island, 1946, U.S. Geol. Surv.Alaskan volcano(Predskazannyi eruption, 1983), }•blcanol.Seismol., 7, 1- VolcanoInvest. Rep., 2, 19-53, 1947. 24, 1988. Conrey,R.M., Geologyand petrology of the Mt. Jeffersonarea, High Lachenbruch,A.H., Depth and spacing of tension cracks, d. CascadeRange, Oregon, Ph.D. dissertationthesis, Wash. State Geophys.Res., 66, 4273-4292, 1961. Univ., Pullman, 1991. Lescinsky,D.T., Lava flow morphology:The roles of external Corte, A., and A. Higashi, Experimentalresearch on desiccation confinement and lava-ice interaction, Ph.D. dissertationthesis, cracksin soil, Cold Reg. Res.and Eng. Lab. Res.Rep. 66, 76 pp., 209 pp., Ariz. StateUniv., Tempe, 1999. Hanover, N.H., 1964. Lescinsky,D.T., and T.W. Sisson,Ridge-forming, ice-bounded lava DeGrafi}J.M., andA. Aydin, Surfacemorphology of columnarjoints flows at Mount Rainier, Washington, Geology, 26, 351-354, and its significanceto mechanicsand directionof joint growth, 1998. Geol. Soc. Am. Bull., 99, 605-617, 1987. Long,P.E., andB.J. Wood, Structures,textures, and cooling histories LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES 23,725

of Columbia River basalt flows, Geol. $oc. Am. Bull., 97, 114- and the Environment,January 1997, p. 5, Gobiemo de Jalisco, 1155, 1986. Unidad Editorial, Puerto Vallarta, Mexico, 1997. Major, J.J., and C.G. Newhall,Snow and ice perturbationduring Sisson,T.W., and M.A. Lanphere, K-Ar ages of selectedMount historicalvolcanic eruptions and the formationof laharsand Rainierlava flows, GSA Data RepositoryItem 9840, 1998. floods, Bull. kblcanoLt,52, 1-27, 1989. Smellie, J.L., Subglacialeruptions, in Encyclopediaof Volcanoes, ManIcy,C.R., Extendedcooling and viscousflow of large,hot edited by H. Sigurdsson,pp. 403-418, Academic, San Diego, rhyolitelavas: Implications of numericalmodeling results, d. Calif., 2000. l,blcanol. Geotherm.Res., 53, 27-46, 1992. Smellie, J.L., and I.?. Skilling, Productsof subglacialvolcanic Marston,R.A., Supraglacialstream dynamics on the JuneauIcefield, eruptions under different ice thicknesses:two examples from Ann. Assoc.Am. Geogr., 73, 597-608, 1983. Antarctica, Sediment. Geol., 91, 115-129, 1994. Mathews, W.H., "Tuyas,"flat-topped volcanoes in northernBritish Smellie, J.L., M.J. Hole, and P.A.R. Nell, Late Miocene valley- Columbia, Am. d. Sci., 245, 560-570, 1947. confined subglacial volcanism in northern Alexander Island, Mathews,W.H., The Table, a flat-toppedvolcano in southernBritish Antarctic Peninsula,Bull. Volcanol., 55,273-288, 1993. Columbia,Am. d. Sci., 249, 830-841, 1951. SmithsonianInstitution, Descriptions of volcanicevents--Westdahl, Mathews, W.H., Ice-dammed lavas from Clinker Mountain, Global VolcanismNetwork Bull., 16 (11), 2; (12), 3, 1991. southwesternBritish Columbia, Am. d. Sci., 250, 553-565, 1952a. Smithsonian Institution, Descriptions of volcanic events-- Mathews,W.H., Mount Garibaldi,a supraglacialPleistocene volcano Veniaminof, Global VolcanismNetwork Bull., 18 (7), 3-4; (10), in southwestern British Columbia, Am. d. Sci., 250, 81-103, 3-4, 1993. 1952b. Smithsonian Institution, Descriptions of volcanic events-- Mathews, W.H., Geology of the Mount Garibaldi map-area, Kliuchevskoi, Global VolcanismNetwork Bull., 19 (8), 11-12; southwesternBritish Columbia, Canada,Geol. Soc. Am. Bull., 69, (9), 2, 1994a. 179-198, 1958. Smithsonian Institution, Descriptions of volcanic events-- McClelland,L. T. Simpkin,M. Summers,E. Nielsen,and T.C. Stein Veniaminof, Global VolcanismNetwork Bull., 19 (2), 6-7; (4), 8- (Eds.), Global Volcanism 1975-1985, 657 pp., AGU, 9; (9), 14, 1994b. Washington,D.C., 1989. Taylor, E.M., N.S. MacLeod, D.R. Shetrod, and G.W. Walker, Meyer, D.F., and D.C. Trabant,Lahars from the 1992 eruptionsof Geologicmap of the Three SistersWilderness, Deschutes, Lane, CraterPeak, Mount Spurrvolcano, Alaska, in The 1992 Eruptions and Linn Counties, Oregon. Scale 1:63,360, U.S. Geol. Surv. of Crater Peak Vent, Mount Spurt Volcano,Alaska, editedby Misc. FieM Ser. Map, MF- 1952, 1987. T.E.C. Kieth, U.S. Geol. Surv. Bull., 2139, 183-198, 1995. Trabant,D.C., R.B. Waitt, andJ.J. Major, Disruptionof Drift Glacier Moreno R., H., and G. FuentealbaC., The May 17-19 1994 Llaima and origin of floodsduring the 1989-1990eruptions of Redoubt volcanoeruption, southern Andes (38ø42'S-71ø44'W), Rev. Geol. Volcano, Alaska, d. Volcanol.Geotherm. Res., 62 (1-4), 369-385, Chile, 21,167-171, 1994. 1994. Mtiller, G., Experimentalsimulation of basaltcolumns, d. kblcanol. U.S. GeologicalSurvey, Mount RainierEast, Washington, U.S. Geol. Geotherm. Res., 86, 93-96, 1998a. Surv 7.5 Min. Topo.Quad., 1971. Mtiller, G., Starch columns:Analog model for basalt columns,d. U.S. GeologicalSurvey, South Sister, Oregon, U.S. Geol. Surv 7.5 Geophys.Res., 103, 15239-15253, 1998b. Min. Topo.Quad., 1988. Naranjo S., J.A., H. MorenoR., and N.G. Banks,La erupci6ndel Vallance, J.W., Holocene tephras and history of Mount Rainier Volc/tn Hudsonen 1991 (46øS), Regi6n Xl, Ais•n, Chile, Serv. volcano,Eos TransAGU, 76 (46), F644, Fall Meet. Suppl.,1995. Nac. Geol. Min. BoL, 44, 50 pp., 1993. Vinogradov,V.N., and Y.D. Murav'ev,Lava-ice interactionduring National ResearchCouncil, Mount Rainier, Active Cascade Volcano, the 1983 Klyuchevskoieruption, Volcanol.Seismol., 7, 39-61, 114 pp., Nat. Acad. Press,Washington, D.C., 1994. 1988. Nemat-Nasser,S., and A. Oranratnachai,Minimum spacingof Vinogradov,V.N., Y.D. Murav'ev, I.M. Nikitina, and A.N. Salmatin, thermally induced cracks in brittle solids, d. Energy Resour. Productionof phreatic explosionsin the interactionof lava and Techol., 101, 34-40, 1979. ice, Volcanol. Seismol., 9, 89-98, 1990. Pierson,T.C., and R.J. Janda,Volcanic mixe/davalanches: A distinct Waitt, R.B., Swift snowmeltand floods (lahars) causedby great eruption-triggeredmass-flow processat snow-cladvolcanoes, pyroclasticsurge at Mount St. Helensvolcano, Washington, 18 Geol. Soc. Am. Bull., 106, 1351-1358, 1994. May 1980, Bull. Volcanol.,52, 138-157, 1989. 'Pierson, T.C., R.J. Janda, J.-C. Thouret, and C.A. Borrero, Walder, J.S., Nature of depositionalcontacts between pyroclastic Perturbationand melting of snow and ice by the 13 November depositsand snow or ice, in Hydrologic Consequencesof Hot- 1985 eruption of Nevado del Ruiz, Colombia, and consequent Rock/SnowpackInteractions at Mount St. Helens Volcano, mobilization, flow and deposition of lahars, d. VolcanoLt Washington,edited by T.C. Pierson,U.S. Geol. Surv.Prof. Pap., Geotherm. Res., 41, 17-66, 1990. 1586, 9-18, 1999. Reeder,J.W., and M. Doukas,Description of volcaniceruptions-- Walker, G.P.L., Morphometricstudy of pillow-sizespectrum among Westdahl,BuiLt l/•lcol. Erupt., 31, 83-86, 1994. pillow lavas,Bull. VolcanoLt,54, 459-474, 1992. Reiter, M., M.W. Barroll, J. Minier, and G. Clarkson, Thermo- Walker, G.P.L., Basaltic-volcanosystems, in Magmatic Processes mechanical model for incremental fracturing in cooling lava and Plate Tectonics,edited by H.M. Prichardet al., Geol. Soc. flows, Tectonophysics,142, 241-260, 1987. Spec.Publ., 76, 3-38, 1993. Ryan, M.P., and C.G. Sammis, Cyclic fracture mechanismsin Walker, G.P.L., and D.H. Blake, The formationof a palagonite coolingbasalt, Geol. Soc.Am. Bull., 89, 1295-1308, 1978. brecciamass beneath a valley glacierin Iceland,Quat. d. Geol. Saemundsson,K., Interglaciallava flows in the lowlandsof southern Soc. London, 122, 45-61, 1966. Icelandand the problemof two-tieredcolumnar jointing, dOkull, Watanabe,K., andY. Katsui,Pseudo-pillow lavas in theAso caldera, 20, 62-77, 1970. Kyushu,Japan, d. Minera. Petrol. Econ. Geol., 71, 44-49, 1976. Scott, K.M., J.W. Vallance, and P.T. Pringle, Sedimentology, Wilding, M., D. Dingwell,R. Batiza,and L. Wilson,Cooling rates of behavior, and hazards of debris flows at Mount Rainier, hyaloclastites:applications of relaxation geospeedometryto Washington,U.S. Geol.Surv. Prof Pap., 1547, 56 pp., 1995. underseavolcanic deposits, Bull. Volcanol.,61,527-526, 2000. Scott, W.E., Holocenerhyodacite eruptions on the flanks of South Wilding, M.C., S.L. Webb, and D.B. Dingwell, Cooling rate Sistervolcano, Oregon, in TheEraplacement of Silicic Domesand variation in natural volcanic glassesfrom Tenerife, Canary Lava Flows, editedby J.H. Fink, Spec.Pap. Geol. Soc.Am., 212, Islands,Contrib. Mineral. Petrol., 125, 151-160, 1996. 35-53, 1987. Wright, A.C., Landformsof McMurdo Volcanic Group, southern Sisson, T.W., An overview of the geology of Mount Rainier's foothillsof the Royal SocietyRange, Antarctica, N. Z. d. Geol. volcanic edifice, Eos Trans. AGU, 76(46), Fall Meet. Suppl., Geophys.,23, 605-613, 1980. F643, 1995. Yamagishi,H., Morphologicaland sedimentologicalcharacteristics Sisson,T.W., and M.A. Lanphere,The growth of Mount Rainier, of the Neogenesubmarine coherent lavas and hyaloclasitesin CascadeArc, USA, in IAVCEI Conferenceon Volcanic Activity southwestHokkaido, Japan, Sediment. Geol., 74, 5-23, 199la. 23,726 LESCINSKY AND FINK: LAVA AND ICE INTERACTION AT STRATOVOLCANOES

Yamagishi, H., Morphological features of Miocene submarine J.H. Fink, Departmentof Geology,Box 871404, ArizonaState coherentlavas from the "Green Tuff" basins:examples from University,Tempe, AZ 85287-1404.(jon.fink•asu.edu) basaltic and andesitic rocks from the Shimokita Peninsula, D.T. Lescinsky,Department of Earth Sciences,University of northernJapan, Bull. Folcanol.,53, 173-181, 1991b. Western Ontario, London, ON N6A 5B7, Canada. Yount, M.E., T.P. Miller, R.P. Emanuel,and F.H. Wilson, Eruption ( dlescins•julian.uwo.ca ) in the ice-filled caldera of Mount Veniaminof, U.S. Geol. Surv. Circ., 945, 58-60, 1985. Zharinov,N.A., S.A. Khubunaya,Y.V. Demyanchyk,V.N. Dvigalo, and V.Y. Kirianov, Description of volcanic eruptions-- (ReceivedDecember 7, 1999;revised May 17,2000; Kliucheskoi,Bull. I/olcanol.Erupt., 33, 79-84, 1996. acceptedJune 7, 2000)