of 14

Explosive submarine eruptions driven by volatile-coupled degassing at Lō`ihi Seamount, Hawai`i

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Explosive submarine eruptions driven by volatile-coupled degassing at Lō`ihi Seamount, Hawai`i
  Explosive submarine eruptions driven by volatile-coupled degassing at L  ō `ihiSeamount, Hawai`i C. Ian Schipper a,d, ⁎ , James D.L. White a , B.F. Houghton b , Nobumichi Shimizu c , Robert B. Stewart d a Geology Department, University of Otago, PO Box 56, Leith St., Dunedin, 9016, New Zealand b Geology and Geophysics, University of Hawai`i at M  ā noa, 1680 East-West Road, Honolulu, HI 98622, USA c Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, MA 02543, USA d Soil and Earth Sciences, INR, Massey University, PB 11-222, Palmerston North, 4474, New Zealand a b s t r a c ta r t i c l e i n f o  Article history: Received 6 November 2009Received in revised form 13 April 2010Accepted 16 April 2010Editor: T.M. Harrison Keywords: Hawai`iL  ō `ihidegassingexplosivesubmarine There is a growing body of information on submarine explosive eruptions, but important questions remainabout their causes and diversity. Magmatic degassing in 󿬂 uences eruption dynamics, but a paucity of appropriate samples from submarine pyroclastic deposits has limited thorough analysis of this process. Wepresent major element and volatile analyses of matrix glasses and olivine-hosted glass inclusions from asubmarine scoria cone (the  “ northern cone ” ) at  ∼ 1060 m below sea level on L  ō `ihi Seamount, Hawai`i, toquantitatively constrain ascent and degassing processes, and infer the magmatic plumbing geometries, thatpermit submarine explosive eruption of basalt in deep water.Olivine crystals form two populations. The  󿬁 rst consists of a single crystal (Fo 86.8 – 88.6 ), bearing inclusions of high-Fe tholeiitic glass (1.18 to 1.28 wt.% H 2 O; 182 to 505 ppm CO 2 ). Major element concentrations indicatethat these high-Fe inclusions represent parental melt that subsequently evolved by crystal fractionation toform the erupted bulk melt. The second population consists of variably zoned crystals bearing low-Feinclusions with lower volatile contents (0.63 to 0.74 wt.% H 2 O; 29 to 176 ppm CO 2 ). Low-Fe inclusions arenot geochemically related to the bulk melt, and their host crystals are interpreted as xenocrysts entrainedfrom shallow storage.The northern cone magma experienced strongly volatile-coupled ( “ closed-system, ”  where bubbles remain inphysical contact with the melt from which they have grown) ascent from at least  ∼ 1.5 km deep in theconduit. Retention of CO 2  in bubbles facilitated strong exsolution of H 2 O (and S). Exsolution of H 2 Odominated the vesiculation and acceleration of ascending melt, accounting for the measured 47 to 65%vesicles in lapilli. CO 2  played an essential chemical role, but negligible physical role, in driving magma ascent.Our results contradict the common interpretation that high hydrostatic pressures require submarineexplosive eruptions to be Strombolian, driven by the accumulation and buoyant decoupling of CO 2 -rich 󿬂 uids.Comparison to volatile data from other explosive and effusive eruption products on L  ō `ihi indicates thatexplosivity is controlled primarily by the style of degassing, rather than magma type or initial volatilecontent. Magmas that ascend directly from source to vent under volatile-coupled conditions retainmaximum explosive potential. Conversely, magmas that degas under open-system conditions, or ascendfrom shallow reservoirs, will have their explosive potential diminished as exsolved volatile  󿬂 uids are lost tohydrothermal systems. This latter path appears to be the most common at L  ō `ihi, causing most magmas toerupt effusively.© 2010 Elsevier B.V. All rights reserved. 1. Introduction 1.1. Degassing in submarine explosive volcanism The study of submarine explosive eruptions is in its infancy, eventhough most basaltic volcanism on Earth is submarine (Crisp, 1984;White et al., 2003). Submersible technology has made directobservation of submarine volcanoes possible, and has revealed thatwidespread clastic deposits are common to depths exceeding 4000 m(LonsdaleandBatiza,1980;Batizaetal.,1984;SmithandBatiza,1989;Gill et al., 1990; Fouquet et al., 1998; Clague et al., 2000; Hekinian etal., 2000; Clague et al., 2003a,b; Eissen et al., 2003; Davis and Clague,2003, 2006; Clague et al., 2008; Sohn et al., 2008). The discovery of these deposits challenged the traditional view that high hydrostaticpressure precludes explosive eruptions at depths exceeding 500 m(McBirney, 1963; Fisher and Schmincke, 1984). Earth and Planetary Science Letters 295 (2010) 497 – 510 ⁎  Corresponding author. Geology Department, University of Otago, PO Box 56, LeithSt., Dunedin, 9016, New Zealand. Tel.: +64 21 137 1422; fax: +64 3 479 7527. E-mail address:  schipper.ian@gmail.com (C.I. Schipper).0012-821X/$  –  see front matter © 2010 Elsevier B.V. All rights reserved.doi:10.1016/j.epsl.2010.04.031 Contents lists available at ScienceDirect Earth and Planetary Science Letters  journal homepage: www.elsevier.com/locate/epsl  Magmatic degassing of CO 2 +H 2 O is widely accepted as a primarycontrol on explosivity, and it is studied routinely at subaerialvolcanoes by a combination of gas emission monitoring and analysisof quenchedvolatiles in erupted products. Emission monitoring is notyet practical at most submarine volcanoes, but submarine pyroclasticdepositsbetterlendthemselves tostudiesof volatilesystematicsthando subaerial ones, since matrix glasses quenched at high hydrostaticpressure typically retain measurable concentrations of volatilespecies, whereas those quenched at atmospheric pressure are oftencompletely degassed. Dissolved volatile species preserved in pheno-cryst-hostedglassinclusionsandmatrixglassesallowdegassingpathsto be evaluated, since they provide estimates of (minimum) parentalandresidualvolatileconcentrations,respectively.Thismethodologyisnot yet widely applied in the study of submarine pyroclastic deposits,with few appropriate samples collected; degassing behaviour is ofteninferred or assumed, and eruptions are usually described by directanalogy to established dispersal-based models for subaerial eruptions(Walker and Croasdale, 1972; Walker, 1973).CO 2  is less soluble than H 2 O in basaltic melts, and exsolves at highpressure, followed by increasing amounts of H 2 O at lower pressure(DixonandStolper,1995;Dixonetal.,1995).IfCO 2 vapourremainsinequilibrium with the melt, it will lower the fugacity of other volatilespecies, facilitating their exsolution at higher pressures (Holloway,1976; Dixon and Stolper, 1995; Dixon et al., 1995; Dixon and Clague,2001; Newman and Lowenstern, 2002). The term  “ closed-system ” describes such chemical equilibrium, but is not physically descriptive.If CO 2 -rich  󿬂 uids exsolved at depth continuously  󿬂 ux throughoverlying magma, they will facilitate H 2 O exsolution (Dixon et al.,1995; Rust et al., 2004; Metrich and Wallace, 2008), but will notproduce bubbles that remain coupled to the melt to assist inaccelerating magma to the surface. Such  󿬂 uxing is a closed-systemoverlargelength(ormagmavolume)scales.To highlight thephysicalrole volatiles play in driving explosive submarine eruptions, we usethe term  “ volatile-coupled ”  to describe closed-system degassing overshort length scales, in which the exsolved volatiles are retained inbubbles that are mechanically coupled to the melt from which theyhave grown.Here we use volatile systematics to demonstrate that scoria coneson the ∼ 1 km deep summit of L  ō `ihi Seamount, Hawai`i, are the resultof strongly volatile-coupled degassing of CO 2  and H 2 O (and S), incontrasttothecommoninferencethatsubmarineexplosiveeruptionsare Strombolian and driven by decoupled volatile volumes. Wepresent new data from one of these cones, the  “ northern cone ” , andcompare it to published datasets from explosive and effusive L  ō `ihideposits, the  “ southern cone ”  (Schipper et al., 2010a,b), and pillowlava KK31-12 (Moore et al., 1982; Dixon and Clague, 2001; Hauri,2002), to constrain the degassing scenarios and magmatic plumbinggeometriesthatpermitexplosivebasalticeruptionsatthisyoungestof Hawaiian volcanoes. 1.2. Location and samples L  ō `ihi Seamount is entirely submarine and lies  ∼ 35 km southeastof the island of Hawai`i (Fig. 1). L  ō `ihi has just entered its tholeiiticshield-building stage (Moore et al., 1982), having grown to  ∼ 5 kmheight over 500,000 years (Garcia et al., 2006). L  ō `ihi's  ∼ 12 km 3 summit plateau (Fig. 1) at  ∼ 1200 mbsl has numerous cone-likelandforms rising to just under 1 km depth. Three pit craters occupythe central summit region, the youngest formed during the mostrecent eruption in 1996 (Loihi Science Team, 1997; Davis and Clague,1998; Garcia et al., 1998).Our data is from samples collected with the Hawaiian UnderseaResearch Laboratory (HURL) Pisces IV submersible (Dives P4-160 toP4-164), in October 2006. The  “ northern cone ” , located at 18°56 ′ 1 ″  N, − 155°15 ′ 5 ″  W (Fig. 1), is  ∼ 80 m high, with a summit at  ∼ 1060 mbelowsealevel(mbsl).Itwasdescribedasa “ scoriacone ” byClagueetal. (2003a; their  “ Deposit J ” ), and textures in coarse lapilli from thesame samples used in this study were analyzed by Schipper et al.(2010a). Selected whole rocks were analyzed for major elements byX-ray  󿬂 uorescence (XRF), glasses and glass inclusions were analyzedformajorelementsbyelectronmicroprobe(EMP),andforvolatilesbyFourier-transform infrared spectroscopy (FTIR) and secondary ionmass spectroscopy (SIMS). For details of sampling and analyticalmethods, see online supplementary material (Appendix A).We compare the northern cone to the type  “ Poseidic ”  southerncone eruption (Fig. 1; Schipper et al., 2010b), de 󿬁 ned by strongvolatile-coupled degassing that causes magma to rapidly ascend andvesiculate, and ultimately fragment by explosive magma – waterinteraction. Effusive L  ō `ihi eruptions are represented by alkalic pillowlava KK31-12, dredged from 2186(±137) mbsl on the western  󿬂 ankof L  ō `ihi (Fig. 1 and inset to Fig. 7) in 1981 (Moore et al., 1982). Major element chemistry and volatile concentrations in KK31-12 matrixglass and inclusions are taken from Dixon and Clague (2001) andHauri (2002), respectively. 1.3. Previous work on L ō `ihi volatiles Volatilespecies(H 2 O,CO 2 ,S,Cl,andF)inL  ō `ihiglassesand/orglassinclusions have been analyzed in several studies (Byers et al., 1985;Garcia et al., 1989; Honda et al., 1993; Garcia et al., 1998; Kent et al.,1999a,b; Clague et al., 2000; Dixon and Clague, 2001; Hauri, 2002;Schipper et al., 2010b). Byers et al. (1985) linked rock types with variability in volatile abundances to demonstrate that they werederived from distinct sources. Garcia et al. (1989) used volatiles inL  ō `ihi tholeiites to demonstrate that Hawaiian volcanoes tap progres-sively depleted sources as they evolve. Dixon and Clague (2001)assessed volatiles in pillow rim glasses for variable composition andapplied solubility/degassing models to develop a heterogeneousplume model for L  ō `ihi. Kent et al. (1999a,b) used volatiles in pillowrims and olivine-hosted glass inclusions to assess the assimilation of seawater-in 󿬂 uencedcomponents. Hauri(2002)comparedvolatilesinolivine-hosted glass inclusions from  󿬁 ve Hawaiian volcanoes, includ-ing L  ō `ihi, to demonstrate heterogeneity in the Hawaiian mantle, andestimate volatile budgets for each volcano. Except for limited volatiledata (2 analyses on a basalt breccia by Garcia et al. (1998), and 7 limuo' Pele shards by Clague et al. (2000)) most of the L  ō `ihi samplespreviously analyzed for volatiles have been matrix glasses from lavascollected by dredging, or have been inclusions in phenocrysts fromthese lavas.The parallel study by Schipper et al. (2010b), was the 󿬁 rstto use volatiles in matrix glasses and inclusions to de 󿬁 ne eruptionstyle for a scoria cone on L  ō `ihi. 2. Results Glasses described as  “ lapilli matrix glass ”  are sideromelane inmedium ( N 8 mm) juvenile lapilli. Glass inclusions were analyzedfrom free olivine crystals hand-picked from the very coarse ash(0 ≤− 1  Φ /1 ≤ 2 mm) fraction of bulk samples. Glasses described as “ selvages ”  are thin, incomplete coatings bordering free olivinecrystals.  2.1. Petrography Northern cone scoria contain  ∼ 2 to 5% euhedral olivine, typically ≤ 1.5 mm long, and sparse, strongly resorbed, clinopyroxene micro-phenocrysts  ≤ 0.25 mm. Glass inclusions are common in olivinephenocrysts. They often contain early-crystallized Cr-spinel (Garciaet al., 1998, 2006), that probably provided geometric imperfectionsfacilitating melt entrapment (Roedder, 1984; Metrich and Wallace,2008). Some inclusions contain vapour bubbles, either from multi-component entrapment of vapour-saturated melt (Hauri, 2002;Metrich and Wallace, 2008) or post-entrapment differential 498  C.I. Schipper et al. / Earth and Planetary Science Letters 295 (2010) 497  – 510  contraction of melt and host olivine (Roedder, 1984). Only inclusionsentirely enclosed by host olivine, hence least likely to have leakedvolatiles (e.g., Cervantes and Wallace, 2003) were analyzed.Northern cone olivines have core compositions varying by lessthan 4%, from Fo 86.8  to Fo 90.4 , with rim compositions of Fo 83.9  to Fo 89.8 (see Table 3), and thereare twodistinct types.The 󿬁 rst is representedby a single observed olivine (L7-04) that is slightly reverse zoned(Fo 86.8 – 88.6 ), with a heavily resorbed, pitted texture, and crystal-spanning cracks (Fig. 2A). All other olivine crystals are very slightlyreverse zoned, to normally zoned, with no cracking or other signs of deformation (Fig. 2B). They have thin manganese coatings a few 10sof microns thick (Fig. 2B). CaO in all olivine crystals (onlinesupplementary material in Appendix A) varies narrowly from 0.24to 0.36 wt.%, typical for L  ō `ihi olivine, indicating crystallization atcrustal depths (Garcia et al., 2006).  2.2. Whole-rock and glass compositions Northern cone scoria have tholeiitic whole-rock major elementcompositions (Table 1, Fig. 3A), and a narrow MgO range, 8.95 to 9.60 wt.%. Slight variations in whole-rock analyses follow olivinefractionation trends for all major elements (Fig. 4). Lapilli matrixglasses and glass selvages on free olivine crystals (Table 2) are alsotholeiitic (Fig. 3A), with a narrow compositional range (Fig. 4); MgO from 6.06 to 7.08 wt.%, and FeO T from 11.7 to 12.2 wt.%, typical forL  ō `ihi glasses(Byersetal.,1985;Garciaetal.,1989,1995,1998;Dixonand Clague, 2001). One exception is selvage L7-04-G, on the reversezoned olivine (Fig. 2A), which is less differentiated (7.72 wt.% MgO)than other matrix glasses. The narrow compositional ranges of wholerock and matrix glasses, and slight variation consistent withfractionation of the observed mineral phases indicate that the Fig. 1.  Location. Main panel: bathymetry of L  ō `ihi Seamount (created by J.R. Smith) showing location of deposits discussed in the text (NC = northern cone, SC = southern cone),dashed box shows location of expanded section in Fig. 7. Inset bottom left: position of L  ō `ihi with respect to the island of Hawai`i. Top right: expanded bathymetry of the northerncone deposit, showing sample location of   “ Deposit J ”  (Clague et al., 2003a), black stars and numbers show location of samples from Pisces IV dive P4-161 used in this study.499 C.I. Schipper et al. / Earth and Planetary Science Letters 295 (2010) 497  – 510  youngest portion of the northern cone from which our samples werecollectedis chemicallymonomict,representinga singlebatchof high-Fe, tholeiitic melt.Northern cone glass inclusions have undergone 6.5 to 14.1% post-entrapmentcrystallization(PEC),asindicatedbyolivine – meltpartitioncoef  󿬁 cients(rawandPEC-correctedcompositionsinTable3;PECdetailsin online supplementary information (Appendix A)). The inclusionshavetwodistinctcompositionaltypes(Table3,Fig.4),correspondingto the two types of host olivine crystals. The  󿬁 rst are  “ high-Fe ”  inclusions,represented by the three inclusions (L7-04-A to L7-04-C) hosted in thesingle,reversezonedolivinecrystal(L7-04;Fig.2A).Thesearetholeiitic(Fig.3A),withMgOrangingfrom7.16to7.65 wt.%beforecorrectionforPEC (12.0 to 12.6 wt.% with the addition of 11.9 to 13.8% olivine). FeO T ranges from 11.0 to 11.6 wt.%. The high-Fe inclusions are withinanalyticalerrorofeachotherforallmajorelements,exceptforL7-04-B,the most differentiated of the suite, which is signi 󿬁 cantly depleted inTiO 2  and enriched in K 2 O. We note that uncorrected high-Fe inclusioncompositions are only slightly less differentiated than matrix glasses.The second group of inclusions includes those from all other olivinecrystals, and is de 󿬁 ned by comparatively low iron. They span alarger compositional range from 5.32 to 7.11wt.% MgO (9.41 to12.3 wt.% MgO, PEC-corrected), with FeO T from 6.95 to 8.82 wt.% (7.30 Fig. 2.  Backscatter electron images of representative olivine crystals, with H 2 O, CO 2 , and FeO T for inclusions and glass selvages (Tables 2 and 3). A: Crystal L7-04, showing fracturedand pitted texture, high-Fe inclusions, and high-Fe glass selvage. B: Crystal NC10-13, showing a low-Fe inclusion, a resorbant high-Fe glass selvage, and incomplete Mn oxide layer.  Table 1 Whole-rock geochemistry.Dive no. P4-161 P4-161 P4-161 P4-161Lapillus 2-A(13) 2-A(18) 2-B(6) 2-B(10)SiO 2  47.7 47.5 47.7 47.5TiO 2  2.65 2.63 2.68 2.66Al 2 O 3  12.0 11.9 12.1 11.9FeO T 11.6 11.6 11.7 11.6MnO 0.17 0.17 0.17 0.17MgO 9.31 9.60 8.95 9.08CaO 10.8 10.7 11.0 10.8Na 2 O 2.11 2.02 1.95 2.16K 2 O 0.69 0.70 0.69 0.70P 2 O 5  0.38 0.22 0.33 0.47LOI% 0.67 0.65 0.36 0.77Total 99.4 98.9 98.8 99.1Mg# 64.1 64.8 63.1 63.6Analyses by XRF. All oxides in wt.%. Analytical procedures given in onlinesupplementary material (Appendix A). Fig. 3.  A: Total alkali – silica diagram. Northern cone (NC) whole rock, matrix glass, andglass inclusion (GI) compositions (Tables 1 – 3) are shown with matrix glasses andinclusions from thesouthern cone (SC; Schipper etal., 2010b) and KK31-12 (Dixon and Clague, 2001; Hauri, 2002). Range of matrix glass compositions from L  ō `ihi (grey  󿬁 eld;Garcia et al., 2006), and range of inclusion compositions from Kent et al. (1999b;stippled  󿬁 eld) shown for comparison. Compositional boundaries after MacDonald andKatsura (1964). B: H 2 O vs. K 2 O. Symbols and patterns as in A, except grey  󿬁 eld andassociated linear best- 󿬁 t line from Dixon and Clague (2001) for L  ō `ihi lavas.500  C.I. Schipper et al. / Earth and Planetary Science Letters 295 (2010) 497  – 510  to 9.49 wt.%, PEC-corrected). Henceforth, references to inclusioncompositions are the PEC-corrected values, unless stated otherwise.The low-Fe inclusions are compositionally atypical for L  ō `ihi. Theyare higher in SiO 2  than any matrix glasses previously reported (grey 󿬁 eld in Fig. 3A; Garcia et al., 2006), although they partially overlap withtherangeofglassinclusioncompositionspresentedbyKentetal.(1999b) (stippled  󿬁 eld in Fig. 3A). They all have iron contents lowerthan previously reportedminimaof 9.55 wt.% (Kentet al.,1999b) and8.89 wt.% (Hauri, 2002; see Fig. 5A) for L  ō `ihi inclusions. Attempts tosubdivide the low-Fe inclusions yield different results depending onwhichmajorelementsareused.This,andhighlyvariableTiO 2 andK 2 Ointhelow-Feinclusions,andevenbetweenmultipleinclusionswithinthe same host olivine crystal (Table 3, and marked by tie-linesin Fig. 4), indicate that they represent entrapment of chemicallyheterogeneous melts.  2.3. Volatile abundances Sulfur typically degasses at shallow levels along with magmaticH 2 O (Gerlach and Graeber, 1985; Dixon et al., 1991), and is routinelymeasured in real-time at subaerial volcanoes (e.g. Stix and Gaonac'h,2000).Glassinsubaerially-eruptedHawaiianbasaltscharacteristicallyhas  b 250 ppm S (Moore and Thomas, 1988; Davis et al., 1991; Clagueet al., 2000). Sulfur in northern cone glasses (Tables 2 and 3, Fig. 5) ranges from 840 to 1525 ppm in lapilli glasses, 742 to 1403 ppm inglass selvages, 1462 to 1766 ppm in high-Fe inclusions, and 1104 to Fig. 4.  Selected MgO variation diagrams for the northern cone. Inclusions corrected for post-entrapment olivine crystallization (PEC). All analyses normalized to 100%, volatile-free,total iron as FeO T . Solid arrows represent fractionation of 15% olivine (Ol) in 5% increments. Dashed arrow represents fractionation of 15% olivine followed by 10% clinopyroxene(Ol+Cpx). Fine solid lines connect low-Fe inclusions from the same host crystal. Trends modeled for equilibrium crystallization from composition of least differentiated (highestMgO) inclusion, using PETROLOG software (Danyushevsky, 2004).501 C.I. Schipper et al. / Earth and Planetary Science Letters 295 (2010) 497  – 510
Related Search
Related Docs
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks