Effect of syneruptive decompression path on shifting intensity in basaltic sub-Plinian eruption:...

Journal of Volcanology and Geothermal Research 198 (2010) 158–176

Contents lists available at ScienceDirect

Journal of Volcanology and Geothermal Research

j ourna l homepage: www.e lsev ie r.com/ locate / jvo lgeores

Effect of syneruptive decompression path on shifting intensity in basaltic sub-Plinianeruption: Implication of microlites in Yufune-2 scoria from Fuji volcano, Japan

Yuki Suzuki ⁎, Toshitsugu FujiiEarthquake Research Institute, Univ. of Tokyo, 1-1-1, Yayoi, Bunkyo-ku, Tokyo, 113-0032, Japan

⁎ Corresponding author. Tel.: +81 3 5841 5746; fax:E-mail addresses: [email protected] (Y. Suzu

(T. Fujii).

0377-0273/$ – see front matter © 2010 Elsevier B.V. Adoi:10.1016/j.jvolgeores.2010.08.020

a b s t r a c t
a r t i c l e i n f o
Article history:Received 25 March 2010Accepted 20 August 2010Available online 24 September 2010

Keywords:basaltic Plinianmicrolitecrystal size distributionoutgassingconduit flow

To constrain the timing and conditions of syneruptive magma ascent that are responsible for shiftingeruption intensity, we have investigated a basaltic sub-Plinian eruption that produced Yufune-2 scoria in Fujivolcano 2200 years ago. We deduced magmatic decompression conditions from groundmass microlitetextures, including decompression path (i.e. evolution in decompression rate) and approximate decom-pression rate, in order to relate them to eruption intensity. The microlites revealed decompression conditionsafter water saturation at 700–1100 m depth.The temporal change in scoria size indicates that the magma discharge rate and resultant eruption intensityincreased from unit a to unit b, and then declined toward ending units d and e. The overall decompression ratein each eruptive unit has a positive correlation with eruption intensity. The variation in decompression ratewas enlarged in the final units, where the maximum remained the same as the peak through the eruption(0.13–0.22 MPa/s for units b and c), while theminimumwas 0.025 MPa/s. The large variation here is due to 1)variation in flow velocity across conduit and 2) part of the erupted magma in unit d experienced remarkablyslow decompression (0.002–0.003 MPa/s) resulting from decreased overpressure in the reservoir followingthemajor eruption of unit b. Furthermore, crystal size distribution (CSD) of microlites implied that the earliesterupted magma (unit a) had once been decompressed slowly (0.005–0.012 MPa/s), having been arrested bymaterial in the conduit–vent system, whichwas followed by an increase in decompression rate due to removalof the material at the initiation of the eruption. In addition, the magma that had been ascending slowly beforethe unit-d eruption may record the increase in decompression rate. This increased rate resulted from beingpushed up by the successive magma at the start of that eruption.Two factors had amajor impact on eruption intensity. First, magma decompression rate determined the degreeof gas-phase separation from ascending magma. Judging from CSD, different decompression rates had beengenerated at least at the start of microlite crystallization. The second factor is the conduit radius that, incombination with magma ascent rate, controlled the magma discharge rate. Before the major eruption of unitb, the conduit radius likely increased, as evidenced by xenoliths of basaltic lava and lithic fragments with thesame petrography as the xenoliths in unit a. In unit e, the conduit radius decreased through inwarddevelopment of high-density magma from the conduit margin.

+81 3 3812 6979.ki), [email protected]

ll rights reserved.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

The syneruptivemagmaascent from the reservoirhas been receivingattention recently, because different conditions at this stage producevariable eruption styles and intensities, even when magmas contain asimilar amount of dissolved water. For felsic magma, Woods andKoyaguchi (1994) proposed that the shift between lava dome formationand Plinian eruption depends on both magma discharge rate, which iscontrolled by magma ascent rate and conduit radius, and on reservoiroverpressure. Therefore, shifting style and intensity should be inter-preted in the context of eruption progress. Among these parameters,

magma ascent rate has been a major target of petrological and texturalstudy as it can be directly estimated from volcanic ejecta.

Slower ascent generally enhances outgassing (i.e. separation frommagma of the gas phase formed in syneruptive decompression) (e.g.Burgisser andGardner, 2004). This tends to causemore effusive eruption(Pioli et al., 2008, 2009, example of basaltic magma). In high-viscosityfelsic magma, bubbles are always coupled with magma. Therefore,outgassing requires permeability development through connection ofvesicles (e.g. Eichelberger et al., 1986). In less viscous basaltic magma,decoupling of bubbles and magma, as exemplified by upwardsegregation of bubbles, occurs rapidly. Basaltic Plinian eruptions,however, are likely to require a coupling between the bubbles andmagma to develop the gas pressure necessary for an explosive eruption(e.g. Sable et al., 2006). Viscosity of basalticmagma in syneruptive ascentcan be increased greatly by rapid crystallization in high temperature

http://dx.doi.org/10.1016/j.jvolgeores.2010.08.020

mailto:[email protected]

mailto:[email protected]

http://dx.doi.org/10.1016/j.jvolgeores.2010.08.020

http://www.sciencedirect.com/science/journal/03770273

Fig. 1. Isopach map for Yufune-2 scoria (Yu-2) erupted in the last summit eruption ofYounger Fuji volcano, with the sampling site in this study. Modified after Miyaji (2007)and Miyaji (1988). Inset shows the location of Fuji volcano in the Izu-Mariana arc. VF(inset), volcanic front.

159Y. Suzuki, T. Fujii / Journal of Volcanology and Geothermal Research 198 (2010) 158–176

magma (e.g. Lofgren, 1980), as pointed in Taddeucci et al. (2004) andSable et al. (2006). This physical property change may shift outgassingmode from bubble segregation to permeability development. Because ofthe above known complexities, studies of particular basaltic eruptionsare required to verify and improve theoretical models.

Groundmassmicrolite texture, characterizedby crystallinity, numberdensity, crystal form, and crystal size distribution, provide informationon subsurface magma behavior (e.g. Cashman, 1992; Cashman andBlundy, 2000; Rutherford and Gardner, 2000; Blundy and Cashman,2008). Studies on natural ejecta are supported by experimental studiesthat quantify the kinetics of decompression-induced crystallization(Hammer and Rutherford, 2002; Martel and Schmidt, 2003; Szramek etal., 2006; Suzuki et al., 2007). The past ten years have seen increasingattempts to relate groundmass texture with eruptive style and intensityand to obtain a better picture of magma flow in the conduit forintermediate to felsic magma eruptions (Gardner et al., 1998; Hammeret al., 1999; Hammer et al., 2000;Noguchi et al., 2006; Clarke et al., 2007;Martel and Poussineau, 2007; Suzuki et al., 2007; Castro and Gardner,2008;Noguchi et al., 2008).On theother hand, basalticmagmaeruptionshave not been the focus of much study until recently (Taddeucci et al.,2004; Polacci, et al., 2006; Sable et al., 2006; Szramek et al., 2006;Andronico et al., 2009; Erlund et al., 2009). This has resulted in a poorknowledge of natural microlite textures and limitation of texturalconstraints on magma decompression rates (estimated only in Szrameket al., 2006; Toramaru et al., 2008).

Furthermore, despite all of these efforts including those for felsicmagma, some important aspects of syneruptive magma ascent have notbeen discussed fully. For example, most previous studies have aimed toreveal the average ascent rate of each magma parcel throughsyneruptive magma ascent, not its ascent path (i.e. evolution in ascentrate, including temporal arrest). If ascent path were known, itscomparison with eruption intensity and style would provide novelinsight into theconditionsandtiming thatdetermined thefinal eruption.Crystallinity, numberdensity and crystal form tend to reflectnot only theintegrated sum of a decompression event but also decompression path.For example, crystallinity and number density data may tell us time-evolution of nucleation and growth and, thus, decompression path.However, this method requires a series of magmas that followed almostthe same decompression path but were quenched at different times.Crystal form may change with time-evolution of magma supersatura-tion, but individual crystals do not elucidate when the change insupersaturation tookplace. Instead, crystal sizedistribution (CSD) canbeused to supplement these approaches. CSD holds clues as to ascent pathbecause its shape reflects the history of crystal nucleation and growth(e.g. Marsh, 1998) and the relative time of change is known from crystalsize. Because of the time-consuming nature of CSD acquisition, effectiveuse of CSD for basalticmagma eruption is found exclusively in Taddeucciet al. (2004) todate. Fortunately, recentdevelopments indataprocessing(Jerram and Higgins, 2007) have made it easier to acquire CSDs for alarge number of samples from a series of eruptive activities.

The present work focuses on a basaltic sub-Plinian eruption of Fujivolcano, 2200 years ago (eruption of Yufune-2 scoria; Fig. 1), in order toconstrain the timing and ascent conditions of syneruptive magma thatare responsible for determining eruption intensity. To make thiseruption most effective for investigating this problem, we tried tocombine microlite textural data with geological records of the eruptionprogress, under the recognition that shifting intensity should beinterpreted in the context of eruption progress. The geological recordsfound in the present study include the opening of the conduit–ventsystem in view of the presence of lithic fragments, the changingmagmadischarge rate, and the temporalwane and cease of the eruptive activity.Microlite textures, including CSD, allowus to document the evolution ofmagmaascent in the conduit through aneruption. For eachmagmaunit,we relate decompression conditions (rate and path) with degree ofoutgassing so that we can examine whether motion of bubbles relativeto magma had a role in changing eruption intensity.

2. Fuji volcano and Yufune-2 scoria

Mt. Fuji, rising 3776 m above sea level, is one of the largest volcanicedifices in Japan. It is situated at the junction of the Northeast Japan arcand Izu-Mariana arc in central Japan (Fig. 1). Subduction of the Pacificplate beneath the Eurasian plate is the primary process of magmageneration. Tsuya (1940) revealed that Mt. Fuji is a compositestratovolcano consisting of Komitake volcano (older than 100 ka),Older Fuji volcano, and Younger Fuji volcano, in decreasing order age.Recent scientific drilling discovered Pre-Komitake volcano (260–160 ka) beneath the Komitake volcano, based on ejecta compositions(Nakada et al., 2007; Yoshimoto et al., 2010). The total volume of ejectafrom Fuji volcano (both Older and Younger) reaches 400 km3. Fujivolcano has issued mostly basaltic magma (e.g. Takahashi et al., 1991;Togashi et al., 1991, 1997; Kaneko et al., 2010), but the basalts areevolved in terms of FeO*/MgO (larger than 1.6) due to high-pressurecrystallization (Fujii, 2007). Older Fuji volcano began its activity atnearly the same position as Younger Fuji on the southern slope of theKomitake volcano at 100 ka (Machida, 1964, 2007). The activity of OlderFuji volcano is characterized by ejection of voluminous pyroclastic falls(Uesugi, 1990; Kaneko et al., 2010). Following previous geological study(e.g. Tsuya, 1968), tephrostratigraphy and 14C age determination (e.g.Miyaji, 1988; Yamamoto et al., 2005), Miyaji (2007) divided the activityof Younger Fuji volcano (past 11,000 years) into 5 stages. The activity ofYounger Fuji volcano is characterized by changing eruption styles (lavaeffusion, explosive activities including pyroclastic flow) and eruptionlocations (summit,flank) at different stages. After theHoei eruption (AD1707), no activity has been recorded.

Yufune-2 scoria, the target of the present study, was produced at theend of Stage-4 activity of Younger Fuji volcano (2.2 ka;Miyaji, 1988). TheYufune-2 scoria is one of the wide-spread scoria fall deposits from thesummit that characterize Stage-4 activity (3.5–2.2 ka). The eruption stylewas sub-Plinian throughout and total volume of ejecta reaches 0.5 km3

160 Y. Suzuki, T. Fujii / Journal of Volcanology and Geothermal Research 198 (2010) 158–176

(Miyaji, 1982). Yufune-2 scoria is distributed over a vast area of theeastern side of Fuji volcano with an eastern dispersal axis (Fig. 1; Miyaji,1988, 2007).Miyaji (1982, 1988) divided the scoria deposit intofiveunits(a–e) based on color and vesicularity. However, details of eruptionprogress have not been discussed for eruption of Yufune-2 scoria.

3. Samples

Sampleswere collected10 kmeastof the summit, on thedispersal axis(Figs. 1 and 2a). In this location, deposits consist dominantly of juvenilescoriae with limited lithic fragments. We first divided the scoria depositinto 5 units (a–e) (Fig. 2b) in accordance with Miyaji (1982, 1988). Eachunit was defined based on scoria size, morphology, and ash abundance.Scoria size increases fromunit a to unit b, but then decreases in the upperunits (Fig. 2c, average of five largest scoriae). Scoriae fromunits b, c, and dare characterized by a weakly elongated and twisted form, butconservation of such form is limited due to fragmentation. Scoriae fromunit e are angular and blocky, while those from unit a possess acauliflower-like, irregular form. These changes in scoria shapemay reflectchanges in eruption conditions. Unit c and the upper part of unit a arereddish in color because they include relatively abundant ash (Fig. 2a).

Furthermore, unitsbande, eachofwhichappears almosthomogenousand has relatively large total thickness, were divided into 3 and 2 sub-units, respectively (lower, middle and upper for b; lower and upper for e;Fig. 2b) for more detailed description. In addition, unit a excluding theupper part with high ash concentration, was divided into two units byslight difference in scoria size (lower and middle in Fig. 2b). Thus, wedescribe the juvenile scoriae according to 9 eruptive units (Fig. 2b).Samples for each eruptive unit were collected evenly from each horizon.

Fig. 2. Photograph (a) and columnar section (b) for the examined outcrop of Yufune-2 scorrelative thickness among eruptive units (a–e) as shown in b, because the photograph was taunit c) with relatively high ash content. c average and ±1σ for diameters of five largest sc

Lithic fragments are limited to a-middle (10% in number) and e-upper (5% in number). Lithic fragments from e-upper are dense,relatively fresh lava, while lithic fragments from a-middle are lessdense, altered lava (e.g. sample No. 13; Fig. 1b Appendix). In theexamination of thin sections, we find less dense lithic fragmentsincluded in the scoriae of unit a (in Section 5.2 and the Appendix).

4. Methods

4.1. Bulk density and sample selection

For each eruptive unit, 10 to 13 scoriae were measured for bulkdensity, with the exception of unit c (4 scoriae), in which fresh scoria islimited. We avoided use of relatively small scoria in order to minimizeanalytical error and to remove the effects of small-scale heterogeneity.Bulk densitywas calculated fromscoriaweight and volume. The volumeof each scoria wasmeasured using 3D Laser Scanner (LPX-1200, RolandDG Corporation Japan) at the Earthquake Research Institute (ERI),University of Tokyo. The scanning pitch was set to 0.1 mm. During thescan, the samplewasfixed to a rotating tablewith clay (Fig. 3, Step 1). Inorder to scan the entire object, we performed two scans with differentsample orientations (Fig. 3, Step 1). The bundled software enables us tosynthesize apolygon fromthedata of twoscans(Fig. 3, Steps 2–3) and tomanually fill holes on the synthesized polygon. The software thencalculates volume of the final polygon which approximates theenveloping surface for a scoria. The precision in volume measurementis approximately±2%,whichwas assessed by takingmeasurementsof aset volume of claymolded in several shapes. Bulk densitywas convertedto bulk vesicularity using a representative vesicle-free (DRE) density

ia (Yu-2). The outcrop locates 10 km east of the summit (Fig. 1). a does not show real,ken gazing up the outcrop. In a, arrows indicate reddish parts (upper part of unit a andoriae which were selected from ca. 50 scoriae randomly sampled.

image of Fig.�2

Fig. 3. Procedures of 3D scanning and data handling, before measuring scoria volume. In Step 1, sample is scanned from two directions, in order to scan entire object. Then, a polygonis synthesized from the two scanning data (Steps 2 to 3). Numbers in Step 2 indicate direction in Step 1.


(2.8 g/cm3). The vesicle-free density was defined based on four sam-ples covering a variety of phenocryst content in erupted magma(Section 5.2) and calculated using proportions of phases (e.g. pheno-cryst, groundmass) and their densities.

The density distribution of each individual eruptive unit was usedas a filter to select a smaller subset of 4–6 clasts, including those withmaximum, minimum, and medium densities. Thin sections of thesesubsets were prepared to investigate phenocryst phases and topreliminarily examine groundmass textures. We fabricated thinsections in a manner that enables us to observe the near-center ofeach scoria by microscopy. For description and quantification ofgroundmass microlites, we selected 3–4 samples from each set of 4–6

Table 1Bulk density, groundmass microlite phase and plagioclase microlite texture.

Sample Microlite phase

Eruptive unit No. Density(g/cm3)

Sample typea Part in sample

e-lower 12 2.27 High – Pl, Ol, Pig, Ox11 1.78 Medium Fine Pl, Ma

Coarse Pl, Ol, Pig, Ox9 1.08 Low – Pl, Ma

d 10 1.17 CGS – (Pl, OlN5 μm), (11 1.08 CGS – (Pl, OlN5 μm), (9 1.46 High – Pl, Ol, Pig, Ox

13 1.05 Medium Fine Pl, MaCoarse Pl, Ol, Pig, Ox

7 1.01 Low – Pl, Mac 2 1.15 High – Pl, Ma

4 1.10 Medium – Pl, Ma, Ox1 1.03 Medium – Pl, Ma, Ox3 0.82 Low – Pl, Ma

b-middle 4 1.26 High – Pl, Ma, Ox1 1.11 Medium – Pl, Ma, Ox5 0.97 Medium – Pl, Ma2 0.75 Low – Pl, Ma

a-middle 1 1.42 High – Pl, Ma,12 1.37 Medium – Pl, Ma, Ox9 1.18 Low – Pl, Ma, Ox

a-lower 10 1.60 High – Pl, Ma, Ox5 1.49 Medium – Pl, Ma, Ox4 0.96 Low – Pl, Ma, Ox

Pl, plagioclase; Ol, olivine; Pig, pigeonite; Ma, mafic phases (Ol and Pig are not distinguishea Based on bulk density, except for coarsely grained sample (CGS) of unit-d. For detail seb Magnification of analyzed BSE images.c Total area of analyzed groundmass.d Based on several images counted.e Phase differs depending on size of microlite. Approximate boundary crystal width is sh

samples, so that the subsets cover the range in bulk density for eachunit. We examined two additional samples for unit d, because thesetwo samples differ considerably from others, as described inSection 6.1.

4.2. Chemical analysis

Bulk-rock compositions of each scoria and lithic sample weredetermined by XRF (PW2400) at ERI, University of Tokyo, using glassbeadswith five parts flux to one part sample. Plagioclase compositionswere analyzed with a JXA-8800R EPMA at ERI, University of Tokyo.Major and trace elements were analyzed using an accelerating voltage

Plagioclase fraction in groundmass Average plagioclase size

Imageb Area(μm2)c

vol.% 1σd Sn(μm)

×800 112,265 35.4 (2.0) 3.94×800 92,046 46.1 (3.1) 3.18×400 247,161 25.5 (3.8) 6.26×800 80,971 40.8 (6.9) 3.06

Pl, Mab5 μm)e ×400 390,845 47.3 (4.7) 4.36Pl, Mab5 μm)e ×400 383,510 42.7 (3.9) 4.45

×800 94,639 43.0 (4.8) 4.37×800 53,877 35.7 (2.5) 3.39×800 80,461 31.4 (5.3) 4.57×800 58,931 39.8 (3.9) 3.63×1000 54,610 44.0 (4.7) 3.32×1000 49,059 50.0 (2.3) 3.19×1000 58,102 43.5 (2.3) 3.59×1000 41,813 43.6 (4.1) 3.25×1000 57,890 43.0 (2.8) 3.32×1000 49,217 41.6 (3.5) 3.68×1000 47,215 35.9 (3.7) 2.97×1000 56,946 32.0 (3.4) 3.45×800 104,768 48.9 (4.7) 3.85×800 71,761 47.2 (4.3) 3.32×800 79,047 45.6 (4.3) 3.54×800 98,748 49.1 (3.1) 4.11×800 96,728 45.4 (2.0) 3.49×800 69,885 35.2 (4.0) 3.31

d in BSE); Ox, Fe–Ti oxides.e text.

own.

image of Fig.�3

Table 2Plagiolase microlite textural data and analytical conditions.

Sample Conditions in acquiring 2D crystal size distribution (CSD) 3D Crystal formb Number densityd

EruptiveUnit

No. SampleType

Part insample

Analysis-1 Analysis-2 Form-1 Form-2 2DN/mm2

3D-1N/mm3

3D-2N/mm3

Image Area (μm2) Targeta Count Image Area (μm2) Targeta Count S:I:L R2 c Targeta S:I:L R2 c Targeta

e-lower 12 High – ×800 14,485 – 331 – – – – 1:4:10 0.64 – – – – 22,851 1,637,825 –

11 Medium Fine ×800 6688 – 305 – – – – 1:4:10 0.78 – – – – 45,602 4,339,070 –

Coarse ×400 94,043 – 612 – – – – 1:3.4:9 0.64 – – – – 6508 267,284 –

9 Low – ×800 10,133 – 442 – – – – 1:4:10 0.84 – – – – 43,619 4,166,020 –

d 10 CGS – ×800 17,397 b10 μm 432 ×400 390,845 N4 μm 445 1:2.7:9 0.64 b5 μm 1:1.7:5 0.73 N5 μm 24,842 1,810,111e 3,479,307f

11 CGS – ×800 11,886 b10 μm 254 ×400 383,510 N4 μm 265 1:2.5:10 0.43 b5 μm 1:2:10 0.73 N5 μm 21,559 1,270,126e 1,042,804f

9 High – ×800 12,293 – 277 – – – – 1:5:10 0.84 – – – – 22,534 2,239,280 –

13 Medium Fine ×800 11,356 – 352 – – – – 1:5:10 0.70 – – – – 30,997 4,188,137 –

Coarse ×800 21,273 – 320 – – – – 1:2.2:8 0.72 – – – – 15,043 725,980 –

7 Low – ×800 16,285 – 492 – – – – 1:4.5:10 0.76 – – – – 30,212 2,365,835 –

c 2 High – ×1000 7759 – 309 – – – – 1:4:10 0.77 – – – – 39,827 4,650,230 –

4 Medium – ×1000 9362 – 461 – – – – 1:4:9 0.79 – – – – 49,241 6,994,210 –

1 Medium – ×1000 9567 – 323 – – – – 1:6:10 0.58 – – – – 33,762 5,396,930 –

3 Low – ×1000 11,976 – 495 – – – – 1:4:10 0.69 – – – – 41,332 4,505,470 –

b-middle 4 High – ×1000 8826 – 344 – – – – 1:6:10 0.77 – – – – 38,976 6,838,350 –

1 Medium – ×1000 12,374 – 380 – – – – 1:6:10 0.57 – – – – 30,711 4,463,500 –

5 Medium – ×1000 8984 – 366 – – – – 1:6:10 0.77 – – – – 40,738 6,281,490 –

2 Low – ×1000 9903 – 267 – – – – 1:4.5:10 0.67 – – – – 26,963 2,955,740 –

a-middle 1 High – ×800 11,359 All size 365 ×800 65,009 N5 μm 198 1:2.2:10 0.69 – – – – 32,898 1,518,091 –

12 Medium – ×800 12,578 All size 532 ×800 59,183 N5 μm 131 1:3:10 0.75 – – – – 42,822 3,538,691 –

9 Low – ×800 13,584 All size 490 ×800 60,619 N5 μm 158 1:3:10 0.72 – – – – 36,338 2,294,199 –

a-lower 10 High – ×800 13,307 All size 388 ×800 85,442 N5 μm 201 1:2.4:10 0.71 – – – – 29,064 1,502,464 –

5 Medium – ×800 14,769 All size 552 ×800 81,959 N5 μm 185 1:3.4:10 0.69 – – – – 37,284 3,552,124 –

4 Low – ×800 6256 All size 190 ×800 51,500 N5 μm 125 1:2.7:8 0.81 – – – – 32,119 2,905,610 –

In acquisition of 2D CSD, we analyzed two sample domains for different crystal size (Analysis-1, 2), if charged withmicrolites of relatively large size variation (unit-a and coarsely grained sample (CGS) of unit-d). If the size variation is large (e.g.CGS of unit-d), different magnification employed. Finally, 2D CSD of one sample was estimated by combining two CSDs from two domains. 2D CSD and 3D crystal form, explained and presented here, were used to determine final 3D CSD inFig. 11.

a Crystal width.b Average form of all crystal size, except for coarsely grained sample (CGS) of unit-d.c Fractional measure of the variation, in application of Morgan and Jerram (2006).d Calculated summing numbers of all size classes in 2D or 3D CSD.e Based on 3D crystal size distribution that was estimated using Form-1(Fig. 11).f Based on 3D crystal size distribution that was estimated using Form-2.

162Y.Suzuki,T.Fujii

/Journal

ofVolcanology

andGeotherm

alResearch

198(2010)

158–176

Fig. 4. Histograms for bulk density of scoriae. Asterisk shows samples that mostly coverbulk density variation for each eruptive unit and were selected for groundmass (GMS)textural analysis (Tables 1 and 2). Square indicates coarsely grained sample (CGS) ofunit d, also analyzed for groundmass texture (Tables 1 and 2; Section 4.1). In calculatingvesicularity, 2.8 g/cm3 was assumed for vesicle-free density.


of 15 kV and a beam current of 12 nA. The beam diameter was set at5 μm. Measuring time for the peak and the two backgrounds variedbetween 15 s and 56 s, depending on element.

Table 3Representative bulk major element compositions for Yufune-2 scoria and basaltic lithic frag

Eruptive unit a-lower a-middle a-middle

Sample no.a No. 4 No. 1 No. 13

Sample type Scoria Scoria Lithic fragment

wt.%c

SiO2 50.56 50.91 50.54TiO2 1.29 1.25 1.36Al2O3 18.38 17.78 17.97FeO 11.09 10.88 11.48MnO 0.19 0.19 0.19MgO 5.62 5.92 5.74CaO 9.69 9.83 9.54Na2O 2.41 2.46 2.37K2O 0.55 0.57 0.57P2O5 0.23 0.22 0.24

a Most samples excluding lithic fragment were analyzed also for groundmass microlite tb Coarsely grained sample (CGS). For detail, see Tables 1 and 2 and the text.c All oxide values are normalized to 100% and total iron is given as FeO.

4.3. Quantitative textural analysis

To characterize groundmass textures, we analyzed plagioclase, themost abundant phase. Backscattered electron (BSE) images were usedto determine volume percent, crystal size distribution (CSD), andnumber density of plagioclase. BSE imageswere taken using a JEOL JXA-8800Rmicroprobe at ERI, University of Tokyo, at conditions of 15 kVand12 nA and magnifications of 1000×, 800× and 400× (Tables 1 and 2).We adjusted the magnification according to size and its variation ofplagioclase microlites (Tables 1 and 2). Image resolutions are ca. 0.1 μmper pixel at 1000× and 800×, and ca. 0.25 μm at 400×. BSE images wereanalyzed using ImageJ software program (version 1.41). For volumepercent estimates, two-dimensional measurements were used becausemicrolites were not oriented. In converting two-dimensional CSD intothree dimensions, we used CSD correction 1.38(Higgins, 2000, 2002).For two-dimensional CSD, we measured the short axis of crystals (i.e.width), not the long axis (i.e. length). Application of this correction toorthogonal plagioclase requires finding the ratios of the short (S),intermediate (I), and long (L) axes of the crystals in 3D. The ratio wasdetermined by the method of Morgan and Jerram (2006), measuringwidth/length ratio of each plagioclase for several hundreds of crystals(Table 2). In running CSD correction 1.38, themodel shapewas set to bea block, and the fabric was set to be massive. Number densities for both2D and 3D (Table 2) were determined by summing counts of all sizeclasses in CSD. Average crystal size (Sn; μm) in 2D was estimated usingthe method of Hammer et al. (1999):

Sn = 1000* φ=Nað Þ1=2 ð1Þ

where φ is crystal fraction in two dimensions and Na (N/mm2) is thenumber of crystals in two dimensions.

5. Bulk density and petrography

5.1. Bulk density

For analyzed populations of each eruptive unit, we found nocorrelation between volume and bulk density. No apparent gap in densitydistribution was found among eruptive units (Fig. 4). However, averagedensitywas found to change systematically; it decreases fromunit a (1.3–1.4 g/cm3 on average) to units b and c (1.0–1.1 g/cm3 on average) andincreases from units b and c to unit e (1.4–1.5 g/cm3 on average). Eacheruptive unit has continuous density distribution. The densities in units band c correspond to the lower tails of distributions of other eruptive units.The density distribution broadens for two units of e, with standarddeviations (1σ) of 0.29–0.37 g/cm3. Samples exceeding 1.8 g/cm3 in

ment coexisting with the scoria in deposit.

b-middle c d d e-lower

No. 1 No. 1 No. 9 No. 10b No. 1

Scoria Scoria Scoria Scoria Scoria

51.03 51.02 50.90 50.76 51.031.28 1.28 1.31 1.27 1.29

18.05 17.83 18.17 17.93 17.8410.81 10.89 10.92 10.92 10.860.19 0.19 0.19 0.19 0.195.52 5.70 5.52 5.92 5.749.77 9.71 9.68 9.72 9.692.53 2.53 2.49 2.50 2.510.60 0.61 0.59 0.58 0.600.23 0.24 0.24 0.23 0.24

exture (Tables 1 and 2).

image of Fig.�4


density are limited to unit ewith a proportion of ca. 20%. Bulk densities of0.7 g/cm3 and 2.3 g/cm3, the lowest and highest observed values,correspond to bulk vesicularities of 75% and 20%, respectively (Fig. 4).

5.2. Petrography

The bulk SiO2 content of the Yufune-2 scoria is nearly constant,independent of eruptive unit (50.5–51.2 wt.%; Table 3). Scoriae fromunits a-lower and a-middle include xenolithic fragments of basaltic lava

Fig. 5. Images for representative plagioclase microlite textures in Yufune-2 scoria. The scaleincreasing bulk density rightward. Two images for “medium-density scoria” from units d andc are not shown, they resemble those of unit a-middle and unit b-middle, respectively. Coa

(Appendix) that constitute less than 10 vol.% of each scoria. Thexenolithic fragments appear ocher in hue within more fresh, blackhost parts (Fig. 1 in Appendix), helping us distinguish the two. A lithic ofbasaltic lavawith the same petrographical characteristics as the xenolithhas a bulk rock composition similar to xenolith-free scoria from othereruptive units (Table 3) explainingwhy unit a bulk rock compositions ofYufune-2 scoriae do not differ from those of other units. Hereafter,we donot describe xenoliths and lithic fragments except as needed. Aside fromthe xenolith fragments, scoria samples are mostly homogeneous

bar at the top is applicable to all. For each eruptive unit, images are listed in sequence ofe-lower correspond to “coarse” and “fine” parts. Although textures for units a-lower andrsely grained sample (CGS) image of unit d is from sample No. 11 (Tables 1 and 2).

image of Fig.�5

Fig. 6. BSE images for heterogeneous groundmass microlite texture (“medium-densityscoria” of unit e-lower). a boundary between “coarse” (C; left) and “fine” (F, right)parts, b magnification of the enclosed part in a. In fact, the “coarse part” is incorporatedwithin the “fine part”. Note the boundary between the two parts is not angular and thetextural change is not gradual. In “coarse part”, groundmass phases are Fe–Ti oxides,olivine, glass, pigeonite, and plagioclase (order of brightness).


regardless of their eruptive unit (e.g. olivine and plagioclase inphenocryst assemblage). All phenocrysts (≤2 mm) are euhedral andlack reaction rims, indicating equilibrium with melt during thesyneruptive stage. However, phenocryst composition and moderevealed that two magmas with slightly different degrees of crystalli-zation (i.e. phenocryst abundance)were eruptedwith orwithoutmixingjust before the eruption to form these scoriae. The less crystallizedendmember with 3 vol.% phenocrysts includes high-Fo (80–76) olivineand high-An (92–85) plagioclase, while the crystallized endmemberwith 17 vol.% phenocrysts is dominated by low-Fo (76–73) olivine andlow-An (85–65) plagioclase. Despite the mixing, the variation ingroundmass compositions among all samples is quite small (51.0–51.7 wt.% in SiO2, estimatedbysubtractingphenocryst composition frombulk composition). Thus, the textural variety of groundmass shownbelow is due to the syneruptive decompression condition and not tochemical variety or crystallization during the approach to equilibriumafter the mixing. The coexistence of olivine and plagioclase inphenocrysts constrains pre-eruptive storage conditions of the basalticmagma.At the real storage condition, botholivineandplagioclase shouldcrystallize from themelt of the groundmass composition (e.g. Table 1 inAppendix) at liquidus. By using MELTS algorithm (Ghiorso and Sack,1995) andchanging thepressure (50–500 MPa) andH2O content (dry tosaturation at each pressure), we find that this condition is fulfilled at lessthan250MPa,H2Ocontentof ca. 1.5–1.8 wt.%, and1100–1120 °C, if QFMbuffer and NNO buffer are assumed. The H2O content indicates thatmagma was saturated with water at ca. 20–30 MPa and at ca. 700–1100m depth during syneruptive ascent.

6. Groundmass microlites and related analysis

6.1. Overall results

For textural analysis, we selected samples representing the bulkdensity distribution from each unit (Low, Medium, and High inTables 1 and 2, Fig. 5). We selected b-middle and e-lower asrepresentatives of units b and e, respectively (Figs. 2b and 4), asthere exists no clear difference in bulk density distribution in each oferuptive units of b and e (Fig. 4). Two additional samples wereexamined for unit d (coarsely grained sample (CGS); Tables 1 and 2,Fig. 5) because the CGSs clearly differ in maximummicrolite size fromthe others (b100 μm vs. b40 μm, in length of plagioclase; Fig. 5). TheCGSs were not identified based on hand samples and account for 40%of 5 unit-d scoriae randomly selected to cover bulk density. Allsamples used for textural analysis are marked in the bulk densitydistributions in Fig. 4.

Groundmass consists of microlites and glass (Figs. 6 and 7).Microlites are clearly distinguished from phenocrysts (Nca. 100 μm)by their smaller size and higher number density, with exception of theCGSs of unit d. Microlites have assemblages of plagioclase+maficminerals±Fe–Ti oxide (Table 1). Plagioclase microlites are alwaysmost abundant, followed by mafic minerals, and Fe–Ti oxide is theleast abundant (Figs. 6b and 7). Mafic minerals include olivine andpigeonite (Figs. 6b and 7c, Table 1).

In most scoriae, microlite textures do not vary within individualsamples. Furthermore, in all scoriae, microlite textures do not have rimto core concentric zoning within individual samples. “Medium-densityscoriae” from unit d and e-lower, however, include domains withcoarsermicrolites (b10 vol.%) (Fig. 6).Wename the twoparts “fine” and“coarse”, respectively (Tables 1 and 2; Figs. 5 and 6), and describe themseparately. Microlite textures are almost similar regardless of bulkdensity in each unit from a-lower to c (Fig. 5). In units d and e-lower,“high-density samples” and the “coarse parts of medium-densitysamples” are embeddedwithmicrolites of larger size and lower numberdensity comparedwithother samples andparts in the sameunit (Fig. 5).The differences are reflected in average size of plagioclase in 2D; 3.94–6.26 μm (former) vs. 3.06–3.63 μm (latter) (Table 1).

In lower units (a-lower to c; Fig. 2), microlite textures of unita-lower resemble those of a-middle, while microlite textures of unit bresemble those of c. Plagioclase widths in unit a-lower andmiddle (upto ca.10 μm) have a wider range than in units b and c (up to ca. 5 μm)(Fig. 5). “Low-density samples” and “fine parts of medium-densitysamples” from units d and e-lower (Fig. 2) resemble samples fromunits b and c (Fig. 5). Coarsely grained samples (CGSs) of unit d arelikely to include two crystal populations; crystal number discontin-uously increases at less than 5 μm (example of plagioclase; Fig. 5). Wedefine these two crystal populations as “large-size” and “small-size”.Differing from other samples, the large-size population includesonly olivine as mafic phase microlite (Table 1). Microlites usually lackclear compositional zoning, but the large-size population in the CGSsof unit d (olivine and plagioclase; Table 1) contains zoning at rims(b5 μm in width).

6.2. Microlite form

In all samples, most plagioclasemicrolites are hopper (i.e. concave,but not swallow-tailed; Figs. 5–7) with a limited number of skeletalcrystals (i.e. swallow-tailed). Equant sections are rare compared toelongated sections except in large-size (≥5 μm in width) populationsof coarsely grained samples (CGSs) of unit d (Figs. 5–7). This indicates

image of Fig.�6

Fig. 8. Three dimensional plagioclase microlite shape, characterized by short/intermediate ratio vs. short/long ratio. Ratios of short, intermediate and long dimensionsin each sample, part (“fine” and “coarse”) and specific crystal size (coarsely grainedsample (CGS) of unit-d) are listed in Table 2. Ifmore thanone plot overlap, number of theplots and their origin are indicated. Schematic diagram at the top illustrates variation ofthe observed plagioclase shapes. With the increase of undercooling at crystallization,plagioclase shape shifts from prismatic to tabular (e.g. Hammer et al., 1999).

Fig. 7. BSE images for variable microlite textures of mafic phases (olivine and pigeonite). awas taken at a magnification of ×1000, while b and c at ×800. In all images, Fe–Ti oxides,mafic phases and plagioclase (order of brightness) are microlites, accompanied byinterstitial glass and vesicle (V). Olivine (Ol) and pigeonite (Pig) are distinguishable only inc. a “medium density sample” of b-middle (sample No. 1; Tables 1 and 2), for showingdendritic texture (Den). b “low density sample” of a-lower, for an example of skeletalcrystal with swallow tail (S.T.). c “high density sample” of e-lower, inwhich both of olivineand pigeonite are short prismatic,without skeletal growth such as swallow tail andhopper.


that 3D shapes of plagioclase microlites are not prismatic becausecrystals are most likely intersected along planes perpendicular to thelongest dimension (Higgins, 1994). Ratios of short (S), intermediate

(I), and long (L) axes of plagioclase microlites in 3D (Table 2) providemore quantitative information. These are average ratios for all crystalsize, except for CGS of unit-d (Table 2). S/L ratios from all analyses aremostly constant in the range of 0.1 and 0.125 (Fig. 8). On the otherhand, S/I ratio varies among samples and sample parts. All samplesfrom unit a, the CGSs of unit d (both N5 μm and b5 μm), and the“coarse parts of medium-density samples” of unit d and e-lowerdisplay relatively high values (0.3–0.6). In the CGSs, the large-sizepopulation has a higher S/I ratio than the smaller one. The overalltrend in Fig. 8 indicates that plagioclase microlite shape variesbetween tabular (low S/I) and prismatic (high S/I) (Top of Fig. 8).

In all samples, Fe–Ti oxide microlites show a blocky and sub-rounded form when present (Figs. 6 and 7c). On the other hand, theforms of olivine and pigeonite microlites vary significantly betweensamples (Figs. 7 and 9). The appearances of the two phases do notdiffer considerably (Fig. 7c), although olivine tends to be moreskeletal than pigeonite. Dendritic morphology is limited to all samplesof units b and c (Fig. 7a) and to the “low-density samples” of units dand e-lower (Fig. 9). Dendritic crystals are sometimes accompaniedby minor, independent skeletal crystals (i.e. swallow-tailed). Skeletalcrystals are dominant in all samples of units a-lower and a-middle(Fig. 7b), in “fine parts of medium-density samples” (units d and e-lower), and in the “high-density sample” of unit d (Fig. 9). Small-size

image of Fig.�8

image of Fig.�7


populations (b5 μm in width) in coarsely grained samples (CGSs) ofunit d also show similar morphology (Fig. 9). In “coarse parts ofmedium-density samples” (units d and e-lower) and the “high-density sample” of unit e-lower, olivine and pigeonite microlites areshort prismatic (Fig. 7c), accompanied by minor hopper crystals (i.e.concave, but not swallow-tailed) (Fig. 9). Large-size (≥5 μm inwidth)olivine microlites in CGSs of unit d have a similar shape (Fig. 9).

6.3. Volume fraction, number density of plagioclase microlites

Volume fraction and number density (in both of 2D and 3D) arelisted in Tables 1 and 2. For plotting, we employ a volume fraction vs.volume number density diagram (Fig. 10), as this enables us tocompare the relative influence of growth and nucleation in crystal-lization (i.e. degree of undercooling) for a series of ejecta (Hammeret al., 2000; Martel and Poussineau, 2007). As a whole, plagioclasemicrolites show maximum 3D number density at high volumefraction (N40 vol.%) and have minimum number density at minimumvolume fraction. However, correlation between number density andvolume fraction is not so clear due to number density variation ofnearly one order of magnitude from middle to high volume fraction(N35 vol.%).

Number densities of units b and c are relatively high (3–7×106/mm3), forming a distinct trend. Number densities of units d and e-lower (3×105–4×106/mm3) are more variable than that of unit a(lower and middle; 1.5–3.5×106/mm3), even at high volume fraction(≥35 vol.%). Compared with units d and e–lower, number densities inunit a seem almost constant regardless of change in crystallinity.Maximum number densities in units d and e-lower are comparable tominimum densities in units b and c. In units d and e-lower, numberdensity is relatively low in the “high-density sample” (clear only forunit e-lower) and the “coarse part of the medium-density sample”.

6.4. Crystal size distributions (CSDs) of plagioclase microlites

In the single logarithmic plot, CSDs in 3D (Table 2) are eitherlinear or consist of two straight segments with an inflection, whichis based on R2 value in the linear fitting (more than 0.96 in mostcases; Fig. 11). The inflected CSDs have higher intercept and slope

Fig. 10. Volume fraction vs. 3D number density for plagioclase microlites. Error bars(±1σ) for volume fraction are based on multiple images measured for each sample.No error bar for number density, because one data was obtained from a CSD of eachsample (Table 2). Plot symbol is the same as in Fig. 8. The mesh represents range forunits b and c. For a given volume fraction, higher number density indicates highercontribution of nucleation and, thus, higher degree of undercooling. CGS indicatescoarsely grained sample of unit d.

Fig. 9.Morphology variation of mafic phase (olivine and pigeonite) microlites. With theincrease of undercooling at crystallization, the morphology evolves in a sequence ofprismatic, hopper, skeletal, and dendritic.

values at small sizes. We find inflected CSDs from all samples of unita (lower and middle) (Fig. 11a) and the coarsely grained samples(CGSs) of unit d (Fig. 11b). The inflection for the CGS is relativelyprominent, as is consistent with the observation that crystal numberdiscontinuously increases at less than 5 μm (Section 6.1 and Fig. 5).Additionally, CSDs from “fine parts of medium-density samples”(units d and e-lower) and from the “low-density sample” of unit care likely inflected.

No significant difference in slope and intercept values is foundbetween CSDs that are smaller than the inflection points (of theinflected CSDs) and the other CSDs (Fig. 11). However, there exists asystematic difference over larger sizes. On the whole, the length ofthe longest plagioclase axis in 3D (horizontal axis in Fig. 11) is largerin CSDs having smaller slope values. For example, the longest axis issmallest in all scoriae of units b and c (~50 μm) with maximumslope value (ca. 200), is relatively large in all scoriae of unit a (bothlower and middle) (~100 μm) with medium slope value (ca. 100),and is greatest in CGS of unit d (~200 μm) with minimum slopevalue (ca. 30). Because the slope and size range of CSD yield numberdensity, these indicate that the difference in number density among

image of Fig.�10

image of Fig.�9


image of Fig.�11

Table 4Fraction and volumetric number density for plagioclase microlite of specific size insamples having inflected CSD.

Sample Target size(μm)b

Fraction ingroundmass

Volumetric

Eruptiveunit

No. Sampletypea

Area ratioof targetsize(%)c

Fractionof targetsize(%)d

Numberdensity(N/mm3)e

d 10 CGS ≥35.2f 85.1 40.3 11,21111 CGS ≥39.2f 85.8 36.6 5126

a-middle 1 High ≥19.6 85.7 41.9 89,09112 Medium ≥29.7 63.8 30.1 27,9919 Low ≥29.7 62.0 28.3 31,899

a-lower 10 High ≥29.7 65.7 32.2 22,2645 Medium ≥19.6 76.5 34.7 89,1244 Low ≥15.7 73.6 25.9 75,610

a Based on bulk density, except for coarsely grained sample (CGS) of unit-d. For detailsee text.

b Sizes at or larger than the infected point (long axis in 3D). For detail, see Fig. 11 andtext.

c Calculated using method in Higgins (2002).d Calculate using total fraction in Table 1 and the area ratio of the target size.e Calculated summing numbers of all size classes in 3D CSD.f Corresponding to crystals having width of ≥ca. 5 μm in 2D.


samples and parts (Fig. 10) is due to difference in CSD at large sizes.Based on the finding of inflected CSDs (Fig. 11), we added data offraction and volumetric number density of plagioclase microliteswhose size is equal to or larger than the inflection point of each CSD(Table 4).

Fig. 12. Schematic diagram for conduit processes and eruption phenomena in the sub-Plinian(Steps 1–5). CGS indicates coarsely grained sample. For detail, see text (Section 7.4.1). Widtwithout approximate estimate.

Fig. 11. Crystal size distribution (CSD) of plagioclase microlites. The horizontal axis representsamples from each eruptive unit. The figures are listed in sequence of increasing bulk densitySample No. is shown, if more than one sample were originally analyzed for each purpose (Tfitting are indicated nearby.

6.5. Compositions of plagioclase phenocryst rims

Plagioclase microlite compositions help to constrain the microlitenucleation condition (Toramaru et al., 2008), if we assume that thecrystallization took place at equilibrium. In the present case,systematic analysis of plagioclase microlites for a series of scoriae isdifficult because some scoriae are charged with fine microlites(e.g. units b and c; Fig. 5). We thus analyzed phenocryst rims thatexhibit the same brightness as the microlites in BSE images and thatare judged to have grown simultaneously with microlite nucleation.Although the rim compositions vary in each sample (An 75–65 mol%),no systematic difference in variation was found among the analyzedsamples. The variation identified probably results from analyseshaving covered different ranges of compositional zoning in therims because of 1) different rim widths, and 2) slight differences inanalytical position. The highest An content of ca. 75 records the timeof initial nucleation of the microlites.

7. Discussion

7.1. Transition of the eruption, inferred from size and component of falldeposit

In the examined outcrop, scoria size increases from unit a-lower tob-middle and decreases in the upper units (Fig. 2). This verticalchange reflects increasing and decreasing magma discharge rates anderuption intensities (Fig. 12), assuming that wind intensity anddirection did not fluctuate significantly. The high concentration of ash(Fig. 2a) may indicate a temporal wane and cease during the

eruption that formed Yufune-2 scoria. The progress of eruption is divided into five stepsh of arrows drawn with solid line reflects relative magma decompression rate, with or

s the longest axis of crystal in three dimensions. a CSDs of low, medium and high densityrightward. b CSDs of coarsely grained samples (CGSs) of unit-d (Tables 1 and 2, Fig. 5).ables 1 and 2). Mathematical expression and the correlation coefficients (R2) in linear

image of Fig.�12


deposition of the upper part of unit a and the deposition of unit c(Fig. 12) (e.g. Cas and Wright, 1987).

The quantity of lithic fragments is highest in unit a-middle (10% innumber; Section 3). Furthermore, lithic fragments are found in mostscoriae from units a-middle and a-lower (xenoliths of basaltic lava;Fig. 1 in Appendix). These characteristics indicate that the formationand/or widening of the vent and/or conduit progressed during theearly stages in this eruption (Fig. 12, Steps 1–2). Referring to thechemical characteristics of ejecta from Fuji volcano (Younger andOlder Fuji) and pre-Fuji volcanoes (Komitake and Pre-Komitake) assummarized in Nakada et al. (2007), we find the bulk composition of alithic fragment with the same petrographical characteristics as thebasaltic xenolith (sample No. 13 in Table 3) matches those of Fujivolcano. For example, TiO2 content of the lithic fragment (1.36 wt.%;Table 3) falls in the range of basalt from Fuji volcano (ca. 1–2 wt.%)rather than that of Pre-Fuji volcano (ca. 0.9–1.1 wt.%). In any case, thebasaltic xenoliths were assimilated into the basaltic magma withinthe volcanic edifice of Mt. Fuji (3776 m high). Lithic fragments are alsoinvolved in the last stage of this eruption (unit e-upper; 5% in numberas in Section 3.), though different lithology from that of unit a.This may be due to the multiplier effect between decreased magmasupply and weakened conduit wall following the climax of theeruption.

7.2. Cause and conditions that controlled groundmasss microlitecrystallization in Yufune-2 scoria

Given that the degree of disequilibrium created by the mixing ofmagma was small (Section 5.2), there exist two possibilities as to thecause of syneruptive crystallization: 1) liquidus increase induced bywater exsolution from melt in decompression (e.g. Hammer andRutherford, 2002) or 2) cooling induced by heat loss in the conduitand after ejection from the vent. With regard to 1), we estimate, basedsimply on liquidus change with pressure, that the basalt of Yufune-2scoria experienced undercooling of ~100 °C if decompressed isother-mally to 1 atm. We preclude the possibility of heat loss in the conduit,following the current model for Plinian eruption (e.g. Papale, 1998).We also consider that the scoriae of the analyzed size (≤3 cm indiameter) were quenched rapidly in the plume and further crystal-lization was limited after leaving the vent. Indeed, we find noconcentric zoning of microlite textures across all scoriae (Section 6.1).In addition, the occurrence of dendritic mafic minerals (Fig. 9),which seem to imply rapid cooling after leaving the vent, are notnecessarily correlated with scoria size; small scoriae that can becooled rapidly sometimes lack dendritic mafic minerals (e.g. unit a-lower and a-middle; Fig. 2). However, heat loss and the resultantcrystallization must be considered for unit-a scoria (both lowerand middle). The basaltic magmas of unit-a scoriae might have cooledat the point of assimilation of basaltic lava (Section 7.1 and Fig. 12,Step 2).

When undercooling results from decompression, the degree ofundercooling increases with the amount of decompression (ΔP) anddecompression rate (ΔP/Δt), as verified in recent decompressionexperiments (Hammer and Rutherford, 2002; Couch et al., 2003;Martel and Schmidt, 2003; Szramek, et al., 2006; Suzuki et al., 2007).ΔP is reflected in the H2O content of groundmass glass, ifdecompression-induced H2O exsolution from melt took place almostin equilibrium. In the present study, H2O content is not availablebecause of the high possibility of hydration in old ejecta. However, ifthe process of generation of Yufune-2 scoriae is considered, theamount of decompression resulting in microlite crystallization shouldhave been nearly identical for the series of ejecta because of thehomogeneity of magma in the reservoir (Section 5.2) and thematching eruption styles. Accordingly, we can infer relative decom-pression rates of basaltic magmas based on the degree of under-cooling reflected in the groundmass microlite textures. The microlite

texture indicates a decompression condition at less than ca. 20–30 MPa (700–1100 m depth; Fig. 12), as the decompression-inducedcrystallization occurs after magma is saturated with H2O duringascent.

7.3. Microlite texture constraints on the magma decompressionconditions

7.3.1. Average decompression rate through syneruptive ascentIn order to estimate the averaged decompression rate, we

consider two characteristics of groundmass microlites; crystal habitand shape (Figs. 8 and 9, Table 2), and the relationship betweenvolume fraction and volume number density (Fig. 10). Coolingexperiments (e.g. Lofgren, 1974, 1980) and decompression experi-ments (Hammer and Rutherford, 2002; Martel and Schmidt, 2003;Szramek et al., 2006; Suzuki et al., 2007) have demonstrated thatcrystal habit evolves in a sequence as degree of undercoolingincreases; euhedral (e.g. tabular+prismatic), hopper, skeletal, den-dritic, and spherulitic. A diagram showing the relationship betweenvolume fraction and volume number density (Fig. 10) reflects therelative influence of crystal nucleation and growth (i.e. degree ofundercooling; Section 7.2) (Hammer et al., 2000; Martel andPoussineau, 2007). For a given volume fraction, higher numberdensity indicates higher contribution of nucleation and, thus, higherdegree of undercooling.

We found no variation in crystal habit for plagioclase and Fe–Tioxide (Section 6.2.), however, mafic phases (olivine and pigeonite)show a variation of dendritic, skeletal, hopper-like, and prismaticshapes (Fig. 9). Dendritic habits, suggesting maximum undercooling,appear in all scoriae from units b-middle and c and in low-densityscoriae from units d and e-lower. Prismatic habits, suggestingminimum undercooling, appear as the primary habit in the “coarseparts of medium-density scoriae” from units d and e-lower, in “high-density scoria” of unit e-lower, and in large-size populations (N5 μm)in the coarsely grained sample (CGS) of unit d.

Degree of undercooling deduced from 3D shape of plagioclasemicrolites mostly matches inferences from crystal habits of maficminerals (Fig. 8). With the increase of undercooling, plagioclasemicrolite shape shifts from prismatic to tabular, as discussed inHammer et al. (1999) based on Pinatubo 1991 ejecta. Samplescharged with dendritic mafic microlites (all scoriae from units b and c,“low-density scoriae” of units d and e-lower) have tabular plagioclasemicrolites without exception (Fig. 8). On the other hand, crystalpopulations having prismatic mafic microlites include plagioclasemicrolites of prismatic or prismatic to tabular shape (“coarse parts ofmedium-density scoriae” from units d and e-lower, large-sizepopulations (N5 μm) in CGSs of unit d; Fig. 8b), with the exceptionof “high-density scoriae” of unit e-lower (Fig. 8b).

Relative degree of undercooling derived from crystal habit andshape is mostly consistent with the relationship demonstrated in thediagram of volume fraction vs. volume number density for plagioclasemicrolites (Fig. 10). For example, unit-b and unit-c scoriae, judged tohave experienced maximum undercooling, have a high numberdensity trend (Fig. 10). On the other hand, samples, sample parts,and crystal populations that are judged to be formed with relativelylow undercooling (“coarse parts of medium-density scoriae”from units d and e-lower, “high-density scoria” of unit e-lower,and large-size (N5 μm) populations of CGSs of unit d) have a relativelylow number density trend (Fig. 10; also see Table 4 for large size(N5 μm) populations of CGSs of unit d). Additionally, the minimumvolume fractions of plagioclase microlites for sample parts withminimum number density (“coarse parts of medium density scoriae”;Fig. 10) indicates that both nucleation and growth rates werelowest for these formations. These are achieved by a lower degreeof undercooling than that of growth rate peak (e.g. Fig. 15 of Couchet al., 2003).


Our conclusion is demonstrated schematically in Fig. 9, whichshows the morphology of mafic phase microlites. We conclude thataverage decompression rate through syneruptive ascent was rela-tively high in the formation of all scoriae of units b and c, and probablyin “low-density scoriae” from unit d and unit e-lower. Averagedecompression rate was relatively low in the formation of “coarseparts of medium-density scoriae” from units d and e-lower, of large-size (N5 μm) populations of CGSs, and probably of “high-densityscoria” of unit e-lower. Quantitative comparison of the variabledecompression rates, the decompression path of eachmagma, and themechanism for the variable rate in units d and e-lower will bediscussed in the following sections.

7.3.2. Decompression path through syneruptive ascent and simultaneouseruption of magmas that experienced diverse decompression path andrate

In order to discuss decompression path (i.e. evolution indecompression rate), we employ crystal size distribution (CSD), theshape of which provides insight into the evolution in crystalnucleation and growth. First, CSDs from a series of samples haveunique slope and intercept values over relatively large crystal sizes(Fig. 11). Because slope and size range of CSD yield crystal numberdensity, the different CSD for large crystal sizes indicates that adifferent decompression rate (Section 7.3.1) was generated at least atthe start of microlite crystallization, after magma was decompressedto ca. 20–30 MPa (700–1100 m depth; Fig. 12).

In interpreting the shape of CSDs further, we follow the batchsystem model (no outflow and inflow, as in syneruptive groundmassformation) proposed in Marsh (1988, 1998) and we assume crystalgrowth rate does not depend on crystal size. Log-linear CSD, found inmost samples (Fig. 11), is a result of an exponential increase innucleation rate with a time-varying degree of undercooling (Marsh,1988, 1998). This does not necessarily require an increase indecompression rate. Liquidus does not increase linearly withdecompression (e.g. Hammer and Rutherford, 2002), and the degreeof undercooling increases at lower pressure. Indeed, constant-ratedecompression experiments produced mostly linear plagioclasemicrolite CSDs (Suzuki et al., 2007). We thus find that most samplesexperienced nearly constant decompression, at least for the durationof microlite crystallization.

CSDs with clear inflections are found in all scoriae of unit a (lowerand middle) and in the coarsely grained sample (CGS) of unit d(Fig. 11). A kinked CSD indicates that the nucleation rate suddenlychanged to a new exponential rate (Marsh, 1988, 1998). This suddenchange in crystal nucleation rate may reflect rapid increase indecompression rate of magma. At the same time, kinked CSD couldbe an artifact of the 2D–3D conversion process when different sizeclasses have different characteristic shapes (Castro et al., 2003).Because the increase in decompression rate of magma is generallyassociated with change in crystal shape (as discussed in Section7.3.1), we cannot find crystal form change alone to be responsible forthe kinked CSD. We thus present the most plausible scenario basedon the model of the rapid increase in decompression rate. Rapidincrease in decompression rate may be caused both by internal andexternal conditions. The internal condition for rapid increase indecompression rate is created when a magma approaches thefragmentation level in an explosive eruption (Wilson et al., 1980;Sparks et al., 1994). However, we cannot attribute the observedincreases in crystal nucleation rate to fragmentation, because kinkedCSDs are not found in all samples. On the other hand, externalconditions are created when materials that filled the vent–conduit–reservoir system are suddenly removed or when magma at lowdecompression rate is incorporated into magma at high decompres-sion rate.

With regard to unit a, all samples have kinked plagioclase CSDswith similar intercepts and inclinations (Fig. 11). This indicates that

each had a similar decompression path and denies the possibility ofincorporation into magma at high decompression rate as a cause ofkinked CSD. Instead, the increase in crystal nucleation rate may havebeen caused by a condition peculiar to the earliest ascendingmagma. One possible such condition would exist at the conduit–ventopening just before the initiation of the eruption (Fig. 12, Step 2),where materials that filled the conduit–vent system are removed.This model is supported by the existence of basaltic lava xenolithsin unit a. Intercept and slope of the CSDs from unit-a scoriaeresemble those of most other scoriae at sizes smaller than theinflection (Fig. 11), indicating that final decompression conditionafter conduit–vent opening was similar to other eruptive phases.The slow ascent before that time would represent magma arrest enroute to the surface (Suzuki and Nakada, 2007; Erlund et al., 2009).The magma erupted earliest would have been arrested becausemagma ascent takes place along with formation of conduit and vent.The other condition to lead to nucleation rate increase is effectivecooling associated with the assimilation of the basaltic xenolith(Fig. 12, Step 2).

For unit d, removal of material that filled the crater–vent–reservoirsystem (one of the external condition) may have been generated asfollows. After discharge of a large volume of magma in the climax(unit b; Section 7.1 and Fig. 12, Step 3), it is probable that general,slow decompression of the reservoir preceded faster decompressionby syneruptive magma ascent. However, this model must berejected because kinked CSD is found in limited ejecta (Fig. 11).The other external condition, incorporation into the magma at highdecompression rate, seems possible because magmas decompressedat different average rates erupted simultaneously in unit d (Section7.3.1). Accordingly, the mechanism for variable decompression ratewill be discussed together with that of kinked CDSs.

There are two mechanisms that lead to simultaneous eruption ofmagmas decompressed at different rates. One is horizontal velocitygradient in a steady magma ascent that develops due to conduit walleffect. In this case, flow velocity is lowest at the conduit margin andincreases toward the center (Taddeucci et al., 2004; Castro andMercer, 2004; Sable et al., 2006). The other mechanism is related tounsteady flow. During the temporal eruption cease and wane, magmamay ascend slowly or stagnate in the conduit. At the restart of theeruption, the stagnant magma would erupt together with thesuccessive magma from the reservoir that decompressed at fasterrate (e.g. Hammer et al., 1999).

First, microlite texture in unit d is not continuous between thecoarsely grained sample (CGS) of unit d and the rest (scoriae of low,medium and high densities) (Fig. 5), suggesting that differentprocesses operated to create two groups. Second, overall variety inmicrolite texture found among the low, medium, and high-densityscoriae, is similar between units d and e-lower (Fig. 5), implying 1)that the processes that created this variety had been operatingcontinuously during the two stages and 2) that the processes are notrelated to eruption cease and wane before the unit d eruption (duringthe deposition of unit c; Section 7.1 and Fig. 12, Step 4). We thusbelieve that the microlite textural variety among scoriae of low,medium, and high densities was formed due variation in flow velocityfrom center to margin (Fig. 12, Step 5). The “medium-densitysamples” have textural heterogeneity within individual scoria(incorporation of coarse part into fine part; Fig. 5), suggesting thatregions with different flow velocities existed in close proximity(“partly mingled” in Fig. 12, Step 5).

On the other hand, coarsely grained samples (CGSs) in unit d(Fig. 5), not found in unit e-lower, may have been generated inrelation to the temporal wane or cease before the unit d eruption.Plagioclase microlites with sizes larger than the CSD inflection (N30–40 μm in long axis in 3D; Fig. 11b) correspond to the large-sizepopulation in 2D (N5 μm in width; Fig. 5), while those with sizessmaller than the CSD inflection correspond to the small-size


population in 2D. The large-size population records one of the slowestdecompressions of all samples (Section 7.3.1). We infer that the large-size population records relatively slow ascent due to decreasedoverpressure of the reservoir after the major eruption of unit b(Fig. 12, Step 3). The small-size population reflects an increase indecompression rate due to the upward force from the successivemagma (low, medium, and high-density scoriae) at the time ofoverpressure recovery (Step 5 in Fig. 12). For CGSs of unit d, theincrease of decompression rate inferred from kinked CSD (Fig. 11b) issupported by the change of microlite form with size (Figs. 8 and 9,Section 7.3.1.).

Finally, the smallest classes of CSD give insight into the lastcondition of groundmass crystallization for all samples. If crystalnucleation ceases because the system approaches equilibrium, nonew crystals will form and the large crystals will grow at the expenseof small crystals (Ostwald ripening). Such an event makes theoriginal log-linear CSD decrease in number for the smallest sizeclasses; the classes deviate downward from a linear state, relative tolarger classes (Marsh, 1988, 1998). If decompression is a primarycause of crystallization, slowed or ceased ascent leads to a cease innucleation. In the case of Yufune-2 scoria, smallest size classes ofCSD do not deviate from the linear trend (Fig. 11). This suggests anabsence of deceleration or arrest long enough to be recorded in theCSD.

7.4. Combined interpretations

7.4.1. Eruption sequenceHere, we combine findings from the outcrop-scale properties

(Section 7.1) and microlite textural records (Section 7.3). In order tohave a quantitative image of magma ascent, water exsolution rate andcorresponding magma decompression rate were estimated using themethod of Toramaru et al. (2008) that can be applied to a wide rangeof melt compositions (for detail of the estimation, see Appendix B).Because this rate meter has not been tested for basaltic magma usingproducts of decompression experiments (see Appendix B), we mightprovide only approximations of exsolution rate and magma decom-pression rate.

The magma discharge rate and eruption intensity increased fromunit a to unit b and then declined toward conclusion (Fig. 12).During the preparation stage and unit a eruption (Steps 1–2; Fig. 12),widening of the conduit–vent system progressed, as evidenced bythe xenolith of basaltic lava (Fig. 1 Appendix). Magma of unit aprobably experienced an increase in magma ascent rate due tosudden conduit–vent opening, from 0.005–0.012 MPa/s (0.05–0.13×10−2 in wt.%/s) to higher values. After a temporal wane orcease, magma ascended at the maximum speed in this eruption(0.13–0.22 MPa/s, 1.3–2.3×10−2 in wt.%/s) to form scoriae of units band c (Step 3; Fig. 12). Being different from unit a, each magmaparcel of units b and c ascended at a mostly constant rate, as far asmicrolite record is considered. Again, eruptive activity waned orceased for a while during unit c scoria deposition, probably due todecreased overpressure of the reservoir after the major eruption inunits b–c. At this stage (Step 4, Fig. 12), magma that wouldsubsequently form the coarsely grained sample (CGS) of unit dascended slowly in conduit (0.002–0.003 MPa/s, 0.02–0.03×10−2 inwt.%/s). As reservoir pressure recovers, successive magma thatwould form “low to high density scoriae” began to rise from thedeeper part of the conduit at a higher ascent rate. This resulted inthe acceleration of the magma which had been ascending slowly atthe upper part of the conduit. These magmas erupted togetherto form unit-d ejecta (Step 5; Fig. 12). The CGSs of unit d cannotbe distinguished from other scoria of unit d by hand samples(Section 6.1), so scoria samples were collected evenly from each unit(Section 3 Sample). Therefore, we cannot judge whether the CGSsform a discrete layer in unit d (e.g. at the boundary with unit c) or

not. During Step 5 (units d–e), a velocity gradient across the conduitdeveloped due to conduit wall effect. The maximum rate in thegradient was comparable to rates in units b and c, while theminimum rate was 0.025 MPa/s (0.3×10−2 in wt.%/s). If we look atthe bulk density distribution of scoria (Fig. 4), we find that magma ofminimum ascent rate, as exemplified by “high density scoria”,accounts for only 20% of all scoriae erupted in the same unit. It islikely that regions with different velocities were partly mingledduring ascent. In view of the microlite record, the variation indecompression rate was similar between unit d and unit e.

The different rates of magma decompression, as described above,had already been generated at least at the start of microlitecrystallization, after magma was decompressed to ca. 20–30 MPa(700–1100 m depth; Fig. 12). As to magmas that ascended at mostlyconstant rates, their H2O exsolution and decompression rate varybetween 0.025 and 0.22 MPa/s and 0.3×10−2–2.3×10−2 wt.%/s,respectively. If we take pre-acceleration rates for unit-a scoriae(0.005–0.012 MPa/s) and the coarsely grained sample (CGS) of unit d(0.002–0.003 MPa/s) into consideration, the two ranges are bothdecreased by one magnitude. These estimated rates are similar tothose for 1986B basaltic sub-Plinian eruption at Izu-Oshima, Japan(Toramaru et al., 2008), 0.012–1.3 MPa/s and 1.1×10−3–1.1×10−1

wt.%/s, respectively.

7.4.2. Outgassing and eruption mechanismBulk density reflects the extent and style of magma vesiculation

and loss of gas phase from magma (outgassing). If we posit a similardegree of water exsolution until magma quenching (Section 7.2), bulkdensities (Fig. 4) directly reflect the relative degree of outgassing;samples with higher bulk density experienced a higher degree ofoutgassing.

In our data, the degree of outgassing is negatively correlated withmagma ascent rate. Samples that experienced highest decompressionrates, such as units b and c scoriae (0.13–0.22 MPa/s; Fig. 12), are allcharacterized by minimum bulk densities (~0.8 g/cm3; Fig. 4). On theother hand, low decompression rates (0.025 MPa/s in minimum;Fig. 12) are inferred for “coarse parts of medium-density scoriae” ofunits d and e-lower. Also, “high-density scoria” of unit e-lower(2.27 g/cm3; Fig. 4) experienced one of the slowest decompressions(Section 7.3.1). Furthermore, magmas of unit-a and the coarselygrained sample (CGS) of unit d, which were decompressed atintermediate rates as evidenced by plagioclase microlite numberdensity (Fig. 10), produced bulk densities ranging from medium toslightly higher than the minimum (Fig. 4).

A proposed model for felsic magmas suggests that the longerresidence time in conduit enhances outgassing (e.g. Burgisser andGardner, 2004). Recent textural studies have reported a similarcorrelation for basaltic magmas erupted in varying styles (Houghtonet al, 2004; Taddeucci et al., 2004; Polacci et al., 2006; Sable et al.,2006; Szramek et al., 2006; Erlund et al., 2009). The outgassingprocess in basaltic magma may occur either by development ofpermeability through connection of vesicles or by the rise of coalescedbubbles in a relatively static magma in conduit. The mode ofoutgassing that takes place may depend on magma viscosity andascent rate. From the overall correlation between magma ascent rateand degree of outgassing in our sample, we infer that there existedalmost no heterogeneity of bubbles in the conduit and that the rise ofcoalesced bubbles did not have a significant role in outgassing. Forfurther discussion of the outgassing process, vesicle textures wouldneed to be studied in greater detail. Yet, we assert that variablemagma ascent rate, as quantified in Section 7.4.1, is partly responsiblefor changing eruption intensity by changing the degree of outgassing.As already noted, the ascent rate of magma had been differed amongmagma parcels at least at the start of microlite crystallization.Experimental determination of the precise timing of the crystalliza-tion start is needed in future study.


From the evolution in bulk density distribution from unit d to unite-lower (including unit e-upper) (Fig. 4), we infer an overall increasein the degree of outgassing in the ending stage of this eruption.The outgassed portion may have developed inward from the conduitmargin, because the magma velocity is lowest at the conduit margin(Step 5; Fig. 12) and magma of minimum ascent rate produces high-density scoria. As discussed in Houghton et al. (2004) and Sable et al.(2006), the presence of outgassed magma at the conduit margin leadsto decrease of effective conduit radius. The question remains as tohow development of the outgassed margin was triggered. The rangesin ascent rate do not differ significantly between unit d and unit e(Sections 7.3.1 and 7.4.1). We thus speculate that the overalldecompression rate decreased toward the unit e eruption, but themicrolite textures do not have enough resolution to distinguish thedifference.

In addition to the degree of outgassing controlled by magmadecompression rates, the conduit radius may have also controlledthe eruption intensity. In combination with magma ascent rate, theconduit radius determines the magma discharge rate. During thepreparation stage and unit-a eruption, magma ascent may havebeen suppressed by the relatively narrow conduit–vent system(Step 1 in Fig. 12), though overpressure of the reservoir wasrelatively high through the eruption, resulting in low magmadischarge rate. In unit e (Step 5 in Fig. 12), magma discharge rateseems to have been suppressed not only by relatively lowoverpressure of reservoir in the waning stage through eruptionand by resultant low magma ascent rate, but also by decreasedconduit radius as already discussed.

Based on the progression of the whole eruption, we discuss thetermination of this eruption. The key observation here is that the finalmagma erupted (in unit e) does not record a significant decrease inascent rate compared with previous stages (Fig. 12). This does notseem odd, however, because the magmas of unit c that were followedby temporal cease in activity also lack a decrease in ascent rate(Fig. 12). After the deposition of unit-e scoriae, magmamay have beenascending in the conduit at a relatively slow speed due to decreasedoverpressure in the reservoir, just as with the magma of the coarselygrained sample (CGS) of unit d (Fig. 12). At this stage, however, theoverpressure did not recover to a level that would allow the eruptionto start again.

8. Conclusion

To constrain the timing and conditions in syneruptive magmaascent that lead to shifting intensity in sub-Plinian eruptions ofbasaltic magma, we have studied a summit eruption that took place inFuji volcano 2200 years ago (eruption of Yufune-2 scoria; Fig. 1).Focusing on the decompression rate and paths (i.e. evolution indecompression rate) reflected in groundmass microlite textures, wehave revealed the evolution of magma flow in conduit through theeruption (Fig. 12), which formed a basis to discuss the interplayamong conduit radius, magma ascent rate, degree of outgassing(separation of gas phase from magma), magma discharge rate, andfinal eruption intensity.

(1) The Yufune-2 scoria deposit in the examined outcrop wasdivided into 5 eruptive units (a–e; lower to upper). Scoria sizeindicates that magma discharge rate and resultant eruptionintensity increased toward unit b and then decreased. Highash concentrations in the uppermost part of unit a and in unitc show temporal wane or cease in the activity. In unit a,xenoliths of basaltic lava in scoriae and lithic fragments withthe same petrography as the xenolith imply formation andwidening of the vent and/or conduit in the earliest stage.

(2) The chemical variation in the groundmass is quite small(0.7 wt.% in SiO2) for a series of scoriae. Thus, textural

variations of groundmass microlites are considered to resultfrom variable syneruptive decompression conditions.

(3) The degree of outgassing, inferred from bulk densities,decreased from unit a to units b and c, and then increasedtoward unit e. Microlite textures were analyzed for sampleswith low, medium, and high bulk densities in each eruptiveunit. Two additional sampleswere analyzed for unit d (coarselygrained samples; CGSs), because they clearly differ from othersin maximum microlite size.

(4) Averaged decompression rate of each magma parcel throughsyneruptive magma ascent increased from unit a to units b andc. Maximum decompression rate in units d and e wascomparable to those in units b and c, but overall variety waswidened due to two mechanisms. One is development ofcentral to marginal variety in flow velocity. The other is co-eruption of magma (CGSs of unit d) that ascended slowly dueto decrease in overpressure in reservoir after the majoreruption of unit b. These variable rates had already beengenerated at least at the start of microlite crystallization, butbranching depth was not specified. The average decompressionrate correlates negatively with degree of outgassing of the finalproduct.

(5) Syneruptive decompression of most magmas took place atconstant rates, except for the following cases. Removal ofmaterials that filled the conduit–vent system just before theeruption initiation led to decompression rate increase asinferred from scoriae of unit a. Before the decompression rateincrease, magma of unit a had been arrested en route to thesurface, probably because magma ascent took place along withformation of conduit and vent. Decompression rate increase,inferred from CGSs of unit d, occurred because magmaascending slowly during the eruption wane or cease waspushed up by the successivemagma at the time of overpressurerecovery.

(6) Decompression rates in the constant-rate decompression havevaried between 0.025–0.22 MPa/s. Incorporation of the ratesbefore probable decompression rate increase for unit-a scoriaeand the CGSs of unit d widens the overall ranges by onemagnitude (0.005–0.012 MPa/s for unit a; 0.002–0.003 MPa/sfor coarsely grained sample). These variable rates are partlyresponsible for the shift in the eruption intensity by changingthe degree of outgassing.

(7) In combination with the magma ascent rate, conduit radiusdetermined the magma discharge rate and resultant eruptionintensity. The conduit radius was enlarged at unit a (1), whilewas decreased by the inward development of the outgassedpart from the conduit margin in unit e.

Acknowledgements

We are indebted to Drs. S. Nakada, T. Koyaguchi, S. Nakai, A.Yasuda, T. Kaneko, F. Maeno and A. Furukawa at the EarthquakeResearch Institute, University of Tokyo (ERI), for valuable discus-sions in a volcanological seminar at ERI. A. Yasuda, T. Kaneko, andA. Furukawa are also thanked for help with field work. We thankM. Sugimori for assistance with XRF analysis. We are grateful toDr. Dougal Jerram (University of Durham) for supplying us withCSDslice database, Excel spreadsheet to calculate the 3D crystalshape. We express our thanks to Dr. A. Toramaru (KyushuUniversity) for instructions on use of his H2O exsolution meter.Finally, the manuscript was greatly improved by insightful com-ments from Dr. Katharine Cashman (University of Oregon) andDr. Julia Hammer (University of Hawai‘i). This work was partlysupported by Grant-in-Aid from MEXT to T. Fujii (No. 16089204).


Appendix A

A. Inclusion of basaltic material in scoria samples

Inclusions of basaltic material are found in most scoriae from a-lower and a-middle (10/12 samples in total) and in a limited scoriafrom e-lower (1/5 samples, with smaller volume fraction and numberthan in unit a). The inclusion is easily identified in thin section by itsocher color (Fig. 1a Appendix) and by its coarser groundmassmicrolites than those of the host. The inclusion has olivine withorthopyroxene reaction rims, plagioclase, and orthopyroxene asphenocrysts, being slightly different from the host (olivine withoutreaction rims and plagioclase; Section 5.2.). The inclusion seems tohave behaved as magma at the engulfment into the host magma,because 1) the outline of the inclusion is not angular, 2) vesicles withsmooth outline exist across the boundary between the inclusion andhost (Fig. 1a Appendix), and 3) no crystal is cut at the boundary. Wethus needed to judge whether the basaltic material originated frommagma which existed beneath the Fuji volcano just before theeruption of present study. However, finding a basaltic lithic fragment(in unit a-middle; Fig. 1b Appendix), with the same petrologicalcharacteristics as the inclusion leads us to conclude that the inclusionoriginated from the country rock. The lithic fragment has an angularoutline (Fig. 1b Appendix), which is different from the inclusion.This may be due to the smaller dimension of the inclusion (Fig. 1Appendix) leading to effective heating by the host magma and furtherremobilization and heating-vesiculation. In addition to the colordifference (Fig. 1 Appendix), Fo content of olivine phenocrysts cores(Fo68–73) being lower in the inclusion than in the host part (Fo72–80)helps us distinguish the two parts.

B. Application of H2O exsolution rate meter by Toramaru et al. (2008)

This rate meter assumes some simplified conditions for microlitenucleation, such as single nucleation event at depth and mostlyhomogeneous nucleation. Regardless of such simplification, thismodel successfully yields water exsolution rates of first-order ap-proximation for products in decompression experiments of felsicmagma (Toramaru et al., 2008). In the tested range (ca. 10–500 MPa/s), the discrepancy between the real decompression rates and the ratescalculated from microlite number density was less than one order ofmagnitude (Fig. 9 of Toramaru et al., 2008). H2O exsolution rate

}

Fig. 1. (Appendix) Basaltic xenolith as inclusion in scoria (a) and a lithic fragment having thepolarized light using microscope (a), and using scanner (b). a. Sample No. 5 from eruptiv(xenolith) and host scoria. b. Sample No. 13 from eruptive unit a-middle. Bulk rock compo

(dCH2O/dt; wt.%/s) can be expressed as a function of microlite numberdensity (MND; N/m3), as follows:

jdCH2O = dt j = kdN2=3 ðA:1Þ

where k is defined by 1/(a2/3 b).a is a constant reflecting SiO2 content in melt (CSi; wt.%) and water

content in melt at crystal nucleation (Cw; wt.%) as follows:

a = 3 × 1015F1 + 0:345ΔCSi−0:65Cw ðA:2Þ

where ΔCSi=CSi−50.b is a constant relating an increasing rate of liquidus (dTL/dt; K/s)

with dCH2O/dt, as follows:

jdTL = dtj = b⋅jdCH2O = dtj ðA:3Þ

We used groundmass composition of unit c (SiO2=51.72 wt.%;Table 1 Appendix) as representative (ΔCSi=1.72 in Eq. (A.2)). First,constant b was set to be 51.0 based on th e slope values for the linearrelationship between CH2O vs. liquidus estimated for the groundmasscomposition using the MELTS algorithm (Ghiorso and Sack, 1995). Forwater content in melt at nucleation (Cw; wt.%), we have comparedobserved anorthite (An) contents and An contents calculated usingthe MELTS algorithm (Ghiorso and Sack, 1995). Maximum An contentof plagioclase rims that have crystallized simultaneously withinitiation of microlite nucleation, is ca. An75 (6.5.). An 75 plagioclasecan be stable in liquidus melt with 0.5 wt.% H2O. We additionally usethe other water content (1 wt.%) to assess its effect on H2O exsolutionrate. We thus estimate

kpl = 6:24d10−13 ðA:4Þ

when Cw=0.5 wt.%, and

kpl = 1:03d10−12 ðA:5Þ

when Cw=1.0 wt.%.The plagioclaseMND of Yufune-2 scoria varies between 0.27×1015

and 7.0×1015 (m−3) (Table 2), which corresponds to H2O exsolutionrate of 0.3–2.3×10−2 (wt.%/s) when Cw is 0.5 wt.% and 0.4–3.8×10−2

(wt.%/s) when Cw is 1.0 wt.%. This means that Cw of this variety does

same petrography as the xenolith (b). Both are thin section images taken under plane-e unit a-lower. Note the vesicles (V) locating across the boundary between inclusionsition of this lithic fragment is shown in Table 3.

Table 1(Appendix) Groundmass composition of a representative Yufune-2 scoria.

Eruptive unit c

Sample no. No.1

(wt.%)SiO2 51.72TiO2 1.58Al2O3 15.72FeO* 12.50MnO 0.22MgO 5.68CaO 8.93Na2O 2.61K2O 0.73P2O5 0.30FeO*/MgO 2.20

Groundmass composition was determined by using bulk rock composition (Table 3),average core compositions of phenocryst phases, and phase proportion. All oxide valuesare normalized to 100% and total iron is given as FeO.


not yield a significant difference in H2O exsolution rate. This result ledus to report results for Cw of 0.5 wt.% in Section 7.4.

Given that the water exsolution proceeds at a rate of equilibriumto the decreasing pressure, decompression rate (dPH2O/dt; Pa/s) canbe related to H2O exsolution rate (dCH2O/dt) as proposed in Toramaruet al. (2008):

jdPH2O = dt j = cd jdCH2O = dt j ðA:6Þ

where c= j dPH2O/dCH2Oj.If water exsolution proceeds in disequilibrium, the decompression

rate estimated by the present method provides the minimum value.Following the Burnham's water solubility model in Holloway andBlank (1994) and assumingmagma temperature of 1100 °C, we definefor our groundmass composition:

PH2O = 9:74 × 106 × C 2H2O ðA:7Þ

Accordingly, we obtain:

c = 1:95 × 107 × CH2O ðA:8Þ

For the H2O exsolution rate and decompression rate for the specifictiming of magma ascent, we have used the following MND values.For the rate beforepossible acceleration of unit-amagma(Section7.3.2),MND for sizes larger than theCSD inflection (2.2–8.9×1013 m−3;mm−3

in Table 4) was used. Similarly, for the possible acceleration ofthe magma that formed coarsely grained samples (CGSs) of unit d(Section 7.3.2), we have used MND for the large-size (N5 μm)populations (0.5–1.1×1013 m−3; Table 4).

References

Andronico, D., Cristaldi, A., Del Carlo, P., Taddeucci, J., 2009. Shifting styles of basalticexplosive activity during the 2002–03 eruption of Mt. Etna, Italy. J. Volcanol.Geotherm. Res. 180, 110–122.

Blundy, J., Cashman, K., 2008. Petrologic reconstruction of magmatic system variablesand processes. Rev. Mineral. 69, 179–239.

Burgisser, A., Gardner, J.E., 2004. Experimental constraints on degassing andpermeability in volcanic conduit flow. Bull. Volcanol. 67, 42–56.

Cas, R.A.F., Wright, J.V., 1987. Volcanic Successions: Modern and Ancient. Allen andUnwin, London. (528 pp.).

Cashman, K.V., 1992. Groundmass crystallization of Mount St. Helens dacite, 1980–1986: a tool for interpreting shallow magmatic processes. Contrib. Mineral. Petrol.109, 431–449.

Cashman, K.V., Blundy, J.D., 2000. Degassing and crystallization of ascending andesiteand dacite. Philos. Trans. R. Soc. Lond. A. 358, 1487–1513.

Castro, J.M., Mercer, C., 2004. Microlite textures and volatile contents of obsidian fromthe Inyo volcanic chain. California. Geophys. Res. Lett. 31, L18605. doi:10.1029/2004GL020489.

Castro, J.M., Gardner, J.E., 2008. Did magma ascent rate control the explosive–effusivetransition at the Inyo volcanic chain, California? Geology 36, 279–282.

Castro, J.M., Cashman, K.V., Manga, M., 2003. A technique for measuring 3D crystal-sizedistributions of prismatic microlites in obsidian. Am. Mineral. 88, 1230–1240.

Clarke, A.B., Stephens, S., Teasdale, R., Sparks, R.S.J., Diller, K., 2007. Petrologicconstraints on the decompression history of magma prior to Vulcanian explosionsat the Soufrière Hills volcano. Montserrat. J. Volcanol. Geotherm. Res. 161,261–274.

Couch, S., Sparks, R.S.J., Carroll, M.R., 2003. The kinetics of degassing-inducedcrystallization at Soufrière Hills Volcano. Montserrat. J. Petrol. 44, 1477–1502.

Eichelberger, J.C., Carrigan, C.R., Westrich, H.R., Price, R.H., 1986. Nonexplosive silicicvolcanism. Nature 323, 598–602.

Erlund, E.J., Cashman, K.V., Wallace, P.J., Pioli, L., Rosi, M., Johnson, E., Granados, H.D.,2009. Compositional evolution of magma from Parícutin Volcano, Mexico: Thetephra record. J. Volcanol. Geotherm. Res. doi:10.1016/j.jvolgeores.2009.09.015.

Fujii, T., 2007. Magmatology of Fuji Volcano. In: Aramaki, S., Fujii, T., Nakada, S., Miyaji,N. (Eds.), Fuji Volcano. Yamanashi Institute of Environmental Sciences, pp. 233–244(in Japanese, with English Abstr.).

Gardner, C.A., Cashman, K.V., Neal, C.A., 1998. Tephra-fall deposits from the 1992eruption of Crater Peak, Alaska: implications of clast textures for eruptiveprocesses. Bull. Volcanol. 59, 537–555.

Ghiorso, M.S., Sack, R.O., 1995. Chemical mass transfer in magmatic processes IV. Arevised and internally consistent thermodynamic model for the interpolation andextrapolation of liquid–solid equilibria in magmatic systems at elevated tempera-tures and pressures. Contrib. Mineral. Petrol. 119, 197–212.

Hammer, J.E., Rutherford, M.J., 2002. An experimental study of the kinetics ofdecompression-induced crystallization in silicic melt. J. Geophys. Res. 107, 1–24.

Hammer, J.E., Cashman, K.V., Hoblitt, R.P., 1999. Degassing and microlite crystallizationduring pre-climactic events of the 1991 eruption of Mt. Pinatubo. Philippines. Bull.Volcanol. 60, 355–380.

Hammer, J.E., Cashman, K.V., Voight, B., 2000. Magmatic processes revealed by texturaland compositional trends in Merapi dome lavas. J. Volcanol. Geotherm. Res. 100,165–192.

Higgins, M.D., 1994. Numerical modeling of crystal shapes in thin sections: estimationof crystal habit and true size. Am. Mineral. 79, 113–119.

Higgins, M.D., 2000. Measurement of crystal size distributions. Am. Mineral. 85,1105–1116.

Higgins, M.D., 2002. Closure in crystal size distributions (CSD), verification of CSDcalculations, and the significance of CSD fans. Am. Mineral. 87, 171–175.

Holloway, J.R., Blank, J.G., 1994. Application of experimental results to C–O–H species innatural melts. Rev. Mineral. 30, 187–230.

Houghton, B.F., Wilson, C.J.N., Del Carlo, P., Coltelli, M., Sable, J.E., Carey, R., 2004. Theinfluence of conduit processes on changes in style of basaltic Plinian eruptions:Tarawera 1886 and Etna 122 BC. J. Volcanol. Geotherm. Res. 137, 1–14.

Jerram, D.A., Higgins, M.D., 2007. 3D analysis of rock textures: quantifying igneousmicrostructures. Elements 3, 239–245.

Kaneko, T., Yasuda, A., Fujii, T., Yoshimoto, M., 2010. Crypto-magma chambers beneathMt. Fuji. J. Volcanol. Geotherm. Res. 193, 161–170.

Lofgren, G., 1974. An experimental study of plagioclase crystal morphology; isothermalcrystallization. Am. J. Sci. 274, 243–273.

Lofgren, G., 1980. Experimental studies on the dynamic crystallization of silicate melts.In: Hargraves, R.B. (Ed.), Physics of Magmatic Processes. Princeton Univ. Press,Princeton, NJ, United States, pp. 487–565.

Machida, H., 1964. Tephrochronological study of Volcano Fuji and adjacent areas.J. Geogr. 73, 293–308 (337–350 (in Japanese, with English Abstr.)).

Machida, H., 2007. Development of Fuji volcano: a review from Quaternarytephrochronology. In: Aramaki, S., Fujii, T., Nakada, S., Miyaji, N. (Eds.), Fujivolcano. Yamanashi Institute of Environmental Sciences, pp. 29–44 (in Japanese,with English Abstr.).

Marsh, B.D., 1988. Crystal size distribution (CSD) in rocks and the kinetics and dynamicsof crystallization: I. Theory. Contrib. Mineral. Petrol. 99, 277–291.

Marsh, B.D., 1998. On the interpretation of crystal size distributions in magmaticsystems. J. Petrol. 39, 553–599.

Martel, C., Schmidt, B.C., 2003. Decompression experiments as an insight into ascentrates of silicic magmas. Contrib. Mineral. Petrol. 144, 397–415.

Martel, C., Poussineau, S., 2007.Diversity of eruptive styles inferred fromthemicrolites ofMtPelée andesite (Martinique, Lesser Antilles). J. Volcanol. Geotherm. Res. 166, 233–254.

Miyaji, N., 1982. Summit eruption of Younger Fuji Volcano in 2000 y. BP and its ejecta.Bull. Volcanol. Soc. Japan 27, 332 (in Japanese).

Miyaji, N., 1988. History of younger Fuji Volcano. J. Geol. Soc. Japan 94, 433–452(in Japanese, with English Abstr.).

Miyaji, N., 2007. Eruptive history, eruption rate and scale of eruptions for the FujiVolcano during the last 11,000 years, Fuji volcano. In: Aramaki, S., Fujii, T., Nakada,S., Miyaji, N. (Eds.), Fuji Volcano. Yamanashi Institute of Environmental Sciences,pp. 79–95 (in Japanese, with English Abstr.).

Morgan, D.J., Jerram, D.A., 2006. On estimating crystal shape for crystal size distributionanalysis. J. Volcanol. Geotherm. Res. 154, 1–7.

Nakada, S., Yoshimoto, M., Fujii, T., 2007. Pre-Fuji Volcanoes, Fuji volcano. In: Aramaki,S., Fujii, T., Nakada, S., Miyaji, N. (Eds.), Fuji Volcano. Yamanashi Institute ofEnvironmental Sciences, pp. 69–77 (in Japanese, with English Abstr.).

Noguchi, S., Toramaru, A., Shimano, T., 2006. Crystallization of microlites and degassingduring magma ascent : constraints on the fluid mechanical behavior of magmaduring the Tenjo Eruption on Kozu Island. Japan. Bull. Volcanol. 68, 432–449.

Noguchi, S., Toramaru, A., Nakada, S., 2008. Relation between microlite textures anddischarge rate during the 1991–1995 eruptions at Unzen. Japan. J. Volcanol.Geotherm. Res. 175, 141–155.


Papale, P., 1998. Volcanic conduit dynamics. In: Freundt, A., Rosi, M. (Eds.), FromMagma to Tephra: Modelling Physical Processes of Explosive Volcanic Eruptions.Elsevier, Amsterdam, pp. 55–89.

Pioli, L., Azzopardi, B.J., Cashman, K.V., 2009. Controls on the explosivity of scoria coneeruptions: magma segregation at conduit junctions. J. Volcanol. Geotherm. Res.186, 407–415.

Pioli, L., Erlund, E., Jonhson, E., Cashman, K., Wallace, P., Rosi, M., Delgado-Granados, H.,2008. Explosive dynamics of violent Strombolian eruptions: the eruption ofParícutin Volcano 1943–1952 (Mexico). Earth Planet. Sci. Lett. 271, 359–368.

Polacci, M., Corsaro, R.A., Andronico, D., 2006. Coupled textural and compositionalcharacterization of basaltic scoria: insights into the transition from Strombolian tofire fountain activity at Mount Etna, Italy. Geology 34, 201–204.

Rutherford, M.J., Gardner, J.E., 2000. Rates of magma ascent. In: Sigurdsson, H. (Ed.),Encyclopedia of Volcanoes. Academic Press, San Diego, pp. 207–217.

Sable, J.E., Houghton, B.F., Del Carlo, P., Coltelli, M., 2006. Changing conditions of magmaascent and fragmentation during the Etna 122 BC basaltic Plinian eruption:evidence from clast microtextures. J. Volcanol. Geotherm. Res. 158, 333–354.

Sparks, R.S.J., Barclay, J., Jaupart, C., Mader, H.M., Phillips, J.C., 1994. Physical aspects ofmagma degassing; 1, Experimental and theoretical constraints on vesiculation. Rev.Mineral. 30, 413–445.

Suzuki, Y., Nakada, S., 2007. Remobilization of highly crystalline felsic magma byinjection of mafic magma: constraints from the middle sixth century eruption atHaruna volcano, Honshu. Japan. J. Petrol. 48, 1543–1567.

Suzuki, Y., Gardner, J.E., Larsen, J.F., 2007. Experimental constraints on syneruptivemagma ascent related to the phreatomagmatic phase of the 2000 A.D. eruption ofUsu volcano. Japan. Bull. Volcanol. 69, 423–444.

Szramek, L., Gardner, J.E., Larsen, J.F., 2006. Degassing and microlite crystallization ofbasaltic andesite magma erupting at Arenal Volcano, Costa Rica. J. Volcanol.Geotherm. Res. 157, 182–201.

Taddeucci, J., Pompilio, M., Scarlato, P., 2004. Conduit processes during the July–August2001 explosive activity of Mt. Etna (Italy): inferences from glass chemistry andcrystal size distribution of ash particles. J. Volcanol. Geotherm. Res. 137, 33–54.

Takahashi, M., Hasegawa, H., Tsukui, M., Nemoto, Y., 1991. Evolution of magma-plumbing system beneath Fuji Volcano: on the view point of whole-rock chemistry.Bull. Volcanol. Soc. Japan. 36, 281–296 (in Japanese, with English Abstr.).

Togashi, S., Miyaji, N., Yamazaki, H., 1991. Fractional crystallization in a largetholeiitic magma chamber during the early stage of the Younger Fuji Volcano.Japan. Bull. Volcanol. Soc. Japan. 36, 269–280 (in Japanese, with EnglishAbstr.).

Togashi, S., Miyaji, N., Yasui, M., Kakuta, A., Asakura, N., Endo, K., Ukawa, M., 1997.Magmatic evolution of the late stage Older-Fuji to early stage Younger-FujiVolcano: petrochemical features of borehole core at Yoshiwara. Bull. Volcanol. Soc.Japan. 42, 409–421 (in Japanese, with English Abstr.).

Toramaru, A., Noguchi, S., Oyoshihara, S., Tsune, A., 2008. MND (microlitenumber density) water exsolution rate meter. J. Volcanol. Geotherm. Res. 175,156–167.

Tsuya, H., 1940. Geological and petrological studies of Volcano Fuji, III: geology of thesouthwestern foot of the Volcano Fuji. Bull. Earthquake Res. Inst. 18, 419–445(in Japanese, with English Abstr.).

Tsuya, H., 1968. Geology of Volcano Mt. Fuji. Explanatory text of geological map 1:50,000 scale. Geological survey of Japan. 24 PP.

Uesugi, Y., 1990. Standard tephra columns for the eastward area of Volcano Fuji. Part 1:S-25 ~Y-141. Bull. Assoc. Kanto Quat. Res. 16, 3–28 (in Japanese).

Wilson, L., Sparks, R.S.J., Walker, G.P.L., 1980. Explosive volcanic eruptions—IV. Thecontrol of magma properties and conduit geometry on eruption column behaviour.Geophys. J. Roy. Astron. Soc. 63, 117–148.

Woods, A.W., Koyaguchi, T., 1994. Transitions between explosive and effusive eruptionsof silicic magmas. Nature 370, 641–644.

Yamamoto, T., Takada, A., Ishizuka, Y., Nakano, S., 2005. Chronology of the products ofFuji volcano based on new radiometric carbon ages. Bull. Volcanol. Soc. Japan 50,53–70 (in Japanese, with English Abstr.).

Yoshimoto, M., Fujii, T., Kaneko, T., Yasuda, A., Nakada, S., Matsumoto, A., 2010. Evolutionof Mount Fuji, Japan: inference from drilling into the subaerial oldest volcano, pre-Komitake. Island Arc 19, 470–488. doi:10.1111/j.1440-1738.2010.00722.x.

Effect of syneruptive decompression path on shifting intensity in basaltic sub-Plinian eruption:...

Documents

Transcript of Effect of syneruptive decompression path on shifting intensity in basaltic sub-Plinian eruption:...