Textures of plagioclase microlite and vesicles within

178 K. NakamuraJournal of Mineralogical and Petrological Sciences, Volume 101, page 178─198, 2006

doi:10.2465/jmps.101.178K. Nakamura, nakamu@kobe－u.ac.jp Corresponding author

Textures of plagioclase microlite and vesicles within volcanic products of the 1914-1915 eruption of

Sakurajima Volcano, Kyushu, Japan

Keisuke NAKAMURA

Graduate School of Science and Technology, Kobe University, Rokko－daicho 1－1, Nadaku, Kobe 657－8501, Japan

We classify volcanic products of the 1914－1915 eruption of Sakurajima Volcano into four types according to vesicularity and the texture of plagioclase microlite. Type－1 is white Plinian fall pumice with high vesicularity (> 60 vol%) and low modal content (< 1.0 vol%) and number density (< 1014 m−3) of microlite. Type－2 is white Plinian fall pumice with high vesicularity (> 55 vol%) and intermediate modal content (1.0－11.0 vol%) and number density (1－10 × 1014 m−3) of microlite. Type－3 is slightly darker Plinian fall pumice with low vesicular-ity (25－50 vol%) and intermediate to high modal content (8－16 vol%) and number density (5 × 1014－2 × 1015 m−3) of microlite. The fourth type of volcanic product is lava flows with low vesicularity (< 25 vol%) and high modal content (> 16 vol%) and intermediate number density (1－5 × 1014 m−3) of plagioclase microlite. Type－1 and type－2 pumices were mainly deposited within the lower and middle parts of the Plinian fall deposit, whereas type－3 pumices were mainly deposited in the upper part. Homogeneous chemical features of preerup-tive melts of all types of pumices indicate that the observed variations in number density and modal content of plagioclase microlite, as well as the decreasing vesicularity from type－1 to type－3 pumices, are related to a decrease in the decompression rate of magma from early to late stages of the Plinian phase. Residual water con-tents and degree of vesicularity of Plinian fall pumices suggest a change in the degassing process during the Plinian phase of eruption from minor degassing during the eruption of type－1 pumices to effective degassing during the eruption of type－3 pumices. Differences in microlite texture between type－1 and type－2 pumices of the early to middle stages of the Plinian phase and late－stage type－1, －2 and －3 pumices reflect changes in the decompression rate, perhaps related to differences between the central and marginal parts of conduits. A sys-tematic increase in the An content of plagioclase microlites from Plinian fall pumice to Taisho Lava reflects a chemical change in the preeruptive melt from silicic to basic composition related to progressive magma mixing.

Keywords: Lava flow, Plagioclase microlite texture, Plinian fall pumice, Vesicularity

INTRODUCTION

Differences in eruption styles, such as Plinian explosion versus lava effusion, are considered to reflect different modes of magma degassing (i.e., gas escape from the sys-tem) (Eichelberger et al., 1986; Jaupart and Allègre, 1991). Jaupart and Allègre (1991) showed numerically that the slow ascent of magma enables efficient degassing that result in effusive eruptions, whereas a rapid ascent of magma prohibits efficient degassing and results in explo-sive eruptions. The eruption style of silicic magma can vary from explosive to effusive within a single eruption,

resulting in various types of volcanic products. Textures of volcanic products, including vesicularity, vesicle tex-ture, and microlite texture, as well as color, and the chem-ical character of the bulk rock, groundmass glass, and individual minerals record various phenomena that occur during an eruption. Magma mixing and mingling pro-cesses during explosive eruptions are recorded by the color and chemical heterogeneity of the bulk rock and/or groundmass glass within volcanic products. For example, chemically and texturally heterogeneous clasts were erupt-ed during the 79 AD eruption of Vesuvius during the Plinian phase of the eruption because of magma mingling (Gurioli et al., 2005). In contrast, following a study of the products of the 1992 eruption of Crater Peak, Alaska, Gardner et al. (1998) proposed that differences in decom-

179Textures of plagioclase microlite and vesicles within volcanic products

pression rate and degassing processes were responsible for differences in vesicularity, color, and microlite texture among rocks of otherwise identical chemical characteris-tics.

Vesicle number density is proportional to the 3/2 power of the decompression rate for homogeneous vesicle nucleation (Toramaru, 1995). The vesicularity of volcanic products changes in response to the degassing process, i.e., closed－system gas expansion or escape of gas from the system (Martel et al., 2000). Furthermore, volatile exsolu-tion from melt increases the liquidus temperatures of melt and consequently can induce the crystallization of micro-lite (Hammer and Rutherford, 2002; Couch et al., 2003; Martel and Schmidt, 2003). According to previous decom-pression experiments, the number density and modal con-tent of microlite change markedly in response to decom-pression rate, which in turn is related to magma ascent rate. This relationship indicates that the texture of vesicles and plagioclase microlite reflects processes of magma ascent and magma degassing.

The 1914－1915 (Taisho) eruption of Sakurajima Volcano, Kyushu, Japan, represents the largest eruption in Japan over the past 1000 years. The eruption continued for about one year, and erupted Plinian fall pumice fol-lowed by the extrusion of lava flows in three stages (Koto, 1916). The total volume of the volcanic products is ca. 2.0 km3 (DRE) (Ishihara et al., 1981). These volcanic prod-ucts are expected to display varying groundmass glass composition, plagioclase microlite texture, and vesicle texture. The purpose of the present study is to document

the transition of the Taisho Eruption in terms of vesicular-ity, plagioclase microlite texture, and chemical features of groundmass glass composition within the volcanic prod-ucts. We first classify the volcanic products of the erup-tion in terms of vesicularity and plagioclase microlite tex-ture. We then discuss the origin of the observed variation in plagioclase microlite texture and vesicularity in terms of chemical features of preeruptive melt compositions and magma ascent processes.

GEOLOGICAL SETTING AND THE 1914-1915 ERUPTION OF SAKURAJIMA VOLCANO

Sakurajima is a volcanic island located on the southern rim of Aira Caldera in southern Kyushu, Japan (Fig. 1). Many tephrochronological and 14C dating studies have documented the activity of Sakurajima Volcano (e.g., Ko-bayashi, 1986; Okuno et al., 1997; Okuno, 2002). Erupt-ive activity at Sakurajima Volcano is classified into three stages: the “Old Kita－dake” stage, the “Young Kita－dake” stage, and the “Minami－dake” stage (Kobayashi, 1986). Activity during the Old Kita－dake stage started about 26 ka BP (Okuno et al., 1997) after a large－scale Plinian eruption and pyroclastic flow eruptions that formed the Aira Caldera (Aramaki, 1984). The Young Kita－dake stage began at 11 ka BP (Okuno et al., 1997), while the Minami－dake stage extended over the period 4.5－1.6 ka BP (e.g., Okuno et al., 1997).

According to historical records, four relatively large eruptions occurred within the last 1000 years of the

Figure 1. Index map showing the loca-tion of Sakurajima Volcano and the distribution of Taisho Lava flows and vents that formed during the 1914－1915 eruption. Modified from Fuku-yama and Ono (1981). Locations of vents are after Koto (1916) and Ishi-hara et al. (1985).

180 K. Nakamura

Minami－dake stage. These are the Bunmei (1471－1476; note, all the following ages are A.D.), An－ei (1779), Taisho (1914－1915), and Showa (1946) eruptions (Fuku-yama, 1978; Fukuyama and Ono, 1981; Kobayashi, 1982). The bulk rock chemistry of these lava flows changed systematically from silicic to basic over time, while the core composition of plagioclase phenocrysts changed from unimodal to bimodal over time. These trends are indicative of ongoing magma mixing within the magma chamber (Yanagi et al., 1991).

The 1914－1915 eruption began with seismic activity that started on January 10, 1914 and continued for several days. An explosive Plinian eruption began on January 12, 1914, with Plinian fall pumices erupting first from the western vent, V.w1 (Fig. 1) (Ishihara et al., 1985), fol-lowed by the eastern vent (Koto, 1916). During the eve-ning of January 13, lava flows of the Taisho 1 Lava began to extrude from both the western and eastern flanks of Sakurajima Volcano; the extrusion continued for approxi-mately two weeks. On the western flank of the volcano, lava firstly extruded from V.w2 and 3, and then from V.w4 to 7, all of which are situated on the western lower flank of V.w1 (Ishihara et al., 1985). The eastern vents of the Taisho 1 Lava flow were buried beneath the following lava flows, and their locations are thus obscured. In

February 1914, the second stage of lava flows (Taisho 2 Lava) erupted from the eastern flank, and continued until the fall of 1914. From the end of March until the begin-ning of April, 1915, secondary lava flows (Taisho 2′ Lava) appeared from the flow fronts of the Taisho 2 flows. Five vents were observed on the eastern flank of the volcano during the 1914－1915 eruption (Koto, 1916). The volume of Plinian fall pumice is 0.6 km3 (not DRE), with 1.34 km3 of lava flows (Ishihara et al., 1981; Kobayashi, 1982).

Textural and compositional analyses of both Plinian fall pumices and lavas are listed in Table 1. Plinian fall pumice samples were collected from the Loc.P1 outcrop, which is located about 4 km east of Minami－dake, in the Nagasakibana Quarry (Fig. 1). The thickness of the Pli-nian fall deposit is ca. 2 m (Fig. 2a). The size of pumice clasts in the lower part of the deposit is slightly larger than those in the upper part, but there are no sharp breaks in deposition within the deposit. Most pumices within the lower to middle parts of the deposit are white pumices, while both white and darker colored pumices are observed in the upper part of the deposit. Accordingly, we collected samples at 30 cm intervals (samples 0809A to 0809G) from the bottom to the top of the deposit (Fig. 2a and Table 1). The diameters of Plinian fall pumices analyzed for bulk density and residual water content are −3ϕ to −4ϕ

Table 1. All sample analyses performed in this study and modal contents of phenocrystic minerals

Pl, plagioclase; Opx, orthopyroxene; Aug, augite; Ol, olivine; Mt, magnetite; Gl, glass; n.a, not analyze.


(8－16 mm), while those analyzed for plagioclase micro-lite textures (number density, modal content, and CSDs (Crystal Size Distributions) (Marsh, 1988) and vesicle texture (number density, vesicularity, and VSDs (Vesicle Size Distributions)) are −2ϕ to −4ϕ (4－16 mm).

The Taisho Lavas are andesitic blocky lavas. The present－day surface textures of the blocky lava flows are classified into two categories: clinkery and massive (Fig. 2b). The massive parts may have cooled slowly during emplacement. To investigate crystallization processes dur-ing magma ascent, we analyzed the clinkery parts of lavas, which represent magma that cooled rapidly. At two localities (L. 1 and L. 8), several samples were collected from clinkery and massive lava to compare differences in microlite textures between the two lava types.

ANALYTICAL METHODS

Textural analyses

X－ray mapping and BSE (back－scattered electron) images for textural analysis were obtained using a JEOL JXA－8900R at the Venture Business Laboratory of Kobe University, Japan. All images were analyzed using MacAspect software from Mitani Co. Ltd. Japan. MacAspect can easily calculate many characteristic features of objects such as area, aspect ratio, and geometric mean diameter, which are extracted in response to the choice of threshold in the gray－scale spectrum. Pict－formatted X－ray map-ping images and photographs of thin sections were con-

verted into binary images for analysis.X－ray mapping images of Si (Fig. 3) were used to

determine VSDs and the vesicularity of Plinian fall pum-ices. Vesicle textures from samples of lava were not ana-lyzed because these vesicles record the coalescence, deformation, collapse, and compaction of vesicles rather than the original vesicle texture. Dwell time at image acquisition was 100 microseconds and resolution of images was 2 μm/pixel to enable a complete analysis of vesicle textures. The total number of pixels is 350 × 350 or 400 × 400, resulting in a total area of 700 × 700 μm2 or 800 × 800 μm2, respectively. At least two images were analyzed for each sample. Vesicles that were apparently coalesced were separated off from the analysis by draw-ing lines of minimum thickness (1 pixel thickness, repre-senting 2 μm). Estimated bulk vesicularity was converted to groundmass vesicularity by using total phenocryst modes estimated using a point counter (1000 points count-ed) (Table 1).

X－ray mapping images of Al (Fig. 3) were used to acquire CSD, modal content, number density, and average length of plagioclase microlites. X－ray mapping analyses were conducted with 15 kV acceleration voltage, 12 nA beam current, and dwell time of 100 microseconds. The resolution of images was 1 μm/pixel for Plinian fall pum-ices and 1, 2, and 4 μm/pixel for lava flows to enable a complete analysis of plagioclase microlite texture. The total areas of 1, 2, and 4 μm/pixel images are 300 × 300 μm2, 600 × 600 μm2, and 1200 × 1200 μm2, respectively. We also analyzed 2 μm/pixel, 350 × 350 μm2 images, for

Figure 2. (a) Columnar section of the Plinian fall pumice deposit at location L. P1 (see Fig. 1 for location). The deposit was divided into 7 subunits of 30 cm thickness. (b) Field photograph of the texture of blocky lava at location L. 8.

182 K. Nakamura

some Plinian fall pumices that contain relatively large microlites. For the CSDs of Plinian fall pumices, at least two images were analyzed for each sample, as some images contained only a few microlites. For CSDs of lava, images of 1, 2, and 4 μm/pixel were analyzed and combined to cover a relatively large size range of micro-lites. In three samples, ×40 photographs of thin sections were used to analyze the CSDs of plagioclase phenocrysts and to distinguish microlite from phenocrysts. Attached crystals were separated off from the analysis by drawing lines of minimum thickness around them.

BSE photo images were used to obtain the modal content of pyroxene and titanomagnetite microlite in the groundmass (Fig. 4).

We also used MacAspect software to measure width－to－length (w/l) ratios, the area of each microlite or vesicle, reference area, modal content or vesicularity, and areal number density (Na) of microlites and vesicles. To convert areal number density per unit diameter range (Na(Dxy)) to volumetric number density per unit diame-

ter range (Nv(Dxy)) for VSDs of Plinian fall pumices, we used the method of Sahagian and Proussevitch (1998). In the present study, all vesicles were approximated as spheres. The geometric mean diameter of vesicles is plot-ted on the X－axes of VSDs. The total volumetric number density was estimated from VSDs as follows:

Nv = ΣNv(Dxy). (1)

To convert areal number density (Na) to volumetric num-ber density (Nv), and to obtain the CSDs of microlite, we used the method of Higgins (2000). We used MacAspect software to measure width－to－length (w/l) ratios, area, modal content, and areal number density of plagioclase crystals, and estimated the aspect ratio and short (S), intermediate (I), and long (L) axes of plagioclase crystals according to the method of Higgins (1994). True crystal length (Lt) and crystal width (Wt) were estimated as fol-lows:


Lt = l*I (2)

and

Wt = w*L/S. (3)

The volumetric number density of crystals in the length interval Lx to Ly (Nv(Lxy)) was obtained as follows:

Nv(Lxy) = Na(Lxy/Hxy), (4)

where Na(Lxy) is the areal number density of crystals in the length interval Lx to Ly, and Hxy is the mean value for the length interval Lx to Ly. We obtained the popula-tion density of crystals (number density per unit volume per unit crystal length) in the length interval Lx to Ly (n(Lxy)) as follows:

n(Lxy) = Nv(Lxy)/(Ly－Lx). (5)

The average size (Sn) of crystals was estimated following the method of Hammer et al. (1999):

Sn = (ϕ/Na)0.5, (6)

where ϕ is the modal content of plagioclase microlite (vol %).

In this study, crystal length (Lt) is plotted on the X－

axes of CSDs for plagioclase. Total volumetric number density (Nv) is obtained as follows:

Nv = Na/Sn. (7)

Glass and mineral composition

Glass and mineral analyses were performed on a JEOL JXA－8900R at the Venture Business Laboratory of Kobe University, Japan. Groundmass glass analyses were con-ducted under the following conditions: 15 kV acceleration voltage, 12 nA beam current, 5－10 μm beam diameter, and 4 s counting time for Na and 20 s counting time for other elements. Mineral analyses were performed under the following conditions: 15 kV acceleration voltage, 12 nA beam current, focused beam, and 10 s counting time for Na and 20 s counting time for other elements. Back-ground corrected intensities were processed via the Bence－Albee method for both glass and minerals.

Measurement of the density and residual water con-tent of Plinian fall pumices

50 pumice samples for each unit (a total of 350 samples) were analyzed for bulk density, while 35 of the 350 sam-ples were analyzed for residual water content. The bulk density of Plinian fall pumices was analyzed at Kobe Uni-versity, Japan, using a helium pycnometer (Micromeritics

Figure 3. X－ray mapping and binary images of Al and Si showing representative microlite and vesicle textural features. (A) Analyses of Plinian fall pumices. (A1) Intermediate modal content and high vesicularity type－2 pumice from the lower part of the deposit (0809A－a; modal con-tent of plagioclase microlite 6.04 vol%; vesicularity 60.2 vol%). (A2) Low modal content and high vesicularity type－1 pumice from the lower part of the deposit (0809B－a; modal content 0.45 vol%; vesicularity 63.2 vol%). (A3) High modal content and low vesicularity pumice from the upper part of the deposit (0809G－a; type－3 pumice; modal content 16.3 vol%; vesicularity 48.2 vol%). (A4) High modal content and low vesicularity pumice from the upper part of the deposit (0809G－b; type－3 pumice; modal content 8.12 vol%; vesicularity 24.7 vol%). (B) Analyses of lavas. (B1) Taisho 1 Lava (0703D; modal content 23.6 vol%). (2) Taisho 1 Lava (1004B; modal content 18.2 vol%). (3) Taisho 2 Lava (0902D; modal content 21.3 vol%). (4) Taisho 2′ Lava (0908B; modal content 27.0 vol%).

184 K. Nakamura

Accupyc 1330, Shimadzu Co. Ltd. Japan). First, the total weight of the pumice was measured. The surface of the pumice was then coated with melted paraffin and the vol-ume of the pumice was measured using the helium pyc-nometer. The bulk density was calculated from the values of pumice weight and volume. Bulk vesicularity was cal-culated from the equation of Houghton and Wilson (1989), assuming that magmatic density (melt + phenocryst) is 2500 kg/m3, and converted to crystal－free vesicularity using the total phenocryst mode.

The residual water contents of Plinian fall pumices were analyzed via the Karl－Fisher titration method using an Aquacounter AQ－200 (Hiranuma Co. Ltd. Japan) housed in Kobe University, Japan. Water within samples (about 0.1－0.5 g) was extracted by two－step heating. The first step was at 120 °C for at least 12 hours, with the aim

of extracting water attached to the surface of the sample. The second step was at 1000 ± 40 °C for approximately half an hour, to extract water exsolved within the sample. Bulk rock water contents were converted to groundmass water contents using total phenocryst modes.

RESULTS

Chemistry of glass and microlite

The phenocryst (> 100 μm in length) assemblage is almost the same in the Plinian fall pumices and lavas (Table 1): plagioclase + orthopyroxene + augite + magne-tite and ± olivine. Olivine phenocrysts are present only in Taisho 2 and 2′ Lavas. Plagioclase phenocrysts are divided into four types: high An plagioclase, low An pla-

Figure 4. BSE images of lavas showing repre-sentative pyroxene and titanomagnetite micro-lite texture. Py, pyroxene; Mt, titanomagnetite. (a) Taisho 1 Lava (0703D). (b) Taisho 1 Lava (1004B). (c) Taisho 2 Lava (0902D). (d) Taisho 2′ Lava (0908B).

Table 2. Average groundmass glass compositions of Plinian fall pumices and lavas

Number in brackets represents one standard deviation.


gioclase, plagioclase with sieved texture, and plagioclase with dusty zoned rims (Yanagi et al., 1991).

The groundmass glass of both Plinian fall pumices and lavas is of rhyolitic composition (Table 2 and Fig. 5). The SiO2 content of groundmass glass from Plinian fall pumices is within the range 68－73 wt%, while ground-mass glass from samples of Taisho 1 Lava has a more evolved composition (SiO2 of 75－78 wt%). The SiO2 con-tent of groundmass glasses from Taisho 2 and 2′ Lavas fall in the range 71－77 wt% and 73－74 wt%, respectively, which are slightly less evolved than glass in Taisho 1 Lava. The CaO, MgO, Al2O3, and FeO contents of ground-mass glass decrease from Plinian fall pumices to Taisho 1 Lava, and slightly increase from Taisho 1 to Taisho 2 and 2′ Lavas. The K2O contents of groundmass glass increase from Plinian fall pumices to Taisho 1 Lava, and decrease slightly from Taisho 1 to Taisho 2 and 2′ Lava.

Microlites (< 100 μm in length) in all samples are orthopyroxene, pigeonite, augite, titanomagnetite, and plagioclase. Most pumices contain pyroxene and titano-magnetite microlites. The modal contents of pyroxene and titanomagnetite increase with plagioclase microlite con-tent. For pumice samples, plagioclase microlite composi-tions were determined from pumices in which microlite was of sufficient size to analyze by electron probe. The anorthite content of plagioclase microlites from Plinian fall pumices ranges from An46 to An50 (Table 3), while the anorthite content in those from Taisho 1 Lava is approxi-mately An50 (Table 3). Plagioclase microlites from Taisho 2 and 2′ Lava are, however, approximately An52－62 (Table 3), which is more anorthitic in composition.

Density and residual water content of Plinian fall pumices

The bulk density (Fig. 6) of pumices varies from the

lower to upper parts of the deposit. Most pumices in the lower to middle parts of the deposit have a density of approximately 700 kg/m3 (vesicularity of 74 vol%). The lightest pumice has a density of 403 kg/m3, corresponding to a vesicularity of 85 vol%. In the upper part of the deposit, from sample 0809F (1.5－1.8 m) to sample 0809G (1.8－2.1 m) (Fig. 2a), the proportion of high－density pumices increases; as a result, the average pumice density exceeds 1000 kg/m3 (vesicularity < 62 vol%). The maxi-mum density of pumice is 1678 kg/m3, corresponding to a vesicularity of 35 vol%. Most of the low－density pumices are white pumices, while high－density pumices of > 950－1000 kg/m3 (vesicularity of 64－62 vol%) are slightly darker in appearance.

The residual water contents of the groundmass within pumice samples are within the range 0.2－0.6 wt% (Fig. 7).

Textures of vesicles and plagioclase microlites within Plinian fall pumices

The vesicularity of Plinian pumices is inversely correlated with bulk density (Fig. 6). The vesicularity of pumices from the lower to middle parts of the deposit, as estimated from mapping images of Si concentration, is within the range 55－75 vol% (Fig. 8a and Table 4). In contrast, the vesicularity of darker pumices collected from the upper part of the deposit is within the range 25－51 vol%. The values of vesicularity estimated from mapping images are not necessarily the same as those estimated from bulk density because of analytical error (Table 4). The values of vesicularity of white pumice estimated from bulk den-sity (400－1000 kg/m3) are in the range 60－83 vol%, while values for darker pumice are in the range 34－60 vol%.

Table 2. (Continued)

Figure 5. SiO2－CaO plot for groundmass glass and preeruptive melt compositions. All data are normalized to total 100 wt%.

186 K. Nakamura

Values of vesicularity estimated from mapping images are 25－75 vol%. The differences in values of vesicularity estimated from bulk density and mapping images result from the facts that (1) vesicularity estimated from map-ping images varies with mapping area and mapping size, resulting in large analytical error (Table 4), and (2) the treatment of apparently coalesced vesicles also causes dif-ferences in estimates of vesicularity. Vesicles that appear-ed to be coalesced were separated off from the analysis by drawing lines of minimum thickness when estimating vesicularity from mapping images; this caused a reduc-tion in the estimated vesicularity. Hence, values of vesicu-larity calculated from bulk density are expected to better represent important features of the vesicularity of Plinian fall pumice. The average vesicularity estimated from bulk density is around 74 vol%. In contrast, values of vesicu-

larity for most of white pumice, as estimated from map-ping images, fall in the range 65－74 vol%. Vesicularity estimated from mapping images, therefore, roughly repre-sents the features of actual vesicularity. The threshold between white and darker pumices is around 950－1000 kg/m3, corresponding to a vesicularity of 64－62 vol%, as estimated from bulk density.

Previously analyzed vesicle number densities of silicic magma are in the range of 3.6 × 1014－2.9 × 1016 m−3 for Mt Mazama Pumice (microlite－free) (Klug et al., 2002), 1.2 × 1013－5.3 × 1014 m−3 for Towada Pumice (Toramaru, 1990), 1.7 × 1014－1.3 × 1015 m−3 for Usu micropumice erupted during the 2000 eruption (microlite－bearing) (Suzuki and Nakada, 2002), and 2.0 × 1015－8.2 × 1014 m−3 for white (microlite－free) and gray (microlite－bearing) pumice, respectively, erupted during the 1980－1986 eruption of Mt St. Helens (Klug and Cashman, 1994). The vesicle number density of Plinian fall pumice erupted during the 1914－1915 eruption of Sakurajima Volcano is within the range 4.2 × 1013－2.7 × 1014 m−3 (Fig. 6a and Table 4), which are relatively small values for silicic magma. The number density of vesicles in pumices from the lower to middle parts of the deposit falls in the range 5.9 × 1013－2.7 × 1014 m−3, while the values for slightly darker pumices from the upper part of the deposit are within the range 4.2 × 1013－2.0 × 1014 m−3; this latter value is smaller than that of white pumice from the lower to middle part of the deposit except for one sample that shows the least vesicularity (Fig. 8a).

The VSDs for all samples show a linear or slightly convex downward correlation between the logarithm of population density and the geometric mean vesicle diam-eter. The VSDs of lower vesicularity samples show lower

Table 3. Average plagioclase microlite composition of Plinian fall pumices and lavas

Number in brackets represents one standard deviation.


population densities at relatively large vesicle sizes com-pared with higher vesicularity samples; this indicates a change in vesicularity due to the difference in the number density of large vesicles (Fig. 8b). Another feature of ves-icle texture is that all pumices analyzed contain some sil-ica minerals within vesicles.

Plagioclase microlite textures vary markedly between different eruptive units (Table 5). From the lower to middle parts of the deposit (from samples 0809A to 0809E), the modal content of plagioclase microlite in the groundmass varies within the range 0.1－10.7 vol%. The volumetric number density (number density herein) of plagioclase varies from 2.5 × 1013－1.1 × 1015 m−3. In the

upper part of the deposit, the modal content of plagioclase microlite falls within the range 5.9－16.4 vol%. The num-ber density of plagioclase microlite is 2.6 × 1014－1.8 × 1015 m−3, indicating that pumices in the upper part of deposit tend to contain more microlite than those in the lower to middle parts of deposit. The average length of plagioclase microlites is within the range 2.7－6.4 μm.

Plagioclase microlite texture within lavas

Plagioclase microlites within lavas are euhedral and gen-erally equant tabular to skeletal fork－shaped. Pyroxene and titanomagnetite microlites are euhedral and some-

Figure 6. Bulk density and vesicularity histograms for Plinian fall pumices from the L. P1 deposit. N is number measured.

188 K. Nakamura

times attached to each other. The number density of pla-gioclase microlite in clinkery lava varies between 1.3 × 1014 and 7.2 × 1014 m−3 (Table 6), which is a smaller range than that for Plinian fall pumices. The modal content of plagioclase microlite in the groundmass is 16－33 vol%, which is higher than that for Plinian fall pumices. The average length of plagioclase microlite in lavas is 7.1－10.7 μm, which is longer than that for Plinian fall pum-ices. The modal contents of pyroxene and titanomagnetite microlite are 4.9－10.5 vol% and 0.5－1.1 vol%, respec-tively (Table 6). The number densities of microlite in clin-kery and massive parts of lavas show a slight increase from surface clinkery parts to inner massive parts in Loc.L1 Taisho 1 Lava (5.9－7.2 × 1014 m−3) and a slight decrease in Loc.L8 Taisho 2′ Lava (1.5－1.4 × 1014 m−3). The modal contents of microlite increase from surface clinkery parts (24 vol% in Taisho 1 Lava and 27 vol% in Taisho 2′ Lava) to inner massive parts of the lava flow (42 vol% in Taisho 1 Lava and 31 vol% in Taisho 2′ Lava).

CSDs for plagioclase microlite within Plinian fall pum-ices and lavas

CSD plots of plagioclase microlite from Plinian fall pum-ices and lava flows are shown in Figures 9－11. There is a gap in the sizes of phenocrysts and microlite of around 100 μm (Fig. 11). CSD plots for most samples, except some white pumices, show a linear correlation between the logarithm of population density and crystal length. Intercept and slope values were estimated by least square fitting. The intercepts of CSDs from Plinian fall pumices

are 1018.2－1020.3 m−4 and show slightly larger variations than the intercepts of CSDs from lava flows, which are 1018.9－1019.5 m−4. The slopes of CSDs from lava flows are shallower than those for Plinian fall pumices, suggesting that size variation of microlite is greater in lava flows that in Plinian fall pumices.

DISCUSSION

Classification of ejecta in terms of vesicularity and plagioclase microlite texture

In this section, volcanic products are classified in terms of vesicularity and plagioclase microlite texture. Plinian fall pumice can be divided into two types in terms of vesicu-larity: white pumice with high vesicularity (low density), and darker pumice with low vesicularity (high density). While the white pumice is deposited throughout the Plinian fall deposit, the darker pumice is mainly deposited within upper part of the deposit from samples 0809F to

Figure 7. Change in vesicularity as a function of water content. Values of vesicularity were calculated from bulk density and phenocryst mode. Values of residual water contents of the groundmass were calculated from water contents and phenocryst mode.

Figure 8. (a) Correlation between vesicularity and vesicle number density for Plinian fall pumices. (b) Vesicle size distribution (VSD) plot for representative vesicles in Plinian fall pumices. Vesicularity and VSD were estimated from mapping images.


0809G.There is no sharp boundary in vesicularity among

white pumices. Highly vesicular white pumice, however, can be classified into two types according to plagioclase microlite texture: type－1 and －2 (Figs. 9b and 12). Hence, Plinian fall pumice can be divided into three types in terms of vesicularity and plagioclase microlite texture. Type－1 pumice is highly vesicular (> 60 vol% estimated from mapping images) white Plinian fall pumice with low modal content (< 1.0 vol%) and number density (< 1014 m−3) of plagioclase microlite. Type－2 pumice is highly vesicular (> 55 vol% estimated from mapping images) white Plinian fall pumice with intermediate modal content (1.0－11.0 vol%) and number density (1－10 × 1014 m−3) of plagioclase microlite.

Differences in plagioclase microlite textures are clearly evident in CSDs for the two types of pumice (Fig. 9b). Type－1 CSDs show minor size variation of less than 30 μm in length and a low population density at small size (less than 20 μm) compared to the other types, due to low modal content and number density. In contrast, type－2 CSDs show large size variation of up to 80 μm, illus-trating the difference between type－1 and type－2 pum-

ices. Darker and poorly vesicular type－3 pumice shows different textural features in terms of vesicles and plagio-clase microlite (Fig. 12). Type－3 pumices show interme-diate to high modal content (8－16 vol%) and number den-sity (5 × 1014－2 × 1015 m−3) of plagioclase microlite. The vesicularity of type－3 pumices, as estimated from map-ping images, is 25－50 vol%.

Plagioclase microlite textures within lavas are differ-ent to those within Plinian fall pumice, suggesting that lavas can be considered separate from the other samples on this basis. Type－4 volcanic material is lava flows with high modal content (> 16 vol%), intermediate number density (1－5 × 1014 m−3) of plagioclase microlites, and low vesicularity. Although the modal content of plagio-clase microlite in lavas is higher than that for type－3 pum-ices, the number density of plagioclase in lavas is smaller than that in type－3 pumices. CSD slopes for lavas are as steep as those for type－2 pumices, and slight gentler than those for type－3 pumices (Fig. 10). The CSDs for lavas, however, show that size variations of microlite in lavas are large, up to 120 μm, resulting in the variation in modal content.

There is no sharp boundary in vesicle texture bet-

Table 4. Groundmass vesicularity and number density of vesicles for Plinian fall pumices, as estimated from mapping images

Vesicularity [= 100 × vesicle/(vesicle + groundmass)].Vesicularitya, estimated from mapping images of Si.Vesicularityb, estimated from mapping images of Al.σ, one standard deviation.

190 K. Nakamura

ween type－1 and type－2 pumices despite the large tex-tural variation in plagioclase microlite (Fig. 12). Highly vesicular white pumices (type－1 and type－2) are depos-ited throughout the Plinian fall deposit. In contrast, type－3 pumices are rare in the lower to middle parts of the deposit, and are mainly deposited within the upper part of the deposit from samples 0809F (1.5－1.8 m) to 0809G (1.8－2.1 m). Type－3 pumices have different vesicle and plagioclase microlite texture compared to type－1 and type－2 pumices. This indicates that textural differences between white and darker colored pumices reflect differ-ences in syneruptive physical changes and differences in the preeruptive chemistry of the magma. Such physical and chemical changes can even occur during a Plinian phase. For example, chemically and texturally heteroge-neous clasts were erupted during the 79 AD eruption of Vesuvius due to magma mingling (Gurioli et al., 2005). In contrast, during the 1992 eruption of Crater Peak, Alaska, changing eruptive and degassing processes produced two

types of texturally different but chemically homogeneous clasts (Gardner et al., 1998). In the case of the 1914－1915 eruption of Sakurajima Volcano, two reasons for the observed textural change from type－1 and type－2 white pumices to type－3 darker colored pumices and/or lavas are considered to be relevant: (1) chemical changes to the magma (e.g., magma mixing), and (2) changes in the decompression rate from early to late stages of the Plinian phase and the lava effusive phase of the eruption.

Preeruptive melt composition within the magma chamber

We estimated the preeruptive melt compositions of lavas and darker colored pumices from the composition and mode of groundmass glass and microlite mineral (Table 7). From this, we evaluate the effect of chemical heterogene-ity (induced by magma mixing) on textural differences observed in the volcanic products. The compositions of

Table 5. Textual data for the groundmass phase and CSD data for plagioclase microlite from Plinian fall pumices

a: 300 × 300 μm2 and 650 × 650 μm2 images were used to analyze textural data and CSDs.ϕpl, modal content of plagioclase microlite.Sn, average size of plagioclase microlite estimated from Eq. (6).Nv, number density estimated from Eq. (7).Number in brackets represents one standard deviation.n.a, not analyze.


groundmass glass and calculated preeruptive melt compo-sitions within the magma chamber are shown in Table 2 and Figure 5. The SiO2 contents of groundmass glass from type－1 and －2 pumices are 67－70 wt%. Preeruptive melt compositions of type－1 and type－2 pumices are sim-ilar to the groundmass glass compositions, because type－1 and －2 pumices show lower modal content of microlite (Fig. 12). The preeruptive melt composition of type－3 pumice and Taisho 1 Lava are 67－70 wt%, indicating that microlite crystallized from chemically homogeneous melt in these rock types. Preeruptive melt compositions of Taisho 2 and Taisho 2′ Lava are 64－69 wt% and are slightly more basic than those of Plinian fall pumice and Taisho 1 Lavas. A change in preeruptive melt composition from silicic to basic is considered to reflect progressive magma mixing (Yanagi et al., 1991), which is also evident in the variable An content of plagioclase microlite within lavas. Hence, these results suggest that textural differ-ences among type－1 and type－2 white pumice, type－3 darker colored pumice, and Taisho 1 Lava are not due to chemical heterogeneity of the preeruptive melt induced by magma mixing, but reflect variability in the decom-pression rate during the Plinian phase to early effusive phase of the eruption. Chemical heterogeneity of the pre-

eruptive melt, as induced by magma mixing during the late stages of the effusive phase of the eruption, is reflected in the An contents of plagioclase microlite rather than in the texture of the plagioclase microlite.

Variation in the decompression rate of magma during the eruption

First, changes in the decompression rate is discussed in terms of microlite texture; however, before discussing the relation between plagioclase microlite texture and decom-pression rate, the effect of cooling－induced crystallization after eruption from the vent must be considered. In the case of Plinian eruptions, relatively small volcanic prod-ucts are quenched rapidly soon after being ejected from the vent; this indicates that crystallization of microlite occurs mainly within the conduit. In contrast, for lava effusion, microlite crystallization can occur during em-placement. We examined textural variation in plagioclase microlite within clinkery parts of blocky lavas, massive parts of blocky lavas, and massive parts of sheet－like lavas at the same locality to determine crystallization pro-cesses during emplacement. Modal contents of microlite in clinkery parts of blocky lava, massive parts of blocky lavas, and massive parts of sheet－like lava differ mark-edly, especially in Taisho 1 Lava at site L. 1, despite the minor variation in number density of plagioclase micro-lite (Table 6). This indicates that the crystallization pro-cess during emplacement mainly involves renewed growth of pre－nucleated microlite.

Previous decompression experiments indicate that crystallization is induced by decompression and that modal content and number density of microlite change in response to the amount of decompression (∆P) and decompression rate (∆P/∆t) (Hammer and Rutherford, 2002; Couch et al., 2003; Couch, 2003; Martel and Schmidt, 2003). Figure 13 shows the logarithm of ∆P/∆t against the logarithm of number density plot for the MSD (Multi－Step Decompression) experiment. Compiled data of MSD experiments are from Couch et al. (2003) (Fig. 13). Number densities were calculated from areal number density and modal contents using Equations (6) and (7). The low number density of microlite is caused by low ∆P/∆t during the MSD experiment and small ∆P. The logarithm of number density is proportional to the loga-rithm of ∆P/∆t in the MSD experiment (Fig. 13). Assuming that initial water content and magmatic temper-ature are 3.3 wt% and 950 °C, respectively, as estimated by Sekine et al. (1979), and that the depth where magma was saturated with H2O is 3.2 km (80 MPa) beneath the Aira Caldera, then the ∆P of Plinian fall pumices of all types, as estimated from residual water contents of ca. 0.4


192 K. Nakamura

wt% (Fig. 7) is approximately 75 MPa. It is expected that the ∆P of lavas is slightly larger than that of Plinian fall pumices. Applying the experimental result to textural changes observed in the volcanic products, the decrease in number density of microlite from type－3 pumice to Taisho 1 Lava is interpreted to represent a decrease in the decompression rate of magma from the late stages of the Plinian phase to the time of effusion of Taisho 1 Lava. However, it is difficult to explain the textural differences between type－1 and type－2 pumices and other rock types. The highest decompression rate in the MSD experiments (13.8 MPa/h, crystallization time is 8 hours in this case) corresponds to an ascent rate of 0.15 m/s, assuming that ∆P/∆t is equal to the ascent rate of the magma. The ascent rate may be larger than this value in the case of the early to middle stage Plinian eruption. Microlite－free pumices observed from the large pyroclastic eruption at Mt Mazana (Klug et al., 2002) or large Plinian eruptions such as the 1980－1986 Mt St. Helens eruption and 1991 Mt Pinatubo eruption (Cashman, 1992; Hammer et al., 1999) are also interpreted as resulting from high ∆P/∆t. Extremely high values of ∆P/∆t result in a low number density, indicating disequilibrium in crystallization. Type－1 pumices had little time to crystallize plagioclase micro-lite during the rapid ascent of magma because of a time lag between H2O exsolution and accompanying crystalli-

zation (approximately < 40 min or 1－4 h) (Hammer et al., 1999; Couch et al., 2003). The intermediate number den-sity of microlite in Taisho 1 Lava is due to the low decom-pression rate of magma despite the large magnitude of overall decompression. In addition, crystals grow during magma ascent and emplacement, resulting in a high modal content of plagioclase microlite in lavas compared with Plinian fall pumices.

While the number density of plagioclase microlite increases from type－1 pumice to type－3 pumice, vesicu-larity decreases. A decrease in microlite number density within type－3 pumice is proportional to the decrease in vesicularity (Fig. 12), expect for the most vesicular sam-ple. These results suggest that vesicle textures are also related to the decompression rate.

Vesiculation and degassing processes within magma

We estimated a change in decompression rate from vesi-cle texture. The number density of vesicles in the volcanic products is proportional to the 3/2 power of decompres-sion rate for homogeneous vesicle nucleation (Toramaru, 1995). In the case that volcanic products preserve the original number density of vesicles, vesicle number den-sity is a useful indicator of changes in the decompression rate. Vesicle textures of all types indicate that the observed

Table 6. Textual data of groundmass phases and CSD data of plagioclase microlite from lavas

a: 300 × 300 µm2, 600 × 600 µm2 and 1200 × 1200 µm2 images were used to analyze textural data and CSDs. ϕpl, modal content of plagioclase microlite.ϕpy, modal content of orthopyroxene, pigeonite and augite microlite. ϕmt, modal content of titanomagnetite microlite.ϕtotal, modal content of microlite in groundmass phase. Sn, average size of plagioclase microlite estimated from Eq. (6).Nv, number density estimated from Eq. (7).Numbers in brackets represent one standard deviation.n.a, not analyze.


number densities do not necessarily record the original number densities at the time of nucleation, suggesting that for such a sample it is not appropriate to discuss relative changes in decompression rate in terms of number den-sity. For example, deformed vesicle shapes in type－1 and type－2 pumice from samples 0809A to 0809E appear to have experienced coalescence and collapse; however, these textural features suggest that the vesiculation and degassing process might vary during the Plinian phase of eruption due to physical changes in the ascending magma. Type－3 Plinian fall pumices collected from upper parts of the deposit show lower vesicularity, thicker vesicle walls, irregular vesicle shapes, and large variations in number densities. The main difference in VSDs between type－3 pumices and other types (type－1 and type－2 pumices) is


Figure 9. (a) Representative crystal size distribution (CSD) plots for plagioclase microlite from Plinian fall pumices. (b) Area of CSD for each type of pumices.

Figure 10. Crystal size distribution (CSD) plots of plagioclase microlite from representative lavas.

Figure 11. Crystal size distribution (CSD) plots of plagioclase phe-nocrysts and microlite from representative lavas.

194 K. Nakamura

mainly the difference in the number density of large vesi-cles. While highly vesicular pumices (type－1 and type－2) contain relatively large vesicles, poorly vesicular type－3 pumices contain fewer large vesicles. These observations

may indicate that large vesicles collapsed or that volatiles escaped from magma.

The H2O content of Plinian fall pumices is about 0.4 wt% (Fig. 7), while the preeruptive H2O content of the magma is estimated to be ca. 3.3 wt% (Sekine et al., 1979). If degassing and expansion of the magma occur in a closed system and under equilibrium conditions, the numerically expected vesicularity (α) is more than 90 vol% (Fig. 7), exceeding the traditionally accepted “frag-mentation vesicularity” of 75－80 vol% (e.g., Sparks, 1978). Here:

α = 100*Vg/(Vg + Vm), (9)

where Vg is the volume of expanded gaseous H2O and Vm is the volume of melt. These values are defined as fol-lows:

Vg = n′RT/P (10)

and

Vm = (1 − n)/ρm. (11)

Here, n is water solubility (wt%), as calculated from the equation of Moore et al. (1998) using the estimated matrix glass composition of Plinian fall pumices. R is the gas constant (8.31 J/mol/K), P is pressure (Pa), n′ is mole wt% of H2O vapor, and T is magmatic temperature. For T, we used 1223 K, as estimated by Sekine et al. (1979). ρm is melt density, for which we used 2400 kg/m3, as esti-

Figure 12. (a) Correlation between vesicularity and modal content of plagioclase microlite. (b) Correlation between vesicularity and number density of plagioclase microlite. Values of vesicularity were estimated from mapping images.

Table 7. Example of the procedure used to estimate preeruptive melt compositions

All data are normalized to total 100 wt%.


mated from the equation of Lange and Carmichael (1990).The values of vesicularity calculated from bulk den-

sity are lower than the calculated values, suggesting that open－system degassing occurred during the eruption (Fig. 7). Additional evidence for open－system degassing is the

existence of silica minerals in vesicles. Hoblitt and Harmon (1993) and Martel et al. (2000) noted that silica minerals in vesicles may precipitate during the degassing of volatiles from vesicles and indicate open－system degassing.

Open－system degassing requires a gas flow from vesicles out of the magma column into basement rocks or upward through the conduit (Eichelberger et al., 1986). Hence, the degree of degassing can be controlled by the decompression rate and relative magma/conduit permea-bility. Rhyolitic melt becomes permeable to gas flow in the vesicularity range of 60－70 vol%, in which vesicles easily collapse (Westrich and Eichelberger, 1994). It is therefore considered that all pumices may have undergone a two－stage vesiculation: the first stage involves closed system expansion of vesicles until vesicularity of 60－70 vol% (around 15 MPa in this case), followed by gas escape from the magma. Magma erupted as type－1 or type－2 pumice certainly reached a vesicularity of 60－70 vol%, at which point it was potentially permeable. Minor degassing occurred during magma ascent and fragmenta-tion occurred upon reaching a vesicularity of 64－83 vol%.

Figure 13. Relation between decompression and number density of microlite from the MSD experiments of Couch et al. (2003). The initial pressure in the experiments was 160 MPa.

Figure 14. Schematic diagram of the sequence of the 1914－1915 eruption of Sakurajima Volcano. See text for details.　(A) Plinian phase　Weak degassing and impaired crystallization of microlite due to high decompression rate during the early to middle stages of the Plinian

phase. Effective degassing and enhanced crystallization of microlite during the late stages of the Plinian phase.　(B) Effusive phase　Effective crystallization due to low decompression rate. The An content of plagioclase microlite changes from sodic to calcic due to magma

mixing.

196 K. Nakamura

In the case of type－3 pumice, effective degassing and foam collapsing of magma occurred upon reaching a vesicularity of around 60 vol%; consequently, final prod-ucts had a vesicularity of 34－64 vol%. Both changes in decompression rate and relative magma/conduit permea-bility may enhance effective degassing. A decrease in decompression rate induced more effective degassing for type－3 pumice, resulting in a decrease in vesicularity; this is consistent with inferences drawn from the texture of plagioclase microlite.

Eruption model

By combining textural data from the volcanic products, stratigraphy of the Plinian fall deposit, and historical records of the eruption, we have reconstructed the erup-tion sequence of the 1914－1915 eruption (Fig. 14).

During the early to middle stages of the Plinian phase, type－1 and type－2 white pumices erupted largely synchronously at the highest decompression rate during the entire phase of the eruption. The high decompression rate of the magma prohibited efficient decompression－induced crystallization of microlite, resulting in low to intermediate modal content and number density of plagio-clase microlite. Textural differences of microlite in type－1 and type－2 pumice reflect minor differences in the decompression rate of the magma. One possible interpre-tation of the origin of the differences between type－1 and type－2 pumices is slight differences in the decompression rate of the magma, perhaps due to differences between the centers and margins of conduits (e.g., Taddeucci et al., 2004; Gurioli et al., 2005) (Fig. 14A). In such a model, type－1 pumices would have ascended in the center of conduit, while type－2 pumices ascended along conduit margins. It is possible that there exists a region of annular flow at the transition from bubbly to particle－laden gas jet flow, where variations in the ascent velocity of magma may occur (Jaupart and Tait, 1990). A slightly lower decompression rate near the margin of the conduit pro-vides more time for the crystallization of microlite, result-ing in a higher modal content and number density of pla-gioclase microlite in type－2 pumices, despite their similar vesicularity to type－1 pumice.

During the late stages of the Plinian phase of the eruption, the proportion of type－1 and type－2 white pum-ices decreased and type－3 darker colored pumice first erupted. The differences between type－3 pumice and type－1 and type－2 pumices reflect a decrease in decom-pression rate rather than chemical heterogeneity related to magma mixing. Lower values of vesicularity in type－3 pumices reflect the degassing of volatiles and are inter-preted as a decrease in the decompression rate of the

magma, consistent with interpretations derived from microlite texture. The decompression rate is proportional to the eruption rate and/or the －4 power of the conduit radius. According to historical records, the intensity of the Plinian eruption decreased during the afternoon of January 13, 1914 (Kobayashi, 1982). During the 1914－1915 eruption of Sakurajima Volcano, many vents were created as the eruption style shifted from explosive to effusive (Koto, 1914). These observations suggest a pos-sible change in eruption rate and/or decrease in conduit radius during the late stages of the Plinian phase of the eruption, resulting in a decrease in the decompression rate and increase in the proportion of type－3 pumices. Type－1 and type－2 pumices erupted at this time may represent magma that ascended rapidly within the central parts of conduits. Pumice that contains the highest number density of microlite records a vesicularity of 40－50 vol%, indicat-ing that the nucleation of microlite occurred effectively during the late stages of the Plinian phase. A decrease in microlite number density within type－3 pumice is propor-tional to the decrease in vesicularity, except for the most vesicular type－3 pumice. The number density of microlite within the least vesicular type－3 pumices is almost the same as that in lavas, indicating that during the late stages of the Plinian phase the decompression rate of type－3 pumices decreased to a rate similar to that of lavas.

During the effusive phase of the eruption, the pre-eruptive melt composition of magma became increasingly basic due to magma mixing. It is therefore not suitable to compare the change in decompression rate with pumices in terms of microlite texture. The effect of chemical het-erogeneity within the preeruptive melt, however, is reflected in the An content of plagioclase microlite rather than in the texture of plagioclase microlite (Fig. 14B).

SUMMARY

1.　Volcanic products of the 1914－1915 eruption of Sakurajima Volcano are classified into four types according to vesicularity and plagioclase microlite tex-ture.

2.　Textural differences among Plinian fall pumices are due to differences in decompression rate. Type－1 and type－2 white pumices, with high vesicularity and low to intermediate modal content and number density of plagioclase microlite indicate that the decompression rate of magma decreased during the early to middle stages of the Plinian phase of the eruption. In the upper part of the deposit, the proportion of type－1 and type－2 white pumices decreases and that of type－3 darker pumice increases, indicating that the decom-pression rate of magma decreased during the late


stages of the Plinian phase of the eruption. Differences in microlite texture between early to middle stage type－1 and type－2 pumices and between type－1, －2 pumice and type－3 pumice during the late stages of the Plinian phase reflect minor differences in the rate of magma ascent, perhaps differences between the central and marginal parts of conduits.

3.　Chemical changes in the preeruptive melt from silicic to basic composition, which were induced by magma mixing during the late stages of the effusive phase of the eruption, are reflected in the An content of plagio-clase microlite rather than in plagioclase microlite tex-ture.

ACKNOWLEDGMENT

I am grateful to Prof. H. Sato for helping me with sam-pling and providing useful comments. I appreciate two anonymous reviewers for their constructive and helpful comments and discussion. I also thank Dr Keiko Suzuki－Kamata and all the members of the Kobe University Vol-canology Group for their discussions.

REFERENCE

Aramaki, S. (1984) Formation of the Aira Caldera, southern Kyushu, 22,000 years ago. Journal of Geophysical Research, 89, 8485－8501.

Cashman, K.V. (1992) Groundmass crystallization of Mount St. Helens dacite, 1980－1986: a tool for interpreting shallow magmatic processes. Contributions to Mineralogy and Petrol-ogy, 109, 431－449.

Couch, S. (2003) Experimental investigation of crystallization kinetics in haplogranite system. American Mineralogist, 88, 1471－1485.

Couch, S., Sparks R.S.J. and Carroll, M.R. (2003) The kinetics of degassing－induced crystallization at Soufrière Hills Volcano, Montserrat. Journal of Petrology, 44, 1477－1502.

Eichelberger, J.C., Carrigan, C.R., Westrich, H.R. and Price, R.H. (1986) Non－explosive silicic volcanism. Nature, 323, 598－602.

Fukuyama, H. (1978) Geology of Sakurajima volcano, southern Kyushu, The Journal of the Geological Society of Japan. 84, 309－316 (in Japanese with English abstract).

Fukuyama, H. and Ono, K. (1981) Geological map of Sakurajima volcano 1:25,000. Geological Survey of Japan, 1, 1－8 (in Japanese with English abstract).

Gardner, C.A., Cashman, K.V. and Neal, C.A. (1998) Tephra－fall deposits from the 1992 eruption of Crater Peak, Alaska: implications of clast textures for eruptive processes. Bulletin of Volcanology, 59, 537－555.

Gurioli, L., Houghton, B.F., Cashman, K.V. and Cioni, R. (2005) Complex changes in eruption dynamics during the 79 AD eruption of Vesuvius. Bulletin of Volcanology, 67, 114－159.

Hammer, J.E., Cashman, K.V., Hoblitt, R.P. and Newman, S. (1999) Degassing and microlite crystallization during pre－climactic events of the 1991 eruption of Mt. Pinatubo,

Philippines. Bulletin of Volcanology, 60, 355－380.Hammer, J.E. and Rutherford, M.J. (2002) An experimental study

of the kinetics of decompression－induced crystallization in silicic melt. Journal of Geophysical Research, 107, 1－24.

Higgins, M.D. (1994) Numerical modeling of crystal shapes in thin sections: Estimation of crystal habit and true size. American Mineralogist, 79, 113－119.

Higgins, M.D. (2000) Measurement of crystal size distributions. American Mineralogist, 85, 1105－1116.

Hoblitt, R.P. and Harmon, R.S (1993) Bimodal density distribu-tion of cryptodome dacite from the 1980 eruption of Mount St. Helens, Washington. Bulletin of Volcanology, 55, 421－437.

Houghton, B.F. and Wilson, C.J.N. (1989) A vesicularity index for pyroclastic deposits. Bulletin of Volcanology, 51, 451－462.

Ishihara, K., Takayama, T., Tanaka, Y. and Hirabayashi, J. (1981) Lava flows at Sakurajima volcano (I), volume of the histori-cal lava flows. Annuals Disaster Prevention Research Institute, Kyoto University, Kyoto, 24(B－1). 1－10 (in Japa-nese with English abstract).

Ishihara, K., Iguchi, M. and Kamo, K. (1985) Lava flows at Sakurajima volcano (II), Numerical simulation of the 1914 lava flows on the western side of Sakurajima. Annuals Disaster Prevention Research Institute, Kyoto University, Kyoto, 28(B－1). 1－11 (in Japanese with English abstract).

Jaupart, C. and Tait, S. (1990) Dynamics of eruption phenomena. Modern methods of igneous petrology: understanding mag-matic processes, (Nicholls. J. and Russell, J. K. Eds.). Reviews in Mineralogy, 24, Mineralogical Society of Ame-rica, Washington DC, 213－238.

Jaupart, C. and Allègre, C.J. (1991) Gas content, eruption rate and instabilities of eruption regime in silicic volcanoes. Earth and Planetary Science Letters, 102, 413－429.

Klug, C. and Cashman, K.V. (1994) Vesiculation of May 18, 1980, Mount St. Helens magma. Geology, 22, 468－472.

Klug, C., Cashman, K.V. and Bacon, C.R. (2002) Structure and physical characteristics of pumice from the climactic erup-tion of Mount Mazama (Crater Lake), Oregon. Bulletin of Volcanology, 64, 486－501.

Kobayashi, T. (1982) Geology of Sakurajima volcano: A review. Bulletin of the Volcanological Society of Japan, 27, 277－292 (in Japanese with English abstract).

Kobayashi, T. (1986) Formation and pyroclastic flow of Sakurajima volcano. Report of research project “Characteristics and disaster of volcanic dry pyroclastic flows. (leader Aramaki, S.) (Monbu Kagakusyou, Tokyo), 137－163 (In Japanese).

Koto, B. (1916) The great eruption of Sakurajima in 1914. Journal of College of Science, Imperial University of Tokyo, Tokyo, 38, 1－237.

Lange, R.L. and Carmichael, I.S.E. (1990) Thermodynamic prop-erties of silicate liquids with emphasis on density, thermal expansion and compressibility. Modern methods of igneous petrology: understanding magmatic processes (Nicholls. J. and Russell, J.K. Eds.), Reviews in Mineralogy, 24, Miner-alogical Society of America, Washington DC, 25－64.

Marsh, B.D. (1988) Crystal size distribution (CSD) in rocks and the kinetics and dynamics of crystallization I. Theory. Con-tributions to Mineralogy and Petrology, 99, 277－291.

Martel, C., Bourdier, J.L., Pichavant, M. and Traineau, H. (2000) Textures, water content and degassing of silicic andesites from recent Plinian and dome－forming eruptions at Mount

198 K. Nakamura

Pelèe volcano (Martinique, Lesser Antilles arc). Journal of Volcanology and Geothermal Research, 96, 191－206.

Martel, C. and Schmidt, B.C. (2003) Decompression experiments as an insight into ascent rates of silicic magmas. Contri-butions to Mineralogy and Petrology, 144, 397－415.

Moore, G., Vennemann, T. and Carmichael, I.S.E. (1998) An empirical model for solubility of H2O in magmas to 3 kilo-bars. American Mineralogist, 83, 36－42.

Okuno, M. (2002) Chronology of Tephra Layers in Southern Kyushu, SW Japan, for the last 30,000 years. The Quaternary research, 41, 225－236 (in Japanese with English abstract).

Okuno, M., Nakamura, T., Moriwaki, H. and Kobayashi, T. (1997) AMS radiocarbon dating of the Sakurajima tephra group, southern Kyushu, Japan. Nuclear Instruments and Methods in Physics Research, B123, 470－474.

Sahagian, D.L. and Proussevitch, A.A. (1998) 3D particle size dis-tributions from 2D observations: stereology for natural appli-cations. Journal of Volcanology and Geothermal Research, 84, 173－196.

Sekine, T., Katsura, T. and Aramaki, S. (1979) Waters saturated relations of some andesites with application to the estimation of the initial temperature and water pressure at the time of eruption. Geochimica et Cosmochimica Acta, 43, 1367－1376.

Sparks, R.S.J. (1978) The dynamics of bubble formation and growth in magmas: A review and analysis. Journal of Vol-canology and Geothermal Research, 3, 1－37.

Suzuki, Y. and Nakada, S. (2002) Vesiculation and magma ascent

process in the Usu 2000 eruption, inferred from texture and size distribution of bubble. Bulletin of the Volcanological Society of Japan, 47, 675－688 (in Japanese with English abstract).

Taddeucci, J., Pompilio, M. and Scarlato, P. (2004) Conduit pro-cesses during the July－August 2001 explosive activity of Mt. Etna (Italy): inferences from glass chemistry and crystal size distribution of ash particles. Journal of Volcanology and Geothermal Research, 137, 33－54.

Toramaru, A. (1990) Measurement of bubble size distributions in vesiculated rock with implications for quantitative estimation of processes. Journal of Volcanology and Geothermal Re-search, 43, 71－90.

Toramaru, A. (1995) Numerical study of nucleation and growth of bubbles in viscous magmas. Journal of Geophysical Re-search, 100, 1913－1931.

Yanagi, T., Ichimaru, Y. and Hirahara, S. (1991) Petrochemical evidence for coupled magma chambers beneath the Sakura-jima volcano, Kyushu, Japan. Geochemical Jounal, 25, 17－30.

Westrich, H.R. and Eichelberger, J.C. (1994) Gas transport and bubble collapse in rhyolitic magma: an experimental app-roach. Bulletin of Volcanology, 56, 447－458.

Manuscript received June 4, 2004Manuscript accepted January 13, 2006Manuscript handled by Makoto Arima

Documents

Textures of plagioclase microlite and vesicles within