dacite-basalt magma interaction at yakedake volcano

194 Y. Ishizaki 195Dacite－basalt magma interaction at Yakedake volcanoJournal of Mineralogical and Petrological Sciences, Volume 102, page 194─210, 2007

doi: 10.2465/jmps.060619Y. Ishizaki, [email protected]－toyama.ac.jp Corresponding author

Dacite-basalt magma interaction at Yakedake volcano, central Japan: petrographic and chemical evidence from the 2300 years B.P. Nakao pyroclastic flow deposit

Yasuo ISHIZAKI

Department of Earth Sciences, University of Toyama, 3190 Gofuku, Toyama 930－0844, Japan

The 2300 years B.P. eruption of the Yakedake volcano in central Japan consisted primarily of lava extrusion and dome growth in the summit area, and a repetitive gravitational collapse of the dome produced a series of block－and－ash flows known as the Nakao pyroclastic flow deposit (NPFD). Based on the geochemistry and mineral-ogy, the juvenile materials in the NPFD can be assigned to five groups: light－colored, porphyritic dacite (white dacite) showing little or no petrographic evidence of magma mixing, dark－colored hybrid andesite (black andesite) with disequilibrium phenocryst assemblages and textures, banded lava with streaks of white dacite and hybrid andesite, basaltic andesitic enclaves (hybrid enclave) having the same disequilibrium phenocryst assemblage as the hybrid andesite, and basaltic enclave (primitive enclave) lacking any evidence of magma mixing. Compositional data from the phenocrysts and whole rocks demonstrate that the juvenile materials of the NPFD preserve a magma mixing/mingling event between a basaltic magma (49.5 wt% SiO2, T ∼ 1075 °C from olivine－melt geothermometry), which is compositionally similar to the primitive enclave, and a dacitic magma (64.6 wt% SiO2, T = 790－800 °C from Fe－Ti oxide geothermometry), which is compositionally similar to the white dacite. The NPFD eruption was caused by an invasion of the basaltic magma into the preexisting, highly crystalline dacitic magma chamber. The composition of the dacitic magma remained constant throughout the eruption. However, heating by the basalt magma increased its temperature locally up to T ∼ 950 °C. Minor disruption and consecutive quenching of the replenishing basaltic magma may have formed the primitive enclave. Simultaneously, the replenishing basaltic magma entrained small amounts of the dacitic magma, pro-ducing a hybrid basaltic andesitic magma layer at the base of the chamber. The quenched part of the hybrid layer was preserved as the hybrid enclave. Finally, the simultaneous ascent of the dacitic magma and liquid interior of the hybrid layer through a common conduit promoted the mixing/mingling of the two magmas and caused the coeruption of diverse lava types.

Keywords: Yakedake volcano, Magma chamber, Magma mixing/mingling, Mafic enclaves

INTRODUCTION

Yakedake volcano is situated ~ 25 km west of Matsumoto City in central Japan and is the newest volcanic center of the Yakedake volcano group (YVG, Fig. 1). The last mag-matic eruption at Yakedake occurred at 2300 14C years B.P. (Oikawa, 2002). The restless nature of Yakedake and the high possibility of future eruptions are indicated by recent periods of seismic upheaval (e.g., earthquake swarms in 1990, 1993－1994, and in 1998) and frequent phreatic erup-tions (e.g., the phreatic eruptions in 1962－1963 and 1995).

This paper documents the petrology of juvenile lava

clasts in the Nakao pyroclastic flow deposit (NPFD). These clasts reveal evidence of interactions between a highly porphyritic dacitic magma and a near－aphyric basalt magma, including the presence of a mafic enclave, a streaky banded lava, and a homogeneous hybrid lava with disequilibrium mineral assemblages and textures. Furthermore, they indicate the coeruption of diverse lava types. The central objective of this work is to answer the following question: “How do the highly porphyritic daci-tic magma and the nearly aphyric basalt magma, a com-mon combination of end members in island arc and conti-nental margin volcanisms (e.g., Pinatubo, see Pallister et al., 1992; Unzen: Nakamura, 1995; Lassen Peak: Clynne, 1999), interact and produce texturally and composition-ally diverse lavas during a single eruption episode?”

194 Y. Ishizaki 195Dacite－basalt magma interaction at Yakedake volcano

Research into such problems requires a systematic analy-sis of the composition and texture of phenocryst minerals, because phenocrysts are known to record episodes of magma mixing/mingling, as well as the pressures and temperatures where two magmas interact (e.g., Nakamura, 1995; Feeley and Dungan, 1996; Clynne, 1999). This paper presents a comprehensive petrologic investigation of the NPFD lava clasts and their mafic enclaves, and dis-cusses the dacite－basalt magma interaction processes that occurred beneath Yakedake volcano.

GEOLOGICAL SETTING

Recent geological studies have revealed that the YVG is composed of three older volcanic edifices (120－70 ka, Warudaniyama, Iwatsuboyama, and Odana) and three younger volcanic edifices (26 ka to the present day, Shirataniyama, Akandana, and Yakedake) (Oikawa, 2002). Presently, Yakedake is the only active center in the YVG. Eruption products of the YVG are composed mainly of thick lava flows/domes and their collapse deposits (block－and－ash flow and debris flow deposits). Pumiceous deposits, indicative of large explosive erup-tions, are rare. Rocks are mainly dacitic and andesitic in composition, and commonly contain large phenocrysts of plagioclase and amphibole.

The last magmatic eruption at Yakedake occurred

about 2300 years B.P. based on 14C dating estimates for the pyroclastic deposits, which effused about 0.3 km3 of magma (Oikawa et al., 2002). During this eruption, a lava dome (the Yakedake Dome) with a dacitic to andesitic composition was formed on a steep slope near the sum-mit. This resulted in a series of block－and－ash flows and the formation of the NPFD. The most suitable exposure for stratigraphic sampling of NPFD rocks occurs at loca-tion YK1 (2.4 km NNW of the summit) in the Ashiarai-dani quarry (Fig. 1), where the NPFD sheet rests on top of the deposits of older debris flows. This can be divided into seven flow units from the interbedded thin ashy lay-ers (Fig. 2). Each flow unit is composed of dispersed lava clasts of gravel to boulder size, set in a sandy matrix. The lack of soil within the deposits suggests that the NPFD was emplaced over a short time interval.

ANALYTICAL METHODS

Two hundred juvenile lava clasts (> 5 cm diameter) were collected from each sampled unit. The rock type of the sampled lava clasts was identified in the field by visual inspection, and subsequently the relative abundance (ex-pressed as a clast number percentage) of each rock type was calculated (Fig. 2C).

Twenty－eight samples were selected for analysis of the modal and whole－rock compositions (Table 1). The

Figure 1. Geologic map and eruption ages of the Yakedake volcano group (Oikawa, 2002).


modal compositions were determined by point counting ~ 3000 points. Phenocrysts were defined as > 0.3 mm in the longest dimension. The whole－rock compositions were measured using a Rigaku RIX 3000 X－ray fluores-cence spectrometer located at Niigata University, Japan. The analytical method used and the accuracy obtained is described in Takahashi and Shuto (1997).

Mineral analysis was carried out on 10 samples using a JEOL JCXA－730 microprobe located at the Center for Instrumental Analysis at University of Toyama, Japan. The operating conditions used were: accelerating voltage = 15 kV, beam current = 20 nA, and count time for each element = 8－20 s. All the analysis was carried out using the oxide ZAF correction method. Represen-tative compositions of the phenocryst minerals are given in Tables 2－4.

ROCK TYPES AND PETROGRAPHY

Four rock types with different compositions and textures were identified at location YK1. In order of abundance, these are: light－colored dacite (white dacite), dark－col-

ored andesite (black andesite), banded lava, and basaltic/basaltic andesitic enclave (mafic enclave) (Fig. 3). The first three rock types occur as a lava clast, while the enclave is found inside the lava clast. A lithologic analysis at location YK1 suggests that the three lava types were deposited together during the entire NPFD eruption (Fig. 2C). Lava clasts of the white dacite and black andesite are consistently predominant (~ 26%－56% of the investigated size), whereas the banded lava is less abundant (< 20%).

The white dacite consists of a moderately vesicular groundmass and 32－51 vol% of phenocrysts of plagio-clase, amphibole, biotite, orthopyroxene (opx), quartz, and Fe－Ti oxides ± olivine. Combinations of the pheno-crysts listed above, except for olivine, are also present as glomeroporphyritic aggregates, indicating an equilibrium cosaturation of these phases within the dacite magma. Most phenocrysts lack disequilibrium reaction textures (except for the breakdown rims on the amphibole). The groundmasses consist of abundant rhyolitic glass and rare microphenocrysts (defined as 0.03－0.3 mm in the long dimension) of plagioclase, opx, and oxides.

The black andesite contains ~ 30－35 vol% pheno-

Figure 2. (A) Outcrop view of the NPFD overlying the pre－NPFD deposits at YK1 shown in Figure 1. Line X－Y denotes the stratigraphic sec-tion shown in Figure 2(B). The person in the circle is shown for scale purposes. (B) Stratigraphic section of the NPFD at YK1 and the selected samples. WD, white dacite; BA, black andesite; BL, banded lava; WB; white band; BB, black band; HE, hybrid enclave; PE, primi-tive enclave. (C) Plot of the clast number percentages of dacitic, andesitic, and banded lava clasts versus the stratigraphic height.


crysts of plagioclase, amphibole, biotite, opx, olivine, quartz, and Fe－Ti oxides. Many of the phenocrysts show various indications of reaction with the surrounding melt, such as the sieve texture on plagioclase, decomposed rims on the amphibole, and reversely zoned rims on the dacite－derived phenocrysts. These phenocryst－melt reactions, as well as the disequilibrium phenocryst assemblages (e.g., quartz + olivine), clearly indicate the hybrid origin of the black andesite. The groundmasses consist of plagioclase, opx, Fe－Ti oxides, clinopyroxene (cpx) micropheno-crysts, and brown glass.

The banded lava contains varying proportions of

light－ and dark－colored bands, which are petrographically and compositionally equivalent to the white dacite and black andesite, respectively. The boundaries between the adjacent white and black bands are sharp, even under the microscope.

Enclaves of phenocryst－poor basaltic andesite and basalt constitute ~ 5 vol% of the material that erupted dur-ing the NPFD eruption. They range in size from submilli-meter－scale blebs identifiable in thin sections to ellipsoi-dal or angular bodies up to 50 cm across. Based on their phenocryst assemblages, these enclaves can be divided into two categories: a hybrid enclave with the same phe-

WD, white dacite; BL, banded lava (WB, white band; BB, black band); BA, black andesite; HE, hybrid enclave; PE, primitive enclave. Each crystal was classified according to its size as follows: phenocryst > 0.3 mm, 0.03 < microphenocryst/mm < 0.3, and micro-lite < 0.03 mm. The areal% was measured from images of thin sections (∼ 6 cm2 per section) using the Scion Image software package.ND, not determined.*** sodic plagioclase and amphibole phenocrysts are not present in the primitive enclave.

Table 1. Whole－rock and modal analysis of the selected NPFD samples


nocryst assemblage as the black andesite, and a primitive enclave with only olivine and calcic plagioclase pheno-crysts (Table 1). Although > 100 enclaves were examined in this study, only one primitive enclave (6ME3) was recovered. All the enclaves have a diktytaxitic－textured groundmass consisting of acicular microphenocrysts of plagioclase, pargasitic amphibole, opx and equant magne-tite, with interstitial glass and vesicles. The groundmass mineral size varies between mafic enclaves. A few en-claves have chilled margins, defined by a decrease in the groundmass grain size at the enclave－host contact.

WHOLE-ROCK CHEMISTRY

Whole－rock major and trace element analyses of the lavas and mafic enclaves are plotted in Fig. 4. According to the classification of Peccerillo and Taylor (1976), the NPFD rocks are medium－K type. The mafic enclaves are plotted in the tholeiitic field of Miyashiro (1974), whereas the lava clasts are plotted in the calc－alkaline field. The trends in the chemical composition of the major and trace ele-ment exhibit remarkable linearity: TiO2, Al2O3, FeO*, MnO, MgO, CaO, P2O5, V, and Sr decrease, but Na2O, K2O, Ba, Rb, and Zr monotonically increase with increas-ing SiO2 content. This strong linearity can best be pro-duced by mixing two end members, as will be discussed later.

The white dacites are virtually identical in whole－rock composition (64.3－64.6 wt% SiO2) and are plotted within the range of compositions of the lava that erupted in the YVG during the last 26 ka. The SiO2 content of the black andesite ranges from 59.8 to 63.9 wt%, bridging the gap between the white dacite and the mafic enclave (49.5－54.2 wt% SiO2). The white bands in the banded lava are compositionally identical to the white dacite, whereas the

black bands are at the silicic end of the compositional range defined by the black andesites. Analysis of seven mafic enclaves yields a plot as a linear extrapolation of the white dacite and black andesite data. However, they are separated from the most mafic black andesite by a gap of ~ 6 wt% SiO2. Primitive enclave 6ME3, the most mafic rock in the NPFD, is a low－MgO type, high－alumina basalt (using the nomenclature of Sisson and Grove, 1993). The primitive enclave is not a primary basalt in the strict sense, judging from its very low compatible－ele-ment contents (e.g., Ni and Cr).

MINERALOGY

Plagioclase

The plagioclase phenocrysts are euhedral, up to 15 mm in length (Fig. 3), and range from An35 to An93 with two pop-ulations centered at An45 and An85 (Fig. 5). The sodic pla-gioclase phenocryst dominates the host lavas, whereas the calcic plagioclase dominates the mafic enclaves. The primitive enclave contains only an An－rich population. Based on the core composition (Fig. 5), zoning patterns (Fig. 7), and textures (Fig. 6), the phenocryst populations can be classified into three types, as follows.

Type 1 phenocrysts typically vary between An38 and An58 with small－amplitude oscillatory zoning from their cores to near their rims. Some have thin calcic rims of An52−63 (Fig. 6A). They constitute more than 90% of the plagioclase phenocrysts in the white dacite, and they com-prise ~ 40%－70% of the plagioclase phenocrysts in the black andesite. However, they are absent in the hybrid enclave.

Type 2 phenocrysts have a clear, resorbed sodic core (An38−56) identical to the Type 1 plagioclase. They are

Figure 3. Sawn slabs of the NPFD samples. Scale bar denotes 2 cm.


mantled by a sieve－textured ring and a clear, normally zoned overgrowth rim (An75−83 at the inner boundary, An50－60 at the margin) with cpx and opx microphenocrysts in the overgrowth rim (Fig. 6B, C). The sieve texture has a marked resemblance to the experimental reaction tex-tures produced when sodic plagioclase is immersed in a melt that is in equilibrium with more calcic plagioclase (Tsuchiyama, 1985). The Type 2 plagioclase is predomi-nant in the hybrid enclave (~ 100% of the sodic plagio-clase phenocrysts), and it comprises ~ 20%－60% of the phenocrysts in the black andesite. However, it is scarce or absent in the white dacite.

Type 3 phenocrysts have a euhedral An－rich core (mostly An85−93), surrounded by a clear overgrowth rim that grades from higher An content (~ An80) to lower An content (~ An50−70) (Fig. 6C). The composition and zoning pattern of the overgrowth rim of the Type 3 phenocrysts are identical to those of the Type 2 phenocrysts, indicat-ing that both the overgrowth rims crystallized in a com-mon environment. The Type 3 plagioclase is predominant in the primitive enclave (~ 100% of the plagioclase phe-nocryst), and it comprises ~ 10%－40% of the phenocrysts

in the hybrid enclave and black andesite. However, it is scarce or absent in the white dacite.

Plagioclase is also an abundant component of the groundmass of lavas as microphenocrysts and microlites, and of enclaves as microphenocrysts. The plagioclase microphenocrysts have broadly overlapping ranges that have intermediate An content between the two phenocryst populations (Fig. 5).

Quartz

Quartz occurs as large (< 12 mm) but always rounded phenocrysts in most of the NPFD rocks. Quartz pheno-crysts with cpx coronas are common in the black andesite and hybrid enclave, and rimmed and unrimmed quartz phenocrysts coexist in these rocks. In contrast, quartz phenocrysts in the white dacite lack cpx coronas.

Amphibole

Amphibole occurs as large (< 15 mm) euhedral pheno-crysts in host lavas. Most phenocrysts are unzoned, and

Figure 4. Whole－rock variation diagrams. The discriminant line in the SiO2－FeO*/MgO diagram is from Miyashiro (1974), and those in the K2O－SiO2 diagram are from Peccerillo and Taylor (1976). The field of the compositional range for the dacite lavas that erupted after 26 ka is also shown.


all belong to magnesiohornblende (using the classification of Leake et al., 1997). In contrast, two compositionally distinct populations, host－derived magnesiohornblende phenocrysts and lath－shaped pargasite microphenocrysts, coexist in the mafic enclave.

The pristine amphibole phenocrysts (Type 1 amphi-bole, Fig. 6G) are predominant in the white dacite. In con-trast, the amphibole phenocrysts in both the black andes-ite and hybrid enclave are partially or completely replaced by fine－grained pyroxene(s), magnetite, and rare plagio-clase (Type 2 amphibole, Fig. 6H, I). Amphiboles can be destabilized by either decreasing the pressure or increas-ing the temperature (Rutherford and Hill, 1993), and the high temperature indicated by the opx－cpx pairs from the breakdown rim of the Type 2 amphiboles suggests that the breakdown reaction had been promoted by an increase in temperature (see the subsection “Temperature”).

Biotite

Biotite occurs as large (< 13 mm), unzoned (Mg# ~ 57－61), and often equant phenocrysts. Some grains are par-tially or completely decomposed into anhydrous minerals. Reacted and pristine biotite phenocrysts occur together in the white dacite and black andesite, whereas only reacted biotite phenocrysts are present in the hybrid enclave.

Orthopyroxene

The orthopyroxene phenocrysts (< 5 mm in length) are euhedral to subhedral, and they have a homogeneous Fe－rich core with an Mg#[= 100 × Mg/(Mg + Fe2+)] between 55 and 64 (Fig. 8). Two phenocryst populations were identified based on the textures and rim compositions (Fig. 9), as follows.

Type 1 phenocrysts are euhedral with flat zoning profiles from their cores to margins with an Mg# fluctuat-ing by ~ 3 units or less. Most grains show little zoning, but a few have a thin magnesian rim with an Mg# ~ 58－63 (Fig. 6D). They constitute more than 80% of the opx phenocrysts in the white dacite and ~ 30%－50% of the opx phenocrysts in the black andesite. However, they are absent in the hybrid enclave.

Type 2 phenocrysts have a homogeneous Fe－rich core with variably resorbed shapes that are surrounded by ~ 20－40 μm thick magnesian opx overgrowth rims (Mg# ~ 70－75) (Fig. 6E). Occasionally, a vermicular aggregate of cpx and opx microphenocrysts is present between the resorbed core and the overgrowth rim (Fig. 6F). Type 2 opx constitute ~ 100% of the opx phenocrysts in the hybrid enclave and ~ 30%－60% of those in the black andesite. However, they are rare or absent in the white dacite.

Clinopyroxene

Clinopyroxene occurs as isolated, sector－zoned micro-phenocrysts in reaction rims on opx, olivine, quartz, and amphibole phenocrysts, and occasionally as inclusions in overgrowth rims of the Type 2 plagioclase phenocrysts. These are variable in composition, with their core Mg# ranging from 65 to 78.

Olivine

The olivine phenocrysts (< 3 mm) are present in most samples, either as single grains or as aggregates with Type 3 plagioclase. All the analyzed olivine phenocrysts are normally zoned with core compositions from Fo79 to Fo73 (Fig. 8). In host lavas, overgrowths on the olivine pheno-

Figure 5. Histograms of the core compositions of the plagioclase phenocrysts and microphenocrysts.


crysts are either thin or absent, and if present, they are composed of opx. In mafic enclaves, overgrowths are diverse, and usually composed of pargasite, or in a few cases, of opx inside and pargasite outside.

Fe-Ti Oxides

Titanomagnetite is present in all rock types as phenocrysts (< 0.7 mm) and microphenocrysts. Based on the core Al2O3 content and the mole fraction of ulvöspinel compo-nent, XUsp, defined using the method of Stormer (1983) (Fig. 10), three populations can be distinguished as fol-lows: a high－Al type (XUsp = 28－44, 3－4 wt% Al2O3), a

low－Ti type (XUsp = 20－27, 1.7－2.2 wt% Al2O3), and a high－Ti type (XUsp = 35－47, 2.0－3.5 wt% Al2O3). The high－Al and high－Ti types occur as microphenocrysts in the mafic enclave and black andesite, respectively. The low－Ti type occurs in both the white dacite and the black andesite as phenocrysts and microphenocrysts. Most low－

Ti type phenocrysts in the white dacite exhibit reverse zoning. This chemical zoning is limited to within a dis-tance of 20 μm from the margin, and the cores of the phe-nocrysts remain constant in their chemistry, irrespective of the rock type. In contrast, most low－Ti phenocrysts in the black andesite have exsolution lamellae along their margins.

Figure 6. A－F: Back－scattered electron micrographs of plagioclase and orthopyroxene (opx) phenocryst textures. The scale bar denotes 100 μm. G－I: Photomicrographs of amphiboles under plane－polarized light. Scale bar denotes 1 mm. A: Type 1 plagioclase in the white dacite, 1W2. This crystal has a dark sodic core and a light calcic rim. B: Type 2 plagioclase in the black andesite, 2B1. This phenocryst has a wide calcic overgrowth rim, which becomes more sodic towards the edge. C: Coexisting Type 2 and Type 3 plagioclase phenocrysts in the hybrid enclave, 7ME2. The Type 3 plagioclase (on the right－hand side) has a light calcic core and a dark sodic rim. D: Unzoned Type 1 opx pheno-cryst in the white dacite, 1W2. Mag, magnetite. E: Type 2 opx phenocryst in the black andesite, 2B1. The thin magnesian rim is darker than the more Fe－rich core. F: Type 2 opx phenocrysts in the black andesite, 2B1. The phenocrysts have wide magnesian rims. DC, dissolved Fe－rich core; VZ, vermicular zone consisting of the magnesian opx and cpx; OM, overgrowth mantle on vermicular zone. G: Type 1 amphibole typical of the white dacite. H: Coexisting Type 1 and Type 2 amphibole phenocrysts in the black andesite, 7B1. I: Large Type 2 amphibole xenocryst and small pargasite microphenocrysts (Prg) in the hybrid enclave, 2ME2. The xenocryst has been pseudomorphed by a fine－grained intergrowth that consists of optically continuous cpx, plagioclase, opx, and titanomagnetite.


Ilmenite (< 5% of the Fe－Ti oxides) is present in all the lava types as phenocrysts and microphenocrysts, but is absent in the mafic enclave. The ilmenite pheno/microphenocrysts are more homogeneous (XIlm = 82－87) than the magnetite phenocrysts are.

END-MEMBER MAGMAS

In the NPFD products, the occurrence of both the mafic enclave and banded lava clast, the disequilibrium pheno-cryst textures (e.g., sieve texture on plagioclase pheno-crysts) and assemblages (e.g., olivine + quartz) in both the black andesite and hybrid enclave, and the linear mixing trends of the variation diagrams provide evidence of mixing/mingling processes having occurred. The black andesite and hybrid enclave lie intermediate and collinear in the major and trace element compositions between the most silicic dacite (4W1, 64.6 wt% SiO2) and the primi-tive enclave (6ME3, 49.5 wt% SiO2) (Fig. 4), suggesting their hybrid origin between two end－member magmas with different compositions. The phenocryst assemblage of Sample 4W1 (sodic plagioclase, magnesiohornblende, biotite, opx, quartz, and Fe－Ti oxides) suggests that little or no mixing with a basaltic magma occurred. Similarly, the phenocryst assemblage of Sample 6ME3 (calcic pla-gioclase and olivine) suggests that no mixing with dacitic magma occurred. Therefore, Samples 4W1 and 6ME3 can be considered as being analogues of the silicic (dacitic)

and mafic (basaltic) end members, respectively.

PHYSICAL CONDITIONS

Temperature

The temperatures of the end members, as well as the ther-mal histories of the dacite－derived phenocrysts, were examined using the single－pyroxene and Fe－Ti oxide geothermometer of the QUILF software program (Ander-son et al., 1993) and the olivine－liquid geothermometer (Sisson and Grove, 1993) (Figs. 11 and 12, Table 3). For the Fe－Ti oxide geothermometer, only the oxide pairs that passed the Mg/Mn partition test of Bacon and Hirschmann (1988) were used. As exsolution textures are developed along the margins of the magnetite phenocrysts in the black andesite, the oxide rim temperatures could only be determined for the white dacite. The pressure dependence of the geothermometers is slight, and a pressure of 200 MPa was assumed in all our calculations.

The orthopyroxene core temperatures stay constant (mostly 770－820 °C), irrespective of either the rock type or the phenocryst type (Fig. 11). The oxide core tempera-tures show a similar result: 797 ± 9 °C (n = 18) for the white dacite and 793 ± 13 °C (n = 16) for the black andes-ite (Fig. 12). The oxygen fugacity (fO2) exhibits a narrow range of NNO + 1 for both lava types (where NNO denotes the nickel－nickel oxide buffer). It is encouraging

Figure 7. Typical zoning patterns of some selected plagioclase phenocrysts.


that two different calculation modes of the QUILF soft-ware program give identical results for the white dacite and the black andesite. The cores of the opx and oxide phenocrysts probably record the temperature of the dacitic end member before mixing occurred.

Unlike the core temperatures, the opx rim tempera-tures are related to the type of phenocryst (Fig. 11). The rim temperatures of the Type 1 and Type 2 opx range between 770 and 950 °C and 890 and 1130 °C, respec-tively. Most oxide phenocrysts in the white dacite are also reversely zoned, so that the rims invariably have a higher temperature of up to 950 °C (Fig. 12).

The temperatures estimated from the micropheno-crystic opx－cpx pairs in the overgrowth and reaction rims of the Type 2 phenocrysts using the opx－cpx geother-mometry of the QUILF software program are as follows (Table 4): 1055－1110 °C for the Type 2 opx, 1060－1120 °C for the Type 2 amphibole, and 1025－1050 °C for the

Type 2 plagioclase (Fig. 11). All the estimated tempera-tures are close to each other, and are > 200 °C above the dacitic end member temperature of ~ 800 °C. This implies that the reactions that formed the Type 2 phenocrysts have been promoted mainly by an isobaric increase in tempera-ture.

An accurate application of the olivine－liquid geo-thermometer (Sisson and Grove, 1993) to the basaltic end member is difficult, because the liquid composition in equi-librium with the olivine phenocryst at the time of crystal-lization (including the melt H2O content) is unknown. Taking a conservative approach, it was assumed that the primitive enclave 6ME3 represented the liquid composi-tion of the basaltic end member and melt H2O content as a parameter. The melt H2O content was estimated based on the plagioclase phenocryst composition, because the exchange of Ca and Na ions between the plagioclase and the liquid [KD

Ca－Na, defined as (Ca/Na)plag/(Ca/Na)liq] is dependent on the melt H2O content (Sisson and Grove, 1993). For the primitive enclave, the value of KD

Ca－Na between the most calcic plagioclase phenocryst (An91) and liquid (6ME3) is KD

Ca－Na = 4.6, corresponding to a melt H2O content of ~ 4－5 wt%. For a range of H2O contents, the olivine－liquid geothermometer yields temperatures of ~ 1050－1075 °C, which are close to those estimated from the microphenocrystic opx－cpx pairs in the Type 2 pheno-

Figure 8. Histograms of the core compositions of olivine and orthopyroxene phenocrysts.

Figure 9. Typical zoning patterns of some selected orthopyroxene phenocrysts.


cryst rims (Fig. 11). It is possible that the temperatures estimated from the olivine－liquid geothermometer are realistic for the basaltic end member before mixing. The value of fO2 estimated from the composition of coexisting Type 3 plagioclase (An88−91) and olivine (Fo76−79) in the glomeroporphyritic aggregates using the plagioclase－oliv-ine－oxygen barometer (Sugawara, 2001) lies just above the NNO buffer curve at the indicated temperature range.

Pressure

The depth of the dacitic end member before mixing was estimated using the Al－in－hornblende geobarometer of Johnson and Rutherford (1989). The white dacite contains all the minerals required for the application of the geoba-

rometer, with the exception of alkali feldspar, but its pres-ence was not found to be critical when using the geoba-rometer (Johnson and Rutherford, 1989). The pheno-crystic amphibole yields pressures of 195 ± 20 MPa (Table 2), which corresponds to a depth of ~ 7 km, assuming a granitic crust. Present－day geophysical evidence, e.g., the presence of a low Q, low velocity, low density, and no seismicity in the area beneath the YVG (e.g., Katsumata et al., 1995; Mikumo et al., 1995), is also consistent with a depth of ~ 7 km for the main part of the dacitic magma chamber.

WD, white dacite; BA, black andesite; HE, hybrid enclave; PE, primitive enclave; PHC, core of phenocryst; PHR, rim of phenocryst; MPHC, core of microphenocryst. * Calculated using the method of Anderson et al. (1993). ** Calculated using the method of Johnson and Rutherford (1989).

Table 2. Representative mafic silicate mineral analyses


DISCUSSION

The dacitic magma chamber

The textural relationships and detailed mineral composi-tional data suggest that the NPFD eruption products origi-nated through an interaction between the low－T dacitic and high－T basaltic end members. Modal data show that the dacitic end member was highly crystalline (~ 50 vol% phenocrysts, Table 1). Therefore, the chain of events lead-ing to the NPFD eruption is likely to have begun when the basaltic end member intruded into the crystallizing dacitic end member in the crustal magma chamber. Lavas that have erupted during the last 26 ka at the YVG exhibit a narrow range in whole－rock chemistry (Fig. 4), imply-ing that they were from the same dacitic magma chamber. This dacitic magma chamber could have survived in the crust for long periods (~ 105 years) as a crystal mush, because the cooling rate of the crystal－rich magma is very low compared to the cooling rate of the aphyric magma (Koyaguchi and Kaneko, 1999). Periodic replenishment by a high－T mafic magma, as evidenced by the presence of the mafic enclave in all pre－NPFD lavas of the YVG (Oikawa, 2002), also helped to prevent the resident dacitic magma from fully crystallizing. The question of whether the dacitic magma(s) of the YVG system was produced mainly by partial melting of crustal material or by frac-tional crystallization of a more mafic parent, and to what

extent they are fractionation products or partial melts, is important and will be discussed in another publication.

The white dacite displays the characteristics of pre-eruption heating in the storage region. The direct evidence is from thermometric data (Figs. 11 and 12), suggesting that the dacitic end member was stored at ~ 790－800 °C. However, it then partially reequilibrated at ~ 950 °C. The most probable cause of the change in the temperature of the dacitic magma was the intrusion of a high－T basaltic end member into the dacitic magma chamber. Generally, a mafic replenishment changes the temperature of the basal part of a silicic magma in a chamber (e.g., Couch et al., 2001). The resultant high－T thermal boundary layer in contact with the mafic magma inevitably attains buoyancy and rises as plumes into the overlying colder interior. Parcels of magma with a similar composition, but with very different temperatures, then progressively mix be-cause of convection within the silicic magma chamber

Figure 10. Al2O3 (wt%) versus XUsp for magnetite pheno/micro-phenocrysts in NPFD rocks.

Figure 11. Temperature versus Mg# for orthopyroxene (opx) phe-nocrysts and microphenocrystic opx－clinopyroxene (cpx) pairs in the white dacite, black andesite, and hybrid enclave. The opx－cpx pairs are denoted as: Am from the breakdown rims of Type 2 hornblende, Pl from the overgrowth of Type 2 plagioclase, and opx from the vermicular zone of Type 2 opx.


(Couch et al., 2001). The wide range of the opx and oxide rim temperatures in the white dacite probably reflects this convective self－mixing process within the dacitic magma chamber beneath Yakedake.

Petrogenesis of mafic enclaves

The primitive enclave shows no evidence of magma mix-ing and can be interpreted as a quenched form of the basaltic end member. In contrast, the hybrid enclave con-tains reacted Type 2 phenocrysts, which originally crys-tallized in the dacitic end member. Thus, the formation of the hybrid enclave requires bulk assimilation of the dacitic end member into the basaltic end member. The mixing calculation suggests that a dacitic end member between 9% and 32% was mixed with the basaltic end member to produce the observed hybrid enclave composi-tions. The experiments of Campbell and Turner (1986) have demonstrated that a mafic magma replenished into a silicic magma chamber entrains a resident silicic magma because of the turbulence of the injection. In such a situa-tion, the physical difficulty in mixing and the hybridiza-tion of compositionally disparate magmas is surmounted by high mafic－to－silicic magma ratios and the turbulent conditions in the replenishing magma (Sparks and

Marshall, 1986).After the hybridization, the resultant hybrid magma

probably pooled at the base of the magma chamber because of its high density. The hybrid magma would then be simultaneously cooled by the overlying low－T dacitic end member and by mixing with the end member, which would have promoted the crystallization of calcic/mafic overgrowths on the Type 2 plagioclase and opx pheno-crysts, and acicular groundmass microphenocrysts. Intrigu-ingly, the size of the enclave groundmass micropheno-crysts is independent of the enclave size. This suggests that the NPFD enclave crystallized before they dispersed within the host magma.

Recently, the enclave－disaggregation model has been widely accepted as the main process that produces a homogeneous hybrid andesitic magma from dacitic and basaltic end members (e.g., Feeley and Dungan, 1996; Clynne, 1999). In this model, fragmentation and disaggre-gation of the hybrid enclave and the mixing of its contents back into the host dacite play an important role in andes-ite formation. In such a case, groundmass minerals that had formed within the mafic enclave should be present in the black andesite groundmasses. However, this is not

Figure 12. Temperatures and oxygen fugacity calculated from mag-netite－ilmenite pairs.

WD, white dacite; BA, black andesite; Mag, magnetite; Ilm, ilme-nite; PHC, phenocryst core; PHR, phenocryst rim; XUlv, mole fraction of ulvöspinel; XIlm, mole fraction of ilmenite. * Calculated using the method of Stormer (1983). ** Calculated using the method of Anderson et al. (1993).

Table 3. Representative analyses of Fe－Ti oxide pairs


observed in our black andesite sample. First, the composi-tion of the groundmass magnetite microphenocrysts in the black andesite differs from that in the enclave (Fig. 10). Second, unlike the enclave, the groundmass amphiboles are absent in the black andesite (Fig. 6G－I). These obser-vations discount the enclave－disaggregation model being applicable to the formation of the black andesite. It is likely that the large difference in the temperatures of the enclave magma and dacitic end member causes a higher crystallinity in the mafic enclave, thereby inhibiting its disaggregation.

Petrogenesis of the black andesite

The black andesite contains two types of dacite－derived phenocrysts: pristine Type 1 and reacted Type 2. It is

believed that the former was crystallized almost exclu-sively in the low－T dacitic magma over the entire lifespan of the phenocrysts. In contrast, the thermometric data (Fig. 11) imply that the Type 2 phenocrysts were initially crys-tallized in the low－T dacitic magma and then engulfed by the high－T mafic magma to form dissolved/decomposed textures and reversely zoned rims. Finally, they returned to, and mixed with, the dacitic magma to form the black andesite magma. The modal data, especially the clear cor-relation between the whole－rock SiO2 content and the rel-ative abundances of the Type 2 phenocrysts (Fig. 13), suggest that the Type 2 phenocrysts in the black andesite were derived from a basaltic andesitic hybrid (~ 54 wt% SiO2). This is compositionally similar to the hybrid en-clave. In addition, the presence of unreacted Type 1 phe-nocrysts in the black andesite suggests that the process of phenocryst recycling and the consequent mixing event for the black andesite formation occurred shortly before erup-tion. This is because reaction rims would be expected on these phenocrysts otherwise. As suggested in the previous subsection on the petrogenesis of the mafic enclaves, a comparison of the groundmass mineral assemblage/com-position confirms the conclusion that the black andesite was not derived from enclave disaggregation but instead represents the liquid－state mixture of the dacitic end

BA, black andesite; HE, hybrid enclave; MPHC, core of micro-phenocryst.* Calculated using the method of Anderson et al. (1993).

　Table 4. Results of 2－pyroxene geothermometry

Figure 13. A: The correlation between the areal% of Type 2 plagio-clase in the total sodic plagioclase and the whole－rock SiO2 con-tent. B: The correlation between the areal% of Type 2 amphi-boles in the total amphibole phenocrysts and the whole－rock SiO2 content. The gray line joins the most silicic dacite, 4W1, and the most silicic hybrid enclave, 2ME2.


member and a basaltic andesitic hybrid, probably a syn-eruptive mixing of the two magmas, as evidenced by the presence of abundant banded lava (Fig. 2C), this being the most likely scenario for the black andesite and banded lava development.

In a series of high－T experiments using dacitic and basaltic melts, Kouchi and Sunagawa (1985) have shown that: (1) the two melts can be mixed partially to form a homogeneous andesitic melt in a period of less than a few hours, (2) the mixing and homogenization can be accom-plished by mechanical attrition and gradual assimilation of a viscous dacitic melt by a convecting basaltic melt at the interfacial region, (3) the basaltic melt changes its composition more readily to andesitic than the dacitic melt does, and (4) a fine banding of the unmodified dacitic melt and andesitic hybrid develops at the interface. These experimental results imply that when the basaltic andes-itic hybrid and dacitic magma flow simultaneously through a common volcanic conduit, three magma types, i.e., the unmodified dacitic magma, the newly created andesitic hybrid magma, and the banded lava with bands composed of the unmodified dacitic magma and newly created andesitic hybrid, are expected to coerupt. Thus, Kouchi and Sunagawa’s experiments provide a potential analogy for the compositional relationship of the three lava types that erupted during the NPFD eruption.

Sequence of events that produced the NPFD rock types

The petrologic evidence discussed above allows us to construct a model for the magma process both before and during the eruption of the NPFD. There appear to have been three distinct stages of magma interaction (Fig. 14).Stage 1: Magma mixing in the dacitic magma cham-ber during basaltic replenishment. Before the NPFD eruption occurred, the basalt end－member magma began to leak into the dacitic magma chamber beneath Yake-dake. The highly crystalline dacitic magma region pro-vided an effective barrier, and furthermore, because the arriving basaltic magma was denser than the dacitic magma, the basaltic magma may have spread onto the chamber floor. Minor disruption of the replenishing basal-tic end member, probably because of a flow front instabil-ity in it, may have formed the primitive enclave. Intruding turbulent fountains of the hot basaltic magma led to mix-ing and the production of the basaltic andesitic hybrid, which subsequently tended to sink and accumulate at the base of the chamber. The high mafic－to－silicic magma ratios and turbulence eddy flow in and around the replen-ished basalt permitted the hybridization of the two dis-similar magmas.Stage 2: Selective solidification and disruption of the

Figure 14. A schematic diagram illustrating the main features of the model discussed in the text.


basaltic andesitic hybrid layer. The newly created basal-tic andesitic hybrid pooled beneath the preexisting dacitic magma. The basaltic andesitic hybrid may have chilled at the interface, forming a solid carapace, although the hot-ter interior remained in the liquid state. The heat supplied from the basaltic andesitic hybrid would increase the tem-perature of the basal part of the dacitic magma, resulting in the formation of a high－T (~ 950 °C) dacitic magma. The activated convection of the dacitic magma promoted self－mixing within the dacitic magma and fragmentation of the solid carapace of the basaltic andesitic hybrid. Subsequently, the fragmented carapaces erupted as the hybrid enclave.Stage 3: Syneruptive mixing/mingling in the conduit. Just before, and during, the NPFD eruption, the dacitic and basaltic andesitic hybrid was flowing through a com-mon conduit. The dacitic magma was torn from the inter-face and trapped in a large volume of the basaltic andes-itic hybrid flowing in the center of the conduit. The central flow readily converged toward the andesitic com-positions and, with time, the compositional gap between the dacitic outer flow and the inner flow diminished. In addition, a banded texture was easily produced at the con-tact points of the two magmas by shear mixing (Kouchi and Sunagawa, 1985). In this way, compositionally and texturally diverse magmas were produced inside the vol-canic conduit.

CONCLUSIONS

Diverse mixing/mingling features in the 2300 years B.P. NPFD products were generated during a continuous inter-action between the highly crystalline dacitic magma body and the replenished basaltic magma. Our observations of the NPFD suggest that mixing and hybridization of a mafic magma and a highly crystallized silicic magma is remarkably efficient, despite the initial large contrasts in composition and viscosity. Petrographic evidence favors a model having two stages of interaction: hybridization dur-ing the basaltic replenishment and syneruptive mingling in a volcanic conduit. Such an interaction would occur in magma chambers and conduits of other island arc volca-noes, where mafic magmas are periodically supplied into a crystallizing dacitic magma chamber.

ACKNOWLEDGMENTS

The author wishes to thank Kenji Shuto, Toshiro Taka-hashi and Osamu Ujike for their technical assistances dur-ing the analytical works, and for the constructive reviews by two anonymous referees and the associate editor, Hirokazu Fujimaki, which led to an improvement in the

draft of this paper. This research was partially funded by the University of Toyama Circum－Japan Sea Project 2004－06.

REFERENCES

Anderson, D.J., Lindsley, D.H. and Davidson, P.M. (1993) QUILF: A PASCAL program to assess equilibria among Fe－Mg－Mn－Ti oxides, pyroxenes, olivine, and quartz. Comput-ers and Geosciences, 19, 1333－1350.

Bacon, C.R. and Hirschmann, M.M. (1988) Mg/Mn partitioning as a test for equilibrium between coexisting Fe－Ti oxides. American Mineralogist, 73, 57－61.

Campbell, I.H. and Turner, J.S. (1986) The influence of viscosity of fountains in magma chambers. Journal of Petrology, 27, 1－30.

Couch, S., Sparks, R.S.J. and Carroll, M.R. (2001) Mineral dis-equilibrium in lavas explained by convective self－mixing in open magma chambers. Nature, 411, 1037－1039.

Clynne, M.A. (1999) A complex magma mixing origin for rocks erupted in 1915, Lassen Peak, California. Journal of Petro-logy, 40, 105－132.

Feeley, T.C. and Dungan, M.A. (1996) Compositional and dy-namic controls on mafic－silicic interactions at continental arc volcanoes: Evidence from Cordon El Guadal, Tatara－San Pedro Complex, Chile. Journal of Petrology, 37, 1547－1577.

Johnson, M.C. and Rutherford, M.J. (1989) Experimental calibra-tion of the aluminum－ in－hornblende geobarometer with ap-plication to Long Valley caldera (California) volcanic rocks. Geology, 17, 837－841.

Katsumata, K., Urabe, T. and Mizoue, M. (1995) Evidence for a seismic attenuation anomaly beneath the Hida mountain range, central Honshu, Japan. Geophysical Journal Inter-national, 120, 237－246.

Kouchi, A. and Sunagawa, I. (1985) A model for mixing basaltic and dacitic magmas as deduced from experimental data. Contributions to Mineralogy and Petrology, 89, 17－23.

Koyaguchi, T. and Kaneko, K. (1999) A two－stage thermal evolu-tion model of magmas in continental crust. Journal of Petrology, 40, 241－254.

Leake, B.E., Woolley, A.R., Arps, C.E.S., Birch, W.D., Gilbert, M.C., Grice, J.D., Hawthorne, F.C., Kato, A., Kisch, H.J., Krivovichev, V.G., Linthout, K., Laird, J., Mandarino, J.A., Maresch, W.V., Nickel, E.H., Rock, N.M.S., Schumacher, J.C., Smith, D.C., Stephenson, N.C.N., Ungaretti, L., Whittaker, E.J.W. and Youzhi, G. (1997) Nomenclature of amphiboles. American Mineralogist, 82, 1019－1037.

Mikumo, T., Hirahara, K., Takeuchi, F., Wada, H., Tsukada, T., Fujii, I. and Nishigami, K. (1995) Three－dimensional veloc-ity structure of the upper crust in the Hida region, central Honshu, Japan, and its relation to local seismicity, Quaternary active volcanoes and faults. Journal of Physics of the Earth, 43, 59－78.

Miyashiro, A. (1974) Volcanic rock series in island arc and conti-nental margins. American Journal of Science, 274, 321－355.

Nakamura, M. (1995) Continuous mixing of crystal mush and replenished magma in the ongoing Unzen eruption. Geology, 23, 807－810.

Oikawa, T. (2002) Geology, volcanic history and eruptive style of the Yakedake volcano group, central Japan. Journal of Geo-logical Society of Japan, 108, 615－632 (in Japanese with

210 Y. Ishizaki

English abstract).Oikawa, T., Okuno, M. and Nakamura, T. (2002) The past 3000

years’ eruption history of Yakedake volcano in the Northern Japan Alps. Journal of Geological Society of Japan, 108, 88－102 (in Japanese with English abstract).

Pallister, J.S., Hoblitt, R.P. and Reyes, A.G. (1992) A basalt trigger for the 1991 eruptions of Pinatubo volcano? Nature, 356, 426－428.

Peccerillo, A. and Taylor, S.R. (1976) Geochemistry of Eocene calc－alkaline volcanic rocks from the Kastamonu area, north-ern Turkey. Contributions to Mineralogy and Petrology, 58, 63－81.

Rutherford, M. and Hill, P.M. (1993) Magma ascent rates from amphibole breakdown: An experimental study applied to the 1980－1986 Mount St. Helens eruptions. Journal of Geophysi-cal Research, 98, 19667－19685.

Sisson, T.W. and Grove, T.L. (1993) Experimental investigations of the role of H2O in calc－alkaline differentiation and sub-duction zone magmatism. Contributions to Mineralogy and Petrology, 113, 143－166.

Sparks, R.S.J. and Marshall, L.A. (1986) Thermal and mechanical constraints on mixing between mafic and silicic magmas. Journal of Volcanology and Geothermal Research, 29, 99－

124. Stormer, J.C. (1983) The effect of recalculation on estimates of

temperature and oxygen fugacity from analyses of multicom-ponent iron－titanium oxides. American Mineralogist, 68, 586－594.

Sugawara, T. (2001) Ferric iron partitioning between plagioclase and silicate liquid: Thermodynamics and petrological appli-cations. Contributions to Mineralogy and Petrology, 141, 659－686.

Takahashi, T. and Shuto, K. (1997) Major and trace element anal-yses of silicate rocks using X－ray fluorescence spectrometer RIX3000. Rigaku Denki Journal, 28, 25－37 (in Japanese with English abstract).

Tsuchiyama, A. (1985) Dissolution kinetics of plagioclase in the melt of the system diopside－albite－anorthite, and origin of dusty plagioclase in andesite. Contributions to Mineralogy and Petrology, 89, 1－16.

Manuscript received June 19, 2006Manuscript accepted December 1, 2006Published online February 17, 2007Manuscript handled by Hirokazu Fujimaki

dacite-basalt magma interaction at yakedake volcano

Documents