Costa_merapi

Journal of Volcanology and Geothermal Research xxx (2013) xxx–xxx

VOLGEO-05055; No of Pages 27

Contents lists available at SciVerse ScienceDirect

Journal of Volcanology and Geothermal Research

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

Petrological insights into the storage conditions, and magmatic processes that yieldedthe centennial 2010 Merapi explosive eruption

Fidel Costa a,⁎, Supriyati Andreastuti b, Caroline Bouvet de Maisonneuve a, John S. Pallister c

a Earth Observatory of Singapore, Nanyang Technological University, Singapore, Singaporeb Geological Agency, Center for Volcanology and Geological Hazard Mitigation, Jl. Diponegoro 57, Bandung 40122, Indonesiac USGS Cascades Volcano Observatory, 1300 SE Cardinal Court, Vancouver, WA 98683, USA

⁎ Corresponding author at: Earth Observatory of Sing6592 2401; fax: +65 6792 2149.

E-mail address: [email protected] (F. Costa).

0377-0273/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.jvolgeores.2012.12.025

Please cite this article as: Costa, F., et al., Pet2010 Merapi explosive..., Journal of Volcano

a b s t r a c t
a r t i c l e i n f o
Article history:Received 20 June 2012Accepted 31 December 2012Available online xxxx

Keywords:MerapiCrystal zoningExplosionGeothermometryIndonesiaIntrusionTime scales

To understand the processes that made the 2010 eruption of Merapi much larger and more explosive thanmost dome-forming eruptions of the past century, we investigated the geochemistry, petrology, andpre-eruptive conditions of magmas erupted in 2006 and 2010. The juvenile rocks of 2010 are plagioclase,two-pyroxene basaltic andesites with seriate textures and minor amounts of reaction-free amphibole, Fe–Ti oxides, and rare crystals of olivine and biotite. The bulk-rock composition, mineral paragenesis, and tex-tures are similar to those of juvenile blocks from the much less explosive eruption of 2006. One of the keydifferences is that most amphiboles in 2010 don't have breakdown reaction rims, whereas those of 2006are largely reacted. We acquired >80 X-ray distribution maps of major and minor elements of large areas(>1 cm2) and single crystals, backscattered electron images, electron microprobe analyse, and compositionaltraverses across crystals. The data reveal that both the 2006 and 2010 samples are heterogeneous at variousspatial scales, with numerous reaction textures between pyroxenes and amphiboles, dissolution textures, andlarge variations of crystal sizes, morphologies, and compositions. These features record open-system mag-matic processes involving the assimilation of carbonate rocks, and interactions between various parts ofMerapi's plumbing system, including a degassed shallow magma system and deep hotter and more volatilerich magma intrusions.The petrological complexity of the samples makes unraveling the pre-eruptive conditions of Merapi magmasa petrological puzzle. We applied five different geothermobarometers and performed thermodynamicmodeling with the MELTS algorithm, and we propose that there are at least three crystallization zones or en-vironments below Merapi. A deep reservoir at about 30 (+/−3) km depth is suggested by some amphibolesand high-Al clinopyroxenes. Here is where the high-Al basaltic andesites fromMerapi are generated probablyby water-rich fractionation of more primitive magmas. Such deep magmas are volatile-rich and atnear-liquidus conditions (≥4–6 wt.% H2O, ≥0.15 wt.% SO2, and an undetermined amount of CO2, at about1050 °C) when they start moving towards the surface. A second crystallization zone is recorded by anothertype of amphibole at about 13 (+/−2) km. Here high-Al clinopyroxene may also grow together withCa-rich plagioclase. Assimilation of limestone may also occur at this level as recorded by the very Ca-rich pla-gioclases found in the cores of some crystals. At this location the water content of the melt must remain highenough to stabilize amphibole (4–6 wt.% H2O) but CO2 and SO2 are probably already degassing and contrib-ute to gas changes observed by the monitoring system at the surface. Finally, a shallower part of the system(b10 km) is recorded by the lower anorthite plagioclase and low-Al in clinopyroxene, and perhaps also inorthopyroxene. This part of the system is probably crystal-rich and largely degassed, and is the likely sourceof the high-temperature fumaroles and the volcanic gas plumes that are commonly seen at Merapi.We propose that the 2006 and 2010 eruptions were driven by basically the same processes and magma types.The main difference is the much larger size of the deep and volatile-rich magma replenishment that tookplace in 2010, which had large effects on the kinetics and dynamics of the plumbing system and processes.In 2006, and perhaps also in most of the typical small dome-forming historical eruptions at Merapi, the directascent of deep and gas-rich magmas towards the surface is slowed down and partially arrested by theshallower crystal-rich zones of left-over magma from previous events. However, this was not possible in2010, where the much larger (up to 10 times) size of the magma intrusions overwhelmed the crystal-richeruption filter. In 2010 the deep magma probably resided for only a short time at intermediate to shallowerdepths which allowed it to proceed to the surface still carrying most of its deep gas cargo. The larger magma

apore, Nanyang Technological University, 50 Nanyang Avenue, Block N2-01a-15, Singapore 639798, Singapore. Tel.: +65

rights reserved.

rological insights into the storage conditions, and magmatic processes that yielded the centenniallogy and Geothermal Research (2013), http://dx.doi.org/10.1016/j.jvolgeores.2012.12.025

http://dx.doi.org/10.1016/j.jvolgeores.2012.12.025

mailto:[email protected]


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


2 F. Costa et al. / Journal of Volcanology and Geothermal Research xxx (2013) xxx–xxx

Please cite this article as: Costa, F., et al., Pet2010 Merapi explosive..., Journal of Volcano

intrusion probably induced higher rates of crustal carbonate assimilation and production of additional CO2

gas at shallow depths. This contributed to the much faster than usual ascent rates and larger explosivitiesin 2010 than in 2006. These inferences are supported by the shorter interaction times calculated from the dif-fusion models of clinopyroxene compositions for the 2010 magmas, by the fact that most amphiboles are notbroken down in 2010 as opposed to 2006, and also by the much shorter times of escalating monitoring sig-nals (seismicity and deformation) in 2010 compared to 2006.A puzzling observation is that despite the multiple explosive phases of the 2010 eruption, pumiceous mate-rials are rare, and were only found in the last part of the eruption. This contrasts with the abundant tephralayers and vesiculated deposits of older historical explosive events like 1872, and suggest that syn-eruptiveprocesses in 2010 were also different from standard models. The rarity of expanded pumices in 2010 maybe due to rapid degassing and re-welding of magma as it ascended from intermediate depths. Given thenear constant bulk composition of Merapi magmas erupted in the last decades, and the similarity of texturesand minerals in 2006 and 2010, our study suggests that most Merapi magmas are intrinsically capable of ex-plosive eruptions. Here we propose that whether they do so or not mainly depends on the degree of interac-tion and magma mass proportions between the upper crystal-rich parts of the system (including carbonates)and the deeper and more gas-rich replenishing magmas. Older historical explosive eruptions at Merapi suchas in 1872 were driven by more mafic magmas than those erupted in 2006 and 2010 and thus might becaused by different processes from those discussed here. The still unanswered and vexing questions remainas to why in 2010 a much larger amount of magma was segregated from depths and whether this will happenagain in the near future.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

One of the most difficult tasks in volcanology is to anticipatewhether the next eruption at an already active volcano will be similarin size and explosivity to the previous one. This can be particularlytroublesome at very active volcanoes like Merapi, because most erup-tions exhibit similar behavior, the “Merapi type” of block-and-ashpyroclastic flows resulting from dome collapse; only rarely muchmore explosive eruptions occur. Over its eruptive history, the direc-tion of pyroclastic flows has varied widely, depending on the openingdirection of the summit crater. Since 1930, pyroclastic flows have oc-curred in all directions except to the north and east (Fig. 1). The factthat the volcano commonly (e.g., last 100 years) shows a particulareruption pattern had generated expectations that the next eruptionwould be similar. However, in late October and early November of2010, Merapi changed its usual eruptive style, producing its mostpowerful, explosive and voluminous eruption in the last 100 years(Gertisser et al., 2011; Surono et al., 2012). Since the 19th century,Merapi has erupted frequently with eruptions every 4–6 years on av-erage, and with most eruptions having explosivity indices ≤VEI 2, al-though moderate VEI 3 (1832, 1849, 1930, 1961) to large VEI 4 (1822,1872, 2010) eruptions have also occurred (Fig. 2).

Other papers in this special volume report the details of the mon-itoring data, unrest signals, deposit stratigraphy, impacts on the land-scape and people, and how the eruption was anticipated and itseffects were mitigated. Here we present the results of a petrologicaland geochemical study of a suite of samples from the 2010 and2006 events. We aim want to provide a petrological image of theplumbing system below Merapi and also identify the main processesthat led to the 2010 explosive eruption. We contrast these findingswith those of the 2006 eruption to contrast processes that occur dur-ing typically small (≤VEI 2) versus large (VEI 3 to 4) Merapi erup-tions. We also attempt to link our petrological observations andmodels with the unrest and monitoring data for both eruptions.

A number of studies have been carried out on the stratigraphy(Berthommier, 1990; Andreastuti et al., 2000; Newhall et al., 2000;Komorovski et al., this issue) and petrology of Merapi Volcano (delMarmol, 1989; Camus et al., 2000; Gertisser and Keller, 2003;Chadwick et al., 2007; Deegan et al., 2011; Borisova et al., this issue;Innocenti et al., this issue-a,b). Berthommier (1990) and Camuset al. (2000) divided the stratigraphy of Merapi into 4 stages, namelyPre-Merapi (0.67 Ma), Ancient Merapi (40,000 yr BP), Middle Merapi(6700 yr BP), and Recent Merapi. Andreastuti et al. (2000) divided

rological insights into the stology and Geothermal Resea

the recent history of Merapi into 3 cycles, >1960 yr BP, 1960 to780 yr BP, and 780 yr BP to the present. From about 3000 years agoto the present day, at least 93 significant eruptions took place.

2. Comparison between 2006 and 2010 eruptions

The 2010 Merapi eruption was much more explosive and larger involume than that of 2006. These two eruptions also showed differentprecursors and eruption styles. Preceding the 2006 eruption, domegrowth took place at the summit at an average rate of ~2–4 m3s−1

(Ratdomopurbo et al., this issue). The first pyroclastic flows were pro-duced on April 26th, 10 days after the dome first emerged at the sur-face. Following an extended period of dome growth and collapse, thepeak of the eruption took place on June 14th, 2006 when a largerdome collapse produced a 5 km altitude plume and pyroclasticflows that traveled up to 7 km from the summit. The total magmaticvolume erupted was on the order of 5×106 m3 (Ratdomopurbo et al.,this issue).

The 2010 eruption contrasted sharply with the 2006, and other‘typical’Merapi eruptions (Surono et al., 2012). Instead of an extendedperiod of dome growth, the 2010 eruption beganwith an explosion on26 October, with an ash plume that rose to 12 km altitude and pyro-clastic flows that reached 8 km from the summit. This initial explosionwas followed by smaller explosive eruptions on 29 and 31 October and1 November, and then by 4 days of extremely rapid dome growth(>25 m3s−1; Pallister et al., this issue) that led up to the climacticeruption on the night of 4–5 November. The 5 November eruptionproduced a plume that rose to 17 km altitude and pyroclastic flowsthat extended 16 km down Kali Gendol (Fig. 1); such an extent of py-roclastic flows was last seen at Merapi during the VEI 4 eruption of1872.

However, in both 2006 and 2010 eruptions, the volcano showedgeodetic and seismic precursors that began many months in advance.In terms of the precursors, the principal difference in 2010 was thatthe rate of deformation, the frequency of earthquakes, the magnitudeof the seismic energy release, and the quantity of gas emissions (espe-cially CO2) were all much greater. The total volume of 2010 magmaerupted is estimated at 0.02–0.05 km3 (20–50×106 m3), most ofwhich is juvenile material in pyroclastic flow deposits (Surono et al.,2012). Merapi eruptions typically produce basaltic–andesitic totrachyandesitic compositions (52–56 wt.% SiO2; e.g., Andreastuti etal., 2000; Gertisser and Keller, 2003). The magmas from the most ex-plosive eruptions at Merapi (e.g., 1872, 1930) tend to have lower

rage conditions, and magmatic processes that yielded the centennialrch (2013), http://dx.doi.org/10.1016/j.jvolgeores.2012.12.025


Kali Gendol5 km

Fig. 1.Map of the southwest flank of Merapi with sampling locations of the 2010 pyroclastic material marked as black pushpins. The pyroclastic flows produced during the 2006 and2010 eruptions descended the south flank of the volcano along the Gendol river (labeled “Kali Gendol”). Pyroclastic flows of November 5th, 2010 traveled as far as 13 km from thesummit, destroying or strongly damaging the villages of Kaliadem, Kinahrejo, Kepuhrajo, and Bronggang. (For a color version of this figure, see the web version of this article.)

3F. Costa et al. / Journal of Volcanology and Geothermal Research xxx (2013) xxx–xxx

SiO2 contents than the magmas produced during typical eruptions(e.g., 2006; Table 2, and Fig. 2). However, this doesn't appear to bethe case for the 2010 eruption. Differences in explosivity have gener-ally been attributed to variations in magma types and/or abundanceof volatiles (Wilson et al., 1980; Neri et al., 1998; Papale et al., 1998;Mangan et al., 2004) and whether this is also the case for the 2010 isone of the topics we pursue in this manuscript. Our 2010 juvenilesamples have relatively uniform bulk compositions with about1 wt.% variation in SiO2 (54.9–55.8 wt.%), and overlap with juvenileblocks from the 2006 eruption at the high-SiO2 end of the range(55.4–55.9 wt.%). Major-element compositions of the 2006 and 2010magma are thus virtually indistinguishable (Fig. 2).

3. Strategy, methodology, and analytical conditions

Eleven juvenile samples from the 2010 and 3 juvenile samplesfrom the 2006 eruptions were collected and analyzed. Most samples

Please cite this article as: Costa, F., et al., Petrological insights into the sto2010 Merapi explosive..., Journal of Volcanology and Geothermal Resear

are pieces of breadcrusted bombs found in pyroclastic deposits, othersare simply denser blocks of the samematerial. The 2010 samples werecollected from the Gendol River and from areas near the village ofKinahrejo. All of the 2006 samples were collected from the GendolRiver (Table 1). The analyses of bulk-rock samples were done atWashington State University using standard X-ray Fluorescence(XRF) methods for major elements minor elements, and Induction-Coupled Plasma Mass Spectrographic methods (ICP-MS) for trace ele-ments (the latter not reported here).

Thin sections of rock samples were observed with an optical micro-scope before being analyzed with an electron microprobe. The samplesfor which we have obtained more detailed data are listed in Table 1.Backscattered electron images and reconnaissance quantitative analysesof the minerals showed that the Merapi 2006 and 2010 samples weretexturally and petrographically complex. Thus, we decided to make alarge number of X-ray maps of major elements (mainly Al, Ca, Mg, Fe,K) at various scales (ranging from a few tens of mm2 to up to 1.5 cm2)

rage conditions, and magmatic processes that yielded the centennialch (2013), http://dx.doi.org/10.1016/j.jvolgeores.2012.12.025


1961

VEI 3 (1930, 1961)VEI 2 or less

VEI 4 (1872)VEI 4 (2010)

VEI 1 (2006)

Basaltic Andesite AndesiteBasalt

medium-K

high-K

low-K

52 56 608584 5450

SiO2 (wt%)

1

2

3

0

K2O

(w

t%)

Fig. 2. Comparison of the whole-rock compositions (SiO2 vs. K2O) of the 2006 and 2010erupted material with that produced during other historic events of similar intensities(VEI). The black contoured and gray shaded areas indicate the compositional trends ofthe high-K series and medium-K series respectively, as defined by Gertisser and Keller(2003). Recent eruptive activity dominantly produced high-K basaltic andesites andandesites, except for the 1961 eruption. The more explosive eruptions (VEI 3: 1930,VEI 4: 1872) seem to have been generated by slightly more mafic magmas than theless explosive eruptions (VEI≤2: 1883, 1904–06, 1906, 1942, 1954, 1957, 1992,1994, 1998). However, this does not appear to be the case of the 2010 eruption. Notethat the primitive sample from 2006 is thought to be a non-juvenile block from oldereruptions (Andreastuti et al., 2007). (For interpretation of the references to color inthis figure, the reader is referred to the web version of this article.)


to guide our quantitative analyses (single points and line traverses). Intotal, 17 X-ray maps including 5 elements (total of 85 element distribu-tion maps), i.e., 6 X-ray maps of matrices, 5 X-ray maps of pyroxenes, 5X-ray maps of amphibole, and 1 X-ray map of apatite crystals wereperformed. Due to the large range of crystal sizes (phenocrysts>500 μm,microphenocrysts 100–500 μm, microlitesb100 μm measured in thelongest direction parallel to a crystal margin) and textural relations(resorbed, anhedral to euhedral), single point analyses were alsoperformed to characterize the full textural variety. We did electron mi-croprobe traverses (rim to rim or rim to core) in plagioclase, pyroxene,amphibole, and apatite. Only some representative analyses and BSEand X-ray images are reported here.

3.1. Quantitative phase and X-ray map analysis conditions

Backscattered electron images, X-ray distributionmaps, and quan-titative analyses of minerals and glass were obtained with a fieldemission gun electron microprobe (JEOL-JXA-8530F) at the NanyangTechnological University (Singapore) using wavelength dispersivespectrometers. An accelerating voltage of 15 kV, current of 15 nA,and spot size of about 1 μm was used for most mineral analyses. Forglass analyses the current was decreased to 10 nA, and spot sizeswere increased to 5 to 10 μm. To minimize the loss of alkalis, Naand K were always analyzed first. Counting times on peak positionswere varied according to the concentration; from 10 s for major ele-ments and up to 120 s for trace elements. Backgrounds on bothsides of the peak were measured for half of the peak time. X-raymaps were obtained at 50 nA with a dwell time of 10–100 ms per

Please cite this article as: Costa, F., et al., Petrological insights into the sto2010 Merapi explosive..., Journal of Volcanology and Geothermal Resea

point, and point spacing of about 1–5 μm. Distances between analysesalong compositional traverses varied between 1 μm and 10 μm. Stan-dards used for calibration were minerals from Astimex (albite, garnet,rutile, pyrite, olivine, sanidine, diopside, celestite, fluorite, biotite,rhodonite, and tugtupite). The calibration was checked against anin-house dacite glass standard analyzed by X-ray fluorescence. Preci-sions vary according to the concentration: major elements have 2 σprecisions of 0.5–1%; precisions for minor elements are 5–10%.

3.2. Mineral end-members, structural formulae, and symbols

Plagioclase (Plag) formulae were calculated according to Deer et al.(1992) with the following end members (in mol%): Anorthite (An)=100×Ca/(Ca+Na+K), Orthose (Or)=100×K/(K+Na+Ca), Albite(Ab)=100×Na/(Ca+Na+K). Pyroxene formulae were calculatedaccording toMorimoto (1989). Clinopyroxene (Cpx) and orthopyroxene(Opx) end-members (inmol%)were calculated as: wollastonite (Wo)=100×Ca/(Ca+Mg+Fe⁎), enstatite (En)=100×Mg/(Ca+Mg+Fe⁎),and ferrosilite (Fs)=100×Fe*/(Ca+Mg+Fe⁎), where Fe⁎ stands fortotal iron. Amphibole (Amph) structural formulae and names werecalculated according to Leake et al. (1997). Biotite (Bt) and apatite(Ap)were also analyzed. For the ferromagnesianmineralswe calculatedthe Fe+3 according to their structural formula and stoichiometricconstraints, from which we obtained the magnesium-number(Mg#)=100×Mg/(Mg+Fe+2) using Fe2+ only.

4. Textures, mineralogy, and petrology of the 2006 and 2010eruption products

4.1. 2006 samples

The juvenile blocks from the 2006 eruption have seriate texturesdominated by large Plag and Cpx phenocrysts (>500 μm) that insome cases form glomerocrysts (Table 1). Both minerals also occuras microphenocrysts (100–500 μm), and Plag is more abundantthan pyroxenes as microlites (b100 μm) in the groundmass (Figs. 3aand 4a). Amph, Opx, and Fe–Ti oxides occur as microphenocrysts,and apatites are found as microphenocrysts, microlites or inclusionsin Cpx (Fig. 3a). Amph is brown and typically has reaction rims(20–30 μm wide) made of pyroxene, Plag, and Fe–Ti oxides. One ofthe 2006 samples we have investigated in detail contains a small(2 mm diameter) rounded xenolith, which is fine grained and con-sists of Plag, Opx, Bt, Fe–Ti oxides, and sulfides (Fig. 3a).

Cpx is occasionally found as rims surrounding anhedral crystals ofOpx. Chemical zoning (most notably in Al, Ca, Mg, Fe) is common andranges from patchy or irregular to concentric (Fig. 5). The zoning alsoincludes numerous dissolution surfaces. The composition of most Cpxphenocrysts varies between En45Wo41 and En36Wo47, with Mg# be-tween 76 and 82 (Figs. 6 and 7; Table 3). The Al2O3 content rangesfrom 1.5 to 6.5 wt.%. Microlites and microphenocrysts have more re-stricted compositions and notably lower Al2O3 and Wo component(Fig. 7). Some Cpx crystals have an abrupt change at the rim (outertens of μm), where they become richer in Al2O3 and Wo component(Fig. 5a, b). Other crystals display a discontinuous high-Al2O3 andhigh-Wo zone 20–30 μm across (Fig. 5c, d). The complexity andvariety of the zoning patterns indicate a dynamic and open magmaticsystem where crystals grew and dissolved multiple times due to tem-perature or compositional variations. Microlites overlap in composi-tion with phenocrysts but tend to be at the high-Mg# end of therange.

Opx crystals are typically anhedral and occur in the interiors of Cpx(Fig. 8) forming glomerocrysts, although some crystals are euhedraland occur as microphenocrysts or microlites in the matrix (Fig. 3a).They range in composition from about En72Wo2.7 to En69Wo3.5, andhave Mg# ~71 to 75 (Fig. 8; Table 3). The Opx microlites tend tohave the highest Wo contents (Fig. 8c). Opx crystals from the small



enclave

cpx

pl

pl

Caapa

ox

ox

Mg

enclave

cpx

opx

ox

ox

2006 - scale bar is 500 µm

a)

cpx

Ca

opx

cpx

pl

pl

apa

Mg

Oct 2010 - scale bar is 500 µm

cpx

ox

hb

opx

cpx ox

hb

b)

Ca

pl

pl

ox

Mg

Nov 2010 - scale bar is 500 µm

hb

cpx

apa

ox

hb

cpxopx

c)

Fig. 3. X-ray maps of Ca (top) and Mg (bottom) in samples from (a) 2006, (b) October 2010, and (c) November 2010 (scale bar is 500 μm). The main mineral are labeled: pl – pla-gioclase, cpx – clinopyroxene, hb – hornblende, opx – orthopyroxene, apa – apatite, ox – oxide. Plag phenocrysts display a wide variety in textures and zoning. Plag microlites havehigher Ca-contents in the 2010 samples compared to 2006. Opx is present as individual phenocrysts or in association with cpx (often within the cores, sometimes along the rims).Apatite occurs as mineral inclusions within cpx or in the matrix. White arrows highlight fragments of broken phenocrysts, which are frequent in the 2010 eruptive products. A small(2 mm diameter) rounded xenolith was found as a xenolith in the 2006 sample (a; present in the right half of the 2006 maps). It is fine grained and consists of plag, pyroxene,biotite, Fe–Ti oxides, and sulfides. Warmer colors mean higher concentrations.


xenolith are significantly different than those in the juvenile material,beingmuch smaller andmore Fe-rich (Fig. 8b, d; Table 3) than those ofthe juvenile material (Mg# between 50 and 60). A few of these crys-tals can also be found in the host where they have a pronounced reac-tion rim.

Plag phenocrysts compositions range from about An90 to An40.Normal, reverse, and oscillatory zoning trends are observed, as wellas dissolution surfaces, sieved textures, and apatite, pyroxene, andFe–Ti oxide inclusions (Figs. 3a, 9a). Despite the compositional andtextural varieties, most Plag phenocryst have narrow compositionalranges at ~An85–90, ~An70 and ~An40 (Fig. 9a), with most rims at~An40 (Figs. 4a, d; 9a, and 10). Microphenocrysts and microlites ofPlag also show a large compositional range but tend to be more Na-and K-rich, ranging from about An50Or05 to An20Or20 in somemicrolites (Fig. 10), and partially overlapping in composition withphenocryst rims (Figs. 4d and 10).

Amphiboles (magnesiohastingites; classification and structuralformula according to Leake et al., 1997) occur mainly asmicrophenocrysts and show reaction rims of a few tens of μm inwidth made of Plag, Opx, and Fe–Ti oxides (Fig. 11). Their Al2O3

contents vary from 10 to 14 wt.%, and Mg# varies from 64 to 67(Figs. 11b and 12). Fluorine contents vary between b0.05 and0.40 wt.%, and Cl ranges from about 0.01 to 0.12 wt.% (Table 4).Some amphiboles are zoned with higher Al2O3 and lower Mg# inthe cores than the rims.


Apatites in the matrix glass and as inclusions in Plag or pyroxeneswere analyzed. Fluorine varies between 2.0 and 4.4 wt.%, Cl between0.6 and 0.9 wt.%, and SO3 has the largest variability between 0.06 and0.70 wt.% (Table 5).

The matrix glass is rhyolitic, with an average of 71 wt.% SiO2 and6 wt.% K2O when renormalized to 100% anhydrous (Table 6). Originaltotals are between 98 and 99 wt.%. The F contents are low(b0.1 wt.%), whereas Cl is around 0.2 wt.%.

Ferromagnesian minerals in the xenolith have higher-Fe contentsthan those in the juvenile material. As mentioned above, Opx is re-versely zoned with Mg# about 50 in the core, increasing to about 60at the rim to (Fig. 8; Table 3). Plag is also zoned over a limitedrange, with most compositions around An40. Biotite has a Mg# of70–72 (Table 5) with about 1.5 wt.% F, and b0.05 wt.% Cl.

4.2. 2010 samples

Samples of breadcrusted bombs from the October and November2010 eruptions have similar textures and mineral assemblages tothose of the 2006 eruption. They also have seriate textures withlarge individual Plag and Cpx crystals and glomerocrysts (Table 1;Fig. 3b, c). Microphenocrysts consist of Plag, two pyroxenes andAmph. The groundmass contains Plag, pyroxenes, apatite, and mag-netite microlites and little interstitial glass (Fig. 4b, c). The November2010 samples have smaller pools of interstitial glass than the October



0

2

4

16

14

12

10

8

6Fre

quen

cy

An (mol%)40 70 10020 50 800 30 6010 90

a - 2006 b - Oct - 2010

c - Nov - 2010

Oct - 2010Nov - 2010

2006

Oct - 2010Nov - 2010

2006

pheno rims

microlites

0

20

40

100

80

60

Cum

ulat

ive

%

d

Fig. 4. BSE images of groundmass and matrix glass of samples from (a) 2006, (b) October 2010, and (c) November 2010 (scale bar is 100 μm). Note the greater vesicularity of theNovember 2010 sample, and the presence of microlites with bright (high-An) cores in the October and November 2010 samples, but not in the 2006 sample. A histogram of themicrolite compositions (d) shows that their An-content progressively increases from 2006, to October 2010, to November 2010 (n=28, 34, 27 respectively). Phenocryst rim com-positions (outer 10 μm), shown as curves of cumulative percentages (n=17, 26, 48 for the 2006, October 2010 and November 2010 samples respectively), overlap with themicrolite compositions. The An-contents of microlites and phenocryst rims reflect the composition, temperature, and volatile content of the interstitial liquid, rather than that ofthe bulk magma, and are therefore indicative of ascent conditions. Microlites and phenocryst rims in the 2006 sample have extremely low-An and high-Or which are indicativeof late-stage crystallization accompanying degassing and slow extrusion, as seen in lava dome-like extrusions (Hammer et al., 2000), whereas, the disappearance of the Or com-ponent and the increase in An-content suggest more rapid ascent and little to no time available for degassing-induced crystallization.


2010 or the 2006 samples, due to the presence of numerous vesicles.Opx is commonly resorbed and surrounded by Cpx in a similarmanner to that in the 2006 samples, or in some cases, resorption iseven more pronounced. In the November 2010 samples we alsofound a few crystals of unreacted biotite in the groundmass and tworesorbed olivine cores within Opx crystals, which are in turn includedin Cpx (referred to here as Opx–Cpx glomerocrysts; Fig. 13e). Amphdoes not show reaction rims (Fig. 14). The 2010 samples have a sig-nificant proportion of broken crystals of Plag and Cpx. These are easilyrecognized by the truncation of the zoning patterns visible on X-raymaps (Fig. 3b, c), a feature not seen in the 2006 samples.

Clinopyroxene phenocrysts in the 2010 samples, like their coun-terparts in the 2006 samples, are complexly zoned, showing oscillato-ry, normal, and reverse zoning and multiple dissolution surfaces(Fig. 13). Cpx crystals also contain small inclusions of glass, apatite,Plag, and Fe–Ti oxides. Most compositions in October 2010 samplesvary between En45Wo42 and En39Wo46, with Mg# between 78 and82 (Fig. 6, Table 3). Most crystals have Al2O3 contents around 2 wt.%but their concentrations increase at the rims, reaching up to 6 wt.%in some crystals (Figs. 7 and 13). The Cpx of the November 2010 sam-ples reach higher Wo contents (En35Wo49) and Al2O3 contents (up to8 wt.%) at their rims than those of the October 2010 samples. Manyhigh Al2O3 and Wo crystals are found surrounding resorbed Opx orCpx with lower Al2O3 and Wo (Fig. 3b, c). Moreover, we find someCpx phenocrysts or microphenocrysts in the November 2010 samplesthat are entirely high-Al2O3 and high-Wo, which is not the case forthe 2006 or October 2010 samples (Fig. 7). Cpx microlites overlap in


composition with phenocrysts, and tend to have higher Al2O3 andWo contents, and lower Mg# than the 2006 microlites (Fig. 7). Asdiscussed later, we think that these textural and compositional fea-tures of the high Al2O3 and highWo Cpx crystals reflect a greater pro-portion of high temperature and deep mafic magma in the Novembersamples.

Opx occur as euhedral microphenocrysts, or as anhedral resorbedcrystals in a reaction relation with Cpx (Fig. 3b, c). In one case,resorbed olivine is included in Opx (Fig. 13e). Compositions of theNovember samples vary between En72Wo2.2 and En68.5Wo2.7, andMg# between 77 and 67 (Fig. 8c, e; Table 3). Opx compositions inthe October 2010 and 2006 samples overlap, whereas those of theNovember 2010 are more variable and extend to lower Mg#(Fig. 8c). However, in general they show a more restricted range inWo component (Fig. 8c). Opx microlites tend to have higher Wo con-tent than the phenocrysts in all samples (Fig. 8c). All Opx crystals inthe 2006 and 2010 samples have much higher Mg# and lower Al2O3

contents than those found in the 2006 xenolith and in the fewresorbed Fe-rich crystals found in the November 2010 samples (e.g.,‘Fe-rich Opx’ in Fig. 8b, d). Thus the xenolith and the isolatedFe-rich crystals probably originated from more evolved and oxidizedmelts, which are not known to have been erupted during the recenthistory of Merapi.

Plagioclase zoning and textures in the October and November 2010samples are variable and include sieve textures, and normal, reverse,and oscillatory zoning in An content. A similar variety to that of the2006 samples is found, with phenocryst compositions varying between



Table 1Sample localities with main field and petrographic observations.

Label Eruptionmonth–year

Field notes Samplelocation

Phenocrysts(>0.5 mm)

Microphenocryst(0.1–0.5 mm)

Matrix/microlites(b0.1 mm)

Texture Other observations Analyses

10_11_21_1A Nov-2010 Dark bread-crusted1-m block from 5Nov awan panas,near terminus150 m N. of bridgeacross Kali Gendol

S 0740.548′ E110 28.095′364 melevation

Plag (25.3%),Amph (0.4%),Cpx (6%)

Opx, Cpx,Amph, Plag

Matrix (68%). Plag,Cpx, glass. Sampleis heterogeneouswith lighter anddarker matrixes.Darker matrix hasa lot more Fe–Tioxides.

Seriate with someglass. Large rangeof crystal sizes.Glomerocrysts ofCpx+Fe–Tioxides+Plag+Opx

Two types of Amph,with reacted rimsand large, and withno reaction rims amicrophenos. Opxcrystals can beresorbed andincluded in Cpx

XRF

10_11_21_1B Nov-2010 Same locality. Hotblock from upper20 cm of awanpanas. Less vesicularvariant of sample A

S 0740.548′ E110 28.095′364 melevation

Cpx, Plag,Amph(pronouncedzoning)

Opx, Cpx,Amph, Fe–Tioxides, Plag(one crystalof phlogopiteclose toglomerocryst)

Plag, Cpx, Amph,Fe–Ti oxides, glass

Seriate with notmuch glass. Largerange of crystalsizes. Glomerocrystsof Cpx+Fe–Tioxides+Plag

Amph occurs alsoas microlites ormicrophenoswithout reactionrims. Opx crystalscan be resorbedand includedin Cpx

XRF

10_11_21_2A Nov-2010 Breadcrusted blockfrom awan panas atKopeng, (dominant,likely juvenilecomponent in awanpanas)

S 07 36.504,E 11027.278,822 melevation

Plag (20.1%),Amph (0.7%),Cpx (5.5%)

Opx, Cpx,Amph, Plag

Matrix (70%). Plag,Cpx, Amph, Fe–Tioxides, glass

Seriate with notmuch glass. Largerange of crystal sizes.Glomerocrysts ofCpx+Fe–Tioxides+Plag


BSE,X-Raymaps,eprobe,XRF

10_11_21_2B Nov-2010 Same locality as 2-A.Prismatic block inawan panas. Denservariant of 2-A; stillhot. Possibly fromdome that formed1-4 Nov. anddestroyed byexplosive eruptionon 5 Nov

S 07 36.504,E 11027.278,822 melevation

Cpx, Plag,Amph

Opx, Cpx,Amph, Plag

Plag, Cpx, Amph,glass. Samplecontains a smallxenolith of oldersomewhatweathered volcanicrock similar to thehost

Seriate with notmuch glass. Largerange of crystal sizes.Glomerocrysts ofCpx+Fe–Tioxides+Plag



1010523 Nov-2010 Reddish brown,breadcrusted blockfrom awan panas atBronggang, possiblejuvenile, 15 km


Opx, Cpx,Amph, Plag

Plag, Cpx, Amph,Fe–Ti oxides, glass.2 type ofhornblende,greenish brown andbrown (need tocheck thecomposition)



XRF

10111123 Nov-2010 Dark gray,breadcrusted blockfrom awan panas atBronggang, at theover bank, possiblejuvenile, 15 km, endof pf, 4 November


Opx, Cpx,Amph, Plag.A lot moreAmph thanprevioussamples

Plag, Cpx, Amph,Fe–Ti oxides, glass

Seriate with notmuch glass. Largerange of crystalsizes. Glomerocrystsof Cpx+Fe–Tioxides+Plag+Amph

Amph occurs alsoas microlites ormicrophenoswithout reactionrims. Opx crystalscan be resorbedand included in Cpx.Cpx rimmed byhornblende Amph


10111121 Nov-2010 Dark gray,breadcrusted blockfrom awan panas atBronggang, at theover bank, possiblejuvenile, 15 km, endof pf, 12 km fromsummit, 3 November

Plag, Cpx,Amph (thinblack rim),Plag

Cpx, Opx,Amph, Plag

Cpx, Amph, Plag,Fe–Ti oxides

Seriate,Glomerocrysts ofPlag, Cpx, Amph,Fe–Ti oxides,moderate to highvesicularity

Opx with rims ofCpx. Amph withreaction rims

XRF

10111123 x Nov-2010 Same sample from10111123, xenolith,from awan panas atBronggang, at theover bank, possiblejuvenile, 15 km, endof pf, 4 November

Cpx, Plag,Amph w/reactionrims

Opx, Cpx,Amph, Plag

Plag, Cpx, Amph,glass. Xenolithsof volcanic rockssimilar to hostbut w/weathering


XRF

11051604-B Nov-2010 Roadcut enroutedown to MbahMarijian house site.Lithic-rich sandyPF/surge of 4–5 Novthat overlies 26 Octdeposit. Juvenile.

S07 34 54.9,E 110 2635.9


Opx, Cpx,Amph, Plag

Plag, Cpx, Amph Seriate texture,glomeroporphyriticpx, (Cpx, Opx, Plag,Amph), lowvesicularity

Opx with rims ofCpx or Amph.Amph withbreakdownreaction rims

XRF

(continued on next page)


Please cite this article as: Costa, F., et al., Petrological insights into the storage conditions, and magmatic processes that yielded the centennial2010 Merapi explosive..., Journal of Volcanology and Geothermal Research (2013), http://dx.doi.org/10.1016/j.jvolgeores.2012.12.025


Table 1 (continued)

Label Eruptionmonth–year

Field notes Samplelocation

Phenocrysts(>0.5 mm)

Microphenocryst(0.1–0.5 mm)

Matrix/microlites(b0.1 mm)

Texture Other observations Analyses

11051605_AD Nov-2010 Same locality. Darkcolored dense rockfragment from thesame upper part ofthe 26 Oct. PF

S 07 3458.0, E 11026 40.4

Plag, Cpx,Opx

Plag, Cpx,Opx, Fe–Tioxides

Plag, Cpx, Fe–Tioxides

Seriate texture,glomeroporphyriticpx, (Cpx, Opx, Plag,Amph), sub ophitic,low vesicularity

XRF

11051605_A/B

Nov-2010 Same locality. Tansub-pumice. From5 cmblock at contactbetween upper andlower units within26 Oct. PF

S 07 3458.0, E 11026 40.4

Plag, Cpx,Opx

Plag, Cpx,Opx, Fe–Tioxides


XRF

110519_3 Nov-2010 Kali Opak area.Prismatic block(p-Mag normal=juvenile) fromNov. 4–5 PF

S 07 3525.8, E 11026 38.5

Plag, Cpx,Opx

Plag, Cpx,Opx


Seriate texture,glomeroporphyriticpx, (Cpx, Opx, Plag,Amph), sub ophitic

XRF

1105605-AL Oct-2010 Main PF sectionin K. Gendol belowKinarehjo. Upper(Atas=A) unit of26 October PF, lightcolored vesicularlapilli. Juvenile.

S 07 3458.0, E 11026 40.4

Plag, Cpx,Opx, Amph

Plag, Cpx,Opx, Amph

Plag, Cpx, Amph Seriate texture,glomeroporphyriticpx, (Cpx, Opx), Plag,low vesicularity,Amph (pale brownand brown)


11051605-C Oct-2010 Same locality. Darkcolored dense rockfragment from thesame upper part ofthe 26 Oct. PF.Juvenile.

S 07 3458.0, E 11026 40.4

Plag, Cpx,Opx, Amph

Plag, Cpx,Opx, Amph

Plag, Cpx, Amph Seriate texture,glomeroporphyriticpx, (Cpx, Opx, Plag,Amph), lowvesicularity

XRF

1406L Jun-2006 Gendol initial, lightsamp. Pmag hot

Cpx, Plag,rare Amphw/reactionrims

Opx, Cpx,Plag

Plag, Cpx, Fe–Tioxides, glass.

Seriate with notmuch glass.Large range ofcrystal sizes

Opx crystals canbe resorbed andincluded in Cpx.


1406D Jun-2006 Gendol initial, dark.Non-juvenile

Cpx, Plag, Opx, Cpx,Plag

Plag, Cpx, Fe–Tioxides, glass

Porphyric vesiculatedwith some glass,vesicular. Large rangeof crystal sizes


XRF

1506_1 Jun-2006 Gendol. PMag hot.Nose of pf.


Opx, Cpx,Plag

Plag, Cpx, glass.Fine grainedxenolithw/Opx+Pl

Seriate with someglass, vesicular.Large range of crystalsizes. Glomerocrystof Cpx–Opx–Amph–Plag–Fe–Ti oxides



1506_L2 Jun-2006 Gendol. P-Mag hot Cpx, Plag,Amph w/reactionrims

Opx, Cpx,Plag

Plag, Cpx, Fe–Tioxides, glass.

Seriate with someglass, vesicular.Large range ofcrystal sizes


XRF

Mineral symbols: Plag=plagioclase, Amph=amphibole, Opx=orthopyroxene, Cpx=clinopyroxene. PMg = measurement of orientation of magnetic field. Modal abundances involume % are vesicle-free and based on counting about 1500 points per section.


about An90 and An40. However, most Plag phenocryst cluster at ~An90,~An70, and ~An40 (Fig. 9b, c). Most phenocryst rims in the October2010 samples are about An40–50, whereas those of the November 2010samples are more calcic at about An60 (Figs. 4d, 9c, and 10). Microlitesalso tend to be more calcic in the 2010 samples compared to the 2006samples; the October 2010 samples have microlites that range fromabout An60 to An40, whereas those in the November 2010 samplesrange from An75 to An50 (Figs. 4d, 10). Thus, a distinctive feature ofthe 2010 rocks is the higher An content of microlites, and an increasein An content from the October to the November eruption. This changetowardsmore calcic plagioclase is mirrored by the higher Al2O3 andWocomponent of Cpx and we think it reflects the effect of the influx of agreater amount of deep, hot and little crystallized magma during theNovember 2010 eruption (see Section 6).

Amph show a significant variety of textures and zoning patterns,particularly in the November 2010 samples (Fig. 14). It mainly occursas unreacted microphenocrysts, although the shapes and crystalmargins range from euhedral to anhedral and in one case resorbed.Crystals can be virtually unzoned (Fig. 14a) or show particularly


pronounced alternations in composition (Fig. 14c). Some Amphcrystals have very large reaction rims of Plag and pyroxenes,but they mostly occur as unreacted microphenocrysts. All aremagnesiohastingites with Al2O3 of ~10–15 wt.% and Mg# of 54–69and cluster at 63–67 (Table 4; Fig. 12). Several crystals are zonedwith alternating bands of Al, Fe, Mg, and sometimes K. Crystal corescan be very complex with patchy zoning and low Mg# compositionsthat fall into the alkaline field. Core to rim traverses show that Mg#increases toward rims, with newly grown rims approaching the aver-age composition of the other crystals (Fig. 14). The 2010 sampleshave amphiboles with the highest Al2O3 contents (Fig. 12). Such vari-ety of amphibole compositions and zoning patterns in the 2010 sam-ples indicate open system processes involving mixing of differentmelts and with a range of crystallization conditions as we will seebelow.

Apatite is typically euhedral and occurs in the matrix or as inclu-sions in Cpx, Plag, Amph, and Fe–Ti oxides. Apatite inclusions in sili-cate minerals are typically found close to dissolution zones and mayreflect the presence of boundary layer melts enriched in incompatible



Table 2VEI and bulk-rock compositions of the 2006, 2010, and some older historical Merapi eruptions.

Sample 11211A 11211B 11212A 11212B 10111123 10110511 MER061406-L MER061406-D MER061506-L-2 MER061506-L MER1961-D MER1942-D MER1872-D-1

Eruption year 2010 2010 2010 2010 2010 2010 2006 2006 2006 2006 1961 1942 1872

VEI 4 4 4 4 4 4 1 1 1 1 3 2 4

wt.%SiO2 55.8 55.4 55.5 54.9 54.9 55.0 55.7 52.4 55.9 55.4 55.7 52.3 52.5Al2O3 19.25 19.16 19.14 19.05 19.26 19.19 19.00 18.60 19.17 19.08 18.42 19.11 18.86TiO2 0.74 0.73 0.74 0.77 0.78 0.78 0.75 0.86 0.74 0.77 0.93 0.89 0.88FeOa 7.79 7.66 7.71 7.96 8.37 8.32 7.50 9.35 7.45 7.77 8.63 9.21 9.51MgO 2.33 2.29 2.36 2.55 2.58 2.60 2.37 3.41 2.36 2.48 2.90 3.25 3.47MnO 0.20 0.20 0.20 0.20 0.200 0.202 0.20 0.21 0.20 0.20 0.19 0.20 0.21CaO 8.27 8.25 8.30 8.62 8.67 8.71 8.25 9.46 8.23 8.33 8.03 9.43 9.55Na2O 3.90 3.89 3.90 3.80 3.74 3.73 3.48 3.04 3.50 3.47 3.46 3.06 3.05K2O 2.16 2.13 2.13 2.06 2.04 2.04 2.15 1.98 2.17 2.14 1.54 1.95 1.98P2O5 0.32 0.32 0.33 0.32 0.32 0.32 0.37 0.36 0.37 0.38 0.38 0.37 0.37Sum 100.8 100.0 100.3 100.2 100.9 100.8 99.8 99.7 100.0 100.0 100.2 99.7 100.4

ppmNi 3 4 5 5 4 5 3 3 3 3 3Cr 5 5 5 5 4 8 b5 b5 b5 b5 b5Sc 13 11 12 14 13 14V 161 156 162 179 177 176 130 157 112 169 109Ba 524 519 522 505 492 506 365 494 559 503 559Rb 52 51 50 47 47 47 35 46 51 42 52Sr 583 582 580 576 568 568 538 637 602 607 603Zr 108 108 110 106 102 102 128 87 110 87 110Y 26 26 26 26 25 27 22 21 22 21 22Nb 4.9 5.0 5.3 5.5 4.4 5.3 4 2 4 2 4Ga 19 19 19 21 20 19 15 15 15 16 13Cu 26 24 24 27 54 30 11 44 15 36 18Zn 94 92 96 96 95 96 58 63 61 73 59Pb 20 24 19 18 18 18La 21 20 16 18 15 18 14 13 10 12 11Ce 41 41 41 42 39 39 33 34 36 34 37Th 5 7 7 6 7 6 b4 b4 b4 b4 b4Nd 22 20 20 19 20 20U 2 1 3 0 1 1 b4 b4 b4 b4 b4

a Total iron. Analyses were done by XRF. VEI=Volcanic Explosivity Index.

9F.Costa

etal./

JournalofVolcanology

andGeotherm

alResearchxxx

(2013)xxx–xxx

Pleasecite

thisarticle

as:Costa,F.,etal.,Petrologicalinsightsinto

thestorage

conditions,andmagm

aticprocesses

thatyieldedthe

centennial2010

Merapiexplosive...,Journalof

Volcanology

andGeotherm

alResearch(2013),http://dx.doi.org/10.1016/j.jvolgeores.2012.12.025


Ca Mg

Al

Distance (µm)0 9070605040302010 80

Wo

En

(wt%

)

02

864

10

40455055

35

(mol

%)

Al2O3(w

t%)

02

864

10

40455055

35

(mol

%) Wo

En

Al2O3

Distance (µm)0 12010080604020

ab

2006 - Px3

2006 - Px-trav

a

bMg

a

b

Caa

b

Al

a

b

c) d)

a) b)

Fig. 5. 2006 pyroxene maps and zoning profiles. (a, c) BSE images, Ca, Al, and Mg X-ray maps, and (b, d) rim to core major-element zoning profiles of two cpx phenocrysts from2006. Wo and En stand for wollastonite and enstatite respectively. Note the high-Al2O3, high-Wo zones present at the crystal rims (bright zones on BSE images).


elements. The F contents of apatites vary between about 2.1 and3.9 wt.%, Cl between 0.4 and 1.4 wt.%, and SO3 between b0.1 and0.96 wt.% (Table 5). There are no systematic changes betweenapatites in different textural positions. The range of volatile contentis very large and comparable to that of the apatites in the 2006samples. X-ray maps of S, F, and Cl in two apatites show that theyare strongly zoned, with S content typically decreasing from core torim (Fig. 15). Such decrease in concentrations could be a record ofgrowth during decompression and degassing. However, the strongzoning also illustrates the potential problems of using only single in-dividual analyses of apatite to understand the volatile contents of themagma.

One euhedral biotite was found in the November 2010 samples. Ithas high TiO2 (about 5 wt.%), low F and Cl (both less b0.5 wt.%), andlow Mg# (about 62; Table 5). This biotite is slightly different andmore evolved than those found in the xenolith of the 2006 sample(see above); however, it is probably from the same source, as wasthe Fe-rich Opx found in the same sample. We suggest that bothFe-rich minerals originated from an older and more evolved magmawithin the Merapi reservoir system.

Two resorbed olivine crystals were found inside an Opx (Fig. 13e).They have slightly different compositions and the most Mg-rich crystalis Fo70, with asmuch as 1.2 wt.%MnO and about 0.1 wt.% CaO (Table 5).

Optical reflected light microscopy showed that pyrrhotite andmagnetite (sensu lato) are the main non-silicate minerals and no il-menite was identified. A few Fe–Ti oxides were analyzed in the Cpxand Opx glomerocrysts only. They are magnetites with about 9 wt.%TiO2, 2.9 wt.% Al2O3 and 2.7 wt.% MnO (Table 5).


Matrix glass was analyzed in the October 2010 sample only becauseglass pools from the November 2010 samples were too small to obtainreliable analyses using a defocused beam (Fig. 4c). Recalculated compo-sitions to 100% anhydrous (original total between 97 and 98 wt.%) arerhyolitic, with about 70 wt.% SiO2 and 5.7 wt.% K2O. The F contentsare low (b0.1 wt.%), and Cl is about 0.2 wt.%. In general these composi-tions are less evolved than the matrix glass of the 2006 sample(Table 6).

Glass inclusions are common in Cpx, Plag, and Amph but most ofthem show clear signs of recrystallization or water loss. We analyzedmore than 50 inclusions in the November 2010 samples (see Table 6for representative analyses). Compositions recalculated to 100 wt.%anhydrous show a significant spread in major element composition,varying from andesite to rhyolite, with SiO2 ranging from 62 to71 wt.% and K2O from 5.7 to 6.3 wt.%. Original totals range from 98to 94 wt.%. The F contents are low (b0.1 wt.%), Cl is between 0.3and 0.4 wt.%, and SO3 is highly variable from b0.05 to 0.45 wt.%.The main difference between the glass inclusions and the interstitialglass is that the former tend to have higher K2O and lower CaO for agiven SiO2 content. This might be due to post-entrapment crystalliza-tion of their host minerals (Cpx, Plag, and Amph).

4.3. Summary of mineral compositions and textures

Two Cpx populations or end-members can be distinguished and arecommon to the 2006 and 2010 samples: (1) low-Al Cpx occur asphenocrysts, microphenocrysts, and more frequently as glomerocrystswith Opx. (2) high-Al Cpx occur as discrete zones within low-Al



Oct 2010

Nov 2010

2006

40 45 50 55Wo (mol%)

45 50 55Wo (mol%)

30

32

34

36

48

46

44

42

40

38

En

(mol

%)

32

34

36

48

46

44

42

40

38E

n (m

ol%

)

a) Cpx-Opx intergrowths

b) Microlites

c) Phenocrystsd) Microphenocrysts

Fig. 6. Plots of Wo versus En contents in Cpx from aggregates with Opx (a), microlites (b), microphenocrysts (c), and phenocrysts (d). The compositions for the 2006, October 2010,and November 2010 samples overlap. Cpx in aggregates show similarly low Wo- and high En-contents. On the other hand, microlites, microphenocrysts, and phenocrysts show alarge range in composition. The highest Wo and lowest En of phenocrysts and microlites gradually increase from 2006, October 2010 and November 2010. En- and Wo-contents inCpx microphenocrysts show some scatter, which could be due to kinetic effects.


phenocrysts, more frequently as rims around low-Al Cpx or Opx,but also as individual phenocrysts within the 2010 magma only.Opx do not show much variety. They have similar compositionswhether occurring as individual phenocrysts, microphenocrysts, oras glomerocrysts with Cpx.

Plagioclase phenocrysts and microphenocrysts are rarely homoge-neous. Three An-compositions dominate within the population ofcrystals. (1) An90 zones are associated with anhedral crystal cores(skeletal or perhaps resorbed) of phenocrysts and microphenocrysts.(2) An40–50 commonly mantle the An90 zones as broad rims or formentire phenocrysts or microphenocrysts. (3) Intermediate An60–70

compositions frequently occur as discrete zones within An40–50

phenocrysts, and constitute most of the sieve-textured parts of thesecrystals.

Three Amph populations can be distinguished on the basis of theirAl-content andMg#. (1) high-Al and relatively high-Mg# compositionsare common as individual crystals in the November 2010 magma.(2) low-Al, higher-Mg# compositions are common within 2006,October 2010 and November 2010 microphenocrysts. (3) A few crystalcores with particularly high Fe-contents (and very low-Mg#) werefound in the November 2010magma. Zoning patternswithin individualcrystals show transitions from one compositional population to theother, often ranging from the high-Al to the low-Al end-member.

Finally, rare Fe-rich minerals (Opx, Amph, and Bt) occur within axenolith in the 2006 samples and as individual crystals in the 2010samples (e.g., ‘Fe-rich Opx’ in Fig. 8). They suggest the presence of adistinct and more evolved magma within the Merapi system.

This large variety of textures, minerals, and chemical zoningpatterns of crystals shows that the 2006 and 2010 magmas had acomplex and protracted magmatic history. This history includedsignificant changes in chemical compositions and probably pressure,


temperature and volatile contents. Open system processes, such asmixing between magmas or assimilation of wall-rocks and cumulatesall probably played a role, as has been found at other subduction-zonerelated volcanoes (e.g., Dungan and Davidson, 2004; Streck et al.,2005; Andrews et al., 2008; Streck, 2008). Below we address insequence the magma storage and pre-eruptive conditions and pro-cesses that may be responsible for the main petrologic features ofthe rocks.

5. Magma storage conditions of the 2006 and 2010 magmas

To identify the conditions at which pre-eruptive processes tookplace we determined the ranges of pressures (P), temperatures (T),and volatile fugacities (e.g., fH2O, fO2

) that the Merapi magmas experi-enced as recorded by geothermometers and geobarometers that arebased on thermodynamic relations or empirical calibrations. Themain difficulty we faced was to identify robust criteria to determinewhich minerals and which compositions are in equilibrium and thuscan be used to reliably determine the P, T, fH2O and fO2

.

5.1. Determination of pre-eruptive pressures and depths of storage

We have used the new empirical calibration of the amphibolegeothermobarometer of Ridolfi and Renzulli (2012) to determine theequilibration pressure (with errors of about 10% relative). Accordingto this method, Merapi amphiboles appear to record a large rangeof pressures from about 300 to 900 MPa (Fig. 16), with two clustersat about 800 MPa and 300–450 MPa. Except for one crystal(Amph 1, 15 points; see Fig. 11), all the 2006 amphiboles show pres-sures at about 300–450 MPa; whereas, magmas from 2010 haveseveral crystals that record higher pressures of about 800 MPa



45 50 55Wo (mol%)

40 45 50 55Wo (mol%)

a) Cpx-Opx intergrowths

b) Microlites

d) Phenocrystsc) Microphenocrysts

Al 2

O3

(wt%

)

0

1

2

3

9

8

7

6

5

4

11

10

Al 2

O3

(wt%

)

1

2

3

9

8

7

6

5

4

11

10

0%

8-9% 15-17%

Oct 2010

Nov 2010

2006

110 MPa (3.5 wt%)

800 MPa (4 wt%)800 MPa (6 wt%)

200 MPa (4 wt%)300 MPa (6 wt%)

Fig. 7. Plots of Wo-content vs. Al2O3 in clinopyroxene (a) from aggregates with orthopyroxene, (b) microlites, (c) microphenocrysts, and (d) phenocrysts. The data overlaps for the2006, October 2010, and November 2010 samples. Clinopyroxenes in aggregates show similarly low Al2O3 and Wo-contents. Microlites, microphenocrysts, and phenocrysts on theother hand show a large range in compositions. Maximum Al2O3 and Wo-contents of the phenocrysts and microlites gradually increase from 2006, to October 2010, to November2010. Al2O3 and Wo-contents in clinopyroxene microphenocrysts show some scatter, which could be due to kinetic effects. Clinopyroxene compositions from the experimentalstudy of Iacono Marziano et al. (2008) on carbonate assimilation by basaltic magmas are shown for comparison in the dashed and gray-shaded areas. Percentages mentionednext to the circled areas indicate the wt.% calcite assimilated. The difference in Cpx compositions could be explained by approximately 10–15 wt.% more carbonate assimilationin November 2010 than in 2006. However, clinopyroxene compositions obtained from thermodynamic modeling using the MELTS algorithm (Ghiorso and Sack, 1995) are alsoshown for comparison, with red (110 MPa), green (200–300 MPa), and blue (800 MPa) lines. Al-content increases with an increase in pressure, and is only mildly affected by achange in water content in the melt (varied between 3.5 and 6 wt.%, as shown in parenthesis in the legend). (For interpretation of the references to color in this figure, the readeris referred to the web version of this article.)


(Fig. 14a–d). The presence of two main pressures suggests that thereare at least twomainmagma-storage zones at different depths belowMerapi.

These pressures can be translated into depths by using a densitystructure of the crust below Merapi. Tiede et al. (2005) deter-mined that rock densities below Merapi range between 2000 and2500 kg/m3. However, these values probably apply only to theshallower parts (b20 km depths), because, for example, the densityof compacted basaltic andesite rocks is about 2700 kg/m3. In addition,the likely presence of crystal cumulates at depth would increase thedensity to about 3000 kg/m3. Given the uncertainties we use amean density of 2800 kg/m3 for the conversion of pressure estimatesto depths. Accordingly, and assuming lithostatic pressure we obtaindepths for Amph growth in a shallower reservoir (400 MPa) atabout 14 (+/−1) km, and in a deeper reservoir (800 MPa) at about29 (+/−3) km. We expect that water degassing and magma crystal-lization also took place in the shallower parts of the system, whereamphibole is not stable given the strong pressure dependence of vol-atile solubility laws.

The picture described above becomes more complex if we con-sider the pressure changes calculated along traverses within singleamphibole crystals (Figs. 11; 14d, f). For example, in one 2006 amphi-bole, the calculated pressure decreases from the core to the rim, aswould be expected if the host magma migrated from a deeper to ashallower reservoir. But other amphiboles from the November 2010eruption show a calculated pressure that oscillates by more than ahundred MPa over a few micrometers (Fig. 14f). Such unrealistic


apparent pressure changes are probably the result of the kinetics ofamphibole growth, which were not keeping pace with the equilibri-um conditions required for the Ridolfi and Renzulli (2012)geobarometer. Despite these potential problems, we think that theclustering of Amph compositions favors two magma reservoirsbelow Merapi, located at about 13 and 28 km depths, and that moreof the 2010 magma originated from the deeper source than in 2006.

A similar range of depths including a multi reservoir storage sce-nario for Merapi magmas was proposed by other authors using petro-logical observations. Chadwick et al. (2007) and Chadwick (2008)argue for an interconnected network of magma bodies betweenabout 3 and 30 km, and Deegan et al. (2010) propose a compositeplumbing system for Merapi with multiple reservoirs at variousdepths between ~7 and 45 km. A significant range of crystallizationpressures have also been inferred for the recent Mt. St Helens erup-tions based on amphibole compositions and phase equilibria experi-mental results (e.g., Rutherford and Devine, 2008; Thornber et al.,2008). Other studies of Merapi based on geophysical data (seismicityand deformation) tend to find evidence for the shallow presence ofmagma between 2 and 8.5 km below the summit (Beauducel andCornet, 1999; Ratdomopurbo and Poupinet, 2000) during older erup-tions than 2000. We suggest that the differences between the petro-logical and geophysical methods can be explained if they arerecording different processes: our determinations are based on thecrystallization of Amph alone, although we also propose degassingand crystallization of an anhydrous mineral assemblage at lowerpressures. Magma depths obtained from geophysical methods are



Wo

(mol

%)

0

2

4

5

3

1

45 50 55 60 757065

En (mol%)

opx

cpx

2006-pxD

opx

cpxcpx

2006-clH

0.1

0.5

1.0

2.0

1.5

Mg#69 71 74 77757370 7672

0

1

2

3

5

4

Al 2

O3

(wt%

)

Mg#45 55 65 75706050

Oct 2010Nov 2010

2006

EnclaveFe-rich Opx (Nov 2010)

a)

b)

d)

c)

e)

2006-clB

opx

cpx

ox

68 69 70 71 72

En (mol%)

2.0

3.0

4.0

3.5

2.5

4.5

Fig. 8. Opx plots and SEM images. (a) BSE images of a typical Cpx-Opx aggregate (left), an Opx microphenocryst rimmed with Cpx (center), and an Opx phenocryst (right). Scale baris 100 μm. Plots of Wo vs. En (b, c) and Mg# vs. Al2O3 concentrations (d, e) in Opx from 2006, October 2010, November 2010 samples and a xenolith within the 2006 sample. Opxfrom the xenolith and a Fe-rich pyroxene in the November 2010 sample (‘Fe-rich Opx’) display extremely low Mg#, Wo- and En-contents and high Al2O3 concentrations comparedto the phenocrysts commonly found in the 2006, October 2010 and November 2010 samples. Enlarged plots of the phenocryst compositions (c, e) show significant overlap betweenthe sampled units and crystal types, although phenocrysts (squares) and microlites (circles) tend to have higher Wo-contents than the crystals in aggregates (diamonds).


derived from observations made during volcanic unrest, and thusmight be related to magma or fluid movement closer to the surfaceand an impending eruption, but not necessarily to the ‘static’magma storage. Also, until recently all of the monitoring of Merapi,with the possible exception of some InSAR, has been focused on theshallow system; i.e., the seismic and EDM networks are locatedclose to the volcano's summit and are therefore insensitive to deeppressurization/depressurization.

5.2. Determination of pre-eruptive temperature

The empirical calibration of the amphibole composition of Ridolfiand Renzulli (2012) is also a geothermometer (Fig. 16), and yieldscrystallization temperatures that range from about 900° to 1000 °C


(+/−about 25 °C). The largest temperature variation is found foramphiboles that record the lower pressures (300–450 MPa). Amphi-boles in the 2006 rocks are mainly at the high-temperature end of therange (950–1000 °C). Temperatures determined along amphibolecompositional profiles showmuch smaller changes than those relatedto pressure (Figs. 11 and 14). They either slightly decrease (for 2006sample; Fig. 11), stay more or less constant, or slightly increase (for2010 sample; Fig. 14). Rim temperatures for many amphiboles ofthe 2010 rocks cluster at about 950 °C.

We have also determined the temperature of coexisting pairs ofOpx and Cpx using the thermodynamic models implemented in theQUILF software (Andersen et al., 1993). We used pyroxenes thatwere touching and thus likely crystallized together; we excludedthe high-Al Cpx because these are typically found in a reaction



An

(mol

%)

40

100

20

80

0

60

16010 100 18060 120 2000 80 14040 220

b) Oct - 2010

Distance (µm)160 28080 200 3200 120 24040 360

An

(mol

%)

40

100

20

80

0

60

c) Nov - 2010

An

(mol

%)

40

100

20

80

0

60

160 28080 2000 120 24040 320

a) 2006

Fig. 9. Plag zoning profiles. Rim to core An zoning profiles in Plag phenocrysts from(a) 2006, (b) October 2010, and (c) November 2010. In all three samples, high-An(~An90) compositional plateaus are found in the cores, and low-An (~An50) plateausare found in the outer zones of the crystals. Rims cluster at An30–40 in the 2006 sample,but range from An40–60 to An40–70 in October 2010 and November 2010 samples,respectively.


relationship with Opx and thus were not in equilibrium. We obtainedin total 56 pairs and found overlapping temperatures for the 2006 and2010 pyroxenes: those from 2006 are 1010°+/−10 °C (error onindividual determination of 40 °C), those from October 2010 are1020°+/−40 °C (error on individual determination of 50 °C), andthose from November 2010 are 1010°+/−20 °C (error on individualdetermination of 50 °C). These temperatures are not very pressuresensitive (we used a pressure of 300 MPa for the T determinations)and they overlap with the higher end of the amphibole temperaturerange. These higher temperatures would suggest earlier initial crys-tallization of the pyroxenes.

Other geothermometers that could theoretically be used includethe empirical equations for pyroxene and liquid equilibria of Putirka(2008) and the hornblende-Plag equilibria of Holland and Blundy(1994). However, it is virtually impossible to determine the meltcomposition that was in equilibrium with individual pyroxenes

Phenocryst Rims

Ab

(mol%)

Microli

20

40

20 40 60

Fig. 10. Compositions of plagioclase microlites (circles) and phenocryst rims (outer 10 μm;nocryst rims in 2006 tend to have low An- and high Or-contents, similar to those measured in(Hammer et al., 2000), as indicated by the gray shaded area. This suggests that time was aAn-content of the microlites and phenocryst rims increases from 2006, to October 2010,rates were greater in 2010, with little to no degassing-induced crystallization occurred.


at the time of crystallization. Consequently and as a test, we usedseveral potential host melt compositions and applied the equilibriumrelations of Putirka (2008). We found that the andesitic to rhyoliticmelt inclusions found in the Merapi crystals are not mafic enoughto be in equilibrium with the Cpx; e.g., the criteria of the Mg/Fe equi-librium of Putirka (2008) are not met. However, a melt with a compo-sition equivalent to that of the bulk rock could be in equilibrium withthe high-Al Cpx and yielded temperatures of about 1050 °Cand pressures between 300 and 600 MPa (with large errors of~300 MPa). The Opx–liquid equilibria formulations of Putirka (2008)also show that the Fe–Mg equilibria conditions are only fulfilledwhen a melt with a composition similar to the bulk rock is used, andin this case apparent temperatures range from 1040° to 1070 °C andpressure from 100 to 600 MPa. This suggests that the high-Al Cpxcrystallized early, when the melt was still close to the bulk-rockcomposition.

To determine the pressure and temperature with the Holland andBlundy (1994) geothermobarometer, it is necessary to know whatPlag composition was in equilibrium with which amphibole. This isalso uncertain from petrographic criteria, because we could not findinclusions of one mineral in the other. Consequently, we used thepressure and temperature determined from the amphibole and theRidolfi and Renzulli (2012) relationships to calculate the compositionof the Plag that would theoretically be in equilibrium withthe amphiboles (Table 4). Using the reaction edenite+albite=richterite+anorthite from the Holland and Blundy (1994) thermody-namic formulation we found that only a relatively narrow range ofPlag can be in equilibrium (An64 to An72). Such compositions arefound in the Merapi rocks as a plateau in crystal traverses, as wellas in some of the microlites from the November 2010 samples.Thus, this indicates that the high-An crystals (e.g., An90) andsome of the low-An crystals (bAn60) were not in equilibrium withthe amphiboles.

5.3. Investigation of intensive variables using the MELTS algorithm

To further investigate potential equilibria between differentmineral–melt compositions a series of calculations using the MELTSthermodynamic algorithm (Ghiorso and Sack, 1995) were conductedwith the bulk rock as a starting composition. Although the texturesand mineral zoning show evidence for open system processes in-cluding limestone assimilation (see discussion in Section 6), wethink that most processes involve mixing and interactions betweencrystals and cumulates of the same magmatic system (e.g., recyclingof 2006 shallow crystal cumulates in the 2010 magma). Moreover,many of these systems are controlled by cotectic and eutectic

An (mol%)

Or (mol%)

An (mol%)

Nov - 2010Oct - 20102006

tes

80

20

40

60 80

triangles) in the 2006, October 2010, and November 2010 samples. Microlites and phe-microlites from domes and pyroclastic material produced during older eruptive eventsvailable for degassing-induced crystallization to occur during magma ascent in 2006.to November 2010, and the Or-component gradually decreases, implying that ascent



Table 3Representative pyroxene compositions (in wt.%), structural formula, and end-members.

Comment 06_amph1a

06_amph1b

06_matrix1_3

oct10_matrix-Cpx1b

oct10_matrix-Cpx2

nov10_matrix1_13

nov10_matrix1_36

nov10_px-matrix2

06_Cpx-OpxA_1

06_px-matrix1

06_Cpx-OpxA_8

2006 2006 oct10_matrix-Opx1

oct10_matrix-Opx3

nov10_Cpx-OpxD_5

nov10_Cpx-OpxAnLine 163

Nov-10

Pyroxene Cpx Cpx Cpx Cpx Cpx Cpx Cpx Cpx Opx Opx Opx Opx Opx Opx Opx Opx Opx Opx bright

mPh mPh matrix matrix matrix matrix mPh matrix CpxOpx matrix CpxOpx xeno xeno matrix matrix CpxOpx CpxOpx

Erupt year 2006 2006 2006 2010-Oct 2010-Oct 2010-Nov 2010-Nov 2010-Nov 2006 2006 2006 2006 2006 2010-Oct 2010-Oct 2010-Nov 2010-Nov 2010-Nov

SiO2 48.48 50.77 53.09 47.69 52.27 52.11 51.76 46.75 54.26 53.81 54.15 51.46 50.20 53.85 53.80 53.75 54.45 49.85Al2O3 5.10 3.19 1.46 6.43 2.10 1.86 3.47 7.07 0.50 1.09 0.66 2.90 3.14 1.22 0.63 1.12 0.85 2.68TiO2 0.76 0.46 0.36 1.38 0.38 0.40 0.43 0.93 0.16 0.16 0.20 0.23 0.27 0.22 0.23 0.23 0.09 0.24FeOa 8.43 7.52 8.27 9.46 8.33 8.72 7.86 9.11 16.55 16.81 17.06 24.06 28.69 17.30 17.45 17.87 18.05 30.82MnO 0.26 0.25 0.59 0.30 0.64 0.66 0.45 0.19 1.22 1.25 1.23 0.68 0.53 1.15 1.33 1.27 1.06 0.74MgO 12.87 14.15 15.99 12.26 15.21 15.23 14.32 11.83 25.56 24.72 25.14 19.95 16.85 24.82 24.77 24.83 24.97 15.08CaO 22.43 22.63 20.32 21.57 20.34 20.31 21.13 22.03 1.39 1.53 1.36 0.56 0.24 1.50 1.77 1.69 1.36 0.74Na2O 0.30 0.25 0.23 0.37 0.42 0.34 0.45 0.36 0.02 0.02 0.04 0.04 0.01 0.03 0.04 0.01 0.01 0.00Total 98.66 99.22 100.36 99.52 99.76 99.65 99.89 98.29 99.66 99.40 99.83 99.89 99.96 100.14 100.07 100.79 100.84 100.16

Si(T) 1.827 1.895 1.955 1.789 1.938 1.937 1.917 1.772 1.982 1.977 1.980 1.939 1.931 1.965 1.967 1.952 1.976 1.938Al(T) 0.173 0.105 0.045 0.211 0.062 0.063 0.083 0.228 0.018 0.023 0.020 0.061 0.069 0.035 0.033 0.048 0.024 0.062Al(M1) 0.053 0.035 0.019 0.073 0.029 0.019 0.068 0.088 0.004 0.024 0.008 0.068 0.074 0.018 -0.005 0.000 0.013 0.061Ti(M1) 0.022 0.013 0.010 0.039 0.011 0.011 0.012 0.026 0.004 0.004 0.005 0.006 0.008 0.006 0.006 0.006 0.002 0.007Fec+3(M1) 0.100 0.063 0.025 0.089 0.044 0.047 0.024 0.115 0.007 -0.008 0.003 -0.015 -0.019 0.010 0.030 0.037 0.007 -0.014Mg(M1+M2) 0.723 0.787 0.878 0.685 0.841 0.844 0.791 0.668 1.392 1.354 1.371 1.120 0.966 1.350 1.350 1.345 1.351 0.874Fec+2(M1+M2) 0.166 0.172 0.230 0.208 0.214 0.224 0.220 0.174 0.499 0.525 0.518 0.773 0.942 0.518 0.504 0.506 0.541 1.016Mn(M2) 0.008 0.008 0.018 0.010 0.020 0.021 0.014 0.006 0.038 0.039 0.038 0.022 0.017 0.036 0.041 0.039 0.033 0.024Ca(M2) 0.906 0.905 0.802 0.867 0.808 0.809 0.839 0.895 0.055 0.060 0.053 0.023 0.010 0.058 0.069 0.066 0.053 0.031Na(M2) 0.022 0.018 0.017 0.027 0.030 0.025 0.032 0.026 0.002 0.001 0.003 0.003 0.001 0.002 0.003 0.001 0.000 0.001

Mg# 81.36 82.05 79.26 76.73 79.70 79.01 78.25 79.32 73.62 72.07 72.56 59.17 50.63 72.27 72.83 72.65 71.40 46.26Wo 47.81 46.96 41.45 46.88 42.37 42.04 44.78 48.31 2.79 3.11 2.75 1.19 0.53 3.02 3.55 3.37 2.71 1.62En 38.17 40.85 45.39 37.08 44.10 43.88 42.23 36.09 71.31 70.13 70.45 58.94 50.88 69.73 69.13 68.84 69.22 45.84Fe 14.02 12.19 13.16 16.05 13.54 14.09 12.99 15.60 25.90 26.76 26.80 39.87 48.59 27.25 27.32 27.79 28.07 52.53

a Total iron as Fe+2. mPh=microphenocryst, CpxOpx=intergrowth of two pyroxenes, xeno=xenolith. Structural formulae after Morimoto (1989). Mg #=100 [Mg/(Mg+Fec+2)]. ‘c’ in subscript indicates calculated value.

15F.Costa

etal./


andGeotherm

alResearchxxx

(2013)xxx–xxx

Pleasecite

thisarticle


thestorage

conditions,andmagm

aticprocesses

thatyieldedthe

centennial2010


Volcanology

andGeotherm



2006 -Hbl1

Al Mg

Ca

Distance (µm)0 2802401601208040 200

T (

°C)

800

1000

900

12

14

16

10

Al 2O

3

(wt%

)

60

65

70

55

Mg#

P (

MP

a)

300

900

500

700P

T

Al2O3

Mg#

a) b)

Fig. 11. 2006 amphibole maps and zoning profiles. (a) BSE image, Ca, Al, and Mg X-ray maps and (b) rim to core major element zoning profiles and associated P–T history of anamphibole phenocryst from 2006. Note the presence of a high-Al, low-Mg core indicative of early crystallization at slightly higher pressures. Note also the presence of reactionrims (20–30 μm wide) made of pyroxene and Plag.


relationships, such that back-mixing of previous cumulate mineralsdoesn't greatly alter the bulk-rock major element composition.We also stress here that our goal with these simulations is to see ifthey agree with other geothermobarometic calculations and to testend-member scenarios. We first explored the range of pressuresobtained from the amphiboles and variable water contents. Wewere also guided by the stability of amphibole in experimental re-sults of intermediate magma compositions (e.g., Barclay et al.,1998; Martel et al., 1999; Pichavant et al., 2002), which require a

10

11

12

13

16

15

14

Al 2

O3

(wt%

)

Mg#50 60 706555

Cores Rims

Oct 2010

Nov 2010

2006

Fig. 12. Plot of Mg# vs. Al2O3 in amphiboles from the 2006, October 2010, and November2010 eruptions. Multiple analyses were donewithin individual crystals. Amphiboles in allthree units show a similar range in Al2O3 andMg#. Three end-member compositionsmaybe distinguished; (1) high-Al, and relatively high-Mg#, (2) low-Al and higher Mg#, and(3) low Mg# and variable-Al, which is only represented by the cores of a few crystals inNovember 2010 samples (diamonds). Amphiboles with the highest Al2O3 are fromNovember 2010.


water content in the melt of about 4 wt.%. Up to about 5 wt.% ofH2O+CO2 was found in Merapi glass inclusions by Borisova et al.(this issue). We considered an oxygen fugacity at the Ni–NiO buffer,as it is standard for arc magmas (e.g., Luhr, 1990). An obvious limita-tion of using MELTS is that it cannot establish the stability of theMerapi amphiboles. Fortunately, the low modal proportions ofAmph in both 2006 and 2010 magmas (b1 wt.%; Table 1) limits thethermodynamic effect of omission of this phase.

We find that at 300 MPa, and 4 wt.% H2O, the liquidus temperatureis 1067 °C. Cpx is the first mineral to crystallize with a compositionclose to the natural high-Al Cpx crystals (Fig. 7). Plag starts crystalliz-ing at 1020 °C, with a composition of An78. At 6 wt.% H2O in the melt,Cpx appears on the liquidus at 1035 °C with a composition similar tothe natural high-Al crystals, and An80 plagioclase starts crystallizingat significantly lower T (945 °C). An80 is the most Ca-rich plagioclasethatMELTS stabilizes at the P and H2O conditionswhich are consistentwith our independent geothermobarometric calculation and with ex-perimental phase equilibria. The An80 composition is found in theMerapi samples, but it is more calcic than that calculated from the am-phibole–plagioclase thermodynamic model of Holland and Blundy(1994). It is also lower in Ca content than the characteristic An90 crys-tals found in many of the Merapi magmas. This result suggests thatthese high Ca-rich plagioclases are probably the product of limestoneassimilation (see Section 6).

The MELTS model runs at 800 MPa (the higher P cluster recordedby the amphiboles) with 4 wt.% H2O stabilize two types of Cpx begin-ning at 985 °C. Their Al2O3 ranges from 6 to 9 wt.% (or higher), andthey coexist with An60 Plag, garnet and Fe–Ti oxides. Increasingthe water to 6 wt.% completely suppresses Plag, although the Cpxhas similarly high Al2O3 of 6–7 wt.%. Thesemodeled high-Al Cpx crys-tal compositions overlap with the highest Al-contents of Cpx found inthe November 2010 samples (Fig. 7), although their temperatures areclearly lower than those derived from other geothermometers. Thushigh pressure (i.e., high water pressure which suppresses plagioclasecrystallization) and moderate temperature, or intermediate pressureand high temperature stabilize high-Al Cpx. Runs at lower pressures(110 MPa) and lower water contents (1–2 wt.%) show that Cpx andOpx with similar compositions to the natural low-Al ones (althoughslightly higher Al2O3 contents) may coexist at about 1000 °C andwith An52 plag.



CaMg

Al

Wo

En

a

b

a

b

Ca

a

b

Mg

a

b

Al

(wt%

)

02

864

10

40

60

80

20(mol

%)

Al2O3

Distance (µm)0 20016040 80 120

ab

Wo

En

(wt%

)

02

864

10

40455055

35

(mol

%)

Al2O3

Distance (µm)0 5040352515105 20 30 45

a

b c

November 2010 - Px5

November 2010 - Px6

November 2010 - PxA

c

a

b

c

ab

Ca

c

ab

Mg

c

ab

Al

ab

Wo

En

(wt%

)

02

864

10

40

60

80

20(mol

%)

Al2O3

Distance (µm)0 800600200 300 500100 400 700

Opx

ol

ol

Opx

a) b)

c) d)

e) f)

Fig. 13. November 2010 pyroxene maps and zoning profiles. (a, c, e) BSE images, Ca, Al, and Mg X-ray maps, and (b, d, f) rim to core major element zoning profiles of two Cpx phe-nocrysts (top) and a Cpx–Opx aggregate (bottom) from November 2010. Wo and En stand for wollastonite and enstatite respectively. Note the high-Al2O3, high-Wo rims of the Cpxphenocrysts (bright zones on BSE images a and c), compared to the uniformly low-Al2O3 in the Cpx of the aggregate (e). Opx is present within the core of Px6, hence the low-Woand high-En (c, d). Olivine is present as an inclusion in Opx in the aggregate (e, f).


5.4. Summary of pre-eruptive conditions and a possible ‘static’ magmaplumbing system of Merapi

Combining the results from MELTS and the different geo-thermometers and barometers we propose a possible magma stor-age scenario for Merapi's 2006 and 2010 magmas. We believe thatthis is the simplest working model for the magma systems, but we


stress that the complexity of the rocks do not allow for a unique ro-bust solution. Regardless, our model provides a useful conceptualsnapshot of the plumbing system during repose times of Merapi(Fig. 17). Similar magmatic models proposed by other authors(e.g., Chadwick et al., 2007; Deegan et al., 2010; Innocenti et al.,this issue-b) tend to validate our multi-stage and multi-depthmodel:



November 2010 - Hbl5

K

Mg

Al

T (°

C)

800

1000

900

12

14

16

10

Al 2O

3

(wt%

)Distance (µm)

0 120100804020 60

60

65

70

55

Mg#

P (

MP

a)

300

900

500

700

P

T

Al2O3

Mg#


Al

Mg

K

Distance (µm)0 2802401601208040 200

T (°

C)

800

1000

900

12

14

16

10

Al 2O

3

(wt%

)

60

65

70

55

Mg#

P (

MP

a)300

900

500

700

P

T

Al2O3

Mg#


T (

°C)

800

1000

900

12

14

16

10

Al 2O

3

(wt%

)

60

65

70

55

Mg#

P (

MP

a)

300

900

500

700P

T

Al2O3

Mg#

Distance (µm)0 800700500200100 400 600300

a) b)

c) d)

e) f)

Fig. 14. November 2010 amphibole maps and zoning profiles. (a, c, e) BSE images, Mg, Al, and K X-ray maps, and (b, d, f) rim to core major-element zoning profiles and associatedP–T history of three hornblende phenocrysts from November 2010. The shapes and crystal rims of amphibole phenocrysts range from euhedral and strait, to anhedral and resorbed,sometimes varying even within a single crystal (a). Crystals can be virtually unzoned (a, b) or show alternating compositions (c, d). Note the absence of reaction rims at the marginsof these crystals.


(a) A deep reservoir at about 29 km below the summit (800 MPa)is suggested by some of the amphibole compositions. Watercontents of 4–6 wt.% are required to stabilize amphibole inthese andesitic magmas (e.g., Barclay et al., 1998; Martelet al., 1999; Pichavant et al., 2002). It is possible that somehigh-Al Cpx grew already at this depth. This near-liquidusmagma is likely derived from basaltic melts that fractionatedat high pressures and sufficient water contents to suppress pla-gioclase crystallization and drive the andesitic liquid towards


the high-Al compositions of the bulk rock.(b) A reservoir at an intermediate depth (about 13 km below the

summit or about 300–450 MPa) is indicated by other amphi-bole compositions. Some of the high-Al Cpx may also havegrown in this reservoir. Temperatures of the magmas in theintermediate reservoir tend to be lower (920 to 1030 °C)although they partially overlap with those of the deepermagma according to geothermometry calculations using Cpxand Amph compositions. Amph and high-Al Cpx would be in



Table 4Representative amphibole compositions (in wt.%), structural formula, pressure and temperature determinations.

Label 2006-hb2aLine 017

2006-hb2aLine 014

oct10-hb3aLine 005

oct10-hb3aLine 007

2010novtrvhbl1Line 004




SiO2 41.48 41.07 41.71 40.76 40.13 41.88 40.44 38.87TiO2 2.52 2.51 2.90 2.85 2.09 2.70 2.36 2.26Al2O3 10.95 11.48 10.97 12.18 14.01 11.10 12.53 14.00FeOa 11.83 11.72 12.22 12.35 11.31 12.37 16.15 15.80MnO 0.36 0.30 0.35 0.27 0.10 0.33 0.36 0.36MgO 14.05 13.98 13.97 13.32 13.96 14.08 11.08 10.61CaO 11.25 11.54 10.97 11.60 12.24 11.06 11.63 11.71Na2O 2.44 2.34 2.43 2.46 2.23 2.41 2.26 2.25K2O 0.92 0.93 0.93 0.91 1.23 0.90 1.07 1.18F 0.22 0.26 0.31 0.15 0.07 0.24 0.01 0.00Cl 0.06 0.06 0.08 0.05 0.01 0.07 0.05 0.04F=O 0.09 0.11 0.13 0.06 0.03 0.10 0.00 0.00Cl=O 0.01 0.01 0.02 0.01 0.00 0.02 0.01 0.01

Si 6.169 6.110 6.150 6.044 5.891 6.144 6.021 5.850AlIV 1.831 1.890 1.850 1.956 2.109 1.856 1.979 2.150AlVI 0.088 0.123 0.056 0.172 0.314 0.064 0.220 0.332Ti 0.282 0.281 0.321 0.318 0.230 0.298 0.265 0.255Fec+3 0.721 0.676 0.815 0.584 0.622 0.866 0.665 0.649Mg 3.113 3.100 3.071 2.944 3.055 3.080 2.460 2.379Fec+2 0.751 0.783 0.692 0.947 0.767 0.652 1.346 1.339Mn 0.045 0.038 0.044 0.034 0.012 0.041 0.045 0.045Ca 1.792 1.840 1.733 1.842 1.924 1.739 1.855 1.888Na 0.703 0.674 0.696 0.707 0.634 0.685 0.651 0.655K 0.174 0.176 0.175 0.173 0.231 0.169 0.203 0.226Mg# 0.806 0.798 0.816 0.757 0.799 0.825 0.646 0.640

P (MPa)$ 317 379 323 406 710 363 402 791T (o C)$ 984 973 995 980 955 987 937 965X An# 66 67 65 70 72 64 65 71

a Total iron as FeO, $ calculated after Ridolfi and Renzulli (2012). X An#=Composition of the plagioclase in anorthite mol% in equilibrium with the amphiboles at the determinedP, T calculated by using Holland and Blundy (1994) equilibria with the reaction edenite+albite=richterite+anorthite Mg #=100 [Mg/(Mg+Fec+2)]. ‘c’ in subscript indicates thecalculated value.


equilibriumwith An70–80 Plag in this reservoir. As in the deeperreservoir, a water content of about 4–6 wt.% was necessary tokeep the amphibole stable. Petrographic relations suggestthat Opx also grew at this depth although we don't havegeobarometric confirmation of this. At this depth the magmaswere beginning to degas, probably mainly CO2 and lesseramounts of SO2 because both have lower solubilities thanwater in silicate melts at these conditions (e.g., Scaillet andPichavant, 2003). The relative abundances of crystalsand melt in this intermediate reservoir is not clear. Thepresence of very high-An Plag is best explained by carbonate

November 2010 - Apa1

Cl F

Fig. 15. November 2010 Apatite maps. BSE image and S, F, and Cl X -ray maps of two apatitwith S content typically decreasing from core to rim, which illustrates the complicated use o


assimilation in the shallow to intermediate parts of the system,as discussed in Section 6 below.

(c) A shallow reservoir-conduit (e.g., 100 MPa) where amphiboleis not stable and magmas degas and crystallize extensively isrequired to explain the presence of low-Al Cpx and Opx,which constitute most of the glomerocrysts, and also the abun-dant lower An Plag. This part of the system is also where theamphiboles break down and the reaction rims form. We pre-sume that the shallow reservoir was much richer in crystalsthan the other reservoirs because of more rapid and extensivewater degassing and cooling.

S

es included in an Amph crystal (scale bar is 20 μm). Both crystals show strong zoning,f single individual analyses of apatite to understand the volatile contents of the magma.



T (°C)

800 900 10501000850 950

300

400

500

600

1000

900

700

800

P (

MP

a)

2006

Oct 2010

Nov 2010

Fig. 16. P–T relations obtained from amphibole and pyroxene thermodynamic equilib-ria. Pressure and temperature estimates based on the compositions of amphiboles(basic shapes symbols, multiple analyses within single crystals) using the geother-mobarometer of Ridolfi and Renzulli (2012). The error on pressure is about 10% rela-tive, whereas that on temperature is ±25 °C. Amphiboles show a large range ofpressures from about 300 to 900 MPa. However, most determinations are around350–400 MPa or 800 MPa. In detail it can be seen that except for one crystal, all the2006 amphiboles show pressures around 350–400 MPa, whereas, those from Octoberand November 2010 have several crystals that record higher pressures of crystalliza-tion of ~800 MPa. The identification of two main pressures suggests that the 2006and 2010 eruptions tapped magmas from two principal magma reservoirs at depthsof ~13 and ~29 km below Merapi. Amphiboles record temperatures from about 900to 1000 °C, with the largest temperature variations found for amphiboles that recordthe lower pressures (350–400 MPa). Amphiboles from the 2006 samples plot mainlyat the high-temperature end of the range (950–1000 °C). For comparison, tempera-tures estimated from coexisting pairs of Opx and Cpx using the thermodynamicmodel QUILF are shown with vertical lines. The horizontal bars shown at the top ofthe plot and attached to each line illustrate the range in temperatures obtained withineach sample (n pairs=29, 19, and 8 for the 2006, October 2010, and November 2010samples, respectively). The error on individual determinations is ~40–50 °C. Thesetemperatures overlap with the higher end of the amphibole temperatures and suggestearly crystallization of pyroxenes.


The presence of an evolved, Fe-rich magmatic component in thesystem is required to explain the distinctive low-Mg# Opx, Amph,and Bt, which are present as individual crystals in the 2006 andNovember 2010 samples, and within the xenolith from one ofthe 2006 samples. These minerals and compositions indicate thepresence of a more evolved magma, although it is not clear howsuch an evolved magma is related to the main Merapi suite of rocks.It is probably located at shallow depths and might be the result ofeven more extensive crystallization or crustal assimilation.

6. Pre-eruptive magmatic processes in the 2006 and 2010 magmas

6.1. Assimilation of crustal carbonates

Merapi volcano is located within the Kendeng basin, which isthought to be underlain by 8 km (de Genevraye and Samuel, 1972)to 11 km (Untung and Sato, 1978) of sedimentary deposits. The ex-posed upper crustal rocks in the region consist of a sequence of Creta-ceous to Tertiary marine limestones, marls, and volcanoclasticsediments (van Bemmelen, 1949). These rocks are frequently foundas thermally metamorphosed xenoliths in recent Merapi lavas(Camus et al., 2000) — a relationship that indicates significantmagma–crust interactions (Chadwick et al., 2007; Deegan et al.,


2010, 2011). The assimilation of such rocks by Merapi magmascould have significant effects on their gas budget, explosivity, andmineral compositions (e.g., Chadwick et al., 2007; Troll et al., 2012;Borisova et al., this issue).

The evidence for the involvement of CO2 from breakdown of car-bonates in Merapi volcano has been recently proposed based on Cisotope compositions of fumarole gases emitted during the 2006eruption. Troll et al. (2012) conclude that about 50–80% of this CO2

could come from crustal carbonates, and they suggest that an increasein the carbonate signature of the gas composition was caused by themagmatic entrainment of limestone following the 2006 earthquake.Dynamic triggering of Merapi's volcanic activity by regional earth-quakes is also discussed by Walter et al. (2007) and Jousset et al.(this issue). However, no marked increase of explosivity was seen im-mediately after the earthquake, and the 2006 eruption remained alow explosivity event (VEI of 2). Thus, although it is clear that CO2

from carbonates contributes to the gas budget coming out of Merapivolcano, it is less clear how this affects the explosivity of theeruptions.

Additional evidence of carbonate involvement in the petrogenesisof Merapi magmas comes from the mineral compositions. Chadwicket al. (2007) found that the concentrations of Fe, Mg as well as Srisotopes of the anorthitic plagioclase overlap with those of thecalc-silicate xenoliths found in the juvenile blocks from July 1998eruptions. The Fe versus An trends we report for the plagioclasefrom the 2006 and 2010 eruptions (Fig. 18) are similar to those ofChadwick et al. (2007). The decreasing concentrations of Fe and Anin crystals bAn85 could be due to crystal fractionation or mixing,and a group of crystal zones with very high An and low Fe concentra-tions could represent the assimilation of carbonate rocks.

The addition of Ca to Merapi magmas by carbonate assimilationcan also potentially affect the composition and stability of Cpx, aslimestone assimilation has been shown to favor crystallization ofCa-rich Cpx in phase equilibria experiments (Freda et al., 2008;Iacono Marziano et al., 2008; Mollo et al., 2010). Thus, some of ourhigh-Al and high-Wo Cpx crystals or crystal rims could be the resultof carbonate assimilation (Fig. 7). However, these Cpx compositionswould require the addition of 10 to 15 wt.% of carbonate; largeamounts of assimilation that would lead to significant changes inthe bulk-rock composition. For example, bulk mixing of only 1 to5 wt.% of carbonate (CaCO3) with the most evolved magma eruptedin 2006 would result in CaO bulk concentrations that are wellabove the general trend of recent magmas that have erupted atMerapi volcano (Fig. 19). Moreover, such large amounts of assimila-tion would also lead to significant changes in the fluid phase, poten-tially decreasing the water fugacity to the extent that amphibolewould no longer be stable. We therefore believe that high-Al Cpx isthe product of moderate-P and high-T crystallization from a ratherprimitive magma, or the combined effect of lesser amounts of carbon-ate assimilation, which was induced by high-T magma replenish-ment. In summary, our dataset confirms the involvement ofcarbonate rocks in Merapi magmas, however, it doesn't enable us todetermine whether this assimilation was more significant prior tothe 2010 eruption than prior to the 2006 eruption. In contrast, a larg-er amount of carbonate assimilation in the 2010 versus 2006 eruptionhas been proposed by Borisova et al. (this issue) based on Sr and ox-ygen isotopic data.

6.2. Mixing and interaction between resident magma and new intrusions

The complexity of textures and chemical zoning patterns of pla-gioclase, clinopyroxene, and amphibole records changing conditionsin the Merapi magmas that involve crystallization, dissolution andcrustal assimilation. We think that mixing/mingling between theintermediate and upper parts of the magmatic system with newmagmas from the deeper reservoir is an important process at Merapi.



Table 5Representative compositions (in wt.%) and structural formula of biotite, apatite, Fe–Ti oxides, and olivine.

Pointmineral

2006enclavtr3 Line 007 2006enclavtr3 Line 010 2006enclavtr3 Line 011 Nov10-phlogo1 Nov10-phlogo2 Nov10-phlogo3 apahbl2r apainglalargetrav Line 003 10-Nov Nov10-Oli

Biotite Biotite Biotite Biotite Biotite Biotite Apatite Apatite Magnetite Olivine

SiO2 37.33 37.01 36.99 36.99 36.43 36.12 SiO2 0.51 0.24 SiO2 0.10 SiO2 37.81TiO2 2.88 1.73 1.49 5.22 4.86 5.15 Fe2O3$ 0.89 0.39 TiO2 9.28 FeOa 25.97Al2O3 16.71 17.48 17.51 14.36 14.55 13.95 CaO 52.30 51.83 Al2O3 2.93 MnO 1.16FeOa 12.38 11.82 11.95 15.24 14.84 15.18 P2O5 39.69 40.83 FeOa 78.01 MgO 34.06MnO 0.08 0.09 0.07 0.25 0.22 0.27 SO3 0.96 0.07 MgO 2.66 CaO 0.12MgO 16.64 17.35 17.06 14.45 14.30 13.94 F 2.38 3.11 MnO 0.90 Total 99.15CaO 0.04 0.02 0.03 0.00 0.01 0.01 Cl 0.62 0.96 Sum 93.88Na2O 1.23 1.41 1.34 0.84 0.87 0.82 Sum 97.34 97.43 Fo mol% 70.05K2O 8.45 8.30 8.26 8.79 8.72 8.58 O-F, Cl 1.14 1.53 Fe2O3c 47.90F 1.67 1.51 1.35 0.26 0.26 0.21 Total 96.20 95.90 FeOc 34.90Cl 0.01 0.00 0.02 0.13 0.12 0.18Tot 97.41 96.71 96.06 96.54 95.19 94.40 Tot 98.68F=O 0.70 0.64 0.57 0.11 0.11 0.09 Si 0.09 0.04Cl=O 0.00 0.00 0.00 0.03 0.03 0.04 Fe 0.13 0.06 Si 0.03Tot 96.71 96.08 95.48 96.40 95.05 94.27 Ca 9.80 9.68 Ti 2.08

P5 5.88 6.02 Al 1.03IVSi 2.71 2.70 2.72 2.69 2.69 2.69 S6 0.13 0.01 Fec+3 10.75IV Al 1.29 1.30 1.28 1.23 1.27 1.22 X F 1.31 1.71 Fec+2 8.70IV Fec+3 0.08 0.04 0.09 X Cl 0.18 0.28 Mg 1.18

Total 17.52 17.81 Mn 0.23VIAl 0.13 0.20 0.23Ti 0.16 0.09 0.08 0.29 0.27 0.29 X OH 0.50 0.00Fec+3 0.20 0.13 0.09 0.38 0.36 0.38 H2O 0.43 0.00Mg 1.80 1.89 1.87 1.57 1.57 1.55 F/Cl 7.15 6.06Fec+2 0.55 0.59 0.65 0.47 0.51 0.48 Tot 18.02 17.81Mn 0.00 0.01 0.00 0.02 0.01 0.02Ca 0.00 0.00 0.00 0.00 0.00 0.00Na 0.17 0.20 0.19 0.12 0.12 0.12K 0.78 0.77 0.77 0.82 0.82 0.81Tot 0.96 0.97 0.97 0.93 0.95 0.93Mg/Fea 2.40 2.62 2.55 1.69 1.72 1.64Mg# 70.56 72.35 71.79 62.84 63.21 62.07AnF 0.38 0.35 0.31 0.06 0.06 0.05An Cl 0.00 0.00 0.00 0.02 0.01 0.02An OH 1.62 1.65 1.68 1.92 1.93 1.93

a Total iron as FeO, $=total iron as Fe2O3; Mg #=100 [Mg/(Mg+Fec+2)]. ‘c’ in subscript indicates calculated value.

21F.Costa

etal./


andGeotherm

alResearchxxx

(2013)xxx–xxx

Pleasecite

thisarticle


thestorage

conditions,andmagm

aticprocesses

thatyieldedthe

centennial2010


Volcanology

andGeotherm



MOHO 30

20

10

0

0.8

0.6

0.4

0.2

0

Assimilation~An90 pl

Volatile saturation4-6 wt% H2O

Pressure(GPa)

Depth(km)

Fe-rich Opx

Fe-rich Amph

Fe-rich Bt

Ort

hopy

roxe

ne

Deep Reservoir

IntermediateReservoir

Shallow‘Reservoir’

?

Pl microlites and phenocryst rims -Ascent rates

Am

phib

ole

high

-Plo

w-P

Am

ph b

reak

dow

n

Pla

gioc

lase

An 40

-50

An 60

-80

high

-Al

Clin

opyr

oxen

elo

w-A

l, cl

uste

rshi

gh-A

l

Fig. 17. The information frommineral textures, chemistry, and geothermobarometry is shown here as a schematic representation of magma genesis and the plumbing system underMerapi volcano. Pressures estimated by Amph geobarometry suggest the presence of at least two main reservoirs. In the deep reservoir, high-Al Amph crystallize in equilibriumwith high-Al Cpx. In the intermediate reservoir, low-Al Amph, high and low-Al Cpx, and intermediate-An Plag (~An60–80) form by crystallization of Merapi magmas with4–6 wt.% H2O. High-An Plag cores (~An90) also originated from those intermediate depths due to the interaction with the host limestones. Shallower storage of magmas is verylikely, however it will not be recorded by amphiboles which break down at low pressures. Opx and low-Al Cpx glomerocrysts originated from crystallization at these shallow depths(b13 km) after substantial degassing and crystallization of ‘left over’ Merapi magmas. Low-An plagioclase microlites and phenocryst rims are formed during magma ascent in theuppermost parts of the system. Their compositions reflect the ascent conditions (e.g., ascent rate) of magma during an eruption. Magma degassing will occur above 8–10 km depth,as Merapi magmas with 4–6 wt.% H2O reach volatile saturation around 200–300 MPa, or deeper if the CO2 is also present An evolved, Fe-rich magmatic component in the system isalso required to explain the rare occurrence of low-Mg# Opx, Amph, and Bt. This evolved magma could be located at shallow depths, as part of a boundary layer surrounding themain shallow magma reservoir.


The dissolution textures of many clinopyroxenes and some amphi-boles (Figs. 5, 11, 13, and 14), and in particular the presence of thehigh-Al Cpx rims growing on resorbed Opx or low-Al Cpx, showthat the Merapi magmas experienced at least one late event of crystaldissolution followed by regrowth. This can be explained by mixingbetween a rather hot and gas-rich magma that may have locallyassimilated limestone, and a more degassed, and probably coolerand more crystal-rich part of the system. This scenario reflects themixing between the degassed materials left over in the plumbingsystem from previous events and the new magma coming fromdepth, as we expect at active volcanoes with eruption frequencies ofa few years and exhibiting permanent degassing behavior (e.g.,“open vent” volcanoes). Similar processes have been proposed atother very active mafic to intermediate volcanoes such as Stromboli(Landi et al., 2004), Arenal (Streck et al., 2005), El Chichón(Andrews et al., 2008), and Llaima (Bouvet de Maisonneuve et al.,2012).

The evidence of magma replenishment and interaction is moreprominent in the 2010 samples than in those from 2006, and in detailwe see more evidence of interactions in the November 2010 than in


the October 2010 samples. This is shown by (i) the increasing Aland Wo contents of Cpx (Fig. 7), (ii) the more common presence ofhigh-Al (high-P) amphibole (Figs. 12 and 16), and (iii) the increasingAn content of the plagioclase microlites (Figs. 4 and 9). It thereforeappears that the total volume of replenishing magma was significant-ly greater in 2010 and we believe that this is the main difference andis reflected in the different eruption behaviors.

6.3. Differences between 2010 and 2006 syn-eruptive processes

Given a more gas-rich magma and explosive eruption, why arethe 2010 Merapi pyroclasts so vesicle-poor (Figs. 3 and 4)? SEM im-ages show the presence of few but relatively large and rounded ves-icles in the 2006 samples. The 2010 samples, and in particular thoseproduced on 4 November, appear to be more vesicular. However, the4 November vesicles are significantly more distorted when large,and they coexist with a fraction of small bubbles. Expanded pumi-ceous samples are only found in one deposit of the 2010 eruption(Komorovski et al., this issue) despite the several Vulcanian tosub-Plinian phases of the eruption and pyroclasts in the tephra



FeO (wt%)0.2 0.6 1.00.4 0.8

a) 2006 c) Nov 2010

FeO (wt%)0.6 1.00.4 0.8

b) Oct 2010

FeO (wt%)1.00.60.4 0.8

An

(mol

%)

30

40

50

60

100

90

70

80

20

Fig. 18. Plots of An-content vs. FeO concentration in Plag from the (a) 2006, (b) October 2010, and (c) November 2010 eruptions. Plag crystals from the three deposits show thesame range in compositions. As described by Chadwick et al. (2007), a slightly negative correlation exists for An-contents >85 mol%, and a positive correlation for An-contentsb85 mol%. Dashed ellipses highlight the two trends defined by Chadwick et al. (2007). The positive correlation could be due to fractional crystallization, whereas the negativecorrelation can be explained by the assimilation of carbonates and/or calc-silicate rocks. Microlites are shown with color symbols and plot at the higher-FeO end of the range.They increase in An-content from 2006 to October 2010, to November 2010. (For interpretation of the references to color in this figure, the reader is referred to the web versionof this article.)


deposits are mainly dense (Pallister, unpublished data). This con-trasts with pyroclasts in deposits from the last comparably explosiveeruption of Merapi in 1872, which have higher vesicularities andlower microlite contents compared to the 2010 eruptions (e.g.,Innocenti et al., this issue-a). Thus, it seems that the syn-eruptiveprocesses that led to the 2010 eruption were also different thanthose of 1872.

The presence of mainly dense pyroclasts in deposits from the 2010explosive eruption may be explained if we consider the results of

VEI 3VEI 42010

2006

10%

8%

1%

SiO2 (wt%)48 5650 52 54

CaO

(w

t%)

14

8

10

12

Fig. 19. Whole-rock compositions (SiO2 vs. CaO) of the pyroclastic material erupted in2006, 2010, and during other historic events of similar intensities (VEI). Data is fromthis study and Table 5 of Gertisser and Keller (2003). The more explosive eruptions(VEI 3: 1930, VEI 4: 1878) have slightly more calcic compositions than the less explo-sive eruptions (VEI≤2: 1883, 1904–06, 1906, 1942, 1954, 1957, 1992, 1994, 1998), ex-cept for the 2010 eruption. The effect of assimilation of 1 wt.%, 8 wt.% and 12 wt.% ofpure carbonate on the CaO content of Merapi magmas is shown with a black arrow.The assimilation of large amounts of carbonates would strongly affect the whole-rockcompositions. (For interpretation of the references to color in this figure, the readeris referred to the web version of this article.)


petrophysical and theoretical studies that have shown that feedbacksbetween magma porosity and permeability can show a hysteresis be-havior (e.g., Rust and Cashman, 2004; Takeuchi et al., 2005; Michautet al., 2009). In the deep parts of the conduit magma will exsolve iso-lated gas bubbles with no connectivity and low permeability. Once athreshold bubble-content is achieved, permeability is establishedand increases with increasing porosity. But once a permeable net-work has formed, the gas can segregate from the magma and leavethe conduit system by porous flow. The expanded magma will thencollapse or compact (Eichelberger, 1989), reducing its porosity, butpotentially retaining the permeability, provided that the processesare fast enough to prevent the destruction of pore connectivity. Theporosity and permeability of a magma therefore depends on its vesic-ulation and degassing history, and some memory of a previouslyhighly vesicular magma can be preserved.

At open-vent volcanoes such as at Merapi, a continuous column ofmagma extends from the near-surface to the shallow parts of thereservoir. Stagnant or slowly ascending magma therefore may vesic-ulate, then undergo vesicle collapse and partially degass, but still re-tain some permeability as described above. This is well exemplifiedby the low porosity of the 2006 samples and other historic Merapilava domes. In contrast, during the November 2010 eruption, ascentrates were greater, which reduced the time available for compactionof the permeable network (e.g., Michaut et al., 2009), hence the nonpumiceous but more vesicular samples. Thus, the rarity of expandedpumices may be due to rapid degassing and maybe partial re-welding of magma as it ascended from intermediate depths.

The October and November 2010 samples contain manymicrolitesthat are broken crystals (Fig. 3). Such observations are not seen in the2006 samples and they probably reflect the faster ascent andexplosivity of the 2010 magma, in agreement with the slightly highervesicularities. We hypothesize that strong shearing in the conduitfragmented the crystals and dispersed them in the matrix as the po-rous network was expanding. And finally, the rapid ascent ofmagma in the conduit, producing the extremely high-rate dome effu-sion on 1–4 November and the 4 November collapse of the new5×106 m3 summit lava dome (Surono et al., 2012; Pallister et al.,this issue) brought more gas-rich, but still mostly compactedmagma to the surface to fuel the main Vulcanian to Plinian eruption.



Table 6Representative compositions (in wt.%) of matrix glass and glass inclusions.

06_matrix1_g4 oct10_matrix1_g7 101123-brownhb-incl1 101121_2amapcpx1 101123-incl2 101123-px2incl1

Matrix Matrix Glass inclusions

2006 Oct-10 Nov-10 Nov-10 Nov-10 Nov-10

SiO2 71.63 69.05 62.30 66.51 69.17 68.60TiO2 0.38 0.46 0.43 0.40 0.36 0.63Al2O3 14.18 15.03 16.66 15.67 16.43 15.22FeO a 2.45 3.43 4.86 3.42 1.26 2.26MnO 0.08 0.11 0.20 0.15 0.05 0.11MgO 0.21 0.44 0.83 0.36 0.19 0.34CaO 0.78 1.37 4.41 1.37 1.95 1.58Na2O 3.78 3.88 4.18 5.22 5.12 5.10K2O 6.21 5.95 5.59 6.70 5.00 5.80F 0.06 0.09 0.07 0.00 0.06 0.01Cl 0.24 0.19 0.42 0.00 0.28 0.32SO3 0.04 0.19 0.12 0.03Original sum 98.6 97.9 97.5 95.7 95.1 94.8

a All Fe as +2. Oxide concentrations recalculated to 100% anhydrous.


7. Time scales of interaction between deep and shallow magmas

We propose that the high-Al Cpx rims formed around low-Al Cpxor Opx cores in the 2006 and 2010 rocks (Figs. 5 and 13) are due tothe intrusion of a hotter, less degassed and crystal-poor magma thatalso assimilated some limestone, into a more degassed, colder andprobably crystal-rich system. The abrupt compositional gradients inFe–Mg–Ca and Al between the Cpx rims and their interiors decaywith time due to intracrystalline chemical diffusion (e.g., Morgan etal., 2004). Thus, it is possible to obtain time estimates for the arrivalof the new magma at the intermediate reservoir, assimilation, anderuption based on known diffusivities and temperature (e.g., Costaet al., 2008; Costa and Morgan, 2010). There have been several deter-minations of major cation diffusivities in clinopyroxene, but mosthave end-members different from those present in Merapi (see

< 2.4 y2006 - px3

Distance (µm)0 4010 20 30

60 10070 80 90

30 7040 50 60

72

74

76

78

70

Mg#

74

76

78

72

Mg#

70

72

74

76

68

Mg#

< 3.0 y2006 - trav2

< 5.0 y2006 - trav1

a)

b)

c)

Fig. 20. Modeling of Fe–Mg diffusion zoning in Cpx from 2006 (a–c) and November 2010 (dthe lower left corner of each panel represent an error of 1% relative on the Fe–Mg concentrathe period between the arrival of new hot and gas rich magmas in the intermediate or shallthan the 2006 event.


Cherniak and Dimanov, 2010 for a recent review of available data).Diffusion in clinopyroxenes involves at least three components (Fe–Mg–Ca) with thermodynamic non-ideality of mixing (e.g., Zhang etal., 2010). There is also a dependency on oxygen fugacity and crystal-lographic orientation (Cherniak and Dimanov, 2010) although theavailable data suggests that these factors do not have a first-order ef-fect on the diffusivities.

We take here a simpler approach of modeling the Fe–Mg gradientsbecause these elements have been best characterized for their diffu-sivities (Dimanov and Sautter, 2000; Dimanov and Windenbeck,2006). At temperatures between 1000 and 1050 °C we obtain diffu-sivities ranging from about 5×10−9 to 5×10−8 μm2s−1; we havechosen to use an average value of 1×10−8 μm2s−1 in our calcula-tions. For the initial conditions we assumed that the crystal boundarywas perfectly sharp before diffusion started. This tends to

< 1.6 y2010 - px5a

0 4010 20 30

Distance (µm)0 4010 20 30

0 4010 20 30

66

70

74

78

66

70

74

78

66

70

74

78

< 2.7 y2010 - px5c

< 1.9 y2010 - px5b

d)

e)

f)

–f) samples. For details about modeling conditions please see main text. Vertical bars intion ratios. The times shown in years are maximum and we interpret them as recordingow reservoirs and the eruption. Note that the times are somewhat shorter for the 2010



Explosive eruption Fragmentation

2010

Large volumeMagma recharge

Magma mixing, cooling,

crystallization

Rapid ascent, little syn-eruptive

degassing,high-An, Or-poor

microlites

NO time for pre-eruptive

magma degassing

Assimilation

MOHO 30

20

10

0

0.8

0.6

0.4

0.2

0

Pressure(GPa)

Depth(km)

2006

time for pre-eruptive

magma degassing

Dome growth and collapse

Small volumeMagma recharge

Magma mixing, cooling,

crystallization

Slow ascent, syn-eruptive degassing,

low-An, Or-rich microlites

MOHO 30

20

10

0

0.8

0.6

0.4

0.2

0

Pressure(GPa)

Depth(km)

Assimilation

Fig. 21. Schematic summary of the possible magmatic processes which led to the 2006 (left) and 2010 (right) eruptions of Merapi. In both cases, the injection of magma from thedeeper reservoir into the shallower parts of the system triggered the unrest. The amount of recharge magma injected in 2010 was substantially larger, as testified by the largervolumes of erupted material as well as the larger amounts of minerals crystallized at greater depths (high-P Amph, high-Al Cpx). Small volumes of recharge magma (as in2006) will strongly interact with the more crystalline magmas stored at shallower depths and thereby cool, crystallize, and start exsolving volatiles. The shallow magmas mayact as a rheological barrier, preventing the rapid extrusion of the newly injected recharge magma, and thereby enabling time for the exsolved volatiles to escape. The resulting erup-tion will have a smaller explosion potential, as part of the volatiles has been lost and overpressures are potentially small (small volumes of injected magma, little to no exsolvedvolatiles in the reservoir). In the case of large volumes of recharge magma (as in 2010), interaction with the more crystalline magmas stored at shallower depths will be less strong,as the rheological barrier will be less effective. In addition, the larger intrusion may lead to more carbonate breakdown and additional liberation of CO2 gas at shallow depths.Magmas will also mix, cool, crystallize, and potentially exsolve part of their volatiles, however the time available for these volatiles to leave the system will be shorter. The resultingeruption will have a larger explosion potential, as a pre-eruptive gas phase is present and overpressures in the reservoir are potentially greater (larger volumes of injected magmaand presence of an exsolved volatile phase). Suggested time scales between magma recharge and eruption are in agreement with estimated time scales from kinetic modeling ofclinopyroxene zoning profiles in 2006 and 2010 erupted materials. Other parameters such as differences in viscosities and crystal content due to varied volumes of magma rechargemay also play a role in controlling the shifts in eruption intensity. In addition, feed-back processes between ascent rates, syn-eruptive degassing, and microlite crystallization willtend to accentuate the initial differences. Slow magma ascent will favor syn-eruptive degassing and microlite crystallization, which in turn will reduce magma ascent rates. Con-versely, fast ascent rates will prevent efficient syn-eruptive degassing and reduce microlite crystallization, which in turn will favor rapid ascent rates. We believe that carbonateassimilation occurred in both 2006 and 2010 however, this was probably not the controlling factor of eruption intensity.


overestimate the time if the profiles are stretched by oblique section-ing of the crystal as the distance between data points is increasedwith respect to the total length of the profile (e.g., Costa andMorgan, 2010), or if the original profile was zoned due to real changesin the liquid (e.g., Costa et al., 2008). Thus, the times we report aremaximum times, and they are not really precise determinations oftimescales and thus include the possible errors noted above.

Three profiles from the 2006 samples give maximum times of2.4 to 5 years, whereas, three profiles from the 2010 samples givesomewhat shorter times between 1.6 and 2.7 years (Fig. 20). Asthese are within the repose periods for Merapi, it is possible thatthe crystals are recording processes associated with each eruptiveevent. Moreover, the calculations suggest that the time since the in-trusion of the deep magma and eruption, may have been shorter forthe 2010 than for the 2006 event. We acknowledge the uncertaintyin making definitive conclusions from the available data and thatadditional work is needed. However, if this relation is robust, itwould explain the different explosivities of the two eruptions; the2010 deep magmas spent less time at shallow depths — insufficienttime for the deep gas cargo to escape and thus powering the explo-sive 2010 eruption.


8. Linking magma plumbing system and processes with eruptiondynamics and monitoring data of the 2006 and 2010 events

The 2010 eruption of Merapi was unusual, and produced Plinianeruptive columns to 17 km altitude, pyroclastic flows to 16 kmdown slope and a volume of magma that was 5–10 times that ofother eruptions in the past few decades. Another distinct featureof the 2010 eruption is that there was an excess gas phase ifwe compare the amount of SO2 measured by satellite and that cal-culated from the volume of the deposits (Surono et al., 2012). Mon-itoring data and unrest times before the 2006 and 2010 show thatthere were seismic crisis months or close to a year before the twoeruptions (Suharna et al., 2007; Surono et al., 2012). The main dif-ferences in the monitoring is that in 2010 much larger increasesin seismicity and deformation were observed in the week precedingthe eruption, and large CO2 emissions began several weeks beforethe eruption (Surono et al., 2012).

A scenario for the 2006 and 2010 eruptions that we think is con-sistent with the observations and our petrologic calculations is thatduring quiescence and passive degassing periods, the upper to inter-mediate Merapi plumbing system progressively loses its gases,




crystallizes and becomes a stiff and crystal-rich mush zone. We inter-pret that both 2006 and 2010 eruptions were the result of new intru-sions from the deep reservoir. This new magma was crystal-poor,hotter, and richer in volatiles than those residing in the upper partsof the system, but its bulk composition was essentially the same. In-teractions and mingling occurred at intermediate to shallow depths(but at least partly within the stability field of amphibole) betweenthe resident crystal-rich magma and the new magma from depth. In-teractions likely also occurred between the magmas and the carbon-ate country rocks, which added CO2 gas to the magma volatilebudget (Fig. 21). At some point after this interaction, when the over-pressure in the reservoir was large enough, the whole system becameunstable and erupted (e.g., Blake, 1981). In this scenario, the upperand potentially more crystalline parts of the Merapi system can beregarded as a filter, or rheological barrier, that in most cases preventsdirect access and eruption of the deeper gas-rich magmas.

Within this framework, the main difference between the 2006 andthe 2010 events is that in 2010 the amount of magma that arrivedfrom depth was 5 to 10 times larger than it had been for the last100 years or so, and this hadmajor consequences for the plumbing sys-tem and eruption dynamics. When the much larger amount of replen-ishment magma reached the shallow and potentially crystal-rich partsof the system it interacted and partially mixed with it, but it did not re-side at the upper levels for long before erupting explosively. Our datasuggest that the main difference between the 2006 and 2010 eruptiondynamics was a shorter residence time for the replenishing magma atshallow to intermediate depths in 2010, the potentially greater thermalbudget it had for carbonate breakdown and crustal CO2 addition, andthe related faster final ascent rates to the surface. We suggest that themuch larger size of the deeply sourced 2010 magma produced an over-pressure of the shallow reservoir that was large enough to trigger theeruption and rapid dome extrusions, without extensive shallow vesicu-lation. Numerical calculations indicate that replenishment volumes of0.1–1.0% of the volume of the resident magma can already be sufficientto trigger an eruption (e.g., Blake, 1981). Thus, after a limited time ofresidence and interaction in the shallow to intermediate levels, themagma, rich in exsolved volatiles, accelerated to the surface anderupted rapidly to initially form lava domes (Pallister et al., this issue)and then explosively as those domes collapsed and unloaded the under-lying conduit (Komorovski et al., this issue).

9. Conclusions

The petrological study of the 2006 and 2010 rocks shows that un-derstanding the causes of shifts in the eruption style at Merapirequires careful and detailed petrologic observations; at first inspec-tion both magmas are quite similar in terms of the bulk compositions,mineral assemblages, and textural features. They are also similar inthe pre-eruptive conditions and magmatic processes. This similarityis in stark contrast with the extremely different eruption dynamicsand volumes of the two events. From our detailed observations, wepropose that the difference is mainly driven by the ascent of a muchlarger amount of hot, deep, and volatile-rich magma in 2010 com-pared to previous years. The plumbing system of Merapi duringperiods of quiescence and degassing is characterized by residentmagmas at intermediate to shallow depths that are probablycrystal-rich and degassed. These can effectively act as a filter or rheo-logical barrier for slowing down the ascent and promoting thedegassing of volatile-rich magmas coming from deeper reservoirs.However, the much larger magma intrusion in 2010 overwhelmedthe upper crystal-rich zone and after a limited period of interactionand limited degassing quickly moved towards the surface. Thisallowed for the volatiles to remain close to the host magma and ledto faster than usual ascent rates and explosivity. Other explosiveeruptions at Merapi that were driven by more mafic magmas (e.g.,the 1872 event) might be caused by different processes. The question


that remains unanswered is what caused the deep magma source ofMerapi to segregate a much larger amount of magma in 2010 and ifthis will happen again in the near future.

Acknowledgments

We thank A. Borisova, an anonymous reviewer, O. Nadeau, and C. R.Thornber for their comments, which helped to clarify various aspects ofthe manuscript. A. Borisova, S. Innocenti, O. Nadeau, and J.-C.Komorowski gracefully shared with us their unpublished data andpaper about Merapi. We also appreciate the work and comments of P.Jousset in editing this manuscript and Special issue. We acknowledgethe support from the Geological Agency of Indonesia (Badan Geologi),the Center for Volcanology and Geologic Hazard Mitigation (PusatVulkanologi dan Mitigasi Bencana Geologi, PVMBG) and the VolcanoResearch and Technology Development Center (Balai Penelitian danPengembangan Tehnik Kegunungapian, BPPTK). These groups led thesuccessful response to the 2010 Merapi eruption crisis and contributedsubstantially to this research project. JSP also acknowledges supportprovided by the USGS and USAID to the Volcano Disaster AssistanceProgram. FC acknowledges funding from the “Magma plumbingsystem” EOS research project. CBdM acknowledges the Swiss NationalScience Foundation for the fellowship PBGEP2_139851 she wasgranted.

References

Andersen, D.J., Lindsley, D.H., Davidson, P.M., 1993. QUILF: A PASCAL Program to AssessEquilibria Among Fe–Mg–Ti Oxides, Pyroxenes, Olivine, and Quartz. Computersand Geosciences 19, 1333–1350.

Andreastuti, S.D., Alloway, B.V., Smith, I.E.M., 2000. A detailed tephrostratigraphic frame-work at Merapi Volcano, Central Java, Indonesia: implications for eruption predic-tions and hazard assessment. Journal of Volcanology and Geothermal Research 100,51–67.

Andreastuti, S.D., Pallister, J., Newhall, C., 2007. The 2006 eruption of Merapi, petrographyand geochemistry of the June14th, 2006 pyroclastic deposits. Proceedings of the32nd and the 26th IAGI Annual Convention.

Andrews, B.J., Gardner, J.E., Housh, T.B., 2008. Repeated recharge, assimilation, andhybridization in magmas erupted from El Chichon as recorded by plagioclase andamphibole phenocrysts. Journal of Volcanology and Geothermal Research 175,415–426.

Barclay, J., Rutherford, M.J., Carroll, M.R., Murphy, M.D., Devine, J.D., Gardner, J., Sparks,R.S.J., 1998. Experimental phase equilibria constraints on pre-eruptive storage con-ditions of the Soufrière Hills magma. Geophysical Research Letters 25, 3437–3440.

Beauducel, F., Cornet, F.H., 1999. Collection and three-dimensional modeling of GPSand tilt data at Merapi volcano, Java. Journal of Geophysical Research 104,725–736.

Berthommier, P.C., 1990. Etude Volcanologique du Merapi (Centre-Java),Tephrostratigraphie et chronologie - produits eruptifs. Ph.D. Thesis. Universite BlaisePascal, Clermont-Ferrand, France, pp. 216.

Blake, S., 1981. Volcanism and the dynamics of open magma chambers. Nature 289(5800), 783–785.

Borisova, A., Martel, C., Gouy, S., Indyo Pratomo, I., Sumarti, S., Toutain, J.-P., Bindeman,I.N., de Parseval, P., Metaxian, J.-P., Surono, this issue. Highly explosive 2010Merapi eruption: evidence for shallow-level crustal assimilation and hybrid fluid.Journal of Volcanology and Geothermal Research.

Bouvet de Maisonneuve, C., Dungan, M.A., Bachmann, O., Burgisser, A., 2012. Insights intoshallow magma storage and crystallization at Volcan Llaima (Andean SouthernVolcanic Zone, Chile). Journal of Volcanology and Geothermal Research 211, 76–91.

Camus, G., Gourgaud, A., Mossand-Berthommier, P.C., Vincent, P.-M., 2000. Merapi(Central Java, Indonesia): an outline of the structural and magmatological evolution,with a special emphasis to the major pyroclastic events. Journal of Volcanology andGeothermal Research 100, 139–163.

Chadwick, J.P., 2008. Magma crust interaction in volcanic systems: case studies fromMerapi Volcano, Indonesia, Taupo Volcanic Zone, New Zealand, and Slieve Gullion,N. Ireland: PhD Thesis, Trinity College Dublin, pp. 181.

Chadwick, J.P., Troll, V.R., Ginibre, C., Morgan, D., Gertisser, R., Waight, T.E., Davidson,J.P., 2007. Carbonate assimilation at Merapi Volcano, Java, Indonesia: insightsfrom crystal isotope stratigraphy. Journal of Petrology 48, 1793–1812.

Cherniak, D., Dimanov, A., 2010. Diffusion in pyroxene, mica and amphibole. Diffusionin minerals and melts. Reviews in Mineralogy and Geochemistry 72, 641–690.

Costa, F., Morgan, D., 2010. Time constraints from chemical equilibration in magmaticcrystals. In: Dosseto, A., Turner, S.P., Van Orman, J.A. (Eds.), Timescales of MagmaticProcesses: From Core to Atmosphere (125–159 pp.).

Costa, F., Dohmen, R., Chakraborty, S., 2008. Time scales of magmatic processes frommodeling the zoning patterns of crystals. minerals, inclusions and volcanic pro-cesses. Reviews in Mineralogy and Geochemistry 69, 545–594.




de Genevraye, P., Samuel, L., 1972. The geology of Kendeng Zone (East Java): IndonesianPetroleum Association. Proceedings 1st Annual Convention, pp. 17–30.

Deegan, F.M., Troll, V.R., Freda, C., Misiti, V., Chadwick, J.P., McLeod, C.L., Davidson, J.P.,2010. Magma–carbonate interaction processes and associated CO2 release atMerapi Volcano, Indonesia: insights from experimental petrology. Journal of Pe-trology 51 (5), 1027–1051.

Deegan, F.M., Troll, V.R., Freda, C., Misiti, V., Chadwick, J.P., 2011. Fast and furious:crustal CO2 release at Merapi volcano, Indonesia. Geology Today 27, 63–64.

Deer, W.A., Howie, R.A., Zussman, J., 1992. Rock Forming Minerals. Longman, London,p. 696.

del Marmol, M.A., 1989. The petrology and geochemistry of Merapi Volcano, Central Java,Indonesia. PhD Thesis, The Johns Hopkins University, Baltimore, U.S.A., pp. 384.

Dimanov, A., Sautter, V., 2000. ‘Average’ interdiffusion of (Fe, Mn)–Mg in naturaldiopside. European Journal of Mineralogy 12, 749–760.

Dimanov, A., Windenbeck, M., 2006. (Fe, Mn)–Mg interdiffusion in natural diopside:effect of fO2. European Journal of Mineralogy 18, 705–718.

Dungan, M.A., Davidson, J., 2004. Partial assimilative recycling of the mafic plutonicroots of arc volcanoes: an example from the Chilean Andes. Geology 32, 773–776.

Eichelberger, J.C., 1989. Are extrusive rhyolites produced from permeable foam erup-tions? Bulletin of Volcanology 51, 72–75.

Freda, C., Gaeta, M., Misiti, V., Mollo, S., Dolfi, D., Scarlato, P., 2008. Magma–carbonateinteraction: an experimental study on ultrapotassic rocks from Alban Hills (CentralItaly). Lithos 101, 397–415.

Gertisser, R., Keller, J., 2003. Temporal variations in magma composition at MerapiVolcano (Central Java, Indonesia): magmatic cycles during the past 2000 years ofexplosive activity. Journal of Volcanology and Geothermal Research 123, 1–23.

Gertisser, R., Charbonnier, S.J., Troll, V.R., Keller, J., Preece, K., Chadwick, J.P., Barclay, J.,Herd, R.A., 2011. Merapi (Java, Indonesia): anatomy of a killer volcano. GeologyToday 27, 57–62.

Ghiorso, M., Sack, R., 1995. Chemical mass transfer in magmatic processes IV. A revisedand internally consistent thermodynamic model for the interpolation and extrap-olation of liquid–solid equilibria in magmatic systems at elevated temperaturesand pressures. Contributions to Mineralogy and Petrology 119, 197–212.

Hammer, J.E., Cashman, K.V., Voight, B., 2000. Magmatic processes revealed by texturaland compositional trends in Merapi dome lavas. Journal of Volcanology andGeothermal Research 100, 165–192.

Holland, T., Blundy, J., 1994. Non-ideal interactions in calcic amphiboles and theirbearing on amphibole-plagioclase thermometry. Contributions to Mineralogy andPetrology 116, 433–447.

Iacono Marziano, G., Gaillard, F., Pichavant, M., 2008. Limestone assimilation by basalticmagmas: an experimental re-assessment and application to Italian volcanoes.Contributions to Mineralogy and Petrology 155, 719–738.

Innocenti, S., Andreastuti, S., Furman, T., Voight, B., del Marmol, M.A., this issue-a. Thepre-eruption conditions for effusive and explosive eruptions at Merapi volcano asrevealed by crystal texture and mineralogy. JVGR this issue.

Innocenti, S., del Marmol, M.A., Voight, B., Andreastuti, S., Furman, T. this issue-b. Texturaland mineral chemistry constraint on evolution of Merapi volcano, Indonesia. JVGRthis issue.

Jousset, P., Budi-Santoso, A., Jolly, A.D., Boichu, M., Surono, Dwiyono, S., Sumarti Sri,Hidayati Sri, Thierry, P., this issue. Signs of magmatic ascent in LP and VLP seismic-ity and link to degassing: an example from the 2010 explosive eruption at Merapivolcano, Indonesia. JVGR.

Komorovski, J.-C., Jenkins, S., Baxter, P.J., Picquout, A., Lavigne, F., Charbonnier, S.,Gertisser, R., Noer, C., Budi-Santoso, A., Surono, this issue. Paroxysmal dome explo-sion during the Merapi 2010 eruption: processes and facies relationships of associ-ated high-energy pyroclastic density currents. JVGR.

Landi, P., Metrich, N., Bertagnini, A., Rosi, M., 2004. Dynamics of magma mixing anddegassing recorded in plagioclase at Stromboli (Aeolian Archipelago, Italy). Contri-butions to Mineralogy and Petrology 147, 213–227.

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, C.N., Ungaretti, L., Whittaker, E.J.W., Youzhi, G., 1997. Nomenclatureof amphiboles: report of the Subcommittee on Amphiboles of the InternationalMineralogical Association, Commission on New Minerals and Minerals Names.American Mineralogist 82, 1019–1037.

Luhr, J.F., 1990. Experimental phase relations of water- and sulfur-saturated arc magmasand the 1982 eruptions of El Chichon volcano. Journal of Petrology 31 (5), 1071–1114.

Mangan, M.T., Sisson, T.W., Hankins, W.B., 2004. Decompression experiments identifykinetic controls on explosive silicic eruptions. Geophysical Research Letters 31,L08605. http://dx.doi.org/10.1029/2004GL019509.

Martel, C., Pichavant, M., Holtz, F., Scaillet, B., Bourdier, J.-L., Traineau, H., 1999. Effectsof fO2 and H2O on andesite phase relations between 2 and 4 kbar. Journal ofGeophysical Research 104, 29453–29470.

Michaut, C., Bercovici, D., Sparks, R.S.J., 2009. Ascent and compaction of gas rich magmaand the effects of hysteretic permeability. Earth and Planetary Science Letters 282(1–4), 258–267.

Mollo, S., Gaeta,M., Freda, C., Di Rocco, T.,Misiti, V., Scarlato, P., 2010. Carbonate assimilationinmagmas: a reappraisal based on experimental petrology. Lithos 114 (3–4), 503–514.

Morgan, D.J., Blake, S., Rogers, N.W., DeVivo, B., Rolandi, G., Macdonald, R.,Hawkesworth, J., 2004. Timescales of crystal residence and magma chambervolume from modeling of diffusion profiles in phenocrysts: Vesuvius 1944. Earthand Planetary Science Letters 222, 933–946.

Morimoto, N., 1989. Nomenclature of pyroxenes. Subcommittee on Pyroxenes. Commissionon New Minerals and Mineral Names. The Canadian Mineralogist 27, 143–156.


Neri, A., Papale, P., Macedonio, G., 1998. The role of magma composition and watercontent in explosive eruptions: 2. Pyroclastic dispersion dynamics. Journal ofVolcanology and Geothermal Research 87, 95–115.

Newhall, C., Bronto, S., Alloway, B., Banks, N.G., Bahar, I., del Marmol, M.A.,Hadisantono, R.D., Holcomb, R.T., MCGeehin, J., Miksic, J.N., Rubin, M., Sayudi,S.D., Sukhyar, R., Andreastuti, S., Tilling, R.I., Torley, R., Trimble, D., Wirakusumah,A.D., 2000. 10000 years of explosive eruptions of Merapi Volcano, Central Java:archaeological and modern implications. Journal of Volcanology and GeothermalResearch 100, 9–50.

Pallister, J.S., Schneider, D.J., Griswold, J.P., Keeler, R.H., Burton, W.C., Noyles, C.,Newhall, C.G., Ratdomopurbo, A., this issue. Merapi 2010 eruption—Chronologyand extrusion rates monitored with satellite radar and used in eruption forecast-ing. Journal of Volcanology and Geothermal Research.

Papale, P., Neri, A., Macedonio, G., 1998. The role of magma composition and watercontent in explosive eruptions: 1. Conduit ascent dynamics. Journal of Volcanologyand Geothermal Research 87, 75–93.

Pichavant, M., Martel, C., Bourdier, J.-L., Scaillet, B., 2002. Physical conditions, structureand dynamics of a zoned magma chamber: Mt. Pelée (Martinique, Lesser AntillesArc). Journal of Geophysical Research 107. http://dx.doi.org/10.1029/2001JB000315.

Putirka, K.D., 2008. Thermometers and barometers for volcanic systems. Reviews inMineralogy and Geochemistry 69, 61–120.

Ratdomopurbo, A., Poupinet, G., 2000. An overview of the seismicity of Merapi volcano(Java, Indonesia), 1983–1994. Journal of Volcanology and Geothermal Research100, 193–214.

Ratdomopurbo, A., Beauducel, F., Subandriyo, J., Nandaka, A., Newhall, C.G., Suharna,Sayudi, D.S., Suparwaka, H., Sunarta, this issue. Overview of the 2006 eruption ofMt. Merapi. Journal of Volcanology and Geothermal Research.

Ridolfi, F., Renzulli, A., 2012. Calcic amphiboles in calc-alkaline and alkaline magmas:thermobarometric and chemometric empirical equations valid up to 1130 °C and2.2 GPa. Contributions to Mineralogy and Petrology 163, 877–895.

Rust, A.C., Cashman, K.V., 2004. Permeability of vesicular silicic magma: inertial andhysteresis effects. Earth and Planetary Science Letters 228, 93–107.

Rutherford, M.J., Devine, J.D., 2008. Magmatic conditions and processes in the storagezone of the 2004–2006 Mount St. Helens dacite. Chap. 31 of In: Sherrod, D.R.,Scott, W.E., Stauffer, P.H. (Eds.), A volcano rekindled; the renewed eruption ofMount St. Helens, 2004–2006: USGS Prof. Pap., 1750.

Scaillet, B., Pichavant, M., 2003. Experimental constraints on volatile abundances in arcmagmas and their implications for degassing processes. Geological Society ofLondon, Special Publication 213 (1), 23–52.

Streck, M.J., 2008. Mineral textures and zoning as evidence for open system processes.Minerals, inclusions and volcanic processes. Reviews inMineralogy and Geochemistry69, 595–622.

Streck, M.J., Dungan, M.A., Bussy, F., Malavassi, E., 2005. Mineral inventory of continu-ously erupting basaltic andesites at Arenal volcano, Costa Rica: implications forinterpreting monotonous, crystal-rich, mafic arc stratigraphies. Journal of Volca-nology and Geothermal Research 140, 133–155.

Suharna, Budisantoso, A., Sapari, Djilal, M., 2007. Statistik dan Analisis SeismisitasMerapi 2006. Edisi Khusus Merapi, 2006, Laporan dan Kajian Vulkanisme Erupsi2006, Pusat Vulkanologi dan Mitigasi Bencana Geologi.

Surono, Jousset, P., Pallister, J., Boichu, M., Buongiorno, M.F., Budisantoso, A., Costa, F.,Andreastuti, S., Prata, F., Schneider, D., Clarisse, L., Humaida, H., Sumarti, S.,Bignami, C., Griswold, J., Carn, S., Oppenheimer, C., 2012. The 2010 explosiveeruption of Java's Merapi volcano — a ‘100-year’ event. Journal of Volcanologyand Geothermal Research 241–242, 121–135.

Takeuchi, S., Nakashima, S., Tomiya, A., Shinohara, H., 2005. Experimental constraints onthe low gas permeability of vesicular magma during decompression. GeophysicalResearch Letters 32 (10).

Thornber, C.R., Pallister, J.S., Lowers, H.A., Rowe, M.C., Mandeville, C.W., Meeker, G.P.,2008. Chemistry, mineralogy, and petrology of amphibole in Mount St. Helens2004–2006 dacite. Chap. 32 of In: Sherrod, D.R., Scott, W.E., Stauffer, P.H. (Eds.),A Volcano Rekindled; The Renewed Eruption of Mount St. Helens, 2004–2006:USGS Prof. Pap., 1750.

Tiede, C., Camacho, A.G., Gerstenecker, C., Fernandez, J., Suyanto, I., 2005. Modelingthe density at Merapi volcano area, Indonesia, via the inverse gravimetricproblem. Geochemistry, Geophysics, Geosystems 6. http://dx.doi.org/10.1029/2005GC000986.

Troll, V.R., Hilton, D.R., Jolis, E.M., Chadwick, J.P., Blythe, L.S., Deegan, F.M., Schwarzkopf,L.M., Zimmer, M., 2012. Crustal CO2 liberation during the 2006 eruption and earth-quake events at Merapi volcano, Indonesia. Geophysical Research Letters 39,L11302. http://dx.doi.org/10.1029/2012GL051307, 2012.

Untung, M., Sato, Y., 1978. Gravity and geological studies in Java. Indonesia: GeologicalSurvey of Indonesia and Geological Survey of Japan: Special Publication, 6 (207 pp.).

van Bemmelen, R.W., 1949. The Geology of Indonesia. Govt. Printing Office, Nijhoff, TheHague (732 pp.).

Walter, T.R., Wang, R., Zimmer, M., Grosser, H., Luhr, B., Ratdomopurbo, A., 2007.Volcanic activity influenced by tectonic earthquakes: static and dynamic stresstriggering at Mt. Merapi. Geophysical Research Letters 34 (5), L05304. http://dx.doi.org/10.1029/2006GL028710.

Wilson, L., Sparks, R.S.J., Walker, G.P.L., 1980. Explosive volcanic eruptions — IV. Thecontrol of magma properties and conduit geometry on eruption column behaviour.Geophysical Journal of the Royal Astronomical Society 63, 117–148.

Zhang, X., Ganguly, J., Ito, M., 2010. Ca–Mg diffusion in diopside: tracer and chemicalinter-diffusion coefficients. Contributions toMineralogy and Petrology 159, 175–186.


http://dx.doi.org/10.1029/2004GL019509

http://dx.doi.org/10.1029/2001JB000315

http://dx.doi.org/10.1029/2005GC000986

http://dx.doi.org/10.1029/2005GC000986

http://dx.doi.org/10.1029/2012GL051307, 2012

http://dx.doi.org/10.1029/2006GL028710


Costa_merapi

Documents

Transcript of Costa_merapi