The 2010 explosive eruption of Java’s Merapi volcano - a’100-year’ eventM. Surono, Philippe Jousset, John Pallister, Marie Boichu, Maria FabriziaBuongiorno, Agus Budisantoso, Fidel Costa Rodriguez, Supriyiat Andreastuti,Fred Prata, David Schneider, et al.To cite this version:M. Surono, Philippe Jousset, John Pallister, Marie Boichu, Maria Fabrizia Buongiorno, et al. The2010 explosive eruption of Java’s Merapi volcano - a ’100-year’ event. Journal of Volcanology andGeothermal Research, Elsevier, 2012, 241-242, pp.121-135. 10.1016/j.jvolgeores.2012.06.018 . insu00723412 HAL Id: /insu-00723412Submitted on 10 Aug 2012HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

The 2010 explosive eruption of Java's Merapi volcano – a ‘100year’ event Suronoa1Philippe Joussetbc2John Pallisterd3Marie Boichue45M. Fabrizia Buongiornof6Agus Budisantosogh7Fidel CostaiSupriyati AndreastutiaFred Prataj8David Schneiderk9Lieven Clarissel10Hanik Humaidag6Sri Sumartig6Christian Bignamf5Julie Griswoldd3Simon Carnm11Clive Oppenheimereno412Franck Lavignep a Center of Volcanology and Geological Hazard Mitigation, Jalan Diponegoro 57,40122 Bandung, IndonesiabBRGM, RIS, 3 Avenue Claude Guillemin, BP36009, 45060 Orléans Cedex 2, FrancecNow at GFZ German Research Center in Geosciences, Telegrafenberg, 14473Potsdam, GermanydU.S. Geological Survey, Cascades Volcano Observatory, 1300 SE Cardinal Court,Vancouver, WA 98604, USAeThe University of Cambridge, Department of Geography, Downing Place,Cambridge CB23EN, United KingdomfIsituto Nazionale di Geofisica e Vulcanolgia, Via di Vigna Murata 605, 00143 Rome,ItalygBPPTK (Balai Penyelidikan dan Pengembangan Teknologi Kegunungapian), JalanCendana 15, Yogyakarta 55166, IndonesiahISTerre, CNRS, Université de Savoie, 73376 Le Bourget du Lac cedex, FranceiEarth Observatory of Singapore, Nanyang Technological University N2-01a-10,Singapore 639798jClimate and Atmosphere Department, Norwegian Institute for Air Research, PO Box100, Kjeller, 2027, NorwaykU.S. Geological Survey, Alaska Volcano Observatory, 4230 University Drive,Anchorage, AK 99508 USAlUniversité Libre de Bruxelle, Unité de Chimie Quantique et Photophysique, Campusdu Solbosch, CP160/09, Avenue F.D. Roosevelt 50, 1050 Bruxelles, BelgiummMTU: Department of Geological/Mining Engineering & Sciences, 1400 TownsendDrive, Houghton MI 49931 USAnLe Studium, Institute for Advanced Studies, Orléans and Tours, FranceoL'Institut des Sciences de la Terre d'Orléans, l'Université d'Orléans, 1a rue de laFérollerie, 45071 Orléans, Cedex 2, France

pLaboratoire de Géographie Physique, 1 Place A. Briand, 92195 Meudon Cedex,FranceAbstractMerapi volcano (Indonesia) is one of the most active and hazardous volcanoes in the world. Itis known for frequent small to moderate eruptions, pyroclastic flows produced by lava domecollapse, and the large population settled on and around the flanks of the volcano that is atrisk. Its usual behaviour for the last decades abruptly changed in late October and earlyNovember 2010, when the volcano produced its largest and most explosive eruptions in morethan a century, displacing a third of a million people, and claiming nearly 400 lives. Despitethe challenges involved in forecasting this „hundred year eruption‟, we show that themagnitude of precursory signals (seismicity, ground deformation, gas emissions) wereproportional to the large size and intensity of the eruption. In addition and for the first time,near-real-time satellite radar imagery played an equal role with seismic, geodetic, and gasobservations in monitoring eruptive activity during a major volcanic crisis. The IndonesianCenter of Volcanology and Geological Hazard Mitigation (CVGHM) issued timely forecastsof the magnitude of the eruption phases, saving 10,000–20,000 lives. In addition to reportingon aspects of the crisis management, we report the first synthesis of scientific observations ofthe eruption. Our monitoring and petrologic data show that the 2010 eruption was fed by rapidascent of magma from depths ranging from 5 to 30 km. Magma reached the surface withvariable gas content resulting in alternating explosive and rapid effusive eruptions, andreleased a total of 0.44 Tg of SO2. The eruptive behaviour seems also related to theseismicity along a tectonic fault more than 40 km from the volcano, highlighting both thecomplex stress pattern of the Merapi region of Java and the role of magmatic pressurization inactivating regional faults. We suggest a dynamic triggering of the main explosions on 3 and 4November by the passing seismic waves generated by regional earthquakes on these days.Highlights First scientific results from largest eruption in 100 years of Merapi volcano. Gasemissions were much higher than recorded at Merapi during past eruptions. Deep influx ofgas-rich mafic magma triggered the eruption. Presence of an exsolved fluid phase coexistent with the pre-eruptive magma body. Eruption warnings by CVGHM andinternational team saved 10,000-20,000 lives.Keywords Merapi;gas emissions;satellite nternational collaboration

1. IntroductionMerapi stratovolcano is located 25–30 km north of the metropolitan area of Yogyakarta,Indonesia (Fig. 1) and the environs are home to around of 1.6 million people. It overlies theJava subduction zone and is composed mainly of basaltic-andesite tephra, pyroclastic flow,lava, and lahar deposits. Eruptions during the twentieth century typically recurred every 4 to6 years and produced viscous lava domes that collapsed to form pyroclastic flows andsubsequent lahars. These eruptions were relatively small, with typical eruptive volumes of 1–4 106 m3 and magnitudes or volcanic explosivity indices (VEI) of 1–3 ( [Andreastuti et al.,2000], [Camus et al., 2000], [Newhall et al., 2000], [Voight et al., 2000a] and [Voight et al.,2000b]), where magnitude (Pyle, 2000) is given by [Me log10(mass of products in kg) – 7].Merapi volcano has been studied extensively by Indonesian and international teams, leadingto improved understanding of the volcano's seismology ( [Hidayat et al., 2000],[Ratdomopurbo and Poupinet, 2000] and [Senschönefelder and Wegler, 2006]), deformation ([Beauducel and Cornet, 1999], [Voight et al., 2000a], [Voight et al., 2000b] and [Young et al.,2000]), potential field geophysics ( [Jousset et al., 2000], [Zlotnicki et al., 2000] and [Tiede etal., 2005]), gas emissions ( [Le Guern and Bernard, 1982], [Nho et al., 1996], [Zimmer andErzinger, 2003], [Humaida et al., 2007], [Toutain et al., 2009] and [Allard et al., 2011]),petrology ( [Gertisser and Keller, 2002], [Gertisser and Keller, 2003], [Chadwick et al., 2007],[Deegan et al., 2010] and [Deegan et al., 2011]), physical volcanology (Charbonnier andGertisser, 2008) and lahar inundation (Lavigne et al., 2000). Merapi's high-temperature(400 – 850 C) summit fumaroles, continuous gas emissions, and frequent small eruptionsindicate an open and hot pathway for magma ascent to the near-surface. At the summit ventlevel, lava domes have typically plugged the uppermost part of the conduit except duringeruptions when magmatic pressure built and new domes composed of mostly degassedmagma extruded and collapsed or much more infrequently, gas-rich explosive eruptionsoccurred.

Fig. 1. Index map showing location of Merapi volcano summit and other features referred to in thetext, e.g., observatory post stations (“Pos” in Indonesian), the Merapi Observatory and TechnologyCenter (BPPTK), major drainages (abbreviated “K.” for “Kali” in Indonesian), short-period permanentseismic stations (full inverted triangles, PUS, DEL,PLA, KLA), temporary broadband stations (emptyinverted triangles, LBH, GMR, GRW, PAS, L56 WOR at summit). Cities and towns are indicated byname. In addition, hundreds of smaller villages are present on the flanks of the volcano. Majorhighways are indicated by heavy dashed-dotted lines and the read arcs at 10, 15, and 20 km radiusdistances from the summit indicate evacuation zones that were put into effect at different timesduring the eruptive activity (see text for details).The lack of large explosive eruptions at Merapi during the several decades preceding 2010 isattributed to extensive degassing during ascent of the magma through the volcano's subsurfaceplumbing system (Le Cloarec and Gauthier, 2003). However, stratigraphic evidence showsthat large explosive eruptions, such as the one that took place in 1872 (Hartmann, 1934) alsooccur. Because of the relatively open-pathway for magma ascent and the lack of explosiveeruptions in the recent past, it was feared that precursors to such a large eruption might only

be modest and inadequately appreciated. The increasing population on the volcano flanksmeant that a large eruption could result in tens to hundreds of thousands of casualties.Fortunately, although of short duration and rapidly escalating, large-magnitude precursorswere recognized and identified in time to issue warnings for the impending large 2010eruption, which had a VEI and Me of about 4.We report on the monitoring techniques, data, and warning issues that came into play andwere gathered during the 2010 eruptive sequence. Main explosive events occurred on 26October ( 10:00 utc), 29 October ( 17:10–19:00 utc), 31 October ( 7:30 and 8:15 utc), 1November ( 3:00 utc), 3 November ( 8:30 utc), 4 November (17:05 utc). We use acombination of petrologic, seismologic, geodetic, and gas emission data, along with remotelysensed observations of changes in morphology and eruption rate to propose a preliminarymodel for this „100-year‟ eruption.In Section 2, we describe technical details of both “traditional” monitoring methods used atMerapi volcano and “state-of-the-art” satellite observations, extensively used during the 2010eruption. In Section 3, we describe the chronology of the eruption and how our geophysicaland satellite observations were interpreted, leading to timely warnings that saved 10,000–20,000 lives. In Section 4, a preliminary eruption model is proposed, based on our analysis ofthe available monitoring signals and petrological data. Finally, we suggest that themanagement and decision-making during the crisis was successful thanks to a combination oflong-term in-country expertise in dealing with volcanic crises and an unprecedented level ofinternational collaboration. We conclude in summarising observations and interpretations onthe eruption dynamics and propose a series of questions that need to be addressed for a betterunderstanding of Merapi's most explosive eruption of the past 100 years.2. Observational methods used during the 2010 Merapi eruptionMerapi has long been monitored using seismology, deformation, gas emission studies andpetrology (Purbawinata et al., 1996) by CVGHM and its observatory and technology center inYogyakarta (Balai Penyelidikan dan Pengembangan Teknologi Kegunungapian, or BPPTK).Under non-eruptive conditions, the rate of inflation/deflation (measured as change in lengthsof Electronic Distance Measurement (EDM) lines between the volcano's summit and flanks)is 0.003 m d- 1; the cumulative seismic energy release is less than 35 MJ d- 1 with dailyaverages of 5 multiphase earthquakes and 1 volcano-tectonic earthquake; the baseline SO2flux is 50–100 Mg d- 1 ( [Nho et al., 1996] and [Humaida et al., 2007]), and the long-termeruption rate is 1.2 106 m3 y- 1 (Siswowidjoyo et al., 1995).2.1. GeodesyDeformation was measured using both tiltmeters near the summit and an Electronic DistanceMeasurement (EDM) network. The Electronic Distance Measurement (EDM) networkutilized reflectors at high elevations on all flanks and measurements were carried out fromfive observation posts (Jrakah, Babadan, Selo, Kaliurang, and Ngepos) at distances of 5–10 km from the summit of Merapi.2.2. SeismologySeismic monitoring and analysis were carried out in real time and used qualitatively duringthe crisis to infer magmatic and eruptive processes. Earthquake activity was monitored with

four short-period (Mark Products L-4 seismometers) permanent stations (PUS, KLA, DEL,and PLA, Fig. 1) and a real-time temporary broadband seismological network of five stations:one Streikesen STS-2 (station LBH) and four Güralp CMG40T sensors (stations GMR, GRW,PAS, WOR) from July 2009 to September 2010, and then station L56 from September 2010).Seismometers installed in July 2009 were part of the MIAVITA (MItigate and Assess riskfrom Volcanic Impact on Terrain and human Activities) European research project (Thierry etal., 2008). Technical problems including poor synchronization (lack of GPS signal) preventeda full analysis in real-time at some stations (GMR, L56, LBH).The seismicity at Merapi volcano during the 2010 crisis revealed that all types of earthquakespreviously identified at Merapi (Ratdomopurbo and Poupinet, 2000) were represented in the2010 activity (Budi-Santoso et al., this issue; Jousset et al., this issue): Volcano-Tectonic (VT)earthquakes, Low-Frequency earthquakes (LF), tremor, “Multiphase” earthquakes (MP),“guguran” rock falls (RF), and Very-Long Period events (VLP). Real Time SeismicAmplitude (RSAM) data ( [Murray and Endo, 1992] and [Endo and Murray, 1999]) played acrucial role in evaluating the status of the volcano activity during the eruptive sequence. Also,as part of the MIAVITA project, a seismic station (CRM) was set-up at about 46 km southfrom the summit close to the Opak fault, source for a M6.3 earthquake that killed more than6000 people during the prior eruption of Merapi volcano in 2006. During 4 November,stations PUS, KLA, and DEL, L56 and PAS were destroyed, and the remaining PLA station(at 6 km) was saturated ( 0.025 mm/s). Consequently, seismic amplitude observations at theCRM station were crucial during the climactic phase on 4 November (see Section 3).Although close range stations have been critical for warnings and research during past smalleruptions at Merapi, this eruption clearly illustrates the value of including distal as well asproximal stations in volcano monitoring networks.To locate events, we performed seismic analysis using the STA/LTA (Short-termAverage/Long-term Average) detection technique and picked P-phases (and when possible Sphases) using an algorithm which includes an estimation of picking uncertainty (e.g., Joussetet al., 2011). We located VT earthquakes using both a linear (Lahr, 1999) and a non-linearlocation iterative technique, which searches for the best fit between observed (picked) traveltimes and synthetic travel times. The latter are computed at regularly distributed points on a3D-grid in the volcanic edifice, where velocity and density are parameterized. Computation isperformed first with a coarse grid and subsequent iterations use a refined grid set-up aroundthe hypocenter location found at the first iteration, and a volume defined by the 68%confidence level surface (e.g., Jousset et al., 2011). This method allows a fast hypocentrecomputation and can be implemented in real-time. Unfortunately, synchronization problemsprevented us from implementing this technique in real-time during Merapi's eruption.Hypocentre positions were calculated as soon as possible after the eruption. Hypocentrepositions are affected by lack of a detailed velocity model for shallow levels of the crust atMerapi ( [Wegler and Luehr, 2001], [Wagner et al., 2007] and [Kulakov et al., 2009]). Theyare located along the length of the conduit down to 8 km below the summit. The frequencycontent of records has been analysed through a variety of signal processing tools and methods(e.g., Lesage, 2009), including Fast Fourier Transform (FFT), complex frequency analysis(Sompi method, e.g., Kumagai et al., 2010), and particle motion analysis.2.3. Satellite SAR, visible, and near-visible imageryA variety of satellite data were utilized including commercial Synthetic Aperture Radar(SAR) from the COSMO SkyMed RADARSAT-2, TerraSAR-X sensors, and when weather

and orbits permitted, thermal infrared from the ASTER sensor and high-resolution visible andnear-infrared data from the GeoEye 1 and WorldView-2 sensors. Cloud cover limitedexploitation of data from optical sensors. However, the radar satellites supplied frequent anddetailed images of the volcano summit crater, rapidly growing lava domes, vent features, andpyroclastic flow deposits (including that of the large flow emplaced on 4 November thatextended towards Yogyakarta; see Section 3). Despite cloud cover, the pyroclastic flow of 26October was also detected by ASTER thermal sensor on 1 November. Images were availablefor analysis by both volcanologists at the USGS Alaska and Cascades Volcano Observatoriesand the Instituto Nazionale di Geofisica e Vulcanologia (INGV) in Italy, typically within 2–6 hours of acquisition, and critical data and analyses were delivered to CVGHM within thesame time periods each day or in some cases twice a day during the crisis. The commercialSAR data were collected with horizontal polarization and with beam resolutions that variedfrom 1–8 m, depending on acquisition mode.2.4. Gas measurementsIn-situ monitoring of volcanic gas emissions (H2O, SO2, CO2, H2S, CO, HCl, H2, O2, andCH4) was carried out by regularly collecting samples from the Woro solfatara. Sampling wasdone by bubbling the gas through NaOH solutions contained in evacuated flasks (Giggenbachand Goguel, 1988). Measurement of insoluble gas in the NaOH solution was carried out bygas chromatography. The dissolved gases were analyzed using spectrometric and volumetricmethods.Ground-based ultra-violet (UV) Differential Optical Absorption Spectroscopy (DOAS)measurements (Galle et al., 2003) proved highly challenging during the eruption because awide area around the volcano was inaccessible (due to the exclusion zone), the plume wasash-rich, and the weather adverse (high humidity and frequent rainfall). Nevertheless, acombination of gas and ash plume remote sensing from the ground and satellites providedcrucial information on degassing during the entire 2010 crisis. Satellite data were especiallyimportant during the most explosive phases of eruption, as they provided measurements ofSO2 emissions and maps of volcanic cloud dispersal, which were used to issue advisories foraviation hazard mitigation by the Volcanic Ash Advisory Centre (VAAC) at Darwin,Australia.Whenever possible, DOAS observations were carried out from Babadan, Ketep andYogyakarta, which are 4, 9, and 28 km from the crater, respectively. Ocean Optics USB2000spectrometers were used spanning a wavelength range of 288–434 nm with a Full WidthHalf Maximum (FWHM) spectral resolution of 0.60 nm. Spectrometers were coupled to asimple quartz-lens telescope mounted on a rotating platform, which enabled scanning ofvertically rising plumes, except on 4 November where the telescope was held in a fixedposition and pointed towards the dense plume. Each UV spectrum was recorded with a totalintegration time of a few seconds. Plume rise speeds were determined from video images,allowing an estimation of the SO2 emission rates. The true SO2 flux was under-estimatedwhen the plume was ash-rich due primarily to hindered UV transmission through the denseplume (especially on 4 and 12 Nov).SO2 burdens in the plume were available daily from satellites, utilizing the infrared (IR) IASIsensor (Infrared Atmospheric Sounding Interferometer, Clarisse et al., 2008) with overpassesat 9:30 AM and 9:30 PM local time, and every 24 h from the UV OMI sensor (OzoneMonitoring Instrument, Carn et al., 2008) with overpasses at 1:30–2:00 PM local time.

Sparse data from the AIRS sensor (Atmospheric Infrared Sounder, Prata and Bernardo, 2007),with overpasses at 1:30 AM and 1:30 PM local time, were also available during theparoxysmal phase. OMI is able to detect SO2 emissions in the lower troposphere whereasIASI and AIRS are restricted to SO2 in the upper troposphere (above 5 km altitude) orhigher, where most plumes traveled during the explosive phases of the eruption. Forsimplicity in IASI and AIRS retrievals, we assumed a plume altitude of 16 km during theentire eruption. Plume altitudes reported by the Darwin VAAC were used to assign theappropriate SO2 altitude for OMI retrievals ( 17 km for 4–5 November, and altitudes in the 5–8 km range after 5 November). Subtracting the SO2 burdens from two consecutive imagesallowed us to evaluate a mean SO2 flux (on 12 or 24 h depending on the sensor), assumingnegligible SO2 depletion in the plume. The OMI detection limit is roughly evaluated at 200 Mg d- 1, based on estimations of the SO2 flux from ground DOAS measurements.Fluxes can be under-estimated when the satellite swath does not span the entire plume, so werestrict our evaluation of fluxes to cases when satellite swaths intersected most of the volcaniccloud. Unfortunately, the presence of a dispersed aged plume in images from 5 to 9 Novemberimpeded accurate estimation of new SO2 emissions from the volcano using IASI images.However, analysis of the area immediately downwind of Merapi with OMI data permittedestimation of SO2 release from new emissions during this period. Prior to 5 November andafter 9 November, IASI could not detect any SO2 emissions, probably due to the low altitudeof the plume.2.5. Petrological methods and electron microprobe analysesSamples were observed first with the optical microscope using reflected and transmitted lightand modes counted. Textures and grain sizes and relations between minerals were recorded.Minerals and glass were analysed for Si, Al, Ti, Fe, Mn, Mg, Na, K, F, Cl, and S in polishedsections using a JEOL-JXA-8530 F electron microprobe (EM) at the Nanyang TechnologicalUniversity (Singapore) using wavelength dispersive spectrometers. An accelerating voltage of15 kV, current of 15 nA, and spot size of about 1 μm was used for mineral analyses. For glassthe current was decreased to 10 nA, and spot sizes increased to 5 to 10 mm. Na and K werealways counted first. Counting times were 10 s peaks and 5 s on backgrounds for the majorelements, and up to 120 s for peaks and 60 s for backgrounds for S. Backscattered electronimages, and X-ray distribution maps were also obtained with the EM. Standards used in thecalibration were minerals from Astimex (albite, garnet, rutile, pyrite, olivine, sanidine,diopside, celestite, fluorite, biotite, rhodonite, and tugtupite). The calibration was checkedagainst an in-house dacite glass standard analysed by X-ray fluorescence. Precisions varyaccording to concentration: major elements have 2-sigma precisions of 0.5–1%; precisions forminor elements are 5–10%.3. 2010 eruption: monitoring, chronology, warnings, and impacts3.1. Alert levels at Merapi volcanoThe early warning system at Merapi is the same as at all volcanoes in Indonesia and is basedon the analysis of instrumental and visual observations. It comprises 4 alert levels: Level Iindicates the activity of the volcano is in normal state, with no indication of increasingactivity, although poisonous gases may threaten the area close to the vent or crater. Level II isset when visual and seismic data indicate that the activity is increasing. Level III is set when atrend of increasing unrest is continuing and there is concern that a dangerous eruption mayoccur. Level IV is set when the initial eruption starts (i.e., ash/vapor erupts which may lead to

a larger and more dangerous eruption). The alert level is declared to the public throughNational Agency for Disaster Management (BNPB) and the local governments. For eachlevel, CVGHM gives recommendations for what the people living around the volcano aresupposed to do. However, orders to the public such as evacuation orders are given by BNPBand local governments, which also organize evacuations.3.2. Intrusion phase (31 October 2009 – 26 October, 2010)EDM (Electronic Distance Measurement) data provided some of the earliest signs ofprecursory unrest in November 2009, when an extended period of deflation that followed the2006 eruption reversed to inflation. Early indications of increased seismic activity includedswarms of volcano-tectonic (VT) earthquakes on 31 October 2009, 6 December 2009, and 10June 2010. In September 2010, marked increases in ground inflation (Fig. 2), earthquakecounts and seismic energy release (Fig. 3), temperature, CO2, and H2S abundances of summitfumaroles (Table 1) were observed. Based on these changes, on 20 September 2010, CVGHMraised the alert from level I (normal background conditions) to level II (increased activity) inanticipation of what many expected to be another small to moderate size eruption.

Fig. 2. Electronic Distance Measurement data for lines between observatory posts and the summit ofMerapi (see Fig. 1). Reflectors near the summit of the volcano were destroyed by the eruption of 26October, preventing further observations. Shortening of EDM lines between the volcano's summitand flanks is indicative of pressurization and inflation of the upper part of the volcano with magma,whereas increasing distances indicates deflation. (a) EDM observations for 3 lines Babadan-summit(West) Jrakah-summit (North) and Kaliurang-summit (South). "Relative Distance" refers to the changein line length with respect to time, reference taken arbitrarily on 1 September 2010. (b) Detail of theKaliurang-summit EDM line, and displacement rate.

Fig. 3. (a) Dayly count of the seismicity recorded at Merapi during 2010 eruption. VT Volcanotectonic; MP Multiphase ( Hybrid earthquake); LF low-frequency; Rockf Rockfall earthquakes;Pyroclastic F Pyroclastic flows; RSAM Real-time Seismic Amplitude Measurement. (b) Location ofearthquake prior and during the eruption.

Table 1. Major-element analyses of juvenile components from pyroclastic flows fromMerapi volcano and gas analyses from the summit Woro fumarole field. n.d. notdetected.* All Fe reported as Fe2O3.** Average (avg.) and standard deviation (s.d.) ofanalyses from 1954, 1957, 1992, 1994 and 1998 of Gertisser and Keller (2003).***H2 O2. Fumarole gas analyses are individual samples on 26 May and 20 October.September averages are for 3 samples analyses. On May analysis, peaks of H2 andO2 Ar can be separated; since August 2010, H2 O2 are analyzed together.1954-1998⁎⁎Year: 2010 2006SiO2 55.8Al2O3 19.2Fe2O3⁎ 7.78MgO 2.33CaO 8.27Na2O 3.90K2O 2.16TiO2 0.74P2O5 0.32MnO 9.553.051.980.880.370.212010 fumarole gas analyses (% mo l)26 May Sept. avg. 20-Oct. 20-Oct.T ( C) 460575575575⁎⁎⁎⁎⁎⁎H2 O2 n.d.n.d.NH30. 71011524CO2/H2S 28221413CO2/HCl 282858125CO2/H2O

The period from 20 September until the initial explosive eruption on 26 October was markedby a dramatic increase in all monitored parameters (Aisyah et al., 2010; Fig. 2,Fig. 3 and Fig. 9). No localized deformation on the northern flank was detected by theNorthern EDM lines. On the contrary the rate of shortening of the line between the summitand south flank of the volcano (indicative of summit inflation) followed an exponential trendfrom 10 mm d- 1 in early September to 500 mm d- 1 just before the eruption on 26 October.The resulting cumulative shortening was 3 m (Fig. 2). Typically preceding eruptions ofMerapi there is significant shortening of EDM lines on the south side of the volcano whileEDM lines on the north side show little change. Consequently, it is generally thought that thenorth side of the volcano is effectively buttressed by the adjacent northern volcano, Merbabu.Prior to the 26 October eruption, however, the seismicity rate increased and SO2 fluxesreached levels comparable to the highest rates observed during past Merapi eruptions (from1992 to 2007) (Fig. 3 and Fig. 9). A remarkable increase in CO2/SO2 and H2S/SO2 ratios wasdetected in fumarole gas composition between the end of September and 20 October(Table 1). The number of both volcano-tectonic (VT) earthquakes corresponding to shearfracturing in the edifice and multiphase events (MP, also called “hybrid” earthquakes)corresponding to magma movement increased exponentially in October 2010 (Fig. 3). Besidesthe sharp increase of VT and MP events, the number and magnitude of rock falls (RF) alsointensified prior to the eruption. From 1 to 18 October, more than 200 very-long-period (VLP)signals were recorded at summit stations, with some large VLP events recorded at allbroadband stations (Jousset et al., this issue).Compared to previous eruptions, the greater frequency of earthquakes, the amplitude ofreleased seismic energy, the rapid and large deformation (from EDM), and significant gasemissions implicated a larger volume of magma than seen in the past decades of Merapi'sepisodic activity. During this period of rapid escalation, on 21 October CVGHM raised thealert from level II to III (indicating a much higher level of unrest and increased likelihood oferuption). On 25 October at 18:00 local time, after seismicity and deformation increased tounprecedented levels, the alert was raised to its highest level IV and CVGHM warn

l Université Libre de Bruxelle, Unité de Chimie Quantique et Photophysique, Campus du Solbosch, CP160/09, Avenue F.D. Roosevelt 50, 1050 Bruxelles, Belgium m MTU: Department of Geological/Mining Engineering & Sciences, 1400 Townsend Drive, Houghton MI 49931 USA n Le Studium, Institute for Adv