Gas and ash emissions associated with the 2010–present activity of Sinabung Volcano, Indonesia

Accepted Manuscript Gas and ash emissions associated with the 2010–present activity of Sinabung Volcano, Indonesia

Sofyan Primulyana, Christoph Kern, Allan Lerner, Ugan B. Saing, Syegi L. Kunrat, Hilma Alfianti, Mitha Marlia PII: DOI: Reference:

S0377-0273(17)30691-1 doi:10.1016/j.jvolgeores.2017.11.018 VOLGEO 6245

To appear in:

Journal of Volcanology and Geothermal Research

Received date: Revised date: Accepted date:

26 September 2016 4 June 2017 20 November 2017

Please cite this article as: Sofyan Primulyana, Christoph Kern, Allan Lerner, Ugan B. Saing, Syegi L. Kunrat, Hilma Alfianti, Mitha Marlia , Gas and ash emissions associated with the 2010–present activity of Sinabung Volcano, Indonesia. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Volgeo(2017), doi:10.1016/j.jvolgeores.2017.11.018

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Gas and ash emissions associated with the 2010 - present activity of Sinabung Volcano, Indonesia Sofyan Primulyana1, Christoph Kern2, Allan Lerner3, Ugan B. Saing1, Syegi L. Kunrat1,4, Hilma Alfianti1, and Mitha Marlia1.

1

IP

T

Center for Volcanology and Geological Hazard Mitigation, Geological Agency, Ministry of Energy and Mineral Resources, Jalan Diponegoro No. 57, Bandung, West Java, 40122 Indonesia 2

CR

Volcano Disaster Assistance Program, US Geological Survey, 1300 SE Cardinal Court, Vancouver, WA, 98683 U.S.A. 3

US

Department of Earth Sciences, University of Oregon, 1275 E 13th Ave, Eugene, Oregon 97403 U.S.A. 4

M

AN

Department of Geology, Portland State University, P.O. Box 751, Portland, Oregon 97207, U.S.A.

Abstract:

ED

Sinabung Volcano (Sumatra, Indonesia) awoke from over 1,200 years of dormancy with multiple

PT

phreatic explosions in 2010. After a period of quiescence, Sinabung activity resumed in 2013, producing frequent explosions, lava dome extrusion, and pyroclastic flows from dome and lava

CE

flow collapses, becoming one of the world's most active volcanoes and displacing over 20,000

AC

citizens. This study presents a compilation of the geochemical datasets collected by the Indonesian Center for Volcanology and Geological Hazard Mitigation (CVGHM) from 2010 current (2016), which provides insights into the evolution of the eruption. Based on observations of SO2 emissions, ash componentry, ash leachate chemistry, and bulk ash geochemistry, the eruption can be split into six distinct geochemical phases. The initial stage of phreatic summit explosions (phase A) occurred from August - September 2010, during which background SO2 emissions averaged ~550 ± 180 t/d (1 s.d.). An eruptive pause (phase B) starting in October 2010 1

ACCEPTED MANUSCRIPT abruptly ended in September 2013 with a resumption of conduit-clearing eruptions (phase C). This third phase had a relatively modest background SO2 emission rate (avg. ~430 ± 310 t/d) and produced ash consisting of accidental ejecta with high S/Cl leachate molar ratios (12.0 ± 8.2, max 34), suggestive of deep-sourced magma and/or the incorporation of hydrothermal sulfur-

T

bearing phases. Lava extrusion at the summit (phase D) began in mid-December 2013, and was

IP

accompanied by relatively low SO2 emission rates (360 ± 200 t/d) and lower, but variable, S/Cl

CR

leachate ratios (6.3 ± 8.5). The most intense phase of the eruption (phase E) occurred from midJanuary to late February 2014 following a major lava dome collapse. This period included

US

increased lava extrusion rates, dozens of large eruptions per day, high SO2 emission rates

AN

(average: 1,680 ± 1,070 t/d, peak: ~3,800 t/d), and a dramatic drop in S/Cl ash leachates to ratios (average 1.4 ± 0.5), consistent with increased degassing from shallow magma and the clearing

M

out of sulfurous phases from the old hydrothermal system. From March 2014 through the time of

ED

writing (September 2016), Sinabung settled into a relatively steady state of lower activity (phase F). Ash emissions now consist of dominantly juvenile andesitic-dacitic material with low S/Cl

PT

leachate ratios (average 1.1 ± 0.6). In August 2016, SO2 emissions started being measured in a

CE

continuous manner using a network of permanent scanning DOAS instruments. Background SO2 emission rates average 450 ± 290 t/d for the entirety of phase F, but have been progressively

AC

decreasing to an average of ~250 - 300 t/d since June 2016. This long-term gradual decline in SO2 emission rates at Sinabung since early 2014 is consistent with an apparent decrease in magma supply and lava effusion rates. Our conceptual degassing model suggests that large explosions and pyroclastic flows could continue in the near-term owing to conduit plugging and dome collapses, remaining a major threat until the magma supply rate decreases further and the eruption ends.

2

ACCEPTED MANUSCRIPT

Introduction Sinabung Volcano is a 2,460 m high, ~5 km diameter composite volcano located in North Sumatra, Indonesia, 40 km northwest of the Lake Toba caldera (Fig. 1). On August 27, 2010,

T

Sinabung experienced a phreatic eruption, the first eruption in over 1,200 years of dormancy

IP

(Iguchi et al., 2012). The Indonesian Center for Volcanology and Geological Hazard Mitigation

CR

(CVGHM) immediately issued a warning, and a second explosion occurred two days later on August 29. This led to the evacuation of over 22,000 people residing at distances of less than

US

6 km from the summit of the volcano. Several additional eruptions ensued, some sending ash up

AN

to 5 – 7 km above the summit.

Phreatic activity ceased in September 2010, and the volcano once again became relatively quiet.

M

Local populations were allowed to return to their homes. However, CVGHM monitoring efforts

ED

ramped up during this period. Sinabung was not monitored prior to 2010 due to its long dormancy, but after the phreatic events, CVGHM installed seismometers, GPS instruments, and

PT

electronic distance measuring benchmarks around the edifice (Gunawan et al., this volume). A

CE

local observatory (pos) was set up near the town of Kabanjahe to monitor activity and interact with the local population.

AC

After three years of quiescence, explosions began occurring again at Sinabung in September, 2013. These explosions eventually sent ash columns as high as 7 - 9 km above the volcano’s summit and produced pyroclastic density currents reaching up to 5 km from the vent. Unfortunately, these flows claimed the lives of 23 people (Associated Press 2014, 2016) who approached the volcano too closely, venturing within the exclusion zones defined by CVGHM. Evacuations ultimately affected over 30,000 citizens (Otniel Ketaren et al., 2016), with entire

3

ACCEPTED MANUSCRIPT villages being permanently relocated as it became recognized that Sinabung’s eruption was likely to continue for multiple years. Sulfur dioxide (SO2) measurements were performed intermittently throughout the period of unrest using temporary scanning Differential Optical Absorption Spectroscopy (DOAS)

T

instrumentation. Direct sampling of gas compositions was impossible due to the hazard

IP

associated with approaching the vent. However, a sequence of erupted ash was collected since

CR

the onset of activity in 2013, and analyzed for bulk composition and adsorbed leachates that were scavenged during ash residence time in eruption gas plumes. Most recently, in August 2016,

US

scanning DOAS instrumentation was installed as part of the Network for Observation of

AN

Volcanic and Atmospheric Change (NOVAC). Here, we report on the results of these geochemical measurements and describe the evolution of the gas and ash emissions during the

ED

M

2010 - current (September 2016) activity at Sinabung Volcano.

PT

Methods Scanning DOAS campaigns

CE

Starting in 2013, SO2 emission rate measurements were performed intermittently using a portable scanning DOAS instrument. This remote sensing technique measures the column

AC

density (in molecules·cm-2 or ppm·m) of SO2 in a cross-section of the volcanic plume by analyzing incident ultraviolet (UV) radiation for absorption by SO2. The SO2 emission rate is then obtained from the product of the SO2 column densities integrated across the plume crosssection and multiplied by the plume speed (Galle et al, 2002). The employed campaign DOAS instrument was based upon an Ocean Optics USB2000 spectrometer with a spectral range of 245 – 385 nm and a spectral resolution of 1.1 nm. The DOAS was equipped with four quartz-glass cells containing known SO2 concentrations between 4

ACCEPTED MANUSCRIPT 82 ppm·m and 980 ppm·m. These cells were used for calibration before and after the measurements were made. The spectrometer and scanner were mounted on a stationary tripod. Measurement location and scan angle were chosen such that the scanning plane was nearly perpendicular to the plume propagation direction (Fig. 2). Scanner alignment ranged from

T

horizontal (to transect buoyant vertical plumes) to nearly vertical scans (to transect horizontal,

IP

i.e. bent, plumes). The azimuthal and scanner elevation angles were respectively measured using

CR

a compass and an inclinometer.

The plume speed is needed for retrieval of the SO2 emission rate. For analysis of the campaign

US

DOAS measurements, the plume speed was measured by recording continuous video footage of

AN

the plume, then tracking individual plume features over time. By estimating the distance to the gas plumes (taken from the GPS location and a map), the plume speed can be calculated

M

(Williams-Jones et al., 2008).

ED

Errors in scanning DOAS measurements have been discussed in detail elsewhere, and are only touched upon briefly here. The main sources of uncertainty in DOAS flux measurements are

PT

generally considered to be linked to light scattering processes (Mori et al., 2006; Kern et al.,

CE

2009) and to the error in the plume speed estimation (Williams-Jones et al., 2008; Nadeau and Williams-Jones, 2009). Since we were able to derive our wind speed using video imagery, we

AC

estimate this error to be relatively small (< 20%). The errors induced by unknown radiative transfer effects are more difficult to judge but, based on the case studies performed by Kern et al. (2009) and the fact that our campaign measurements were performed at distances of up to 5 km, it is possible that we are underreporting the SO2 emission rate by up to about 40% at times. Our reported values therefore likely represent a lower limit of the true SO2 emissions (Kern et al., 2009).

5

ACCEPTED MANUSCRIPT During each day that the campaign scanning DOAS instrument was deployed, measurements were made for several hours and typically quantified the level of background degassing. Unfortunately, generally very few of the campaign DOAS measurements captured degassing associated with individual explosions. As such, the data cannot be used to calculate the total SO2

T

emissions over time, but do provide constraints on typical passive, inter-eruptive degassing

CR

IP

during different phases of the eruption.

Ash sampling

US

Volcanic ash was collected starting in September 2013, when continuous volcanic activity first

AN

began. Ash was collected at various locations around the volcano, depending on prevailing wind direction. Whenever possible, tephra was collected immediately following observed explosions.

M

Care was taken to avoid contamination by soil and exposure to rain. Tephra was mostly collected

ED

using plastic sheets (Witham et al., 2005), though in some cases dry ash was also sampled from plant leaves. We consider the selected samples representative of fresh‐fallen ash, and overall

PT

uncontaminated and unaltered by postdeposition processes. Rare cases in which non-volcanic

CE

material (e.g., rust flakes) became mixed in with volcanic material were recognized by anomalous trace element signatures in X-ray fluorescence (XRF) analyses (see Appendix B), but

AC

only two of 26 ash samples were affected. Similarly, ash leachate samples are expected to be minimally affected by non-volcanic contamination. However, the source of the ash is somewhat ambiguous, and we cannot rule out contamination of our samples by ash previously emplaced on the flanks of the edifice and remobilized by wind, explosive events, pyroclastic density currents, or rockfall avalanches spalling off of a continuously extruding lava dome.

6

ACCEPTED MANUSCRIPT Ash bulk geochemistry Whole-rock major and trace element compositions of bulk ash samples were determined by XRF at CVGHM’s Chemical Laboratory in Yogyakarta, Indonesia. A total of 26 ash samples collected on separate days between September 2013 and January 2016 were analyzed. The

T

sampling rate was higher during the resumption of activity in late 2013, and diminished after

IP

activity became steadier in 2014. Ash samples were dried for 12 hours at 80 °C, then passed

CR

through a 74 μm sieve (200 mesh). Eight grams of fine-grained sample were mixed with 1 g of adhesive before XRF analysis as pressed powders. Japan Standard Andesite (standard codes JA-

US

1, JA-2, JA-3) and Japan Rhyolites (JR-1) were used as analytical standards.

AN

A subset of ash samples (washed in water and sieved to > 250 μm) was investigated in plain and polarized binocular light to assess the ash componentry and determine whether the bulk ash

M

analyses reflect older edifice material (accidental ejecta) or fresh magmatic material associated

ED

with the current Sinabung magma (juvenile material).

PT

Ash leachates

The chemical composition of volcanic gases emitted into the atmosphere contains information

CE

about processes occurring at depth (e.g., Allard et al., 2005; Burton et al., 2007; Aiuppa et al.,

AC

2007). Unfortunately, the hazards associated with the Sinabung activity prevented us from directly measuring gas compositions in the plume. However, some information about gas composition can be gained by studying a time series of the leachates adsorbed on fine-grained tephra particles. Fine ash can adsorb, and therefore rapidly scavenge, volatile elements such as sulfur, halogens, and metal species in the form of soluble salts. Analysis of such water‐soluble surface materials represents an indirect source of information on volcanic gas composition that can be obtained at volcanoes inaccessible to direct measurements (Nogami et al., 2000; Armienta 7

ACCEPTED MANUSCRIPT et al., 2010). In particular, the S/Cl ratio in the water-soluble materials adhering to the ash surfaces is broadly related to the relative concentrations of S and Cl in the emitted volcanic gas, though other processes such as the leaching of ash constituents and the entrapment of detrital lithic fragments can also contribute (Bagnato et al., 2011).

T

We measured the S/Cl ratio in the water soluble components adsorbed to fine-grained ash

IP

samples collected on 64 separate days from August 2013 to June 2016. Each ash sample was

CR

passed through a 74 μm sieve and 0.2 - 0.3 g of the fine material was immersed in 80 oC distilled water for eight hours to dissolve adsorbed water-soluble ions. The ash leachate-containing

US

solution was filtered through a 0.45 μm Millipore filter, and diluted with distilled water back to

AN

50 ml. Concentrations of chloride (Cl-) and sulfate (SO42-) ions in the water leachates were then determined by ion chromatography in CVGHM’s Chemical Laboratory in Bandung, Indonesia

M

following the procedure in Nogami et al. (2001). Ash leachate S/Cl values are reported as molar

PT

Continuous scanning DOAS

ED

ratios.

CE

Due to the insufficient repeat-interval of our temporarily deployed scanning DOAS measurements, we generally did not capture the SO2 flux associated with large explosive events

AC

from Sinabung, which often produce plumes that are only measureable for tens of minutes. During the 2013 - 2016 period of observations, explosions numbered 0 - 10 events/day, with most days having < 5 explosions, and many having < 2 events/day. An exception was the period of heightened explosive activity in January 2014, where explosions peaked at 25 - 70 events/day (see Observations section), so that campaign DOAS measurements during this time likely do include SO2 releases related to explosive events. However, the vast majority of campaign DOAS measurements represent an estimate of the emissions associated with background degassing 8

ACCEPTED MANUSCRIPT activity in between the sporadic eruptions; while still useful, this data cannot be used to retrieve the total integrated SO2 output during the eruptive crisis. In order to better capture the SO2 emissions occurring during individual explosions, and to improve overall SO2 measurement frequency, a system of continuous scanning DOAS

T

instruments was installed at Sinabung in August 2016. Three NOVAC stations were installed in

IP

the NE, E, and SE sectors of the volcanic edifice, each approximately 6 km from the volcanic

CR

vent. These instruments employ a 60-degree conical scanner and scan the sky around the volcano from one horizon to the other (Galle et al., 2010). The scanners operate continuously when

US

sunlight is adequate for UV measurements (typically 7:30 – 17:30 local time), with individual

AN

scan durations being automatically adjusted to UV light conditions to ensure appropriate signal intensities. Entire horizon-to-horizon scans last ~5 minutes at mid-day and extend up to 15

M

minutes during reduced light levels in the early morning and late afternoon. The scanners were

ED

installed at the top of 5 m poles to provide adequate sight lines to the horizons. Incident UV radiation is coupled into a fiber optic cable and transferred to a spectrometer installed in an

PT

enclosed box on the ground (Fig. 3). The instruments are powered by four 55 W solar panels

CE

coupled to four 75 Ah lead acid batteries. The spectrometers analyze the UV spectrum for the absorption of light by SO2 in the volcanic

AC

plume. The acquired spectral data are stored on-site, then telemetered to the local CVGHM observatory where the analysis takes place in near real-time. The retrieval method is similar to that of the above-described DOAS campaign instrument except that a reference gas cell is not used. Instead, the high-resolution absorption cross-sections of SO2 (Vandaele et al., 2009) and ozone (Bogumil et al., 2003), convolved with the instrument line shape, are fit to the measured optical depth. A Ring spectrum is included in the evaluation to take inelastic scattering in the

9

ACCEPTED MANUSCRIPT atmosphere into account (Grainger and Ring, 1962). The total SO2 burden in a cross-section of the volcanic plume is calculated by integrating the SO2 column densities across each individual scan. A measure for plume completeness is then derived by fitting a Gaussian curve to the progression of SO2 column density as a function of scan angle. The plume completeness

T

parameter is calculated from the ratio of the area under the Gaussian curve inside the scan

IP

horizon vs the area behind the horizon (Johansson, 2009). The scanning data can then be filtered

CR

to include only scans that captured the complete plume, and exclude scans where large portions of the plume were missed.

US

Global Forecast System (GFS) wind data were retrieved from the National Oceanic and

AN

Atmospheric Administration’s (NOAA) Air Resources Laboratory Real-time Environmental Applications and Display system (READY). Wind forecasts were downloaded each morning and

M

input into the NOVAC evaluation software, which then used these to determine SO2 emission

ED

rates throughout the day. For all days, the volcano’s summit elevation was assumed as the plume height. At the time of writing of this manuscript, the NOVAC stations had been running for

Observations

CE

PT

approximately one week (August 15 - 22, 2016), collecting continuous SO2 emission data.

AC

SO2 emission rates, S/Cl ash leachate molar ratios, seismic and visual observations, and satellitederived lava extrusion rates from Pallister et al. (this volume) are presented in Appendix A.

Pre-eruptive conditions Prior to the onset of explosive activity in 2010, Sinabung’s last eruption occurred in ~800 CE (Iguchi et al., 2012), and Sinabung was classified by CVGHM as a “Type B” volcano (a designation for volcanoes with no eruptions since 1600 CE) (Gunawan et al., this volume). 10

ACCEPTED MANUSCRIPT Cumulative Holocene activity at Sinabung has produced a youthful volcano geometry, with basaltic-andesite to andesitic lava flows and pyroclastic deposits on all sides of the edifice (Prambada et al., 2010; Iguchi et al., 2012). Long-lived sulfur-bearing fumaroles have been active on Sinabung’s southern upper flanks, and sublimated sulfur was actively mined until the

T

early 2000’s, when apparently a decrease in S-emissions rendered S-mining uneconomical

IP

(Sutawidjaja et al., 2013). However, no emission measurements were made at these fumaroles

CR

prior to the onset of the current eruptive crisis, so the pre-eruptive degassing rates of Sinabung are unknown. Edifice inflation of ~2 cm/yr from 2007 - 2010 has been retroactively recognized

US

at Sinabung (Chaussard and Amelung, 2012; Chaussard et al., 2013, González et al., 2015). The

AN

deformation source for this inflation was modeled to be ≤ 2.6 km beneath the summit, suggesting a shallow pressurizing magmatic source (Chaussard and Amelung, 2012) or hydrothermal system

M

(González et al., 2015). Other signs of pre-eruptive unrest included reports of dead fish and

ED

“white liquid” in streams on Sinabung’s lower east flank, dying vegetation, and low-energy steam plumes from the summit (Novi Indrastuti, CVGHM internal report; Pallister et al., this

PT

volume). Pre-eruptive inflation culminated in explosive activity on August 27, 2010, initiating a

CE

multi-year eruptive crisis that continues to the time of writing (September 2016).

AC

Based on our geochemical measurements, the 2010 - current eruptive crisis at Sinabung can be broadly divided into six phases. We note that companion studies in this volume on the Sinabung eruptive crisis (e.g., Gunawan et al.; Pallister et al.) divide the 2013 - present eruptive period differently, based on geologic and seismic observations (Table 1). To reduce confusion with these geologically/seismically defined phases (which the authors refer to as phases 1 - 5), we refer to our geochemically defined phases alphabetically, as phases A – F. Unless explicitly

11

ACCEPTED MANUSCRIPT mentioned, all phases referred to in this study are our geochemically defined phases, and all dates and times are in local time (WIT; UTC + 9).

IP

T

Table 1. Chronology of 2010 – present (2016) eruptive activity at Sinabung Volcano summarizing major changes in volcanic activity. Dates mark the start of notable events related to Sinabung eruption and monitoring; Dates are in local time (WIT). Geologic phase distinctions recognized by Gunawan et al. (this volume), which differ from the geochemical phase distinctions described in this study. LF = Low frequency earthquakes, PDC = Pyroclastic density current. Major event descriptions are from companion papers in this volume: Gunawan et al.; Pallister et al.; McCausland et al.

CR

Major Events

US

Date

Geological Phases

First phreatic explosions

October, 2010

Onset of quiescence

July 2013

Onset of increased seismicity

September 15, 2013

Onset of phreatic-phreatomagmatic explosions, moderate SO2 emission rates, high S/Cl ash leachates

December 7, 2013

Increase in deep, LF, and hybrid earthquakes, low SO2 emission rates

December 18, 2013

Onset of dome extrusion on surface, low SO2 emission rates, declining S/Cl ash leachates

January 1011, 2014

First major dome collapse, large increase in SO2 emission rates, development of lava flow, PDCs to the NE of lava flow, low S/Cl ash leachates

(Gunawan et al.)

(this study)

--

Phase A

--

Phase B

AC

CE

PT

ED

M

AN

August 27, 2010

Geochemical Phases

February 28, 2014

Decrease in SO2 emission rates, followed by decrease in lava extrusion rates, S/Cl ash leachates remain low

midSeptember 2014

Onset of second lava dome formation, PDCs to the SW of 2014 lava flow. SO2 emissions gradually declining, S/Cl ash leachates remain low 12

Phase 1

Phase C

Phase D Phase 2

Phase E Phase 3

Phase F Phase 4

ACCEPTED MANUSCRIPT July 2015

Second lava dome collapse, lava flow and PDCs to SE, SO2 emissions continue gradual decline, S/Cl ash leachates remain low NOVAC SO2 system installation

September 2016

Time of writing

Phase 5

IP

T

mid-August 2016

CR

Phase A - 2010 phreatic eruptions (August 2010 – September 2010)

The first phase of the Sinabung eruptive crisis (phase A) spans from August 2010 to September

US

2010, when phreatic explosions occurred at Sinabung for the first time in ~1,200 years (Iguchi et al., 2012). The first phreatic explosion occurred on August 27, 2010 and ejected a small ash

AN

cloud that drifted southeast. A total of six additional explosions ensued between August 29 and

M

September 7, some of them sending ash as high as 5 - 7 km above the volcano’s summit,

ED

oftentimes in two distinctly separate plumes (Fig. 4) (Gunawan et al., this volume; Pallister et al., this volume). Activity was accompanied by shallow and deep swarms of volcano-tectonic (VT)

PT

earthquakes, as well as distal VT events (Hendrasto et al., 2012)

CE

Volcanic ash erupted in these explosions contained no juvenile material, consisting entirely of variably altered old lava fragments and hydrothermal minerals including quartz, magnetite,

AC

alunite, anhydrite, kaolinite, and pyrite (Iguchi et al., 2011; Sutawidjaja et al., 2013; Nakada, personal communication). During this phase, the passive SO2 emission rate measured between eruptions on 13 individual days averaged about 550 metric tonnes/day (t/d), with a standard deviation (s.d.) of 180 t/d (hereafter, averages are all reported as ± 1 s.d.) (Fig. 5). It is unclear whether this represents an increase over emission rates prior to phase A as, to our knowledge, no pre-eruptive SO2 emission rate measurements exist.

13

ACCEPTED MANUSCRIPT Phase B – Eruptive quiescence (October 2010 – September 14, 2013) The phreatic explosions in August and September of 2010 were followed by a period of repose which we describe as phase B. During this second phase of activity, spanning from October 2010 to September 2013, no additional explosions occurred. The local populations were allowed back

T

into areas that had previously been evacuated. Following the 2010 activity, Sinabung was

IP

considered a “Type A” volcano by CVGHM (a designation for volcanoes with explosive activity

CR

within the past 400 years), and monitoring infrastructure was consequently improved. A network of seismometers and GPS instruments was installed, and electronic distance measuring (EDM)

US

instruments were used to monitor deformation of the main edifice (Gunawan et al., this volume).

AN

Despite the eruptive quiescence, seismicity remained elevated (McCausland et al., this volume) and Sinabung’s upper edifice was undergoing continuous slow extension (Hotta et al., this

M

volume). Additionally, low levels of white, steam-rich gas emissions visibly persisted during this

ED

phase, although no measurements of SO2 emission rates were made.

PT

Phase C – Renewed phreatic/phreatomagmatic eruptions (September 15, 2013 – December 6, 2013)

CE

Phase C began on September 15, 2013, when explosions once again occurred at the north flank

AC

of Sinabung’s summit (Gunawan et al., this volume). These explosions sent columns of ash 7.5 km above the volcanic edifice, which led to significant ash fall in proximal areas (Pallister et al., this volume). Several villages around the volcano’s perimeter were evacuated. In October 2013, the eruptive center moved to the volcano’s summit crater. At this point explosions were occurring up to 5 times per day, with ash column heights again reaching up to 7.5 km above the volcano’s summit (Fig. 4). Similar activity continued through the month of November, with explosions generally occurring 3 - 8 times per day. Explosive activity spiked on November 24 –

14

ACCEPTED MANUSCRIPT 25 with 29 recorded explosions and eruption columns up to 9 km above the summit. Subsequently, explosive activity lessened until January 2014, with < 2 explosions per day and columns heights below 2 km above the summit. Ashes from this phase of the eruption are similar to ash from the 2010 eruptions, consisting of accidental ejecta from the old Sinabung edifice.

T

The ash components include material from a hydrothermal system with sulfide minerals, quartz

IP

veins, vuggy silica, and altered lavas (Fig. 6). XRF analyses (see Appendix B) of bulk ash

CR

samples collected during this time period (n = 11) are the most primitive of the eruptive period (e.g., low SiO2 [57 - 61 wt%] and high MgO [2.2 - 4.2 wt%]), and contain very high levels of

US

chalcophile trace metals (Cu, Pb, Mo, Zn) compared with phases E and F. These measurements

AN

are consistent with the ash being sourced from a mixture of old basaltic-andesite and andesitic edifice lavas (Iguchi et al., 2012; Nakada et al., this volume) and the sulfide-bearing

M

hydrothermal system (e.g., Barnes and Ripley, 2016) (Fig. 7). We find that ash samples from this

ED

period contain little to no juvenile material (Fig. 6), although rhyolitic glass shards were identified by Nakada et al. (this volume) and Gunawan et al. (this volume) in ashes erupted in

PT

November 2013. However, it is unclear whether this is truly juvenile material, as these glass

CE

shards are geochemically very similar to glass from the extensive 74 ka Youngest Toba Tuff deposits that underlie Sinabung (Gunawan et al., this volume). Further geochemical analyses are

AC

required to definitely show whether these early glass shards in Sinabung ash are juvenile material or are incorporated glass from the underlying Toba deposits (Gunawan et al., this volume). Despite the heightened eruptive activity during this phase, passive SO2 emission rates averaged only 430 ± 310 t/d (39 measurement days), though the syn-explosive emissions were probably significantly higher. S/Cl molar ratios measured in ash leachates collected downwind of the vent on 19 different days during this eruptive phase averaged 12.0 ± 8.2, with maximum S/Cl ratios of

15

ACCEPTED MANUSCRIPT 20 - 34. These S/Cl ratios are notably higher than leachate ratios during any other phase in the eruption, as all ash leachate S/Cl ratio measurements since mid-January 2014 have been < 3 (Fig. 5). While most ash leachate studies worldwide find S/Cl molar ratios ≤ 4 (Witham et al., 2005), very high S/Cl ash leachate ratios (> 20) have been recognized in other eruptions interacting with

T

S-rich hydrothermal systems (hosting gypsum [CaSO4∙2H2O], anhydrite [CaSO4], and sulfide

IP

minerals), such as the 1995 - 1996 Ruapehu (Cronin et al., 1998) and 2012 Tongariro eruptions

CR

(Cronin et al., 2014). We note that anhydrite was recognized in accidental ejecta from Sinabung’s 2010 phreatic eruptions (Sutawidjaja et al., 2013), and consider it possible that

US

anhydrite was also present in the phase C ejecta and contributed to the high S/Cl leachate ratios.

AN

Unfortunately, our water-based ash cleaning would have quickly dissolved any present anhydrite, thus precluding any petrographic identification of anhydrite in the ash. Overall, it is

M

likely that the high S/Cl ratios during phase C of the Sinabung eruption were due to some

ED

combination of a deeply exsolving magmatic gas, which would exsolve more S than Cl gas, the thermal or fugacity-induced breakdown of sulfides and Ca-sulfates in the relict hydrothermal

CE

PT

system, and the possible dissolution of hydrothermal sulfates during leaching.

Phase D – Lava ascent and extrusion (December 7, 2013 – January 10, 2014 )

AC

Starting December 7, 2013, the number of deep and hybrid earthquakes occurring at Sinabung began increasing dramatically (Fig. 5). These seismic signals are thought to herald the movement of magma through Sinabung’s conduit (McCausland et al., this volume). Hybrid earthquake swarms continued intensifying through mid-December and were accompanied by 1.5 cm of vertical summit inflation and 125 m of lateral movement to the S-SE (Gunawan et al., this volume). However, during this presumed period of lava ascent, summit explosions were

16

ACCEPTED MANUSCRIPT infrequent (0 - 3/day) and relatively weak, and SO2 emission rates were some of the lowest measured during the entire 2010 - present eruptive crisis (260 ± 130 t/d; n = 7). Hybrid earthquakes peaked at > 1,600 events/day from December 15 - 17, and the southern summit area laterally moved over 100 m just before the onset of lava dome extrusion at

T

Sinabung’s summit on December 17 – 18 (Gunawan et al., this volume; Pallister et al., this

IP

volume). Once lava breached the surface, summit explosions ceased and low rates of lava

CR

extrusion (1 - 3 m3/s) produced a lava dome confined to the summit (Pallister et al., this volume). The large number of hybrid earthquakes accompanying lava ascent and dome extrusion lead to

US

an association at Sinabung between hybrid earthquakes and magma and/or fluid movement

AN

(McCausland et al., this volume). SO2 emission rates increased only slightly following lava extrusion on the surface (440 ± 210 t/d; n = 9). Overall, SO2 emission rates during this phase of

M

initial lava ascent and extrusion were surprisingly low (360 ± 200 t/d SO2; n = 16). Summit

ED

explosions began again in January, with dozens of eruptions occurring each day. Ashes collected from these explosions in early January had lower, but highly variable, S/Cl ratios of 6.3 ± 8.5 (n

PT

= 5). No XRF measurements were made on ashes erupted during this phase of early lava

CE

extrusion.

AC

Phase E – Peak magmatic activity (January 11, 2014 – February 27, 2014) The fifth phase (phase E) of activity began after a large collapse of the summit lava dome on January 10, 2014, where 80 - 90% of a 3 Mm3 lava dome collapsed, creating extensive pyroclastic flows (Pallister et al., this volume). Volcanic activity increased substantially following this collapse, with the period from January 10 through late February being the eruption’s most highly active eruptive phase. Powerful summit explosions (plumes reaching over 7 km above the vent) occurred 20 – 60 times per day in mid-January, then decreased to < 5 per 17

ACCEPTED MANUSCRIPT day late January, and finally ceased altogether by mid-February. SO2 emission rates increased immediately after the dome collapse, rising from 200 – 800 t/d during the three weeks before the collapse to 1,200 - 3,800 t/d in the week following the collapse (see Fig. 5). SO2 emission rates remained elevated until late February, averaging 1,680 ± 1,070 t/d during entirety of phase E (n

T

= 21). In contrast to the immediate change in SO2 emission rates, lava extrusion rates remained

IP

relatively low (1.0 – 2.5 m3/s) for over a week after the dome collapse, then increased markedly

CR

to 6 – 18 m3s-1 from January 18 through March 6 (Pallister et al., this volume). On January 13, the lava dome began extruding down a breached southeast portion of the summit,

US

rapidly advancing down the southeast flank (Fig. 4). By February 4, this lava flow extended 1.9

AN

km from the summit to the base of the edifice, with a calculated volume of 13 Mm3 (Pallister et al., this volume). Collapses from the lava dome and lava flow margins led to numerous

M

pyroclastic flows descended down the southeast flank of the volcano, particularly during the

ED

dome growth and early lava flow periods in early to mid-January. Ash samples from this phase (n = 4) include definitive juvenile material for the first time in the

PT

eruption, although we assume that juvenile material would have also been present in ashes from

CE

phase D (lava ascent and initial extrusion), however no samples were obtained during that time. The ash volumetrically contained 50-80% juvenile material, which consisted of 60 - 70% fresh

AC

glass, 25 - 30% feldspar, and 5 - 10% mafic components (pyroxene and Fe-oxides) (Fig. 6). XRF analyses of bulk ash from this eruption phase show that this ash had a slightly more silicic bulk composition (60.2 ± 0.7 vs 58.9 ± 1.0 wt% SiO2), with significantly lower levels of trace metals than the phreatic/phreatomagmatic phase C (~40% less Cu, 50% less Pb and Zn, and 70% less Mo in phase E vs phase C ash) (Fig. 7), consistent with appearance of juvenile material and diminishing, though still significant, inputs of accidental material (old edifice lavas and

18

ACCEPTED MANUSCRIPT hydrothermal components). As mentioned, the highest measured SO2 emission rates of the entire eruptive crisis occurred during phase E, with emissions peaking at ~3,800 t/d (see Fig. 5). Given the great frequency of explosive events during the early stages of this phase, it is likely that campaign DOAS measurements in mid-January captured more of the explosive gas releases than

T

in any other phase. However, DOAS measurements during this time most likely still represent an

IP

underestimate of the maximum SO2 emission rates, as DOAS measurements still missed a large

CR

number of individual explosions. The increase in SO2 emission rate is most likely attributable to an increased supply of magma to the shallow system, as also evidenced by the increased lava

US

extrusion rates during this phase. Interestingly, the ash leachate S/Cl ratios dropped to 1.4 ± 0.5

AN

during this most vigorous phase of eruptive activity (ashes collected on six different days). This decrease would be consistent with the degassing of a shallower magmatic source, as the Cl

M

solubility in melt at low pressures is higher than that of S, and Cl is therefore only emitted at

ED

very low pressures and shallow depths as lava is extruded (Edmonds et al., 2001; Spilliaert et al., 2007; Johnson et al., 2010). It is also plausible that a magma slug moving up in the conduit could

PT

have previously partially degassed its sulfur, therefore also leading to a lower S/Cl ratio as it

CE

reaches near-surface pressures. Also contributing to, and possibly even dominating, these decreased S/Cl ratios is the greatly reduced input from Sinabung’s old sulfide and sulfate-bearing

AC

hydrothermal system which, as ash componentry and the active lava dome/flow shows, had been largely cleared out by this stage.

Phase F – Steady magma extrusion (Late February 2014 – present) The most recent phase of the eruption (phase F) began in late February 2014 and continues to the time of writing (September 2016). SO2 emission rates decreased fairly dramatically from > 1,000 t/d to < 300 t/d by late February to early March, and the lava effusion rate decreased from ~15 19

ACCEPTED MANUSCRIPT m3/s to ~6 m3/s about a week later (Pallister et al., this volume). SO2 emission rates increased modestly for a short period in April 2014, but never returned to the scale of phase E emissions. Subsequently, aside for some minor fluctuations, SO2 emission rates and lava extrusion rates (Pallister et al., this volume; Nakada et al., this volume) have been gradually declining through

T

the present time of writing. Activity has generally been less explosive during this ongoing phase,

IP

consisting mainly of small explosions (0 - 2 per day through September 2015, then generally < 8

CR

per day through the time of writing). Pyroclastic flows, caused by failure of the lava dome and lava flow fronts, continue down the south and southeast flanks. By early March, 2014, the lava

US

flow from the growing dome had traveled about 2.4 km downslope to the base of the edifice

AN

(Pallister et al., this volume). The lava flow subsequently stalled and thickened as the effusion rate became relatively slower. Eventually the large lava flow lobe became inactive, and new,

M

smaller lava flows began extending anew from the summit dome. Clearly incandescent at night,

ED

one of these near-summit flow lobes remains a prominent feature at Sinabung through the present time. The last lava extrusion rate measurements available show a long-term extrusion

PT

rate from August 2015 – January 2016 of 3.1 m3/s (calculated from Pallister et al., this volume).

CE

The lava dome and flow both retain or develop overpressure post-extrusion (potentially caused by ongoing crystallization processes within the lava) (Sparks, 1997; Massol and Jaupart, 2009;

AC

Boudon et al., 2015), and failures lead to pyroclastic activity, with flows traveling up to 5 km from the summit in the southeast sector of the volcano, and producing ash plumes over 4 km in height (Fig. 5) (Pallister et al., this volume). Various large lava dome collapses and significant restructuring of lava flows and pyroclastic flow directions have occurred since February 2014, leading Gunawan et al. and Pallister et al. to divide this 2014 – present period into multiple distinct phases of volcanic activity (see Table 1). However, SO2 emissions, S/Cl leachate ratios,

20

ACCEPTED MANUSCRIPT and bulk ash chemistry have remained relatively constant since March 2014, and therefore we consider this to be geochemically a single eruption phase. Ash from summit explosions, pyroclastic flows, and remobilized during rockfalls off the lava flows during this phase are volumetrically 70 - 80% juvenile glass and feldspar, 10 - 20%

T

juvenile Fe-oxides and pyroxene, and < 10% non-juvenile accidental material (n = 11) (Fig. 6).

IP

Considering that ashes erupted during this phase are > 90% juvenile material, the bulk ash XRF

CR

analyses during phase E are our closest measured approximations of the juvenile erupting magma composition. However, we note that “crystal depletion” (i.e. the density-driven fallout of

US

mineral phases from the eruptive column) likely caused the ash to be depleted in denser phases

AN

(Fe-oxides, pyroxene) and enriched in low-density phases (glass and feldspar) relative to the bulk magma. Lava fragments entrained in pyroclastic flows were sampled by Nakada et al. (this

M

volume) indeed show that the bulk lava composition has ~3-5 wt% lower SiO2 and 0.5 – 1.0 wt%

ED

greater total alkalis (Na2O + K2O) than ashes erupted during the same time interval. The juvenile ash we measured during this phase is andesitic-dacitic, while Nakada et al. find that the bulk lava

PT

composition is andesitic, consistent with crystal depletion affecting the ashes. The juvenile ashes

CE

are more evolved than earlier ash samples, which consisted of accidental ejecta (62.5 ± 1.8 wt% SiO2 for phase F juvenile ashes vs. 58.9 ± 1.0 wt% SiO2 for phase C accidental lithic ashes). Iron

AC

and other chalcophile metals are much less abundant in phase F ashes than in phase C ashes (~25% less Fe; 50% less Cu, Pb, and Zn; 90% less Mo), consistent with greatly reduced contributions from the sulfide/sulfate-bearing hydrothermal system as the eruptive sequence progressed (Fig. 7). Ash leachate measurements for S and Cl continue to the present day, with a total of 34 ash samples analyzed from different days during this current eruptive phase. S/Cl leachate ratios

21

ACCEPTED MANUSCRIPT during this current phase are very low, averaging 1.1 ± 0.6. These low S/Cl ratios are similar to phase E ashes. All ash leachate measurements since the collapse of the first lava dome in midJanuary 2014 have been < 3, in stark contrast to S/Cl ratios of 10 to > 30 during the earlier phreatic/phreatomagmatic phase C (Fig. 5). The low S/Cl ratios since January 2014 are likely a

T

result of an increased contribution from shallow magma and the clearing out sulfide/sulfate-

IP

bearing materials of the pre-existing hydrothermal system. Additionally, the low S/Cl ratios may

CR

be partially due to the phase F ash samples being sourced from a combination of summit explosions, pyroclastic flows, and lava flow rockfall debris. Ash produced from pyroclastic

US

flows and rockfall debris would have significantly degassed S within in the lava dome/flow, but

AN

more Cl would be retained in the cooling magma owing to Cl’s greater magmatic solubility at low pressures. The collapse of lava domes and lava flow fronts would therefore generate

M

pyroclastic flows with more Cl-rich plumes.

ED

The campaign DOAS measurements indicate that passive SO2 emissions during the entirety of phase F averaged about 450 ± 290 t/d (149 individual measurement days), with higher emission

PT

rates likely during explosive events (see continuous DOAS measurements described below).

CE

However, as has been mentioned before, there is considerable difficulty in interpreting these values due to the relatively low time resolution of the campaign measurements and the resulting

AC

inability to capture the SO2 emissions associated with explosive activity. Only after a continuous NOVAC scanning DOAS network was installed at Sinabung in August 2016, did more detailed information on the nature of the current degassing activity become available. The first week of SO2 emission rate data measured by the NOVAC scanning DOAS instruments (August 15 - 22, 2016) is shown in Fig. 8. Different colored filled points represent data collected from three different instruments in the northeast (NE), east (E) and southeast (SE) sectors of the

22

ACCEPTED MANUSCRIPT volcanic edifice. The data are filtered such that only scans that are assumed to have captured at least 80% of the volcanic plume are shown; scans during which the plume was out of range of a particular instrument were omitted (see Methods section). Most of the displayed data are from the station at Sukandebi (SKDB), as this station is east of the volcano and the predominant wind

T

direction was from the west during the displayed timeframe.

IP

The continuous DOAS data show that the emission rate was generally quite low during the first

CR

week of NOVAC operation. If we assume that degassing activity was of passive nature whenever the emission rate was below 450 t/d (the average passive degassing rate for phase F measured by

US

campaign DOAS), the daily “background” SO2 emissions averaged 250 ± 20 t/d for the one-week

AN

period consisting of 635 NOVAC scans (here and below, the given uncertainty represents the standard deviation of the daily averages, not of individual measurements or the measurement

M

error). This emission rate is similar to the most recent (at time of writing) campaign DOAS

ED

measurements, which averaged 320 ± 100 t/d during June, 2016. However, several gas bursts were detected by the NOVAC DOAS instruments, during which the measured emission rate

PT

briefly climbed to several thousand tonnes per day (650 - 3,300 t/d peak emission rates). Most of

CE

these peaks were associated with summit explosions, which were observed visually during clear conditions and also detected by the Sukanalu seismic station 3.7 km NE of the summit. The

AC

timing and relative seismic amplitude of each eruption is depicted by red asterisks in Fig. 8. The integrated amount of SO2 gas released during individual explosive pulses was between 10 and 50 tonnes. These estimates might be slightly skewed towards low values due to the possibility of ash slightly obscuring parts of the plume from view by the DOAS scanners, but we did not find any specific evidence for such an effect in the spectral data (see Kern et al., 2012 for examples).

23

ACCEPTED MANUSCRIPT Though most of the measured gas bursts were associated with explosions, this is not the case for the event detected in the afternoon of August 18, 2016. This particular gas pulse was the largest SO2 release measured during the first week of NOVAC network operation, and about 73 tonnes of SO2 were released and passed through the DOAS scanning plane 6 km east of the vent over a

T

period of about 1.5 hours. Interestingly, no explosion or “gas emission” seismic events were

IP

registered at this time. “Gas emission” seismic events have been recognized at Sinabung, and

CR

have a distinct seismic signature (an impulsive or emergent onset, and a long-lasting broadband frequency content) and are typically accompanied by an observed gas plume emission from the

US

crater (see McCausland et al. [this volume] for details of seismic characterizations at Sinabung).

AN

However, no visually obvious gas plume (i.e., white, steam-rich) was recognized at the time of this major SO2 release, nor was any “Gas emission” seismic signature. Rather, the gas pulse was

M

immediately preceded by the largest “hybrid” seismic event measured during the reporting

ED

period. Hybrid events such as these have been associated with brittle failure and resonance in fluid-filled cracks (Chouet and Matoza, 2013). In some cases, hydraulic fracturing has been

PT

identified as the cause of the hybrid events (Foulger et al., 2004), but other processes can cause

CE

similar signals (Chouet and Matoza, 2013). At Sinabung, multiple years of observations have shown that hybrid events appear to be associated with enhanced lava dome extrusion, and hybrid

AC

earthquakes are therefore considered a proxy for magma ascent in the conduit (McCausland et al., this volume). Including the SO2 emission rates measured during all gas bursts (explosive and non-explosive), the average daily SO2 emission rate measured during this week-long reporting interval was 320 ± 40 t/d (711 individual NOVAC scans). Consequently, the SO2 emitted during discrete explosive and non-explosive degassing events accounted for about 20% of the total SO2

24

ACCEPTED MANUSCRIPT output from Sinabung during this measurement period, with continuous passive emission accounting for the remaining 80%.

Summary and Conclusions

T

The somewhat sparse geochemical data collected during the course of this eruption does not

IP

provide conclusive evidence for a specific model of volcanic activity valid for the entire six-year

CR

reporting period. However, a few general conclusions can be drawn. For one, the observed average SO2 emission rate of 550 ± 180 t/d during the first phase of phreatic activity (phase A)

US

indicates a high volume of magmatic gases relative to that of hydrothermal water, or the presence

AN

of dry pathways, which allow these magmatic gases to reach the surface without complete scrubbing by the shallow hydrothermal system. When volcanic gases pass through aquifers or

M

wet hydrothermal systems of significant size, the process of scrubbing can efficiently remove

ED

sulfur-bearing compounds (Symonds et al., 2001). Oftentimes, SO2 emissions are not detected until the hydrothermal system at a given volcano has been at least partially dried out and a dry

PT

pathway exists for gas to escape (for example, see the onset of the 2004 - 2008 eruption of

CE

Mount St. Helens [Gerlach et al., 2008]). The relatively high SO2 emission rate associated with Sinabung’s phreatic explosions in 2010 could point towards pressurization of the shallow

AC

hydrothermal system by influx of magmatic volatiles sourced from depth, rather than shallow magma coming into contact with the hydrothermal system itself. In this slow pressurization model, volcanic gas containing SO2 could build up in a dry area beneath the summit’s sealed lithic cap until pressure exceeded a failure threshold, thus releasing SO2-rich gas and highly altered, phreatic tephra. A gradual cap-sealing hypothesis could also be consistent with the apparent decrease in fumarolic emissions in the early 2000’s that ended S-mining activity

25

ACCEPTED MANUSCRIPT (Sutawidjaja et al., 2013), and the shallowly sourced edifice inflation in the years preceding the eruption (2007 - 2010, and possibly earlier) (Chaussard et al., 2013; González et al., 2015). Additionally, this mechanism could help explain the 2010 - 2013 repose interval (phase B), as the volcano could return to relative dormancy after the pressure was released and until the

T

magma reached shallower depths several years later.

IP

Secondly, it is interesting to note that when explosive activity resumed at the summit crater in

CR

September 2013 (phase C), the system initially ejected materials from the old Sinabung edifice and phreatic material from the sulfide/sulfate-bearing hydrothermal system. A pathway was not

US

yet established for juvenile material to be erupted. The high S/Cl values (> 10 - 30) found in the

AN

water-soluble ash leachates from this period are also consistent with the inclusion of hydrothermally sourced S in the ash, and with the magmatic source having not yet reached the

M

surface to strongly exsolve Cl gases and produce more magmatic, low S/Cl leachate ratios.

ED

The hydrothermal system appears to have started clearing out of the vent pathways during the initial lava dome extrusion in mid-December 2013 (phase D), and was completely cleared out by

PT

early 2014 (phases E and F). In mid-January 2014, juvenile glassy material and fresh feldspar

CE

crystals became visible in the ash samples, a lava dome/flow was extruding, and the SO2 emission rate increased sharply from an average of 360 t/d to several thousand tonnes per day.

AC

Explosions were frequent and vigorous during this stage of activity, making it more likely that campaign measurements captured degassing during explosions rather than strictly background passive degassing. SO2 emission rates as high as 3,800 t/d suggest that volatiles were able to escape without significant interaction with a hydrothermal system, thus practically eliminating any scrubbing that may have been occurring previously. The observed drop in S/Cl ash leachate ratios during this phase could be caused by increased Cl emissions from lava that was at, or very

26

ACCEPTED MANUSCRIPT close to, the surface. Also, with the increased acidification of any remaining liquid phase by large amounts of gas passing through the conduit, Cl would have been removed less efficiently from the gas phase and would be more readily released to the gas phase plume (Symonds et al., 2001). Additionally, and perhaps most importantly, the old S-rich hydrothermal material was

T

largely cleared out by this stage in the eruption, removing a potentially large non-juvenile S

IP

source, and contributing to the lower S/Cl ratio. We conclude that the low S/Cl leachate ratios (<

CR

3) reflect the juvenile magmatic contribution, and that elevated S/Cl ratios are largely due to external contributions of S from pre-existing hydrothermal minerals. This is consistent with

US

findings at other volcanic systems (Diller et al., 1998; Bagnato et al., 2011; Cronin et al., 2014),

AN

and argues for caution when attempting to relate ash leachate measurements to compositions of magma or eruptive plumes whenever pre-existing volatile-bearing material may be entrained.

M

Another interesting observation, stemming from the long term multi-parametric monitoring of

ED

Sinabung’s eruption, is the relationship of SO2 emission rates and lava extrusion rates reported by Pallister et al. (this volume). While these dataset comparisons are somewhat limited by

PT

measurement frequencies and the uncertainty related to incomplete capturing of SO2 releases

CE

during explosive events, broad associations can still be drawn. Overall, we find that passive SO2 emission rates and lava extrusion rates generally co-vary throughout the eruption (see Fig. 5).

AC

The highest SO2 emission rates (1,000 – 3,800 t/d) and greatest lava extrusion rates (6 – 18 m3/s) occurred during the most active phase of the eruption from mid-January – February 2014 (phase E). However, within the periods of dome initiation and peak activity (phases D and E: December 2013 – March 2014), a few notable associations stand out. One is that the onset of hybrid earthquakes (indicating magma movement in a conduit [McCausland et al., this volume]) ~10 days before lava dome extrusion at the surface (December 18, 2013), is associated with very low

27

ACCEPTED MANUSCRIPT SO2 emission rates (260 ± 130 t/d, December 7 - 17, 2013). Once the lava dome began to be slowly extruded on the surface (1 – 3 m3/s extrusion rate), SO2 emissions increased slightly (440 ± 220 t/d, December 18, 2013 - January 10, 2014), but were still relatively low, indicating that the slowly ascending dome-forming magma may already have been largely degassed.

T

We also note interesting short-term delays between changes in SO2 emission rates and lava

IP

extrusion rates. Immediately following the large dome collapse on January 10, 2014, SO2

CR

emission rates increased markedly and remained elevated for weeks. In contrast, while lava extrusion rates temporarily increased immediately after the dome collapse (from < 2 m3/s to ~5

US

m3/s within a few hours after dome collapse), lava extrusion rates quickly returned to ~ 2 m3/s

AN

and did not dramatically increase until ~10 days after the dome collapse (Pallister et al., this volume). Pallister et al. (this volume) note that the dome collapse and changes in summit

M

topography allowed subsequent extruded lava to migrate downslope as a lava flow rather than

ED

accumulating as a lava dome. We surmise that the transition from a thick lava dome to a lava flow decreased the overburden pressure on the magma in the conduit. This conduit

PT

decompression, in turn, caused increased gas exsolution and release from depth, leading to

CE

elevated SO2 emission rates. This onset of enhanced SO2 degassing over a week before an increase in lava extrusion rate points towards a decoupling of gas migration at depth, relative to

AC

ascending magma. A similar delay between changes in SO2 emissions and lava extrusion rates occurred in late February 2014, when SO2 emissions declined markedly from ~ 1,300 t/d in February 23 – 26, to 300 t/d by February 28, and then remained relatively low, marking the start of phase F. Lava extrusion rates, however, remained elevated (15 – 18 m3/s) for 7 - 10 days after the decrease in SO2 emissions, finally decreasing to < 6 m3/s by March 7, and subsequently staying low during the rest of phase F (Pallister et al., this volume). Similar decoupling of SO2

28

ACCEPTED MANUSCRIPT emission rates and lava extrusion rates has been recognized during other dome forming eruptions (e.g., Soufrière Hills Volcano [Christopher et al., 2015]), indicating gas-magma separation during magmatic ascent may be a common occurrence. Lastly, we reflect on Sinabung’s current degassing and volcanic behavior. SO2 output has been

T

in a long-term decreasing trend since February 2014. This trend was confirmed by the

IP

continuous SO2 emission rate measurements from the NOVAC scanning DOAS instruments,

CR

with passive background emissions having dropped to about 250 t/d by mid-late August 2016 (Fig. 5 and Fig. 8). However, as of September 2016, explosions are still occurring, typically

US

multiple times per day, and the continuous DOAS measurements show that SO2 emissions

AN

associated with explosions are substantially above background degassing rates. This pattern indicates that the volcanic vent is probably partially sealing in between Vulcanian eruptions

M

(Nakada et al., this volume). Though the seal is not perfect and some gas is still able to escape,

ED

sealing is sufficient for pressure to build beneath it. Pressure builds up to a failure threshold at which point an explosion occurs and the pressure is relieved (Edmonds and Herd, 2007; Cassidy

PT

et al., 2015).

CE

The gas release event that occurred on August 18, 2016 is the one exception to this model. This release was immediately preceded by a large hybrid earthquake, likely caused by dynamics in the

AC

shallow system and possibly related to movement/extrusion of the lava dome. Whatever the exact process was, it appears to have fractured the gas seal and thus relieved the pressure before it reached the failure threshold that would have caused explosion. However, this hypothesis needs further testing as more data is collected by the continuous scanning DOAS network. Despite the apparent gradual decrease in SO2 emissions, the eruptive hazards at Sinabung Volcano remain great. Comparing Sinabung with other dome-forming eruptions from around the

29

ACCEPTED MANUSCRIPT world, it is likely that despite Sinabung’s apparent slow decrease in activity, the volcano may continue to extrude lava and thus generate pyroclastic flows for many years to come (Wolpert et al., 2016). Though the gradual decrease in SO2 emission rate most likely indicates a decrease in magma supply to the surface, this does not necessitate an immediate decrease in explosivity of

T

the eruption. Instead, a lower magma supply rate could promote formation of denser magma

IP

plugs in the upper conduit, as longer residence times increase crystallization and densification by

CR

degassing and vesicle collapse (Melnik and Sparks, 1999; Diller et al., 2006; Nakada et al., this volume). In the near term, this model could lead to a decrease in frequency of explosions, but

US

also a potential increase in explosion strength, as the pressure failure threshold is increased due

AN

to the strengthening of the shallow, degassed plug. Only after the magma supply rate drops to the point that the plug’s failure threshold can no longer be overcome would the pressure-induced

M

explosions cease. At the same time, gravity-induced failure of the extruding dome will also

ED

continue to cause collapse-induced pyroclastic flows until dome extrusion ends. For these reasons, it is important to realize that the volcanic hazards associated with the eruption, in

PT

particular pyroclastic flows, may decrease in frequency but perhaps not in magnitude in the near-

CE

term. These low-frequency, high-impact events could pose a challenge for local emergency

AC

managers as the crisis evolves and until magma supply ceases entirely.

Acknowledgements

The authors would like to thank John Pallister, Kasbani, Muhammad Hendrasto, and Gede Suantika who have supported in making this paper. Also thanks to Hendra Gunawan, Sulus Setiono, Armen Putera, Aaron Rinehart, and Martin LaFevers for the support provided prior to, and during the installation of the NOVAC scanning DOAS instruments at Sinabung. Thanks also

30

ACCEPTED MANUSCRIPT to Arif Cahyo Purnomo and Deri Alhidayat who have helped with the SO2 campaign measurements and ash sample collection. This paper greatly benefitted from thoughtful reviews by Taryn Lopez and an anonymous reviewer. Thank you to Wendy McCausland for editorial

T

handling.

IP

References

CR

Aiuppa, A., Moretti, R., Federico, C., Giudice, G., Gurrieri, S., Liuzzo, M., Papale, P., Shinohara, H., Valenza, M., 2007. Forecasting Etna eruptions by real-time observation of

US

volcanic gas composition. Geology. 35, 1115. doi:10.1130/G24149A.1.

AN

Allard P., Burton M., Muré F., 2005. Spectroscopic evidence for a lava fountain driven by previously accumulated magmatic gas. Nature. 433, 407-410. doi:10.1038/nature03246.

M

Armienta M.A., de la Cruz-Reyna S., Soler A., Cruz O., Ceniceros N., Aguayo A., 2010.

ED

Chemistry of ash-leachates to monitor volcanic activity: An application to Popocatépetl volcano, central Mexico, Appl Geochem, 25, 1198-1205. doi:10.1016/j.apgeochem.2010.05.005.

PT

Associated Press, 2014. Indonesian volcano Mount Sinabung eruption causes 16 deaths. In

CE

Independent, http://www.independent.co.uk/news/world/indonesian-volcano-mount-sinabungeruption-causes-16-deaths-9102479.html. Published February 2, 2014.

AC

Associated Press / CBS, 2016. Indonesian volcano erupts again, killing several people. In CBS News, http://www.cbsnews.com/news/mount-sinabung-volcano-eruption-several-deadindonesia/. Published May 22, 2016. Bagnato E., Aiuppa A., Andronico D., Cristaldi A., Liotta M., Brusca L., Miraglia L., 2011. Leachate analyses of volcanic ashes from Stromboli volcano: A proxy for the volcanic gas plume composition?, J. Geophys. Res. 116, D17204. doi:10.1029/2010JD015512.

31

ACCEPTED MANUSCRIPT Barnes S.-J., Ripley E. M., 2016. Highly siderophile and strongly chalcophile elements in magmatic ore deposits. Rev. Mineral. Geochem. 81, 725–774. doi:10.2138/rmg.2016.81.12. Bogumil K., Orphal J., Homan T., Voigt S., Spietz P., Fleischmann O., Vogel A., Hartmann M., Bovensmann H., Frerick J., Burrows J., 2003. Measurements of molecular absorption spectra

T

with the SCIAMACHY Pre-Flight Model: instrument characterization and reference data for

IP

atmospheric remote-sensing in the 230– 2,380 nm region. J. Photoch. Photobio. A.: Chem. 157,

CR

167–184. doi:10.1016/S1010-6030(03)00062-5.

Boudon G., Balcone-Boissard H., Villemant B., Morgan D.J., 2015. What factors control

US

superficial lava dome explosivity? Sci. Rep. 5, 14551. doi:10.1038/srep14551.

AN

Burton M., Allard P., Mure F., La Spina A., 2007. Magmatic Gas Composition Reveals the Source Depth of Slug-Driven Strombolian Explosive Activity. Science 317, 227-230.

M

doi:10.1126/science.1141900.

ED

Cassidy M., Cole P.D., Hicks K.E., Varley N.R., Peters N., Lerner A.H., 2015. Rapid and slow: Varying magma ascent rates as a mechanism for Vulcanian explosions. Earth Planet Sc. Lett.

PT

420: 73-84. doi:10.1016/j.epsl.2015.03.025.

CE

Chaussard, E., Amelung, F., 2012. Precursory inflation of shallow magma reservoirs at west Sunda volcanoes detected by InSAR. Geophys. Res. Lett. 39, L21311.

AC

doi:10.1029/2012GL053817.

Chaussard, E., Amelung, F., Aoki, Y., 2013. Characterization of open and closed volcanic systems in Indonesia and Mexico using InSAR time series. J. Geophys. Res. Solid Earth. 118, 3957–3969. doi:10.1002/jgrb.50288.

32

ACCEPTED MANUSCRIPT Chouet, B. A.; Matoza, R. S., 2013. A multi-decadal view of seismic methods for detecting precursors of magma movement and eruption. J. Volcanol. Geotherm. Res. 252, 108–175. doi:10.1016/j.jvolgeores.2012.11.013 Christopher, T.E., Blundy, J., Cashman, K., Cole, P., Edmonds, M., Smith, P.J., Sparks, R.S.J.

T

Stinton, A., 2015. Crustal‐scale degassing due to magma system destabilization and magma‐gas

IP

decoupling at Soufrière Hills Volcano, Montserrat. Geochem Geophy Geosy. 16, 2797-2811.

CR

doi: 10.1002/2015GC005791

US

Cronin S. J., Hedley M. J., Neall V. E., Smith R. G., 1998. Agronomic impact of tephra fallout from the 1995 and 1996 Ruapehu Volcano eruptions, New Zealand. Environ. Geol. 34, 21–30.

AN

doi:10.1007/s002540050253

Cronin, S. J., Stewart C., Zernack V., Brenna M., Procter J.N., Pardo N., Irwin M., 2014.

M

Volcanic ash leachate compositions and assessment of health and agricultural hazards from 2012

ED

hydrothermal eruptions, Tongariro, New Zealand. J. Volcanol. Geotherm. Res. 286, 233–247. doi:10.1016/j.jvolgeores.2014.07.002

PT

Diller K., Clarke A.B., Voight B., Neri A., 2006. Mechanisms of conduit plug formation:

CE

Implications for vulcanian explosions. Geophys. Res. Lett. 33, 1-6. doi:10.1029/2006GL027391. Edmonds M., Pyle D., Oppenheimer C., 2001. A model for degassing at the Soufriere Hills

AC

Volcano, Montserrat, West Indies, based on geochemical data, Earth Planet Sc. Lett. 186, 159– 173. doi:10.1016/S0012-821X(01)00242-4. Edmonds M. and Herd R.A., 2007. A volcanic degassing event at the explosive‐effusive transition. Geophys. Res. Lett. 34, 21. doi:10.1029/2007GL031379.

33

ACCEPTED MANUSCRIPT Foulger, G. R., Julian, B. R., Hill, D. P., Pitt, A. M., Malin, P. E., Shalev, E., 2004. Non-doublecouple microearthquakes at Long Valley caldera, California, provide evidence for hydraulic fracturing. J. Volcanol. Geotherm. Res. 132, 45–71. doi:10.1016/S0377-0273(03)00420-7 Galle B., Oppenheimer C., Geyer A., Mcgonigle A.J.S, Edmonds M., Horrocks L., 2002. A

T

miniaturised ultraviolet spectrometer for remote sensing of SO2 fluxes: a new tool for volcano

IP

surveillance. J. Volcanol. Geotherm. Res. 119, 241-254. doi:10.1016/S0377-0273(02)00356-6.

CR

Galle B., Johansson M., Rivera C., Zhang Y., Kihlman M., Kern C., Lehmann T., Platt U., Arellano S., Hidalgo S., 2010. Network for Observation of Volcanic and Atmospheric Change

US

(NOVAC)—A global network for volcanic gas monitoring: Network layout and instrument

AN

description. J. Geophys. Res. 115, D05304. doi:10.1029/2009JD011823. Gerlach T.M., Mcgee K.A., Doukas M.P., 2008. Emission Rates of CO2, SO2, and H2S,

M

Scrubbing, and Preeruption Excess Volatiles at Mount St. Helens, 2004 – 2005. In: Sherrod D.,

ED

Scott W., Stauffer P., eds. A Volcano Rekindled: The Renewed Eruption of Mount St. Helens 2004-2006, U.S. Geological Survey Professional Paper 1750, 543-571.

PT

http://pubs.usgs.gov/pp/1750/

CE

González P. J., Singh K. D., Tiampo K. F., 2015. Shallow Hydrothermal Pressurization before the 2010 Eruption of Mount Sinabung Volcano, Indonesia, Observed by use of ALOS Satellite

AC

Radar Interferometry. Pure Appl. Geophys. 172, 3229–3245. doi:10.1007/s00024-014-0915-7. Grainger J.F. and Ring J., 1962. Anomalous Fraunhofer Line Profiles. Nature. 193, 762. Gunawan H., Surono, Budianto A., Kristianto, Prambada O., McCausland W., Pallister J.S., Iguchi M., this volume. Overview of the eruptions of Sinabung 2010 and 2013-present and details of the phreatomagmatic phase. J. Volcanol. Geoth. Res.

34

ACCEPTED MANUSCRIPT Hendrasto M., Surono, Budianto A., Kristianto, Triastuty H., Haerani N., Basuki A., Suparman Y., Primulyana S., Prambada O., Loeqman A., Indrastuti N., Andreas A.A., Rosadi U., Adi S., Iguchi M., Ohkura T., Nakada S., Yoshimoto M., 2012. Evaluation of volcanic activity at Sinabung volcano, after more than 400 years of quiet. J. Disaster Res., 7, 37-47.

T

doi:10.20965/jdr.2012.p0037

IP

Hotta K., Iguchi M., Ohkura T., Hendrasto T., Gunawan H., Rosadi U., Kriswati E. this volume.

CR

Process of magma intrusion and effusion at Sinabung volcano, Indonesia, during the eruptive activity from 2013 to 2016, as revealed from continuous GNSS observation data. J. Volcanol.

US

Geoth. Res.

AN

Iguchi M., Ishihara K., Surono, Hendrasto M., 2011. Learn from 2010 Eruptions at Merapi and

http://hdl.handle.net/2433/151071.

M

Sinabung Volcanoes in Indonesia. Annu. Disas. Prev. Res. Inst., Kyoto Univ. 185–194.

ED

Iguchi M., Surono, Nishimura T., Hendrast, M., Rosadi U., Ohkura T., Triastuty H., Basuki A., Loeqman A., Maryanto S., Ishihara K., Yoshimoto M., Nakada S., Hokanishi N., 2012. Methods

PT

for eruption prediction and hazard evaluation at Indonesian volcanoes. J. Disaster Res. 7, 26–36.

CE

doi:10.20965/jdr.2012.p0026.

Johansson M., 2009. Application of Passive DOAS for Studies of Megacity Air Pollution and

AC

Volcanic Gas Emissions, PhD Thesis, Chalmers University of Technology, 64 pp. Johnson E. R., Wallace P. J., Cashman K. V., Delgado Granados, H., 2010. Degassing of volatiles (H2O, CO2, S, Cl) during ascent, crystallization, and eruption at mafic monogenetic volcanoes in central Mexico, J. Volcanol. Geotherm. Res. 197, 225–238. doi:10.1016/j.jvolgeores.2010.02.017.

35

ACCEPTED MANUSCRIPT Kern C., Deutschmann T., Vogel L., Wöhrbach M., Wagner T., Platt U., 2009. Radiative transfer corrections for accurate spectroscopic measurements of volcanic gas emissions. Bull Volcanol. 72, 233-247. doi:10.1007/s00445-009-0313-7. Kern C., Deutschmann T., Werner C., Sutton A.J., Elias T., Kelly, P.J., 2012. Improving the

T

accuracy of SO2 column densities and emission rates obtained from upward-looking UV-

IP

spectroscopic measurements of volcanic plumes by taking realistic radiative transfer into

CR

account, J. Geophys. Res. 117, D20302, doi:10.1029/2012JD017936.

Massol H. and Jaupart C., 2009. Dynamics of magma flow near the vent: Implications for dome

US

eruptions. Earth Planet Sci. Lett. 279, 185-196. doi:10.1016/j.epsl.2008.12.041.

AN

McCausland W.A., Gunawan H., White R.A., Indrastuti N., Patria C., Suparman Y., Putra A., Triastuty H., Hendrasto M., this volume, Using a process-based model of pre-eruptive seismic

M

patterns to forecast evolving eruptive styles at Sinabung volcano, Indonesia, J. Volcanol. Geoth.

ED

Res.

41. doi:10.1038/46950.

PT

Melnik O. and Sparks R.S.J., 1999. Nonlinear dynamics of lava dome extrusion. Nature. 402, 37-

CE

Mori T., Mori T., Kazahaya K., Ohwada M., Hirabayashi J., Yoshikawa S., 2006. Effect of UV scattering on SO2 emission rate measurements. Geophys Res Lett. 33, 1-5.

AC

doi:10.1029/2006GL026285.

Nadeau P.A. and Williams-Jones G., 2008. Apparent downwind depletion of volcanic SO2 flux—lessons from Masaya Volcano, Nicaragua. Bull. Volcanol. 71, 389-400. doi:10.1007/s00445-008-0251-9. Nakada S., Zaennudid A., Yoshimoto M., Maeno F., Suzuki Y., Hokanishi N., Iguchi M., Ohkura T., Gunawan H., Triastuty H., this volume. Growth process of the lava dome/flow

36

ACCEPTED MANUSCRIPT complex during the 2013-2016 at Sinabung Volcano, North Sumatra, Indonesia. J. Volcanol. Geoth. Res. Nogami K., Hirabayashi J.I., Ohba T., Ossaka J., Yamamoto M., Akagi S., Ozawa T., Yoshida M., 2001. Temporal variations in the constituents of volcanic ash and adherent water-soluble

T

components in the Unzen Fugendake eruption during 1990–1991, Earth Planets Space, 53, 723–

IP

730. doi:10.1186/BF03352400

CR

Otniel Ketaren S., Sudibyato H.A., Hasan W., Purba A., 2016. Environmental Health Aspect In Health Emergency Management (A Case Study : Sinabung Vulcanous Eruption). Int. J. Appl.

US

Nat. Sci. 5, 47-56.

AN

Pallister J.S., Wessels R., Griswold J., Kartadinata N., Gunawan H., Budianto A., Primulyana S., this volume. Monitoring eruptive activity, estimating effusion rates, forecasting collapse events

M

and mapping the distribution of pyroclastic deposits from Sinabung volcano 2013-2016 using

ED

optical, IR and satellite radar imagery. J. Volcanol. Geoth. Res. Prambada O., Zaennudin A., Iryanto., Basuki A., Yulius C., Mulyadi, Rohaeti E., Kisroh,

PT

Suparmo, 2010. Report of geological mapping of Sinabung volcano, regency of Karo, Province

CE

of North Sumatera. Center for Volcanology and Geological Hazard Mitigation. Unpublished report (in Indonesian).

AC

Sparks R.S.J., 1997. Causes and consequences of pressurisation in lava dome eruptions. Earth Planet Sci. Lett. 150, 177-189. doi:10.1016/S0012-821X(97)00109-X. Spilliaert N., Métrich N., Allard P., 2006. S-Cl-F degassing pattern of water-rich alkali basalt: Modelling and relationship with eruption styles on Mount Etna volcano, Earth Planet. Sci. Lett. 248, 772–786, doi:10.1016/j.epsl.2006.06.031.

37

ACCEPTED MANUSCRIPT Sutawidjaja I.S., Prambada O., Siregar D.A., 2013. The August 2010 Phreatic Eruption of Mount Sinabung, North Sumatra, Letusan Freatik Gunungapi Sinabung Agustus 2010, Sumatra Utara. Indones. J. Geol. 8, 55-61. Symonds R.B., Gerlach T.M., Reed M.H., 2001. Magmatic gas scrubbing: implications for

T

volcano monitoring. J. Volcanol. Geotherm. Res. 108, 303-341. doi:10.1016/S0377-

IP

0273(00)00292-4

CR

Vandaele A.C., Hermans C., Fally S., 2009. Fourier transform measurements of SO2 absorption cross sections: II. Temperature dependence in the 29000-44000 cm-1 (227-345 nm) region, J.

US

Quant. Spectrosc. and Radiat. Transfer, 110, 2115-2126, doi:10.1016/j.jqsrt.2009.05.006.

AN

Williams-Jones G., Stix J., Hickson C., 2008. The COSPEC Cookbook: Making SO2 Gas Measurements at Active Volcanoes, IAVCEI, Methods in Volcanology, 1.

M

Witham C.S., Oppenheimer C., Horwell C.J., 2005. Volcanic ash-leachates: a review and

ED

recommendations for sampling methods, J. Volcanol. Geoth. Res. 141, 299–326. doi:10.1016/j.jvolgeores.2004.11.010.

PT

Wolpert R.L., Ogburn S.E., Calder E.S., 2016. The longevity of lava dome eruptions. J.

CE

Volcanol. Geoth. Res., 121, 676–686. doi:10.1002/2015JB012435.

AC

Collected Figure Captions

Figure 1 – Location and photograph of Sinabung Volcano in North Sumatra, Indonesia (3.17 °N, 98.392 °E). The photo of the southeast flank of Sinabung taken on August 16, 2016 from the CVGHM observatory, ~8 km south-southeast from Sinabung’s summit. Features formed during the current eruptive period are labeled: persistent S-bearing fumaroles that were previously mined and remained active throughout the eruptive episode, the large 2014 lava flow, the

38

ACCEPTED MANUSCRIPT currently active lava dome and flow, and the extensive pyroclastic deposits (2013 - present). See Gunawan et al. and Pallister et al. (this volume) for specific details about formation of these

T

eruptive features.

IP

Figure 2 – SO2 emission rates were initially measured by campaign deployment of scanning

CR

DOAS instruments. The scanners were set up at various locations around the volcano at a distance of about 5 km from the vent such that scans could be made perpendicular to the plume

AN

US

propagation direction (in this case, vertical DOAS scans through a horizontal, ‘bent’ plume).

M

Figure 3 – Installation of a NOVAC continuous scanning DOAS site at a location 6 km east of

ED

the Sinabung summit on August 14, 2016. The scanner, seated within a black cylindrical housing and mounted on a 5 m pole, scans the sky from one horizon to the other in search of the gas

CE

PT

plume. A relatively small explosion from Sinabung is occurring in the background.

AC

Figure 4 – Photographs of Sinabung Volcano during various stages of activity in the current eruptive crisis. Phreatic activity began in August 2010 and lasted for several weeks. This was followed by a period of relative dormancy until explosions once again began in September 2013. Activity ramped up after the emergence of a lava dome (December 18, 2013) and peaked following a large dome collapse in mid-January 2014. With continued lava extrusion and regrowth of the lava dome, pyroclastic flows became common, and a substantial lava flow

39

ACCEPTED MANUSCRIPT (clearly visible in February 4, 2014 photo) occurred down the SE flank during January 2014 until mid-spring 2014. Though the size and frequency of occurring explosions has decreased since late February 2014, explosions are still occurring today and are often accompanied by pyroclastic density currents. Lava dome and lava flow extrusion continue today as well, but at substantially

CR

IP

T

lessened rates compared to early 2014 (see Gunawan et al. and Pallister et al., [this volume]).

Figure 5 – SO2 flux, seismic activity, and dominant ash characteristics during the ongoing

US

volcanic crisis at Sinabung. We recognize six phases of the eruption crisis, which are highlighted

AN

here in white or gray backgrounds, and described in the text. The top plot shows the SO2 emission rate in metric tonnes per day, measured during campaign scanning DOAS deployments

M

(red circles). These values are daily measurement averages that represents the effusive degassing

ED

rate between eruptions and generally don’t include gas emissions during major explosions. The NOVAC SO2 flux data in this plot (yellow circles, August 2016) are average background

PT

degassing rates, which exclude high SO2 pulses associated with explosions, so that these fluxes

CE

are comparable to the DOAS campaign measurements. This plot also shows lava extrusion rates from Pallister et al. (this volume); dashed line is a calculated long term average from August

AC

2014 – January 2016. The second panel shows S/Cl molar ratios measured in ash leachates. Note the significant decrease in S/Cl ratios in January 2014 following dome extrusion at the surface. The lower panels show seismic and visually observed activity recorded by CVGHM Sinabung networks and classified into signals associated with various types of volcanic activity (see Gunawan et al.; McCausland et al., [this volume]). We note that near-continuous seismic activity from lava collapses impeded event classifications between mid-February – June 2014.

40

ACCEPTED MANUSCRIPT

Figure 6 – Photos of Sinabung ash particles > 250 μm (plain polarized light) from a subset of ash samples during 2010 – 2014 activity show the progression from phreatic to magmatic

T

eruptive products. September 7, 2010, September 15, 2013, and September 17, 2013 ash consist

IP

entirely of accidental ejecta, including non-altered lavas (dark to pale grey), altered lavas (orange

CR

to reddish), vuggy silica or silica veins that are occasionally sulfide-bearing (milky white +/golden-yellow sulfides), and free-floating plagioclase, pyroxene and Fe-Ti oxide minerals; no

US

juvenile material is present. January 14, 2014 ash was produced after dome extrusion began. This

AN

sample consists of the same types of accidental ejecta, but contains 50 - 60% juvenile material, including clear-grayish glass, clear euhedral feldspar, and vitreous black Fe-Ti oxides and

M

pyroxene. December 7, 2014 ash consists almost entirely of juvenile glass and fresh crystals

ED

(feldspars, Fe-Ti oxides, and pyroxene), with < 10% accidental ejecta. Ash samples from late 2014 are considered mostly juvenile, and therefore ash bulk geochemistry is a close

CE

PT

approximation of the Sinabung magma composition.

AC

Figure 7 – Ash componentry and XRF major and trace elements chemistry of bulk ash from Sinabung Volcano during the progression of phreatic to magmatic eruptions from 2013 - 2016. Ash produced during Phase C eruptions consist of accidental material: altered material, old lavas, and sulfide-bearing hydrothermal material being cleared from the edifice, and have low SiO2, high Fe2O3 (total Fe), MgO, and metals. No ashes were analyzed from phase D. Ash produced during phase E is a mixture of old, accidental material and juvenile glass and crystals, becoming

41

ACCEPTED MANUSCRIPT more SiO2 rich, slightly more K2O and Ba rich, and Fe and trace-metal poor as the contribution of old hydrothermal material decreases. Ash from phase F is largely sourced from the juvenile magma and lava dome, with < 10% accidental material; this late ash is largely representative of

IP

T

the juvenile andesitic-dacitic magma that is fueling present Sinabung eruptive crisis.

CR

Figure 8 – SO2 emission rates measured with the continuous NOVAC scanning DOAS instruments during their first week of operation: August 15 - 22, 2016. DOAS measurements can

US

only be made during daylight hours; the grey windows are periods of low light intensity

AN

(evening/night) that precluded DOAS SO2 measurements. Red squares, blue circles, and yellow triangles correspond to values measured by stations in the NE, E, and SE sectors, respectively.

M

Only scans that registered a plume completeness of ≥ 80% are included, to ensure that the

ED

emission rate reflects measurements through the majority of the plume. Also shown are observed summit eruptions (red asterisks) and hybrid seismic events (green stars), both of which are

PT

associated with large SO2 releases (see text). Seismic signal amplitudes were measured at the

CE

Sukanalu seismic station, 3.7 km to the NE of the summit. Only major seismic events with amplitudes > 25 mm measured from an analog drum recorder are included here. Dates and times

AC

are in local time (WIT; UTC + 9).

42

Figure 1

AC

CE

PT

ED

M

AN

US

CR

IP

T

ACCEPTED MANUSCRIPT

43

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

AN

US

CR

IP

Figure 2

44

US

CR

IP

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

AN

Figure 3

45

M

AN

US

CR

IP

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

Figure 4

46

AC

CE

PT

ED

M

AN

US

CR

IP

T

ACCEPTED MANUSCRIPT

47

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

AN

US

CR

IP

T

Figure 5

48

ED

M

AN

US

CR

IP

T

ACCEPTED MANUSCRIPT

AC

CE

PT

Figure 6

49

AC

CE

PT

ED

M

AN

US

CR

IP

T

ACCEPTED MANUSCRIPT

50

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

AN

US

CR

IP

T

Figure 7

51

M

AN

US

CR

IP

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

Figure 8

52

ACCEPTED MANUSCRIPT

Highlights

T

IP

CR

US AN M



ED



PT



CE



After over 1,200 years of dormancy, Sinabung Volcano became active in 2010 and has been continuously erupting since 2013. Ash bulk geochemistry and S/Cl leachates show the progression of Sinabung activity from early vent-clearing activity in 2010-2013 to a full magmatic eruption by January 2014 Sinabung SO2 emissions generally mirrored lava extrusion rates, but changes in SO2 were found to precede marked changes in lava extrusion by  10 days. SO2 emissions peaked at 1,000-3,000 t/d during January 2014, and have been gradually decreasing from ~500 t/d in early 2014 to 250-300 t/d by mid-2016 Continuous SO2 monitoring since August 2016 with a NOVAC scanning DOAS network shows both explosive and non-explosive gas releases during Sinabung activity

AC



53