RESEARCH ARTICLE
10.1002/2017GC006892
ACO
2
-gas precursor to the March 2015 Villarrica volcano
eruption
Alessandro Aiuppa
1,2
, Marcello Bitetto
1
, Vincenzo Francofonte
2
, Gabriela Velasquez
3
,
Claudia Bucarey Parra
3
, Gaetano Giudice
2
, Marco Liuzzo
2
, Roberto Moretti
4
,
Yves Moussallam
5
, Nial Peters
5
, Giancarlo Tamburello
1,6
, Oscar. A. Valderrama
3
, and
Aaron Curtis
7
1
Dipartimento DiSTeM, Universit
a di Palermo, Palermo, Italy,
2
Istituto Nazionale di Geofisica e Vulcanologia, Sezione di
Palermo, Palermo, Italy,
3
Observatorio Vulcanol
ogico de los Andes del Sur (OVDAS), Servicio Nacional de Geolog
ıa y
Miner
ıa, Temuco, Chile,
4
Dipartimento di Ingegneria Civile Design, Edilizia e Ambiente Seconda Universit
a degli Studi di
Napoli, Naples, Italy,
5
Department of Geography, University of Cambridge, Downing Place, Cambridge, UK,
6
Istituto
Nazionale di Geofisica e Vulcanologia, Sezione di Bologna, Bologna, Italy,
7
Department of Earth and Environmental
Science, New Mexico Institute of Mining and Technology, Socorro, New Mexico, USA
Abstract We present here the first volcanic gas compositional time-series taken prior to a paroxysmal
eruption of Villarrica volcano (Chile). Our gas plume observations were obtained using a fully autonomous
Multi-component Gas Analyser System (Multi-GAS) in the 3 month-long phase of escalating volcanic activity
that culminated into the 3 March 2015 paroxysm, the largest since 1985. Our results demonstrate a
temporal evolution of volcanic plume composition, from low CO
2
/SO
2
ratios (0.65-2.7) during November
2014-January 2015 to CO
2
/SO
2
ratios up to 9 then after. The H
2
O/CO
2
ratio simultaneously declined to
<38 in the same temporal interval. We use results of volatile saturation models to demonstrate that this
evolution toward CO
2
-enriched gas was likely caused by unusual supply of deeply sourced gas bubbles. We
propose that separate ascent of over-pressured gas bubbles, originating from at least 20-35 MPa pressures,
was the driver for activity escalation toward the 3 March climax.
1. Introduction
Recent advances in instrumental monitoring of volcanic gas compositions have markedly improved our abil-
ity to track preeruptive degassing of magmas, and therefore to interpret and predict transition from quies-
cence to volcanic eruption [Edmonds, 2008; Oppenheimer et al., 2014; Aiuppa, 2015; Fischer and Chiodini,
2015]. The advent of the Multi-component Gas Analyzer System (Multi-GAS) [Aiuppa et al., 2005; Shinohara,
2005] in the past decade has enabled systematic measurement of volcanic CO
2
/SO
2
gas ratios at high tem-
poral resolution, and represents a major breakthrough for volcanic gas studies. The temporally resolved vol-
canic gas time-series contributed by permanent Multi-GAS networks have allowed capturing the passive
degassing of CO
2
-rich gas prior to eruption of mafic arc volcanoes [e.g., Aiuppa et al., 2007]. High temporal
resolution gas measurements initially focused on Italian volcanoes, where the first permanent MultiGAS net-
works were installed [Aiuppa et al., 2009, 2010a, 2010b], and where a peculiar CO
2
-rich magmatism [M
etrich
et al., 2004; Kamenetsky et al., 2007] makes gas CO
2
/SO
2
ratios a particularly suitable monitoring parameter.
Recent work at Redoubt in the Aleutians [Werner et al., 2013], Bezymianny in Kamchatka [Lopez et al., 2013],
and Turrialba in Costa Rica [de Moor et al., 2016], all belonging to the category of CO
2
-poor (Group 1) volca-
noes of Aiuppa et al., [2015, 2017], indicates that precursory changes in the volcanic gas CO
2
/SO
2
ratio do
occur in wide-ranging volcano contexts.
The Southern Volcanic Zone (SVZ) of the southern Andes [Hildreth and Moorbath, 1988; Stern, 2004] is an arc
segment where volcanic gases have been found especially poor in carbon [Shinohara and Witter, 2005; Tam-
burello et al., 2014, 2015], possibly due to marginal limestones in the subducted sedimentary succession
[Plank, 2014]. Geochemical studies on erupted magmas [Jacques et al., 2013; Wehrmann et al., 2014] are also
consistent with a relatively modest addition of sediment-derived slab fluids to SVZ magmas, at least relative
to other arc segments (e.g., the Central American Volcanic Arc [Sadofsky et al., 2008]). Villarrica volcano
in Chile (39.428S, 71.938W; Figure 1) is the strongest volcanic gas source within the SVZ, and is
Key Points:
We present the first volcanic gas
compositional time-series taken prior
to a paroxysmal eruption of Villarrica
volcano (Chile)
We find evidence for a gas CO
2
/SO
2
ratio precursor to eruption of a
carbon-poor arc magma
We interpret preeruptive evolution
toward CO
2
-enriched gas as caused
by supply of deeply sourced gas
bubbles
Correspondence to:
A. Aiuppa,
alessandro.aiuppa@unipa.it
Citation:
Aiuppa, A., et al. (2017), A CO
2
-gas
precursor to the March 2015 Villarrica
volcano eruption, Geochem. Geophys.
Geosyst., 18, 2120–2132, doi:10.1002/
2017GC006892.
Received 27 FEB 2017
Accepted 10 MAY 2017
Accepted article online 22 MAY 2017
Published online 14 JUN 2017
V
C
2017. American Geophysical Union.
All Rights Reserved.
AIUPPA ET AL. CO
2
-GAS PRECURSOR TO VILLARRICA ERUPTION 2120
Geochemistry, Geophysics, Geosystems
PUBLICATIONS
world-renowned for its persistent open-vent lava-lake activity [Moreno et al., 1994; Witter et al., 2004; Palma
et al., 2008]. Owing to its glaciated nature and relatively difficult accessibility, Villarrica has been targeted by
sporadic volcanic gas surveys, but not systematically monitored with permanent instrumentation [Witter
et al., 2004; Mather et al., 2004; Shinohara and Witter, 2005; Palma et al., 2008; Sawyer et al., 2011; Moussallam
et al., 2016]. Unfortunately, no volcanic gas report exists that characterizes degassing activity prior to the
paroxysmal explosions that occasionally interrupt the normal, mild lava lake activity.
Here, we present new Multi-GAS-based observations of the Villarrica volcanic gas plume, taken in the
months/days before the volcano’s latest paroxysm in early March 2015 [Global Volcanism Program (GVP),
2016]. Our observations, covering a period of 3 months, improve the currently sparse volcanic gas data
set for this unique but remote volcano. On a broader scale, our results contribute to a better understanding
of lava-lake degassing dynamics [Sawyer et al., 2008a, 2008b; Oppenheimer et al., 2009, 2011; Martin et al.,
2010; Moussallam et al., 2014, 2015, 2016; Allard et al., 2015, 2016; Molina et al., 2015].
2. Villarrica Volcano
Villarrica is a 2847 m high, partially glaciated strato-volcano [Moreno et al., 1994], famous for hosting one of
the few examples of open-vent lava lakes on Earth [Francis et al., 1993]. The Villarrica lava lake is typically
20–30m wide, and is located at depths of 50 to >150m within a funnel-shaped summit crater (Figure 1).
The lake underlies a overhung spatter roof, grown from repeated accumulation and agglutination of ejected
spatter [Palma et al., 2008; Goto and Johnson, 2011]. The regular Villarrica activity is dominated by continu-
ous quiescent outgassing, sustained by convective over-turning of the shallow lava lake reservoir [Witter
et al., 2004; Ripepe et al., 2010]. A variety of bubble bursting activities are also observed at the lake surface,
including small strombolian bursts and 10–20m high lava fountains [Calder et al., 2004; Palma et al., 2008].
Intensity of such bubble bursting activity fluctuates over time, and vigorous phases are typically associated
with higher levels of seismic tremor, SO
2
flux and seismo-acoustic activity [Palma et al., 2008; Richardson
et al., 2014].
This regular activity is occasionally interrupted, at intervals of every few decades on average [Van Daele
et al., 2014], by paroxysmal eruptive periods, punctuated by VEI 2 explosive events and lahars. Lake core
sediment records indicate that at least 22 such paroxysmal lahar-forming eruptions have occurred during
the last 600 years [Van Daele et al., 2014], most recently in 1971, 1984–1985, and March 2015 [Johnson and
Palma, 2015]. The 3 March 2015 paroxysmal eruption interrupted a period of > 20 years of regular activity
following the 1984–1985 and 1991 eruptions [GVP, 2013]. The paroxysmal phase was preceded by
Figure 1. Image sequence showing evolution of Villarrica volcano throughout December 2014 to March 2015. Dates are in the yyyy/mm/dd format. (a) Quiescent degassing activity in
mid-December 2014; (b) and (c) more vigorous lava lake activity (seething magma) in early to mid-February 2015; (d) intense strombolian activity on 2 March 2015; (e) the paroxysmal
lava fountaining activity in the night of 3 March; (f) the post-paroxysm summit of Villarrica on 4 March.
Geochemistry, Geophysics, Geosystems 10.1002/2017GC006892
AIUPPA ET AL. CO
2
-GAS PRECURSOR TO VILLARRICA ERUPTION 2121
approximately 1 month of escalating bubble bursting activity at the crater starting in early February, associ-
ated with high levels of seismicity and infrasound ([GVP, 2016] based on reports by Observatorio Volca-
nol
ogico de los Andes del Sur, OVDAS, Servicio National de Geolog
ıa y Miner
ıa). Due to a progressive
increase in reduced displacement and amplitude of volcanic tremor, the Villarrica alert level was raised to
yellow on 6 February. On 3 March, strombolian activity increased further in vigour and then transitioned at
3:00 AM (local time) to lava fountain activity, with fountains reaching 1500 m above vent altitudes [GVP,
2016]. This violent VEI 2 paroxysm lasted about half an hour, but produced intense tephra fallout, scoria
flows and a 20 km long lahar [Johnson and Palma, 2015].
3. Material and Methods
A fully autonomous Multi-GAS was deployed on 12 November 2014 on the eastern outer rim of Villarrica
crater, at an altitude of 2870 m a.s.l (coordinates: 39825’15’’ S, 71856’21’’W) (Figure 2a). Deployment was con-
ducted by UniPa-INGV-OVDAS personnel (M.B. V.F. and C.B) as part of the Volcanic Deep Earth Carbon
Degassing (DECADE) project of the Deep Carbon Observatory (https://deepcarbon.net/dco_project_sum-
mary?uri5http://info.deepcarbon.net/individual/n7907. The Multi-GAS was housed in a small waterproof-
case (Figure 2b), also containing a Moxa embedded computer (model 7112plus) that commanded the
Multi-GAS operations. The waterproof-case was fit inside a metal box along-side with 4 batteries (Figure 2c),
and the telemetry system (FGR2-PE 900 MHz Industrial Ethernet Radio; from FreeWave Technologies). The
cover of the metal box was fit to a 120 W solar panel and an antenna (Figures 2d and 2e), pointed toward a
radio master OVDAS station on the volcano’s eastern slope. This telemetry system granted regular data
streaming to OVDAS servers where data were stored.
The Villarrica Multi-GAS employed the same sensors as in previous work [e.g., Aiuppa et al., 2014]. In particu-
lar, we used a Gascard EDI030102NG infra-red spectrometer from Edinburgh Instruments (accuracy, 61.5%),
Figure 2. Images demonstrating the Multi-GAS deployment on 12 November 2014. (a) Location of the Multi-GAS site on the Villarrica sum-
mit (the inset is a Goggle Earth map of Villarrica volcano, also showing the location of the VN2 seismic station, which data are used in this
study); (b) the Multi-GAS housed in a waterproof-case; (c) metal box housing 4 batteries and the waterproof-case shown in the previous
image; (d and e) solar panel and antenna.
Geochemistry, Geophysics, Geosystems 10.1002/2017GC006892
AIUPPA ET AL. CO
2
-GAS PRECURSOR TO VILLARRICA ERUPTION 2122
2 electrochemical sensors for SO
2
(T3ST/F - TD2G-1A) and H
2
S (T3H - TC4E-1A) both from City Technology
(repeatability, 1%) and a KVM3/5 Galltec-Mela T/Rh sensor. The Multi-GAS operated from 13 November
2014 (0 AM local time) to 1 March 2015 (0.30 AM local time), when data flow to OVDAS servers ceased. A
reconnaissance survey carried out on the crater days after the 3 March paroxysm confirmed the Multi-GAS
had been buried underneath a thick cover of spatter deposited during the vigorous explosive activity of 1–
3 March (Figure 1). In the 13 November to 1 March interval, the Multi-GAS worked continuously, excepted
for a single data gap between 20 December and 11 January, due to heavy snowfall and consequent icing of
the instrument’s inlet (as verified in a reconnaissance survey in late-December). During operation, the Multi-
GAS measured (at 0.1 Hz rate) the in-plume concentrations of CO
2
,SO
2
and H
2
S during 4 daily measurement
cycles (0–0.30; 6–6.30; 12–12.30; 18–18.30; all Local Time). Ambient pressure, temperature and relative
humidity were also measured, which allowed calculation of in-plume H
2
O concentrations using the Arden
Buck equation [Buck, 1981]. The volcanic H
2
O signal (of a few thousands ppmv) was resolved from the over-
whelming background (ambient) air H
2
O concentration (2500–10,000 ppmv) from analysis of co-acquired
SO
2
concentrations (being only 20 ppb in ambient air). In particular, a polynomial function was fit to sets of
H
2
O readings with contemporaneous SO
2
0[Aiuppa et al ., 2010c]. The so-obtained background air H
2
O
content was subtracted from H
2
O readings to obtain volcanic H
2
O. No H
2
S was detected above the 13%
cross-sensitivity of the H
2
S sensor to SO
2
[Tamburello, 2015].
We postprocess the acquired CO
2
and SO
2
concentration data using the Ratiocalc software [Tamburello,
2015] to obtain time-averaged CO
2
/SO
2
ratios calculated over 30 min long Multi-GAS acquisition windows.
The obtained results are listed in Table 1. Results are only reported for temporal windows in which the SO
2
concentration was above a 5 ppmv threshold, and in which high correlations (R
2
0.7) between CO
2
and
SO
2
concentrations were observed. Each temporal window included 50 to 200 measurements, and the size
of the window was automatically adapted in Ratiocalc to maximize the correlation coefficient. Based on lab-
oratory text, we assess the overall errors at 15% and 30% for the derived CO
2
/SO
2
and H
2
O/CO
2
/ratios.
4. Results
The CO
2
and SO
2
concentrations measured by the Villarica summit crater Multi-GAS instrument are illus-
trated in the temporal plot of Figure 3b. The Multi-GAS almost continually detected SO
2
concentrations well
above the background atmosphere level (20 ppb), implying persistent volcanic gas plume fumigation at
our measurement site (Figure 3b). CO
2
concentrations also typically exceeded the normal atmospheric lev-
els (400 ppmv), and fluctuated around a 420–470 ppmv baseline (Figure 3b). Median concentrations of
SO
2
and CO
2
of 5.4 6 5 and 476 6 23 ppmv, respectively, were observed between 13 November 2014 and 5
February 2015 and represent typical background degassing (Figure 3b). The median SO
2
and CO
2
concen-
trations then increased between 6 February and 1 March 2015 to 9.2 6 14 and 488 6 54 ppmv, indicating
elevated degassing (Figure 3b). The peak SO
2
and CO
2
concentrations were also far larger after 6 February
(respectively of 122 and 1043 ppmv) than in the 13 November 2014 to 5 February 2015 period (45 and 550
ppmv, respectively). Interestingly, gas concentrations increased simultaneously with an abrupt increase in
seismicity (Real-time Seismic-Amplitude Measurement, RSAM; Figure 3a), that induced OVDAS to raise the
volcano’s alert level to yellow on 6 February.
The time-averaged CO
2
/SO
2
ratios, calculated in individual 30 min-long Multi-GAS acquisition windows, are
illustrated in the temporal plot of Figure 3c. Based on the temporal record of volcanic gas CO
2
/SO
2
ratios,
we identify three distinct phases. During Background degassing Phase I (13 November 2014 to 25 January
2015), the derived CO
2
/SO
2
ratios were systematically lower than 3 (range 0.65–2.7), and mostly comprised
between 1 and 2 (all ratios reported here and below are on molar basis). Starting from January 26, in what
we refer to as Phase II (26 January 26 to 5 February), the CO
2
/SO
2
ratios fluctuated more widely, and peak
values as high as 8.3 were noticed. Both visual observations and measurements (e.g., stable pressure read-
ings on the Multi-GAS) indicate that no closing/icing of the inlet or any other malfunctioning occurred,
implying the CO
2
/SO
2
difference between Phase I and II is real, and not an instrumental drift effect. The
mean CO
2
/SO
2
ratio for Phase II was 2.1 (Figure 4), or slightly higher than in Phase I. Fluctuating, high (up to
9.1) CO
2
/SO
2
ratios persisted also during the Phase III (same as in Figure 3b) that preceded the 3 March par-
oxysm. The mean CO
2
/SO
2
ratio for Phase III was 2.7, or the highest between the 3 periods.
Geochemistry, Geophysics, Geosystems 10.1002/2017GC006892
AIUPPA ET AL. CO
2
-GAS PRECURSOR TO VILLARRICA ERUPTION 2123
Table 1. Multi-GAS Derived Volcanic Gas Plume Ratios From Villarrica Volcano
a
Date CO
2
/SO
2
H
2
O/CO
2
Date CO
2
/SO
2
H
2
O/CO
2
Date CO
2
/SO
2
H
2
O/CO
2
13/11/2014 00:04 1.2 19.5 31/01/2015 06:24 3.0 14.8 18/02/2015 18:09 4.3 5.4
14/11/2014 00:27 1.5 42.2 31/01/2015 12:06 1.9 10.9 19/02/2015 06:25 3.2 6.8
14/11/2014 06:14 2.1 7.5 31/01/2015 18:13 2.2 9.6 19/02/2015 12:06 2.5 6.2
14/11/2014 12:14 1.3 24.5 01/02/2015 00:04 3.4 11.6 19/02/2015 18:07 4.0 n.d.
15/11/2014 06:24 1.4 19.1 01/02/2015 06:22 2.4 10.3 20/02/2015 00:12 3.0 8.4
15/11/2014 18:26 1.2 79.4 01/02/2015 12:03 7.1 8.3 20/02/2015 06:04 3.7 n.d.
16/11/2014 00:18 0.8 35.5 01/02/2015 12:03 3.6 16.4 20/02/2015 12:13 3.1 n.d.
19/11/2014 12:10 1.2 n.d. 01/02/2015 12:26 2.6 23.2 20/02/2015 18:16 1.7 23.6
19/11/2014 18:28 0.6 n.d. 02/02/2015 00:12 2.0 29.2 21/02/2015 00:06 1.4 13.5
23/11/2014 00:18 2.1 n.d. 02/02/2015 06:25 3.3 n.d. 21/02/2015 06:14 1.9 22.0
23/11/2014 04:35 1.2 n.d. 02/02/2015 18:24 0.7 131.9 21/02/2015 12:09 2.5 14.6
23/11/2014 06:23 1.5 n.d. 02/02/2015 18:27 1.7 n.d. 21/02/2015 18:05 3.1 5.3
24/11/2014 12:25 1.2 n.d. 03/02/2015 18:06 2.1 n.d. 22/02/2015 12:16 2.1 9.1
24/11/2014 18:25 1.5 n.d. 04/02/2015 00:12 3.3 25.3 22/02/2015 18:17 1.9 13.0
25/11/2014 00:15 1.9 n.d. 04/02/2015 06:19 3.8 21.3 23/02/2015 00:15 1.5 12.9
25/11/2014 06:24 1.2 n.d. 04/02/2015 12:12 5.9 n.d. 23/02/2015 06:04 3.4 19.6
25/11/2014 12:26 0.7 8.7 04/02/2015 18:05 2.6 16.0 24/02/2015 00:00 1.9 11.5
26/11/2014 12:22 1.1 12.2 04/02/2015 18:25 2.4 n.d. 24/02/2015 06:15 1.6 12.8
27/11/2014 12:15 1.9 17.9 05/02/2015 06:13 3.7 12.4 24/02/2015 12:14 2.8 4.7
28/11/2014 18:10 1.7 11.2 05/02/2015 06:22 3.3 n.d. 25/02/2015 00:12 2.4 8.4
29/11/2014 18:11 1.2 28.2 05/02/2015 18:04 1.5 36.6 25/02/2015 06:12 2.8 6.0
30/11/2014 12:01 1.9 21.2 05/02/2015 18:07 2.3 n.d. 25/02/2015 12:11 3.1 n.d.
01/12/2014 06:09 1.1 9.0 05/02/2015 18:17 0.9 n.d. 25/02/2015 18:06 2.1 8.1
02/12/2014 00:26 0.9 18.8 06/02/2015 00:13 2.0 12.0 26/02/2015 00:06 1.7 15.1
03/12/2014 06:14 1.3 15.4 06/02/2015 00:19 2.4 n.d. 26/02/2015 06:00 2.7 5.2
03/12/2014 18:07 1.3 16.8 06/02/2015 06:05 2.3 38.6 26/02/2015 12:11 2.8 7.6
04/12/2014 00:11 1.4 14.3 06/02/2015 06:13 1.9 n.d. 26/02/2015 18:08 2.4 6.6
04/12/2014 06:21 1.2 19.6 06/02/2015 06:22 1.4 n.d. 27/02/2015 06:10 3.4 3.6
04/12/2014 18:07 1.2 48.3 06/02/2015 12:14 1.6 16.0 27/02/2015 12:16 2.9 12.3
05/12/2014 00:19 1.7 47.8 06/02/2015 12:23 1.4 n.d. 27/02/2015 18:14 3.8 n.d.
05/12/2014 18:05 2.0 17.9 06/02/2015 18:06 1.5 53.3 28/02/2015 00:09 2.2 14.4
06/12/2014 00:14 1.6 14.0 07/02/2015 00:12 1.9 16.9 28/02/2015 12:20 2.5 n.d.
06/12/2014 06:04 1.9 n.d. 07/02/2015 06:06 1.9 12.4 28/02/2015 18:24 1.3 32.5
08/12/2014 00:13 1.3 n.d. 07/02/2015 06:10 1.6 n.d. 01/03/2015 00:25 2.4 2.8
08/12/2014 12:16 1.3 16.9 07/02/2015 12:03 1.9 11.3
08/12/2014 18:16 0.8 62.1 07/02/2015 12:11 3.1 n.d.
09/12/2014 00:26 1.5 5.7 07/02/2015 12:24 5.0 n.d.
10/12/2014 00:01 1.8 n.d. 07/02/2015 18:06 2.2 13.5
10/12/2014 06:06 2.6 7.1 07/02/2015 18:16 2.9 n.d.
10/12/2014 12:23 2.7 9.3 07/02/2015 18:27 2.3 n.d.
11/12/2014 06:25 2.7 n.d. 08/02/2015 00:06 3.8 5.5
12/12/2014 12:24 1.5 16.6 08/02/2015 00:18 3.2 n.d.
12/12/2014 18:19 1.5 25.6 08/02/2015 06:05 4.3 4.2
13/12/2014 00:14 1.5 17.1 08/02/2015 06:10 3.6 n.d.
14/12/2014 18:15 1.6 28.4 08/02/2015 06:15 5.3 n.d.
15/12/2014 00:16 2.2 n.d. 08/02/2015 12:08 3.4 6.7
15/12/2014 06:27 2.3 n.d. 08/02/2015 12:17 3.6 n.d.
16/12/2014 06:15 2.4 5.8 08/02/2015 12:25 3.3 n.d.
16/12/2014 12:09 1.9 12.4 09/02/2015 00:06 3.6 19.5
16/12/2014 18:03 2.1 17.9 09/02/2015 00:12 3.3 n.d.
17/12/2014 06:16 1.7 10.0 09/02/2015 00:15 3.0 n.d.
17/12/2014 06:22 1.5 n.d. 09/02/2015 00:23 5.3 n.d.
18/12/2014 06:21 1.6 14.5 09/02/2015 00:23 4.3 n.d.
14/01/2015 00:13 1.3 n.d. 09/02/2015 06:03 5.0 n.d.
14/01/2015 06:27 1.8 n.d. 09/02/2015 06:08 6.7 n.d.
14/01/2015 12:08 2.3 8.3 09/02/2015 06:25 5.3 n.d.
14/01/2015 18:02 1.2 n.d. 09/02/2015 18:07 5.0 5.4
15/01/2015 12:19 1.4 n.d. 10/02/2015 00:03 2.7 10.0
15/01/2015 18:24 1.9 n.d. 10/02/2015 06:09 2.4 23.9
17/01/2015 12:19 1.7 14.9 10/02/2015 12:05 3.2 7.5
22/01/2015 18:18 1.1 29.6 10/02/2015 18:10 1.9 25.2
23/01/2015 00:17 2.4 5.6 11/02/2015 00:10 2.0 5.5
23/01/2015 06:16 1.7 20.6 11/02/2015 06:09 1.3 28.3
23/01/2015 12:10 1.1 31.8 11/02/2015 12:15 1.8 6.6
23/01/2015 18:03 1.7 22.6 11/02/2015 18:12 2.6 38.2
24/01/2015 00:17 1.9 17.9 12/02/2015 00:23 1.2 32.7
24/01/2015 06:10 2.0 18.7 12/02/2015 06:23 2.1 0.7
25/01/2015 00:15 2.6 n.d. 12/02/2015 12:26 1.6 10.4
Geochemistry, Geophysics, Geosystems 10.1002/2017GC006892
AIUPPA ET AL. CO
2
-GAS PRECURSOR TO VILLARRICA ERUPTION 2124
![Figure 3. Temporal records of (a) Real-time Seismic-Amplitude Measurement (RSAM) [Endo and Murray, 1991] taken from the VN2 station, located 4 km northeast of the Villarrica crater (see inset in Figure 2a). Seismic data were filtered in the 0.4–9 Hz range to maximize the signal/noise ratio. This 0.4–9 Hz frequency range includes the majority of seismic energy irradiated by Villarrica, and is poorly affected by nonvolcanic (meteorological) effects [Palma et al., 2008]. (b) CO2 and SO2 concentrations at the Villarrica Multi-GAS site and (C) derived CO2/SO2 ratios. In Figure 3a, the sequence of alert levels issued by OVDAS is also indicated. Each point in Figure 3c represents the mean CO2/SO2 ratio obtained from processing of CO2 and SO2 concentration data (using the Ratiocalc software [Tamburello, 2015]) acquired in an individual 30 min-long Multi-GAS acquisition window (data from Table 1). Three distinct degassing periods (Phases I, II, and III) are identified (see text).](/figures/figure-3-temporal-records-of-a-real-time-seismic-amplitude-91tpq68m.webp)
