Journal of Volcanology and Geothermal Research 257 (2013) 12–20

Contents lists available at SciVerse ScienceDirect

Journal of Volcanology and Geothermal Research

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

Continuous monitoring of fumarole temperatures at Mount Etna (Italy)

Paolo Madonia, Andrea L. Rizzo ⁎, Iole S. Diliberto, Rocco FavaraIstituto Nazionale di Geofisica e Vulcanologia, Sezione di Palermo, Via Ugo La Malfa 153, 90146 Palermo, Italy

⁎ Corresponding author. Tel.: +39 91 6809407; fax: +E-mail address: a.rizzo@pa.ingv.it (A.L. Rizzo).

0377-0273/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.jvolgeores.2013.03.001

a b s t r a c t

a r t i c l e i n f o

Article history:Received 10 September 2012Accepted 6 March 2013Available online 14 March 2013

Keywords:Continuous monitoringMount EtnaFumarole temperatureMeteorological parametersVolcanicDegassing

In this paper we present the first data of temperature continuously recorded in two fumarole fields (designat-ed VOR and HOR) located in the summit area of Mount Etna volcano (Italy). The time series embraces twodistinct periods: (1) October 2007 to November 2009, during which an effusive eruption occurred fromMay 2008 to July 2009, and (2) November 2011 to June 2012, characterized by the occurrence of strong par-oxysms (fire fountains and lava flow). The analysis of the temperature signal in both the time and frequencydomains, and its comparison with meteorological observations allowed us to separate the exogenous influ-ences from the effects of variations in the activity state of the volcano. The acquired data were weakly affect-ed by seasonal cycles of the air temperature and strongly affected by the rainfall. Optimization of siteconditions (i.e., sensor depth and soil permeability) markedly reduced meteorological disturbances. The dis-tance from the main degassing and/or eruptive fractures was crucial to maximizing the probability of thetechnical survival of the monitoring apparatus, which was seriously affected by the emission of acidicgases, tephra fallout, and lava flows. Apart from the exogenous influences, the most appreciable variationwas observed at VOR, where a huge increase in fumarole temperature was detected immediately after theonset of the 2008–2009 eruption. Such an anomalous increase was attributed to the rapid ascent of magmafeeding the eruptive fracture. Another abrupt increase in temperature was recorded at HOR in March andApril 2012. During this period the frequency of paroxysm occurrence increased markedly, and this led usto hypothesize that the thermal anomaly was due to the intrusion of a new batch of magma in the conduitsof the southeast crater. Medium- to long-term monitoring (weeks to months) of fumarole temperatures re-vealed variations that were attributed to pressurization/depressurization phases of the shallow volcanic sys-tem, which varied between the various monitored sectors of the volcano. Our observations suggest thatcontinuous monitoring of fumarole temperature can give useful information about the activity of MountEtna. Moreover, due to the complexity of its shallow plumbing system, we conclude that the monitoring sys-tems should be extended to cover the entire fumarole network of the summit area.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

One of the most exciting goals during the study of a volcanic sys-tem is the evaluation of the state of activity aimed to eruption fore-casting; in doing this, the multi parametric approach is crucial torecognize and followmagma dynamics in the plumbing system.More-over, a high-frequency data acquisition is essential to avoid the risk ofmissing information. As a consequence, volcanic monitoring has beenmostly based on geophysical techniques. Indeed, gases released fromvolcanoes may be sampled and analyzed only discontinuously (e.g.,Hirabayashi, 1986; Giggenbach and Goguel, 1989; Pecoraino andGiammanco, 2005; Martelli et al., 2008; Liotta et al., 2010, 2012;Paonita et al., 2012), because during eruptive periods or extrememeteorological conditions the approach of summit areas is very haz-ardous, allowing only an incomplete understanding ofmagma dynam-ics. In addition, the delay (from days to months) between sampling

39 91 6809449.

rights reserved.

and analysismay represent a further issue to a timely evaluation, lead-ing very often to an “a posteriori” assessment of activity state.

Only recently geochemical monitoring has been integrated withtechniques that permit a high-frequency data acquisition of some pa-rameters and it has been thus accepted by the scientific community asan essential tool to integrate geophysical background. One of the firstapplications regarded the routine monitoring of SO2 flux from plumegases by remote sensing, which has allowed data acquisition at safedistances from summit craters and volcanic activity to be followedduring eruptive periods. At Kilauea (Hawaii) SO2 flux has been moni-tored since 1979 (Sutton et al., 2001), while in the following yearsthese measurements have been extended to other volcanoes such asfor instance Stromboli (Italy) (Allard et al., 1994), Mount Etna (Italy)(Caltabiano et al., 1994) and Soufrière Hills (Montserrat) (Young etal., 1998). Today this parameter is monitored nearly real-time inmost of the volcanic systems locatedworldwidewithmore innovativetechniques. The chemical composition of plume gases is routinely in-vestigated by FTIR measurements, which allow to record remarkableshort-period (seconds) variations, such as for instance at Stromboli

13P. Madonia et al. / Journal of Volcanology and Geothermal Research 257 (2013) 12–20

(Burton et al., 2007). In addition, Aiuppa et al. (2007) have reportedthe first systematic set of simultaneous H2O, CO2 and SO2 acquisitionin the Etnean gas plume during the 2006 eruption. In the last decades,technological innovation has also concerned the investigation ofchemical composition of gases released from fumaroles. It has startedfrom the measurement of a single gas component, e.g. O2 (Benhamouet al., 1988), SO2 (Bluth et al., 1994) and Rn (Cigolini et al., 2001). From1995 to 1998, Shimoike and Notsu (2000) have continuously moni-tored several components at Izu-Oshima (Japan) volcano; similarstudies have been performed at Galeras (Colombia) (Faber et al.,2003), Merapi (Indonesia) (Zimmer and Erzinger, 2003) and Nisyros(Greece) (Teschner et al., 2007). Geochemical, volcanological, andgeophysical observations have recently been mutually integrated,giving new insights on magma dynamics at considerable depths andpermitting to assess the state of activity of a volcano with increasingreliability (e.g., Andronico et al., 2005; Caliro et al., 2005; Harris etal., 2005; Gottsmann et al., 2007; Aiuppa et al., 2010; Carniel et al.,2010; Cannata et al., 2011).

The investigation of chemical composition of volcanic gases maybe coupled with continuous monitoring of fumarole temperature, be-cause the latter is closely related to variations in the fluxes of deep andhot gases (e.g., Tedesco et al., 1991; Connor et al., 1993; Alparone et al.,2004; Brusca et al., 2004; De Gregorio et al., 2007; Rizzo et al., 2009;Cannata et al., 2011; Diliberto, 2011). In detail, when steam flux in-creases, as a consequence of an enhanced degassing during magmaupraise toward the surface, a gradual and/or a sharp rise of fumaroletemperatures may be recorded at the surface. The thermal anomalymay precede the onset of eruptive activity, as observed at VolcànPoàs (Costa Rica) in 1980 (Barquero, 1983, 1988). On the contrary,when the discharge of volatiles reduces, it is reasonable to expect a de-crease of fumarole temperatures, coherentwith previous observationsin other volcanic systems (Stoiber et al., 1975; Connor et al., 1993).Other precious information comes from thermal imaging. Calvari etal. (2004) obtained helpful results during the 2002–2003 eruptionsof Mt Etna and Stromboli, individuating the opening of fissure systemsrelated to the increasing volcanic activity. Thus, this parameter can beconsidered useful in evaluations of volcanic activity, even thoughanomalies caused by variations of the local tectonic stress field (viasoil permeability changes) may interfere with the pure volcanogenicsignal, as observed at Merapi (Indonesia) (Zimmer and Erzinger,2003; Richter et al., 2004), at Stromboli (Italy) (De Gregorio et al.,2007) and at Vesuvius (Italy) (Madonia et al., 2008). Fumarole tem-peratures are also subject to changes due to exogenous causes, mostlyconnected to meteorological parameters that must be contemporarilymonitored. Indeed, Connor et al. (1993) have demonstrated that baro-metric pressure has an inverse correlation with fumarole temperatureand mostly affects low-temperature emissions also with a low massflow. The direction and velocity of the wind may also change the fu-marolic system, as observed by Faber et al. (2003) and Liotta et al.(2010), while rainfall is the most critical cause of temperature drops(e.g., Zimmer and Erzinger, 2003; Friedel et al., 2004; Richter et al.,2004).

At Mount Etna only few studies based on discrete temperaturemonitoring have been carried out. The greatest variations were ob-served during 1991–1994 by Bonfanti et al. (1996), who measuredan increase in water temperature in springs and wells prior to theonset of the 1991–1993 eruption. Alparone et al. (2004) and Pecorainoand Giammanco (2005) performed discrete monitoring of low-temperature fumaroles located in the upper part of the volcano, butthey found no clear correlation with volcanic activity. Finally, Liottaet al. (2010) studied the fluid circulation in the summit area for a fewmonths by monitoring variations of a high-temperature fumarole nearVoragine crater (hereafter called VOR).

In the present study we show the results of the first continuousmonitoring of two low-temperature fumaroles located in the summitarea of Mount Etna volcano during two distinct periods: (1) from

October 2007 to November 2009 and (2) from November 2011 toMay 2012. The former period was characterized by a new long-lasting eruption that occurred between May 2008 and July 2009,while the latter was characterized by frequent paroxysmal activity.We investigate how exogenous (i.e., meteo-astronomical) and endog-enous (e.g., degassing and volcanic activity) parametersmay influencefumarole temperature, both on short (hours to days) and medium(months) timescales, comparing thermal data with geodeticmeasure-ments recorded during the same period.

2. Methods and site descriptions

Soil temperatures were acquired hourly at depths of 10 and 30 cmby automated stations, based on Onset Microstation four-channeldata loggers, each connected to two 12-bit digital temperaturesmart sensors from the same manufacturer. Sensors operate in therange −40 °C to 100 °C, with a resolution of 0.03 °C and an accuracyof ±0.2 °C. Temperatures were measured every 4 min and theirhourly averages were stored in the permanent memory of the logger.

Two measuring sites (Fig. 1) were selected from among the loca-tions compatible with safety conditions for the installation and main-tenance of instruments, on the basis of geochemical investigationscarried out in the summit area, but also considering the main activestructures feeding soil degassing of magmatic fluids and recent erup-tive activities.

The firstmeasuring site is located at 3260 m a.s.l. on the northeast-ern inner rim of VOR, which is one of the central craters, and wasmonitored from October 2007 to November 2009. Previous geochem-ical studies were performed by Martelli et al. (2008) and Liotta et al.(2010, 2012) on a high-temperature fumarole (named F4, up to500 °C) located close to our monitoring site. From these studies thecomposition of the fumarole — even though highly contaminated byair—was only slightly affected by postmagmatic processes (i.e., scrub-bing and hydrothermal contributions), and thus reflected the degas-sing of pristine magmatic gases, like those characterizing the plume.In addition, Liotta et al. (2010) continuously recorded its temperaturevariations for a short period (July–September 2008), which revealed arelationship with wind speed due to the Venturi effect acting on theopen steam-feeding cracks. To avoid this effect, we installed our datalogger and thermal sensor at a small distance from the hottest frac-tures, where the exposed surface appeared coherent and unfractured,thereby protecting the system from damage due to acidic corrosionand overheating. Considering that the soil around the fumaroles isheated by steam discharged through advective degassing, the param-eter we monitored was the steam-heated soil temperature (hereafterSHST).

The second site is located at about 3000 m a.s.l. next to the“hornitos” formed during the 1999 eruption (hereafter called HOR)and positioned south of the southeast crater (hereafter SEC). Thisfumarole field is located on the southeast rift that represents one ofthe main degassing fractures in the Etnean summit area, which hasbeen known since the 1787 eruption and named “Vulcarolo” (Ponte,1927; Aubert, 1999; Alparone et al., 2004). This field is covered bylava and tephra that has erupted from SEC since 1989, but its hugedegassing has persisted over time. In the first decades of the 20th

century the high steam flux from Vulcarolo has led to place a condens-er in order to collect water for feeding the neighboring observatory(Ponte, 1927). Anomalous degassing from this fumarole field wasidentified by Alparone et al. (2004), who investigated the CO2 concen-tration of soil along a transect crossing the main fracture. A prelimi-nary study showed that the soil temperatures were generally below90 °C, which is the boiling temperature of water at that altitude, andthe gas is directly emitted through advective degassing. During ourstudy HOR was monitored during two periods: from November 2008to November 2009 and from November 2011 to May 2012; the datafrom December 2009 to October 2011 are missing because the first

Fig. 1. Digital elevation map of Mount Etna craters modified from report no. 35/2011 of INGV-CT (www.ct.ingv.it). Sketch map of Italy and the Etnean areas modified from Liotta etal. (2012) and Paonita et al. (2012), respectively. VOR, Voragine crater; BN, Bocca Nuova; NEC, northeast crater; SEC, southeast crater; NSEC, new southeast crater. The acronymsVOR and HOR (in red) represent the locations of the monitored fumaroles (see text for details). Track A–B indicates the cross section reported in Fig. 7. (For interpretation of thereferences to color in this figure legend, the reader is referred to the web version of this article.)

14 P. Madonia et al. / Journal of Volcanology and Geothermal Research 257 (2013) 12–20

station was lost due to the entire area being covered by more than1.5 m of volcanic scoria that erupted in that period. A new automaticsystem was installed in November 2011.

The temperature data of fumaroles were compared with meteoro-logical parameters such as air temperature, wind speed, atmosphericpressure, and rainfall. Air temperature, wind speed, and atmosphericpressure variation at 700 hPa (corresponding to ~3000 m a.s.l.) wereobtained from the MARS archive of the European Centre for Medium-Range Weather Forecasts (http://data-portal.ecmwf.int) under licensefrom the Italian Air Force Meteorological Service. Data of the daily rain-falls were acquired at Pedara, located at 810 m a.s.l. on the southernflank of Mount Etna, and furnished by the Sicilian Regional Agro-Meteorological Service (http://www.sias.regione.sicilia.it).

The power spectra of the acquired data were obtained using anFFT algorithm implemented in freeware software (Ky-Plot).

3. Mount Etna activity

Mount Etna is the most active volcano in Europe characterized byhuge degassing activity that largely occurs at the main craters [NEC,SEC, and central craters (VOR, Bocca Nuova 1, and Bocca Nuova2); Fig. 1]. SEC does not show a continuous quiescent degassing, butin recent times it has been themost active, with intermittent explosiveand effusive activity occurring during the last decade that produced inthe last 2 years a new cone on the eastern side of the old crater. Thisnew SEC (hereafter called NSEC; Fig. 1), which was initially named a“pit crater” due to its smallness, is periodically characterized by parox-ysmal activity (mainly fire fountains and lava flows) that generallylasts only a few hours (see internal report at www.ct.ingv.it).

During our monitoring the volcano was characterized by variableactivity (both explosive and effusive), which may be summarizedin three phases: (1) from October 1, 2007 to May 10, 2008; (2) fromMay 13, 2008 to November 30, 2009; and (3) from November 2011to May 2012. The first phase includes two episodes of lava fountains(on November 23 and 24, 2007 and May 10, 2008) which occurred

from a pit crater located on the eastern SEC flank. These eruptiveevents represented the sixth and seventh paroxysms of a series thatstarted on March 29, 2007 after a period of quiescence that followedthe 2006 eruption (Bonaccorso et al., 2011a and references therein).Each episode, which generally lasted a few hours, was characterizedby variable activity moving from intense Strombolian explosions tofire fountains and tephra eruptions, and also accompanied by an east-ward lava flow. The paroxysm that occurred on May 10, 2008 wasamong the most violent explosive crisis recorded since 2003, and oc-curred 3 days before the 2008–2009 effusive eruption (Di Grazia etal., 2009; Bonaccorso et al., 2011a,b).

The second period of activity started on May 13, 2008, when astrong shallow seismic swarm preceded the onset of the 2008–2009eruptive crisis (Di Grazia et al., 2009). The main feature of this erup-tion was a long-lasting lava flow expanding within the upper Valledel Bove. The initial phases of the crisis were also characterized bysustained Strombolian activity at the fracture, probably due to the as-cent of a more volatile-rich magma, as indicated by the compositionaldata of the products that erupted 1 month after the onset of the erup-tion (Bonaccorso et al., 2011b). Thereafter an almost continuous lavaeffusion lasted until July 6, 2009, when the eruption ceased. Thefollowing months were characterized by passive degassing from thecraters and from the old eruptive fracture. Strombolian activity tem-porarily resumed on November 6, 2009 on the eastern flank of SEC(see internal report at www.ct.ingv.it).

The third period was characterized by eight paroxysms that oc-curred at NSEC during a passive degassing phase of the volcano. Thetemporal sequence of the events was as follows: November 15, 2011,and January 5–27, February 9, March 4–18, and April 1–12–24, 2012.The paroxysm that occurred on February 9, 2012 was preceded byStrombolian activity lasting 13 days (see internal report at www.ct.ingv.it). All of the eruptive episodes (similarly to those that occurredin previous years) lasted a few hours and where characterized by aninitial Strombolian activity followed by fire fountains and a lava flowmoving eastward toward Valle del Bove.

15P. Madonia et al. / Journal of Volcanology and Geothermal Research 257 (2013) 12–20

4. Data presentation

The fumarole temperatures recorded at VOR andHOR are shown inFigs. 2 and 3, together with the rainfall andmonthly averages of the airtemperature. Areamovements recorded by Aloisi et al. (2011) at sum-mit elevations are also plotted, and their variations are discussedbelow. We present data acquired at a depth of 30 cm because theywere less influenced by the exogenous noise. This can be deducedfrom Fig. 4, which shows the power spectra of signals acquired atVOR at the two depths (i.e., 10 and 30 cm). The spectra are quite sim-ilar, with the main difference being that the 24-hour component isclearly visible at 10 cm but completely absent at the greater depth, in-dicating that the effect of the daily solar thermal cycle vanishes (at thislatitude and elevation) within 30 cm from the surface.

SHSTs ranged from 24.4 °C to 67.7 °C at VOR, from 0 °C to 63 °C atHOR installed in 2008, and from 16 °C to 25 °C at HOR reinstalled in2011, displaying both high- and low-frequency variations during theinvestigated period.We describe themedium- to long-term variations(lasting months) in Section 4.1, and the short-term ones (lasting fromhours to days) in Section 4.2.

4.1. Medium- to long-term variations

The main exogenous factor influencing soil temperature on theseasonal timescale is the variation of solar irradiance, whose bestproxy is the air temperature. We determined whether or not this ef-fect occurred by applying a frequency analysis to fumarole tempera-ture data obtained at VOR (Fig. 4). We split the entire investigatedperiod into four consecutive time windows of 6 months each,according to the alternation of the warmer and dryer months (May–October) with the colder and wetter months (November–April) ofthe hydrological year. We also plotted for comparison the spectra ofa temperature signal acquired during 2005–2007 at the same depth(i.e., 30 cm) in an inactive volcanic area of Sicily (Mt. Fossa delle

Fig. 2. From top to bottom: hourly temperature data acquired at 30-cm depth at HOR (thin bsummit area of Mount Etna (light gray line; sketch from Aloisi et al., 2011), hourly temperad) indicating periods of activity variation (see text for details), and monthly running averagelines indicate onset and end of the 2008–2009 eruption.

Felci, ~900 m a.s.l., Salina Island), where the main variations are ex-clusively due to air temperature and solar irradiance. Under normalconditions the spectra of soil temperatures should show high-frequency components (peaking at 24 h) whose amplitudes arelower during the cold season than during the warmer season, whenthe solar irradiance is stronger and more effective in modifying thethermal state of the ground. This behavior was found in the datafrom Salina, plotted at the top of Fig. 4, where the 24-hour componentand its second-order harmonic at 12 h are clearly visible, and whoseamplitudes are higher in spring and summer than in autumn andwin-ter. A similar behavior was discussed by Connor et al. (1993), who ob-served a daily fluctuation of fumarole temperature at Colima volcano(Mexico).

The VOR signal did not contain the 24-hour component, whichwasreplaced by short-period noise characterized by a progressive andcontinuous increase in the signal from the first (May–October 2008)to the last (November 2008 to April 2009) selected period (Fig. 4),which can be attributed to the existence of a variable heat source(volcanogenic steam flow) that is not affected by season. The continu-ous increase in this noise during our observational period can beinterpreted as a progressive increase in the endogenous heat flux.

Themedium- to long-term observations of VOR data in the time do-main seem to confirm the frequency analysis, with the seasonal cycleshowing a weak effect on fumarole temperature (Fig. 2), whose trendcan be subdivided in four phases labeled from (1) to (4) in the figureas follows:

1) From October 2007 to April 2008 the fumarole temperature grad-ually decreased from about 40 °C to about 25 °C while air temper-ature was also decreasing; after January 2008 the air and fumaroletemperatures displayed opposite trends, suggesting a decouplingof the two signals. A slight increase in the SHST was recorded atVOR only after April 2008, probably as a response to a change inthe thermal regime.

lack line) and its third-degree polynomial fit (thick dark gray line), deformation of theture data acquired at 30-cm depth at VOR (black line) with arrows (labeled from a toof air temperature (thin dark gray line) and daily rainfall (vertical bars). Vertical dotted

Fig. 3. Hourly temperature data acquired at 30-cm depth at the new HOR sampling station (black line). Six-hour averages of air temperature (gray line) and daily rainfall (gray bars)are also plotted. Paroxysmal activity that occurred during this period is indicated by gray stars (see text for the dates).

16 P. Madonia et al. / Journal of Volcanology and Geothermal Research 257 (2013) 12–20

2) The second period was from May to November 2008. The SHSTincreased by about 25 °C within 5 days of the onset of the2008–2009 eruption (i.e., after May 15), which constitutes almost60% of its total variability range observed during our monitoringperiod. After this the variation temperatures progressively de-creased to reach values comparable to those observed immediatelybefore the eruption onset. Also during this period the correlation

Fig. 4. Top plot is the power spectrum (obtained by FFT analysis) of soil temperaturedata acquired at 30-cm depth at Mt. Fossa delle Felci (Salina Island, Sicily). Middleplot is the same analysis for VOR, subdivided into autumn-and-winter and spring-and-summer consecutive subperiods (see text for details). Bottom plot shows the anal-ysis of the entire series at depths of 10 and 30 cm. Each zero baseline has been arbitrari-ly shifted to avoid superposition.

with air temperature was missing, with the exception fromSeptember to November when both the fumarole and air temper-atures contemporarily decreased.

3) FromNovember 2008 to June 2009 a new change in the thermal re-gime was recorded at VOR, with the temperature doubling fromabout 30 °C to about 60 °C. This increasing trend stopped a fewdays before the end of the eruption on July 6. The average SHSTvalues recorded in summer 2009 were higher than those recordedduring 2008, as already observed in the frequency analysis (Fig. 4),while the air temperature showed similar values during theseyears. We therefore assume that an increase in the steam flux oc-curred that was unrelated to the eruption ending and independentof meteorological effects.

4) During the fourth period, from June to November 2009 (the end ofour observations), VOR temperatures did not show relevant varia-tions, with a stabilization or a slight decrease in values. However,there were peaks superimposed on this main trend, indicating theoccurrence of short-term variations of the steam flux.

The SHSTs at HOR measured during the two distinct periods arenot directly comparable because a new sampling station was installedto replace the previous one that was lost due to the deposition of1.5 m of scoria in the area between December 2009 and October2011. Although the new station was at the same location, soil perme-ability at the measuring site had changed, which would have locallymodified the heat flux. As illustrated in Fig. 2, during the first moni-toring period the thermal signal was much more noisy at HOR thanat VOR, reflecting differences in soil permeability between the twosites: VOR is characterized by fine-sized ash whereas the scoriapresent at HOR is more permeable, so favoring the thermal exchangebetween the subsoil and the atmosphere via aeration (e.g., cooling)and/or the infiltration of meteoric precipitation. As a consequence,temperature variations at HOR were driven by a seasonal cycle, withminima in March and maxima in July, which would have significantlyaffected the fumarole temperatures. Nevertheless, a general trend ofdecreasing values can still be observed during the monitored period:the temperatures recorded in November 2009 were lower than thosemeasured in November 2008, when the peaks exceeded 60 °C, asexpected due to the end of the eruption. No marked differenceswere recorded in the average air temperature (which was only 1 °Clower in November 2009 than in November 2008), indicating thatthe observed decreasing trend was probably due to an endogenousfactor rather than a meteorological effect.

During the second period of monitoring (i.e., November 2011 toNovember 2012) HOR showed a very different behavior relative to

17P. Madonia et al. / Journal of Volcanology and Geothermal Research 257 (2013) 12–20

the previous observations (Fig. 3). The meteorological noise affectingthe SHST data, which was mainly due to rain, appeared to be dramat-ically reduced both in amplitude and frequency. The evidence of aseasonal cycle was absent, and was substituted by a general trend ofdecreasing SHST (from 23 °C in November 2011 down to 19 °C inMay 2012), interrupted by a sudden increase in the values fromMarch and lasting 2 months, during which the fumarole field reachedthe maximum temperature of this period (~25 °C); it is remarkablethat four of the observed paroxysms occurred during this anomalousperiod.

4.2. Short-term variations

High-frequency variations of the SHST can be driven either byrapid changes in volcanic activity or by external causes, such as mete-orological parameters; we investigated the air temperature, atmo-spheric pressure, wind speed, and rainfall. Daily changes in airtemperature did not induce appreciable variations in the monitoredtemperature (Fig. 4). In order to evaluate the effect of the other mete-orological parameters we selected a period (July–September 2008)during which data from VOR were compared with measurementsmade in the same fumarole field by Liotta et al. (2010), who moni-tored the hottest part of the fracture system. We applied a frequencyanalysis to our data, as presented in Fig. 5, where VOR amplitude spec-tra are plotted together with those of air temperature, atmosphericpressure, wind speed, and high-temperature data by Liotta et al.(2010). While VOR did not display appreciable short-term variationsthat were clearly related to meteorological parameters, the high-temperature fumarole monitored by Liotta et al. (2010) appears tohave been affected by atmospheric parameters, with wind speed con-tributing themost. Meteorological disturbanceswere also observed byConnor et al. (1993) in fumaroles (T > 350 °C) from Colima volcano(Mexico) and, in particular, a daily fluctuation of temperature inverse-ly correlated with barometric pressure. This could imply that high-frequency variations are more appreciable in high-temperature fuma-roles. Based on the above-mentioned considerations, the VOR signalcould be considered as the output of a natural low-pass filter of the

Fig. 5. Power spectra of data acquired every 4 h for the high-temperature fumarole monitorsure (dashed line), air temperature (thin black line), and hourly temperature acquired at VOshown as in Fig. 4.

thermal regime characterizing the high-temperature fumarole fieldlocated in the summit area of Mount Etna volcano.

The lack of any significant correlation between high-frequencyvariations and meteorological effects is also evident in the time do-main (Fig. 6). In addition, the comparison with high-temperaturedata from Liotta et al. (2010) reveals that the two signals exhibit sim-ilar behaviors and are characterized by an initial increase in thevalues during July 2008, followed by a slow and constant decreaseuntil the end of the selected period. The only appreciable differenceis that high-temperature data are affected by large increases drivenby wind during phases of low atmospheric pressure, as already iden-tified and explained by Liotta et al. (2010).

Comparing data obtained during the entire acquisition period forVOR and HOR (Fig. 2) with the rainfall distribution reveals several cy-cles of rapid cooling followed by a resumption of the pristine temper-atures that were driven by precipitation, as has also been observed inother volcanic areas such as Merapi (Java, Indonesia) (Zimmer andErzinger, 2003; Friedel et al., 2004; Richter et al., 2004), Stromboli(Italy) (De Gregorio et al., 2007), Nisyros (Greece) (Teschner et al.,2007) and Vesuvius (Italy) (Madonia et al., 2008). These cooling epi-sodes are even more evident at HOR during the first recorded period(Fig. 2), probably due to the higher permeability of the deposits ampli-fying the response of the thermal regime to the infiltration of meteoricwater in the subsoil. Indeed, the sampling station newly installed atHOR, positioned at a less permeable site, did not show the same fre-quency of temperature decreases induced by rain (Fig. 3).

5. Discussion and concluding remarks

The continuous monitoring of fumarole temperature in the Etneansummit area has revealed that the effects of meteorological parame-ters on the acquired data vary with the site conditions. The amplitudeof the daily thermal cycle in the SHST record depends on the depth atwhich the thermal sensor is buried, while placing the sensor in soilwith a low permeability is crucial for ensuring adequate sealing thatwill reduce the effects of rain and wind.

ed by Liotta et al. (2010) (thick gray line), wind speed (dotted line), atmospheric pres-R at 30-cm depth (thick black line) during July–September 2008. The zero baselines are

Fig. 6. Data of atmospheric pressure (dotted line) and wind speed (thin black line) measured every 6 h for the high-temperature fumarole (thick gray line) from Liotta et al. (2010),hourly temperature data acquired at VOR (30-cm depth; thick black line), hourly data of air temperature (gray line), and daily rainfall (gray bars) during July–September 2008.

18 P. Madonia et al. / Journal of Volcanology and Geothermal Research 257 (2013) 12–20

The two monitoring sites (VOR and HOR) were active during2007–2009 when a new subterminal eruption occurred at MountEtna, while HOR was also monitored in 2011 and 2012 when severalparoxysms occurred. Our monitoring revealed both short-term andmedium-to-long-term variations of soil temperature that can be as-cribed to endogenous changes in the steam flux. The occurrence ofthis volcanic activity provided the opportunity to better evaluate theobserved variations. In the short term, the most appreciable variationwas recorded at VOR immediately after the onset of the 2008–2009eruption (i.e., on May 15), when the temperature increased by 25 °Cwithin a few days (Fig. 2). Such an increase corresponds to almost60% of the total variability range observed during our monitoring pe-riod, and can be directly ascribed to the eruptive event. In order to cor-rectly interpret the reason for the delay in recording the onsets of theeruption and the thermal anomaly, we point out that the eruptive ac-tivity occurred from a fracture located eastward of the original SECand propagating from 3000 to 2600 m a.s.l. (Bonaccorso et al., 2011aand references therein). Magma ascending toward the surface origi-nated from a depth of 1600 m below the summit area (Bonaccorsoet al., 2011a) and propagated eastward bypassing the central conduits.The intrusion of a dike toward the NEC conduit could be caused by theoverpressure of magma in the SEC conduit, even though no trigger oferuptive activity from this summit crater was recorded (Bonaccorsoet al., 2011a). This means that the 2008 dike originated in strict rela-tion with the dynamics occurring at the shallow levels of the SEC con-duit, although the eruptive fracture was observed next to NEC (Aloisiet al., 2009, 2011; Bonaccorso et al., 2011a; Alparone et al., 2012).We recall that the monitored fumarole field is close to VOR (one ofthe central craters) that is fed by the central conduits (Fig. 7), whichwere not directly involved in the 2008–2009 eruption (Bonaccorsoet al., 2011a and references therein). Therefore, in our opinion thedelay between the eruption onset and the thermal anomaly at VORcould be due to the distance of our monitoring point from the summitvents that were directly involved in the eruption.

After the onset of the 2008–2009 eruption, the SHST at VOR grad-ually decreased until November 2008 (Fig. 2), which could beaccounted for by preferential degassing from the eruptive fracturethat reduced the volatile flux from central conduits. The continuousGPS network operating at Mount Etna recorded an abrupt deflationof the summit area during the same period (Fig. 2), which was caused

by the output of magma that accompanied the first months of theeruption (Aloisi et al., 2011). A similar deformation pattern after theonset of the crisis was reported by Aiuppa et al. (2010). The SHST atVOR increased from November 2008 to June 2009, when the valuesstabilized or decreased slightly until November 2009 (Fig. 2). This pos-itive trend can be attributed to a new progressive increase in thesteam flux associated with the reactivation of degassing through theVOR feeding fractures. During this period the GPS signal indicated agradual change from the above-described deflation phase to an infla-tion of the summit area (Fig. 2), indicating that the plumbing systemof the volcano was starting a new recharging phase (Aloisi et al.,2011). This probably enhanced the degassing throughout the centralconduits, even if such inferences appear to contradict the surface ob-servations at the end of the eruption (on July 6, 2009). However, ourshort monitoring period means that it is not possible to establish adirect relation between the deformation of the summit area of the vol-cano and changes in the steam flux in the fumaroles, and further in-vestigations are necessary.

The SHST was monitored at HOR for only 1 year (from November2008 to November 2009), but this still revealed a trend opposite tothat recorded during the same period at VOR, where there was a gen-eral decrease in the values (Fig. 2). We recall that HOR is located at thesouthern base of SEC (Figs. 1 and 7) and is fed by the southeast rift.This suggests that the temperature variations observed at HOR couldbe related to the steam flux through a fracture system directly con-nected to the SEC conduit (unlike the situation at VOR; Fig. 7). Basedon this interpretation, the trend of decreasing temperature recordedat HOR would be explained by the progressive depletion of magmasupplying the SEC conduit, leading to the end of the 2008–2009eruption.

The last data recorded at HOR in 2011 and 2012 displayed a generaldecrease in temperature compatible with a reduced steam flux. Thistrend was temporarily interrupted during March and April 2012,when the SHST increased abruptly (>5 °C). As observed at VOR inMay 2008, this variation corresponds to almost 60% of the total vari-ability range during our monitoring period, and thus can be directlyascribed to an increased steam flux due to endogenous factors. Unfor-tunately, neither geochemical nor geophysical data are available fromthis period for comparison with our data. However, we may ascribethe recorded variation to a change in the activity of the volcano,

Fig. 7. Cross section (NNW–SSE) of the Etnean summit area along the track (A–B) plotted in Fig. 1, reporting a sketch of the shallow plumbing system. Branching depth of NEC fromcentral conduits (i.e., below 2000 m a.s.l.) is from La Spina et al. (2010), while that of SEC (~2000 m a.s.l.) is from Bonaccorso et al. (2011a). Rough position of investigated fuma-roles is also reported.

19P. Madonia et al. / Journal of Volcanology and Geothermal Research 257 (2013) 12–20

which is also supported by the paroxysm frequency at NSEC increasingabruptly relative to the previous months (see internal report at www.ct.ingv.it). Therefore, considering that HOR is directly fed by thesoutheast rift and is very close to the crater feeding these paroxysms(i.e., NSEC; Fig. 1), it is reasonable to assume that the anomalous in-crease in the SHST was related to enhanced volatile degassing from anew batch of magma that entered the SEC conduits (Fig. 7). Thisphase lasted until April 24, when the last paroxysm occurred, afterwhich no further eruptive events were recorded and the SHST de-creased in accordance with the general trend that had been observedsince November 2011. This behavior appears to be consistent with ageneral reduction of Etnean activity from 2011 to 2012 (see internalreport at www.ct.ingv.it).

In conclusion, it is reasonable to suppose that the fumarole fieldslocated in the crater area of Mount Etna reflect the degassing of frac-tures that are directly fed by different portions of the shallow plumb-ing system (Fig. 7). Because most of the degassing occurs through thethree main craters [i.e., NEC, SEC, and central craters (VOR, BoccaNuova 1, and Bocca Nuova 2); Figs. 1 and 7] that are fed by the corre-sponding conduits diverging at depth, we infer that continuous mon-itoring of soil temperature in fumaroles located around the craterscould aid evaluations of the state of activity of Mount Etna. Future im-provements of this method could arise from the implementation ofthe observation network, by installing new stations in the proximityof the other active vents (i.e., NEC and Bocca Nuova) and from thecomparison with volcanic tremor data.

Acknowledgments

We thank Marcello Liotta for useful discussions of the meteorolog-ical data and Stefano Branca of INGV-CT for providing the DEM used inFig. 1. We are also grateful to Tobias Fischer and to an anonymousreviewer for helpful comments that improved the manuscript.

Appendix A. Supplementary data

Supplementary data associated with this article can be found inthe online version, at http://dx.doi.org/10.1016/j.jvolgeores.2013.03.001. These data include Google maps of the most important areas de-scribed in this article.

References

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 observa-tion of volcanic gas composition. Geology 35 (12), 1115–1118. http://dx.doi.org/10.1130/G24149A.

Aiuppa, A., et al., 2010. Patterns in the recent (2007–2008) activity of Mount Etnavolcano investigated by integrated geophysical and geochemical observations.Geochemistry Geophysics Geosystem 11, Q09008. http://dx.doi.org/10.1029/2010GC003168.

Allard, P., Carbonelle, J., Métrich, N., Loyer, H., Zettwoog, P., 1994. Sulphur output andmagma degassing budget of Stromboli volcano. Nature 368, 326–330.

Aloisi, M., Bonaccorso, A., Cannavò, F., Gambino, S., Mattia, M., Puglisi, G., Boschi, E.,2009. A new dike intrusion style for the Mount Etna May 2008 eruption modelledthrough continuous tilt and GPS data. Terra Nova 21, 316–321.

Aloisi, M., Mattia, M., Ferlito, C., Palano, M., Bruno, V., Cannavò, F., 2011. Imaging themulti-level magma reservoir at Mt. Etna volcano (Italy). Geophysical Research Let-ters 38, L16306. http://dx.doi.org/10.1029/2011GL048488.

Alparone, S., Andronico, D., Giammanco, S., Lodato, L., 2004. A multidisciplinary ap-proach to detect active pathways for magma migration and eruption at Mt. Etna(Sicily, Italy) before the 2001 and 2002–2003 eruptions. Journal of Volcanolologyand Geothermal Research 136, 121–140.

Alparone, S., Barberi, G., Cocina, O., Giampiccolo, E., Musumeci, C., Patanè, D., 2012. In-trusivemechanism of the 2008–2009Mt. Etna eruption: constraints by tomograph-ic images and stress tensor analysis. Journal of Volcanolology and GeothermalResearch 229–230, 50–63.

Andronico, D., et al., 2005. A multi-disciplinary study of the 2002–03 Etna eruption:insights into a complex plumbing system. Bulletin of Volcanology 67, 314–330.http://dx.doi.org/10.1007/s00445-004-0372-8.

Aubert, M., 1999. Practical evaluation of steady heat discharge from dormant active vol-canoes: case study of Vulcarolo fissure (Mount Etna, Italy). Journal of Volcanolologyand Geothermal Research 92, 413–429.

Barquero, J., 1983. Termometria de la fumarola del Volcano Poas. Bulletin of Volcanol-ogy 13, 11–12.

Barquero, J., 1988. Changes in fumarole temperatures at Volcan Poas. In: Newhall, C.G.,Dzurisin, D. (Eds.), Historical Unrest at Large Calderas of the World: U.S. Geol. Surv.Bull., 1855, pp. 878–879.

Benhamou, G., Allard, P., Sabroux, J.C., Vitter, G., Dajlevic, D., Creusot, A., 1988. Oxy-gen fugacity of gases and rocks from Momotombo Volcano, Nicaragua: applica-tion to volcanological monitoring. Journal of Geophysical Research 93 (B12),14872–14880.

Bluth, G.J.S., Casadevall, T.J., Schnetzler, C.C., Doion, S.D., Walter, L.S., Krueger, A.J.,Badruddin, M., 1994. Evaluation of sulphur dioxide emission from explosivevolcanism: the 1982–1983 eruptions of Galungung, Java, Indonesia. Journal ofVolcanolology and Geothermal Research 63, 243–256.

Bonaccorso, A., Bonforte, A., Calvari, S., Del Negro, C., Di Grazia, G., Ganci, G., Neri, M.,Vicari, A., Boschi, E., 2011a. The initial phases of the 2008–2009 Mount Etna erup-tion: a multidisciplinary approach for hazard assessment. Journal of GeophysicalResearch 116, B03203. http://dx.doi.org/10.1029/2010JB007906.

Bonaccorso, A., Cannata, A., Corsaro, R.A., Di Grazia, G., Gambino, S., Greco, F., Miraglia,L., Pistorio, A., 2011b. Multidisciplinary investigation on a lava fountain preceding aflank eruption: the 10 May 2008 Etna case. Geochemistry Geophysics Geosystem12 (7), Q07009. http://dx.doi.org/10.1029/2010GC003480.

20 P. Madonia et al. / Journal of Volcanology and Geothermal Research 257 (2013) 12–20

Bonfanti, P., D'Alessandro, W., Dongarrà, G., Parello, F., Valenza, M., 1996. Medium-term anomalies in groundwater temperature before 1991–1993 Mt. Etna eruption.Journal of Volcanolology and Geothermal Research 73, 303–308.

Brusca, L., Inguaggiato, S., Longo, Madonia, P., Maugeri, R., 2004. The 2002–2003 eruptionof Stromboli (Italy): evaluation of the volcanic activity by means of continuous mon-itoring of soil temperature, CO2 flux, and meteorological parameters. GeochemistryGeophysics Geosystem 5 (12). http://dx.doi.org/10.1029/2004GC000732, 1525-2027.

Burton, M.R., Allard, P., Muré, F., La Spina, A., 2007. Magmatic gas composition reveals thesource depth of slug-driven Strombolian explosive activity. Science 37, 227–230.

Caliro, S., Chiodini, G., Galluzzo, D., Granieri, D., La Rocca, M., Saccorotti, G., Ventura, G.,2005. Recent activity of Nisyros volcano (Greece) inferred from structural, geo-chemical and seismological data. Bulletin of Volcanology 67, 358–369.

Caltabiano, T., Romano, R., Budetta, G., 1994. SO2 flux measurements at Mount Etna,Sicily. Journal of Geophysical Research 99 (D6), 12809–12819.

Calvari, S., Lodato, L., Spampinato, L., 2004. Monitoring active volcanoes using a handheldthermal camera. In: Burleigh, Douglas D., Elliott Cramer, K., Raymond Peacock, G.(Eds.), Thermosense XXVI. : Proceedings of SPIE, Vol. 5405. SPIE, Bellingham, WA,pp. 199–209.

Cannata, A., Diliberto, I.S., Alparone, S., Gambino, S., Gresta, S., Liotta, M., Madonia, P.,Milluzzo, V., Aliotta,M., Montalto, P., 2011. Multiparametric approach in investigatingvolcano–hydrothermal systems: the case study of Vulcano (Aeolian Islands, Italy).Pure and Applied Geophysics. http://dx.doi.org/10.1007/s00024-011-0297-z.

Carniel, R., Jolis, E.M., Jones, J., 2010. A geophysical multiparametric analysis of hydrother-mal activity at Dallol, Ethiopia. Journal of African Earth Sciences. http://dx.doi.org/10.1016/j.jafrearsci.2010.02.005.

Cigolini, C., Salierno, F., Gervino, G., Bergese, P., Marino, C., Russo, M., Prati, P., Ariola, V.,Bonetti, R., Begnini, S., 2001. High-resolution radon monitoring and hydrodynam-ics of Mount Vesuvius. Geophysical Research Letters 28, 4035–4038.

Connor, C.B., Clement, B.M., Song, X., Lane, S.B., West-Thomas, J., 1993. Continuousmonitoring of high-temperature fumaroles on an active lava dome, Volcan Colima,Mexico: evidence of mass flow variation in response to atmospheric forcing. Jour-nal of Geophysical Research 98 (B11), 19713–19722.

De Gregorio, S., Madonia, P., Gurrieri, G., Giudice, G., Inguaggiato, S., 2007. Contemporarytotal dissolved gas pressure and soil temperature anomalies recorded at Strombolivolcano (Italy). Geophysical Research Letters 34, L08301. http://dx.doi.org/10.1029/2007GL029578.

Di Grazia, G., Cannata, A., Montalto, P., Patanè, D., Privitera, E., Zuccarello, L., Boschi, E.,2009. A multiparameter approach to volcano monitoring based on 4D analyses ofseismo-volcanic and acoustic signals: the 2008 Mt. Etna eruption. Geophysical Re-search Letters 36, L18307. http://dx.doi.org/10.1029/2009GL039567.

Diliberto, I.S., 2011. Long-term variations of fumarole temperatures on Vulcano Island(Italy). Annals of Geophysics 54, 2. http://dx.doi.org/10.4401/ag-5183.

Faber, E., Morà, N.C., Poggenburg, J., Garzà, N.G., Teschner, M., 2003. Continuous gasmonitoring at Galeras volcano, Colombia: first evidence. Journal of Volcanolologyand Geothermal Research 125 (1–2), 13–23.

Friedel, S., Byrdina, S., Jacobs, F., Zimmer, M., 2004. Self-potential and ground temper-ature at Merapi volcano prior to its crisis in the rainy season of 2000–2001. Journalof Volcanolology and Geothermal Research 134, 149–168.

Giggenbach, W.F., Goguel, R.L., 1989. Collection and analysis of geothermal and volca-nic water and gas discharges. New Zealand Department of Scientific and IndustrialResearch, Chemistry Division, Report CD 2401, Petone.

Gottsmann, J., Carniel, R., Coppo, N., Wooller, L., Hautmann, S., Rymer, H., 2007. Oscilla-tions in hydrothermal systems as a source of periodic unrest at caldera volcanoes:multiparameter insights from Nisyros, Greece. Geophysical Research Letters 34.http://dx.doi.org/10.1029/2007gl029594.

Harris, A.J.L., Carniel, R., Jones, J., 2005. Identification of variable convectives at Erta' AleLava Lake. Journal of Volcanolology and Geothermal Research 142, 207–223.

Hirabayashi, J., 1986. Sampling of fumarolic gases at Vulcano (Italy) and analytic re-sults. Geothermics 15, 201–203.

La Spina, A., Burton, M., Salerno, G.G., 2010. Unravelling the processes controlling gasemissions from the central and northeast craters of Mt. Etna. Journal of Volcanolologyand Geothermal Research 198, 368–376.

Liotta, M., Paonita, A., Caracausi, A., Martelli, M., Rizzo, A., Favara, R., 2010. Hydrother-mal processes governing the geochemistry of the crater fumaroles at Mount Etnavolcano (Italy). Chemical Geology 278 (1–2), 92–104.

Liotta, M., Rizzo, A., Paonita, A., Caracausi, A., Martelli, M., 2012. Sulfur isotopic compo-sitions of fumarolic and plume gases at Mount Etna (Italy) and inferences on theirmagmatic source. Geochemistry Geophysics Geosystem 13 (5). http://dx.doi.org/10.1029/2012GC004118.

Madonia, P., Federico, C., Cusano, P., Petrosino, S., Aiuppa, A., Gurrieri, S., 2008. Crustaldynamics of Mount Vesuvius from 1998 to 2005: effects on seismicity and fluid cir-culation. Journal of Geophysical Research 113, B05206. http://dx.doi.org/10.1029/2007JB005210.

Martelli, M., Caracausi, A., Paonita, A., Rizzo, A., 2008. Geochemical variations of air-freecrater fumaroles at Mt Etna: new inferences for forecasting shallow volcanic activity.Geophysical Research Letters 35, L21302. http://dx.doi.org/10.1029/2008GL035118.

Paonita, A., Caracausi, A., Iacono-Marziano, G., Martelli, M., Rizzo, A., 2012. Geochemicalevidence for mixing between fluids exsolved at different depths in the magmaticsystem of Mt Etna (Italy). Geochimica et Cosmochimica Acta 84, 380–394.

Pecoraino, G., Giammanco, S., 2005. Geochemical characterization and temporalchanges in parietal gas emissions at Mt. Etna (Italy) during the period July 2000–July 2003. TAO 16 (4), 805–841.

Ponte, G., 1927. Il vulcarolo sull'Etna e la utilizzatzione del suo vapore aequeo. Bullettinomensile delle sedute della Accademia Gioenia di Scienze Naturali 57 (1–2), 14–17.

Richter, G., Wassermann, J., Zimmer, M., Ohrnberger, M., 2004. Correlation of seismicactivity and fumarole temperature at the Mt. Merapi volcano (Indonesia) in2000. Journal of Volcanolology and Geothermal Research 135, 331–342.

Rizzo, et al., 2009. Geochemical evaluation of observed changes in volcanic activityduring the 2007 eruption at Stromboli (Italy). Journal of Volcanolology and Geo-thermal Research 182, 246–254.

Shimoike, Y., Notsu, K., 2000. Continuous chemical monitoring of volcanic gases inIzu-Oshima volcano, Japan. Journal of Volcanolology and Geothermal Research101, 211–221.

Stoiber, R.E., Rose, W.I., Lange, I.M., Birnie, R.W., 1975. The cooling of Izalco volcano (ElSalvador) 1964–1974. Geologisches Jahrbuch 13, 193–205.

Sutton, A.J., Elias, T., Gerlach, T.M., Stokes, J.B., 2001. Implications for eruptive processesas indicated by sulfur dioxide emissions from Kilauea Volcano, Hawaii, 1979–1997.Journal of Volcanology and Geothermal Research 108, 283–302.

Tedesco, D., Toutain, J.P., Allard, P., Losno, R., 1991. Chemical variations in fumarolicgases at Vulcano Island (Southern Italy): seasonal and volcanic effects. Journal ofVolcanology and Geothermal Research 45, 325–334.

Teschner, M., Faber, E., Poggenburg, J., Vougioukalakis, G.E., Hatziyannis, G., 2007. Con-tinuous, direct gas-geochemical monitoring in hydrothermal vents: installationand long-term operation on Nisyros Island (Greece). Pure and Applied Geophysics164, 2549–2571.

Young, S., Francis, P.W., Barclay, J., Casadevall, T.J., Gardner, C.A., Darroux, B., Davies,M.A., Delmelle, P., Norton, G.E., Maciejewski, A.J.H., Oppenheimer, C., Stix, J.,Watson, I.M., 1998. Monitoring SO2 emission at the Soufrie're Hills volcano: impli-cations from changes in eruptive conditions. Geophysical Research Letters 25, 3681.

Zimmer, M., Erzinger, J., 2003. Continuous H2O, CO2, 222Rn and temperature measure-ments on Merapi Volcano, Indonesia. Journal of Volcanology and Geothermal Re-search 125, 25–38.