Introduction

Mt Etna (Italy) is an active volcano where eruptions occur mainly from craters located at the edifice summit (e.g. Corsaro et al. 2017). However, flank eruptions sometimes occur along three main rift - zones, (south, west and northeast rifts; NE Rift in Fig. 1) due to the lateral propagation of shallow dikes from the central conduit (Neri et al. 2011, 2016; Ruch et al. 2012, 2013). The eruptive activity is characterized by a gaseous plume, strombolian and less-frequent paroxysmal explosions, and lava flows (e.g. Bombrun et al. 2016; Bonaccorso and Calvari 2013; Calvari et al. 2011). In many other volcanic areas, groundwater, diffuse gas emissions, and the fumarole fields are used to trace hydrological circuits and chemical-and-physical exchanges with solid structures (e.g. Heasler et al. 2009). The approach, based on fluid geochemistry research, includes a lot of experimental research activities, and the references are too broad and numerous to be cited individually. Paonita et al. (2016, 2012) showed examples for Etna’s groundwater flow and degassing areas, based on a wide geochemical dataset, collected since the year 2000. Some examples from other sites are: Maldonado et al. (2017) and Federico et al. (2017) for Columbian volcanoes, Peiffer et al. (2015) for El Chichon (Mexico), Inguaggiato et al. 2011 for Stromboli; Byrdina et al. 2009, for Nepal; Paonita et al., (2016), Federico et al. 2010, for Vulcano; Matsushima et al. 2003 for Iwodake; and Lardy and Tabbagh, 1993, for Vanuatu. The structural weakness of Etna’s volcanic edifice has already been suggested by analyses of deformation patterns, tectonic lineaments, and other geophysical investigations (e.g. Cannata et al. 2015; Mattia et al. 2015). Ground-based thermal monitoring can also be used to follow the real-time evolution of surface effects and to better define the geochemical fluid leakages (e.g. Mattia et al. 2015; Revil et al. 2011; Mazot et al. 2011; Byrdina et al. 2009; Lowenstern et al. 2006; Bonaccorso et al. 2003; Patanè et al. 2003; Oskarsson et al. 1984; Sato and McGee, 1982). The hydrothermal activity on the flanks of Mt Etna changes both spatially and temporally in response to variations in magmatic pressure (e.g. Cannata et al. 2015) within the main feeding system of the volcano which, in turn, drives lava to the different eruptive vents (Paonita et al. 2012).

We here apply a method to continuously measure the shallow vertical temperature profile in a zone of steam-heated soil (SHS) on the upper northeast slope flank of Mt Etna (Italy) 4 km from the summit craters. This shows how variations in heat flux can be used to track connections between this fracture system and the ongoing volcanic activity. We set up a new monitoring site (BTL station in Figs. 2–3), and here we present the recorded variations in soil temperature and the application of the SHS monitoring method. Data (amounting to nearly 26,000 data points) were acquired hourly from September 2009 to September 2012 along a vertical soil profile (down to a depth of 60 cm) in the direction of the main component of the surface heat flow.

Fig. 1
figure 1

Map of Mount Etna volcano showing the locations of sites mentioned in the text. The box outlined in double red indicates the monitored area. The red triangle locates the WS-EN, weather station at the Etna North. Inset bottom-left is the location of Mt. Etna in the southern Mediterranean region (modified from Azzaro et al., 2012). BTL, VOR, NEC, SEC TDF, and BLV stand, respectively, for Bottoniera Lines, Voragine, North East Crater, South East Crater, Torre del Filosofo, and Belvedere

Full size image

Geothermal anomalies at Mt. Etna: Geochemical and thermal monitoring

Etna’s peripheral and summit areas have been monitored geochemically for more than 20 years (e.g., Allard et al. 1991; Bonfanti et al. 1996a,b). Large variations in geochemical signals were recorded in groundwaters during the 1991–1993 eruption (Bonfanti et al. 1996a,b). D’Alessandro et al. (1997), for example, identified a significant increase in groundwater temperatures that was related to the huge volume of magma (> 200 × 106 m3) erupted between 1991 and 1993 (Stevens et al. 1997; Tanguy et al. 1996; Barberi et al. 1993). Conversely, over the last two decades, no direct relationship has been found between groundwater temperatures across the lower flanks of the volcano and eruptive activity. In a geochemical investigation of the summit area, Madonia et al. (2013), however, showed a clear relationship between changes in surface thermal activity and the eruptions. Tabbagh and Trezeguet (1987) performed the earliest measurements of surface heat flux in the summit area of Mt. Etna, near Torre del Filosofo (TDF, Fig. 1). Afterwards, Aubert (1999) applied the same methodology to the same fissure system, to monitor time variations of the surface thermal release. Along the active structures, a continuous tectonic strain enhances the natural steam loss from magma and hydrothermal sources (Maucourant et al. 2014), but at the same time prevents the self-sealing phenomenon caused by mineral deposition of silica, carbonates, and/or salts that precipitate from the fluid phase (Chiodini et al. 1995; Facca and Tonani 1967). Therefore, while newly erupted products at Etna, such as lava flows and pyroclastic deposits can bury volcano-tectonic structures, the surface fluid flow can reveal structurally controlled zones beneath these superficial layers that still act as pathways for natural fluids ascending from deeper levels, variations on which can inform on the dynamics of the degassing system. As a result, Rizzo et al. (2013) were able to analyze the composition of the fluids associated with hydrothermal circulation at Etna’s summit zone to obtain geochemical information about the feeding sources. Moreover, many researchers have related the time variations in diffuse CO2 degassing from the summit and flank areas of Mt. Etna to the eruptive activity (Allard et al. 1991; D’Alessandro et al. 1997; Giammanco et al. 2007; Camarda et al. 2012; De Gregorio et al. 2014). In contrast, less attention has been given to the heat released by the steaming areas present at various altitudes and around the rim of the summit craters. It could be due to anomalies of the soil temperature—resulting in weak heat fluxes—being difficult to distinguish from the normal background when only discontinuous sampling is performed (e.g., Bonneville et al. 1985; Bonneville and Gouze 1992; Madonia et al. 2013; Maucourant et al. 2014). The thermal surface has though been measured using satellite data firstly by Bonneville et al. (1985) and by Bonneville and Kerr (1987) who also compared the thermal images with shallow ground temperatures (depth 0.03 m). Moreover, the association of thermal areas with anomalous degassing of CO2 suggests a common hydrothermal source (e.g., Macourant et al. 2014). The isotopic signature and chemical composition of emitted fluids can reveal the character of such a hydrothermal source that is the relative contribution of magmatic, meteoric, and/or marine fluids (e.g., Aizawa et al. 2016; Liotta et al. 2010). The delays between geochemical signals and volcanic events can reveal underground connections within, and between, the plumbing and hydrothermal system (i.e., Alparone et al. 2004; Pecoraino and Giammanco 2005; Behncke et al. 2014; Giammanco et al. 2016).

With the aim of detecting a connection between an active fracture system and ongoing volcanic activity, we investigated a steam-heated soil (SHS) on the upper northeast flank of Mt. Etna close to fractures and eruptive fissures belonging to the northeast rift zone (NE rift, Fig. 1) (Neri et al. 2011, 2016). This SHS is located close to the eruptive fissure (BTL fumaroles in Figs. 2 and 3) that opened in 2002 at about 2500 m a.s.l. We set up a new monitoring site (BTL station in Figs. 2 and 3), and here we present the recorded variations in soil temperature and the application of the SHS monitoring method. Data (amounting to nearly 26,000 data points) were acquired hourly from September 2009 to September 2012 along a vertical soil profile (down to a depth of 60 cm) in the direction of the main component of the surface heat flow. The heat flux anomalies from the ground have been evaluated over a 3-year period. This allows changes in the hydrothermal release to be related to levels of volcanic activity to be assessed, discussed, and compared with published geophysical data. Figure 3 gives a sketch of a 3D section of the system tracked, where dikes depart from the central conduits. The sketch suggests a connection between the surface thermal activity around the fumaroles along the active NE rift and the eruptive vents on the southeast flank.

Fig. 2
figure 2

Satellite image (Google Earth) of the monitoring site BTLs and fumarole (BTLf) sampled by Pecoraino and Giammanco (2005) and Liotta et al. (2010). The thick green line indicates a zone of low-temperature fumaroles

Full size image
Fig. 3
figure 3

Schematic model of the plumbing system with volcanic features. Refer to caption of Figs. 1 and 2 for the acronyms

Full size image

Surface steam loss around Mt Etna

Steam is emitted from many areas on the slopes of Mt. Etna, some of which have been studied geochemically in terms of temperature and the chemical and isotopic compositions of CO2 (Giammanco et al. 1998; Pecoraino and Giammanco 2005). However, the locations of many of the steaming zones are still unknown. Individual steaming areas can appear or disappear after every eruption on the upper slope as well as at the top of the craters (e.g., Bonneville et al. 1985; Bonneville and Kerr 1987). New steaming zones are caused by the active tectonic strain allowing hot fluid to escape through the permeable framework of fractured rocks and pyroclastic layers.

The most evident steaming areas at Mt. Etna that have already been investigated geochemically or by combined approaches are as follows:

  1. a)

    An area located on the south rift near TDF in Fig. 1 (Pecoraino and Giammanco 2005; Giammanco et al. 1998) at an average altitude of 2900 m a.s.l. The south rift consists of a group of parallel fissures running in a north-south direction, starting from the summit crater on the southeastern flank of Mt. Etna. The heat released in 1988 was evaluated for the first time by Aubert (1999) using the SHS method. Tests were subsequently performed in 2009 and 2012 (Diliberto, unpublished data), and the recorded temperature profiles suggested that this location was not suitable for continuous monitoring using the SHS method proposed by Aubert (1999) due to the presence of a liquid phase. Coarse and fresh scoria has been frequently added to the soil profile by recent explosive activity from the summit craters. The inhomogeneity of the shallow permeability due to the presence of coarse and fresh scoria in this area allows the steam to expand and condense very close to the ground surface. Other authors have also chosen this area as a test site for obtaining information on the potential reactivation of the summit craters (Maucourant et al. 2014).

  2. b)

    The Belvedere steaming area (BLV in Fig. 1), which is located on the rim of Valle del Bove on the south rift at an altitude of about 2720 m a.s.l., is characterized by significant CO2 fluxes (Pecoraino and Giammanco 2005; Di Martino et al. 2013; De Gregorio et al. 2014).

  3. c)

    The rim of the Voragine crater at 3260 m a.s.l. (VOR in Fig 1). In this area, Harris et al. (1997) already observed a highly radiating surface, north of the active lava field, while they were retrieving surface temperatures by satellite data from very high resolution radiometer (AVHRR), during the March to July 1985 eruption. That thermal anomaly corresponded to the cooling surface from a previous activity, dated April to October 1984 (Harris et al. 1997). Later on, in the same area, the SHS thermal monitoring revealed advective degassing (Madonia et al. 2013), still active and correlated to the 2008–2009 subterminal eruption from the SEC crater (Bonaccorso et al. 2011a,b). However, the location of this unstable area of the volcanic cone makes it extremely dangerous to either monitor changes in the heat flow by direct measurements or set up a long-term program of geochemical investigations.

  4. d)

    An area named “hornitos” located at 3000 m a.s.l., southeast of Voragine (VOR in Fig. 1), which has been covered by lava and tephra erupted by the southeast crater (SEC in Fig. 1). This area lies on the south rift in the proximity of the Vulcarolo fumarole field (Fig. 1) and is actively degassing (Alparone et al. 2004; Madonia et al. 2013).

  5. e)

    The NE rift (Fig. 1) is a volcano-tectonic feature visible on the northeast flank of the volcano which is subject to (1) frequent magma emplacement events involving dike intrusion and eruption along the rift (Patanè et al. 2005) (2) seismic energy release resulting from magma emplacement (Patanè et al. 2003, 2006), and (3) transient slipping of the Pernicana fault system (Ruch et al., 2013). A line of craters formed during the 2002–2003 eruption (BTL in Figs. 2 and 3) and these are aligned with the NE rift, and the gas emissions from some low-temperature fumaroles in the proximity of the upper part of this crater row have been sampled and analyzed (Pecoraino and Giammanco 2005). Studies of these steam emissions suggest that the emitted fluids are connected to the main feeding system of the volcano (Pecoraino and Giammanco 2005; Liotta et al. 2010). Samples collected by us at the BTL fumaroles (BTLf in Fig. 2) in 2009, before the eruptive period discussed in this paper, showed a gas mixture of CO2 (predominating) and CH4 with high degrees of air contamination.

Notwithstanding the different locations, both the south rift and NE rift are fed by dikes extending from the main magma conduit (Fig. 3), rather than being connected to different, individual shallow magmatic chambers individually (Giammanco et al. 2016 and references therein).

With the aim of detecting a potential connection between an active fracture system and the ongoing volcanic activity, we set up a monitoring site on the NE rift at an altitude of 2480 m a.s.l., i.e., in a summit-proximal location (BTL station in Fig. 2). It is worth noting that this site was not affected by surface changes during the period of acquisition, since it is on the flank opposite the main direction of the pyroclastic fall, that tend to be in the direction of the prevailing wind (E to SE), and lava flows erupted by the new southeast crater (New Southeast Cone in Fig. 4). Any physical changes in coverage are thus simply due to seasonal variations in snow cover and melt.

Fig. 4
figure 4

View of Mt. Etna showing plume dispersion from the erupting crater (5 January 2012) and the monitored site (4 km to the N). In the top-right corner, the rose diagram indicates the main wind direction. Photo credit (http://www.youreporter.it/foto Eruzione_5_gennaio_2012)

Full size image

Monitoring method

Tabbagh and Trezeguet (1987) introduced the theoretical background for evaluating and monitoring thermal releases, showing a first set of field measurements made in the summit area of Mt. Etna (TDF in Fig. 1) during 1982. Afterwards Aubert (1999) applied the same monitoring method through a field deployment of seven buried thermocouples separated by 0.1 m of soil, down to a depth of 1.2 m in 1998. The measurements by Aubert (1999) on the southern flank of Mt. Etna (TDF in Fig. 1) showed heat flux values ranging from 10 to 300 W × m−2. The SHS method has been applied continuously under other very different conditions: at New Hebrides (Tabbagh and Lardy 1993; Lardy and Tabbagh 1999), on Mt. Etna during eruptive activity (Madonia et al. 2013), and on Vulcano Island in a close conduit volcanic system away from the main fumarole field (Diliberto 2017; Cannata et al. 2012; Aubert et al. 2008).

The best site for monitoring heat flux using the SHS method is where the condensation zone occurs just beneath the monitored profile. Within the upper layer (from Z1 up to the ground level in Fig. 5a, b), whenever the ascending steam is completely condensed, the convective component of the heat transport becomes negligible, as does any process implying heat production or consumption. Below the Z1 depth, the soil temperature is enhanced by the steam condensation which dominates over temperature conductive exchanges between the ascending fluids and the soil. Figure 5a shows a sketch of the soil profile, and Fig. 5b shows a model of the temperature gradient present in the monitored soil. The layer between depth Z1 and the surface (CdHT in Fig. 5a) is generally dry and is dominated by the conductive transfer law (Fourier’s Law), and so δT/δZ is linear (Fig. 5b). The heat flux is essentially convective below level Z2 (CvHT in Fig. 5a) and the temperature of the steam is buffered by the boiling point, which is about 91 °C at this altitude, corresponding to the temperature measured at the nearby fumaroles (Fig. 2). A liquid phase dominates in the layer between Z2 and Z1 due to the condensation of hydrothermal steam, so that the heat transfer here is partially conductive (CdHT in Fig. 5a) and partially convective (CvHT in Fig. 5a). The success of this monitoring approach is thus crucially dependent on the site location and physical conditions of the ground. Two conditions need to hold when estimating the heat flux from the temperature gradient: (1) the site must be far from the boiling temperature and (2) the heat flux should be unidirectional along the monitored profile (the Z direction and depth over Z1 in Fig. 5b) and conductive at the top.

Fig. 5
figure 5

a System model of the monitored soil profile with the main heat transfer processes. b Hypothetical temperature gradient at BTL station: conductive transfer dominates, and δT/δZ is linear between Z1 and the surface (redrawn from Diliberto 2017)

Full size image

Figure 6 shows the temperature distribution along the monitored profile at three different times. Each set of data shows a high linear correlation calculated among the values recorded at the four monitored depths at the same time (R in Fig. 6). The linear temperature gradient has been extrapolated to the boiling temperature to evaluate the lowest limit of the conductive layer. The limit of the conductive layer is deeper than the last extrapolated values shown in Fig. 6. Above these limits (i.e., above the Z1 depths Figs. 5 and 7), the simplified heat flow equation (ϕH) can be applied up to the ground level (Aubert 1999; Tabbagh and Trezeguet 1987):

Fig. 6
figure 6

Examples of the temperature distribution along the monitored profile during the three different eruptive phases. The profiles refer to three phases of volcanic activity at Mt. Etna: October 3, 2009 for the pre-eruptive period (pre), October 6, 2011 for the eruptive period (er), and April 10, 2012 for the post-eruptive period (post). The respective correlation coefficients (R) for the profiles are shown. The Z1 depths are extrapolated by the best fit equations

Full size image
Fig. 7
figure 7

a Time variations of temperature at four depths (in meters) within the shallow layer governed by conductive heat transfer (linear correlation, R > 0.990; a1, a2, a3). Detail of the ground temperatures recorded during the first week of August in 2010, 2011, and 2012 as an example of the pre-eruptive, eruptive, and post-eruptive periods, respectively. b Time variations of air temperature recorded at the WS-EN meteorological station. c Rainfall recorded at the WS-EN meteorological station

Full size image
$$ {\phi}_{\mathrm{H}}=\lambda \left({t}_4-{t}_2\right)/\left({z}_{\mathrm{t}4}\hbox{--} {z}_{\mathrm{t}2}\right) $$
(1)

where t4 and t2 are the temperatures (in kelvin) measured at the different levels in the ground, zt4 and zt2 are the corresponding reference depths (in meters), and λ is the thermal diffusivity of the ground.

In the following calculation, the uppermost monitored level (t1 located at depth of 0.15 m below the ground surface) was excluded to avoid the external influence of the daily variation due to solar heating. Moreover, λ was assumed to be constant, at λ = 0.8 W m−1 K−1 (Aubert 1999; Tabbaugh and Trezeguet 1987). This assumption applies only to dry ground conditions (i.e., in the absence of water and steam), which fulfills the necessary conditions highlighted by the linear correlation of temperatures (R > 0.990) among the four monitored levels. Sixty-six percent of the temperature data fulfilled this condition and have been considered in further processing.

The monitoring site was located in the proximity of a line of fumaroles (indicated as BTLf in Fig. 2) along the NE rift, at 2470 m a.s.l. The BTLs (Fig. 2) monitoring station is about 120 m NW direction from the low-temperature fumaroles on top of the 2002–2003 eruptive fissure. The shallow temperature appeared homogeneous around the selected point at the time of the survey. At the same time, and prior to the excavation, the temperatures measured along a vertical profile indicated a positive heat flux toward the surface, even at the time of maximum solar radiation (11:00 a.m. during summertime), with temperatures of 20 and 32 °C at depths of 0.15 and 0.60 m, respectively (the shallowest and deepest positions measured). These temperature measurements were made in the undisturbed soil and could be accounted for by an effective bottom-sourced (hydrothermal and/or magmatic) heat flux and fulfilled both of the required conditions for heat flux monitoring. The temperature increased with depth, which is opposite to the normal condition expected at the time of the first survey (noon on a clear-sky day in summer). A vertical line with four PT1000 sensors (range − 40 to 150 °C, accuracy ± 0.2 °C, resolution ± 0.1 °C) was connected to an EBRO EBI 2T-313 four-channel data logger which was then buried at a depth of 0.60 m within the loose volcanic breccia. The temperature variations within this undisturbed ground profile have been recorded hourly for about 3 years at depths of 0.15, 0.30, 0.45, and 0.60 m.

Results

Temperature data

The recorded temperatures varied markedly throughout the analyzed period (Fig. 7a). The position chosen at the starting time appeared to be a ground with surface temperature influenced mainly by solar radiation, and the temperature range along the monitored vertical axis was far below the boiling temperature. Overall, the BTLs site can be considered to be only weakly heated by steam, based on comparisons with temperature profiles recorded elsewhere. At the shallowest monitored level (0.15-m deep; Fig. 7a), the temperature ranged from − 9 to 46 °C, while it ranged over a narrower range (0–34 °C) at a depth of 0.60 m. The recorded temperatures peaked at all monitored levels during summer 2011 (Fig. 7a). Higher temperatures were generally measured at the deepest level, except in August 2011 and after April 2012. Moreover, during these two periods, a very high daily fluctuation was only recorded at a depth of 0.15 m (Fig. 7a).

Our time series can be compared with the local meteorological parameters by considering the records from two weather stations located at two ski resorts on Mt. Etna: (1) weather station Etna south, located at UTM 499936 E and 4172571 N, and 1935 m a.s.l. and (2) weather station Etna north (WS-EN in Fig. 1), located at UTM 503250 E and 4182972 N and 1825 m a.s.l. Data from these two weather stations do not indicate large differences in the time variations of temperature and rainfall during the analyzed 3-year period. The BTL and WS-EN sites are on the same flank and have no physical barriers between them, and so we assumed that atmospheric variations would affect the two sites in the same way. The air temperature at BTL should be about 4.2 °C lower than that recorded at WS-EN (Fig. 7b), based on the average adiabatic thermal gradient 6.5 °C km−1 (e.g., Tomasi et al., 1997).

A visual comparison of air and soil temperatures (Fig. 7a, b) reveals the presence of a direct relationship between them, such as in common cyclic behavior due to seasonal and daily solar radiation effects. However, panels a1, a2, and a3 in Fig. 7 show that under the same weather conditions (i.e., maximum solar radiation and lowest rainfall), different years show very different temperature fluctuations in the ground. As we expected, the atmospheric conditions affected the shallowest monitored level (i.e., t1 at 0.15 m depth) more than lower levels. The external perturbations, due to occasional combinations of certain physical features (e.g., rainfall, wind, and melting ice), exert unpredictable effects, mostly on the short-term variations and during the winter season (Fig. 7c). However, the temperature distribution in the ground appears related not only on the atmospheric agents but also to other causes. These data could thus reveal a relationship with changes of the bottom-sourced heat flux that in turn depend on the local dynamics of volcanic activity.

Limit of the conductive layer

In accordance with the models in Fig. 5 and the examples shown in Fig. 6, the linear extrapolation of the temperature distribution along the monitored profile has been used to trace the depth variations of steam ascending beneath BTL. The Z1 depths are the uppermost limits suggested by the δT/δZ slopes for the boiling conditions at TBL site (91 °C; 0.74 atm). As a result, Z1 can be considered as the reference depths for the upper limit of a convective front (CvHT; Z2 in Fig. 5). To obtain the Z1 depths shown in Fig. 8, we first filtered out the presence of water in the layer, by excluding the temperature profiles having linear correlation coefficients R ≤ 0.990. Then, we extracted the data recorded at nighttime (hour 0.00) from the residual subset of data (17,160 profiles), so as to filter out the daily modulation caused by the diurnal cycle of solar radiation. Finally, the linear extrapolation was calculated on the temperature records from the second, third, and fourth levels only, to further reduce the thermal effects of atmospheric factors affecting essentially the shallowest level.

Fig. 8
figure 8

Time variations of the Z1 depths extrapolated from the temperature profiles, only extrapolations with R > 0.990 are considered. The three colored arrows highlight the Z1 values extrapolated during the month of April in 2010 (blue), 2011 (red), and 2012 (green). In the legend “pre-” stands for the pre-eruptive period, “syn-” stands for the eruptive period, and “post-” is the post-eruptive period. The red triangles at the bottom indicate the periods of active lava flow

Full size image

Table 1 shows the descriptive statistics of the Z1 depths extrapolated during different phases of volcanic activity, and Fig. 8 shows the relative temporal variations. The volcanic phases have been assigned using the volcano observation network (Behncke et al., 2014; INGV-Ct). At the monitored site (BTL in Figs. 2 and 3), the limit of the conductive layer varied within the range from 1.5 to about 30 m of depth below the ground level (Depth bgl, Fig. 8). During the first eruptive period, that is from January 2011 until September 2011, the range and the variance of Z1 depths have not changed much in respect to estimations concerning the pre-eruptive period (Table 1). However the relative time variations of Z1 depth show that the convective front had been less stable in the pre-eruptive phase. Especially from October 2009 to June 2010, Z1 depths frequently had values greater than 6 m (Fig. 8). While the conductive layer has remained confined within 4 m of the surface during the eruptive periods, such as from January to March 2011 and from October 2011 to March 2012 (Fig. 8). The general deepening of the base of conductive layer since March 2012 suggested a relative contraction of the convective front, after the eruption end (Fig. 8; Table 1).

Table 1 Descriptive statistics for the Z1 depths extrapolated during the three eruptive phases
Full size table

The profiles have thus been analyzed during three different phases of volcanic activity: pre-eruptive, eruptive, and post-eruptive (Fig. 8). The Z1 depth limits were shallowest during an eruption and deepest after an eruption. The shallowest Z1 depths were found during the eruptive period, indicating ascent of the condensation level and an increase in heat flux from the diffusive layer directly related to magma ascent. In contrast, during the pre-eruptive and post-eruptive phases, deeper levels of condensation and lower heat fluxes were registered. Even if the end of the linear trend and the relative Z2 depths are not properly constrained, it is clear from Fig. 8 that the δT/δZ slope increased during the productive period of the eruption, being associated with an increase in the heat flux. The temperature trends recorded from September 30, 2009 to September 14, 2012 at four levels along a vertical profile frequently showed linear correlations (i.e., R close to 1) adding confidence to our temporal trends and interpretations.

Heat flux from the soil

The heat flux from the soil was calculated at the BTL station by applying the simplified heat flow equation (Eq. 1) to the same subset of data described in the previous section. Three different ranges of heat flux from the soil can be distinguished in Fig. 9, during the pre-eruptive, eruptive, and post-eruptive phases. Figure 10 compares the 30 day-period moving averages of the SHS heat flux with the cumulative eruption rates of lava and pyroclasts (Behncke et al. 2014) and with the monthly average air temperatures. Differences among the range of heat fluxes from the steaming ground (white arrows 1–4 in Fig. 10) are still evident in the smoothed curve. Moreover, the higher rate of erupted material is associated with the main increase in the SHS heat flux (arrow 3 in Fig. 10) that began on July 1, 2011, reaching the highest recorded value in November 2011. In contrast, there is a clear inverse correlation between SHS flux and air temperature during the summit activity in 2011 and 2012, although such a correlation was less evident during the pre-eruptive period in 2009 and 2010.

Fig. 9
figure 9

Heat flux variations in SHS retrieved from our thermocouple data. The blue, red, and green arrows and dashed boxes indicate the ranges of heat flux (daily averages) for the pre-eruptive, eruptive, and post-eruptive phases. Only extrapolations with data validated by R > 0.990 are considered

Full size image
Fig. 10
figure 10

Heat flux (Hf) variations in SHS and cumulative volume of erupted material (CVP, values on the right axis, Behncke et al. 2014; Bonaccorso and Calvari 2013). The white arrows indicate the ranges represented by the mean term variations. The dashed arrows give the SHS trends. The thin lines indicate the minimum values of heat flux observed during the different phases

Full size image

Three different phases of volcanic activity (pre-eruptive, eruptive, and post-eruptive) are also identified in Fig. 10 from the cumulative volume of erupted material. The range of heat fluxes peaked during the eruptive period, from January 2011 until April 2012 (red arrow in Fig. 9 and arrows 2 and 3 in Fig. 10). An intermediate range characterized the pre-eruptive period (blue arrow in Fig. 9 and arrow 1 in Fig. 10) from September 2009 to December 2010. During the post-eruptive activity, a narrower range and the lowest heat flux were observed (green arrow in Fig. 9 and arrow 4 in Fig. 10). At this point, the SHS had probably reached the local background range. In this last period, the modulation of heat flux from the ground was dominated by external factors.

Discussion

A geochemical approach based on long-term monitoring makes it possible to follow the changes in energy released from hydrothermal and/or magmatic sources in real time (e.g., Inguaggiato et al. 2018; Paonita et al. 2013; Chiodini et al. 2005; Matsushima et al. 2003; D’Alessandro et al. 1997; Bonfanti et al. 1996a,b). However, systematic observational timescales and a wide multidisciplinary approach are necessary to interpret the observed temporal variations in volcanic heat flux (e.g., Harris 2013 and references therein; Stix and Gaonach 2000). With this in mind, coupled with the low intensity of the thermal anomaly involved, we monitored soil temperature changes at a specific site in order to identify variations in the bottom-sourced heat flux. Our results are the first application of the method proposed by Tabbaugh and Trezeguet (1987) and Aubert (1999) during a complete volcanic cycle. During the studied period starting from January 2011, the eruptive activity of Mt. Etna generated the New Southeast Crater on the flank of the SEC (Behncke et al. 2014). This eruption was characterized by high eruption rates of lava and Strombolian activity until April 24, 2012, which was followed by a period of collapse (Behncke et al. 2014). The BTL monitoring site (Figs. 2 and 3) is in proximity to an eruptive fissure (which last erupted during 2002–2003), about 4 km from the vents that erupted during the monitored period (SEC, Fig. 1). Although the depth limit of the convective heat transfer regime has not been fully constrained, we used the recorded data to detect variations in the depth of convective front (i.e., the depth of Z1). Such variations should reflect changes in fluid pressure in the shallow hydrothermal or magmatic system. Following the δT/δZ slope changes, the upper convective front moved during the different volcanic phases (Fig. 8). In particular, the limit of the conductive layer (Z1) tended to rise before the start of a new eruption. The lowest variance in the conductive layer has been observed between mid-August 2011 and March 2012, while the volcano was in continuous eruption. The deepest positions of the conductive layer were observed after the end of the eruptive period (Fig. 8), and this result can be interpreted as a contraction of the convective system feeding the fumaroles of the NE rift. On the NE rift, the shallow hydrothermal circuit, which apparently was not involved in the eruption, thus showed fluctuations of the convective front related to the eruptive activity, resulting from heat flux variations of the soil. The increases of the intensity of the thermal anomaly around low-temperature fumaroles reflected, at the same time, the ascent of the condensation level driven by the increased pressure of hydrothermal fluids during periods of activity.

We thus find a direct relationship between the low-temperature fumaroles and the eruptive activity. We infer that the increase of fluid pressure in the magmatic source feeding the impending eruption provoked an increase in steam release along the Northeast Rift and consequently enhanced SHS heat flux. The energy discharge of magmatic origin, which was associated with eruption at the southeast crater, then resulted in high and steady levels of thermal release along the NE rift. Variations in fluid pressure associated with volcanic activity thus result in thermal effects over a wide area. Tracing its changes by remote monitoring is thus a valid tool for assessing the activity levels of the system in an eruptive and pre-eruptive phase.

Conclusion

We applied an empirical approach to compare different time series of data in order to identify variations related to the energy released by an active geothermal, hydrothermal, and magmatic system. The continuous monitoring system was based on the temperature monitoring method applied to soil heat fluxes by Tabbagh and Lardy (1993), Lardy and Tabbagh (1999), Aubert (1999), Aubert et al. (2008), and Diliberto (2017). Our temperature monitoring station yielded reliable time series of conductive heat transfer in which we could identify different ranges related to the different stages of activity: pre-eruptive, eruptive, and post-eruptive.

Low-temperature fumaroles are buffered at the boiling point. This buffering of temperature makes the temperature of fumaroles unsuitable for monitoring, because only strong decreases in temperature can be detected. The present method, applied to the SHS around fumaroles, yielded a method capable of overcoming the limit imposed by the buffering condition. In SHS method, the steam condensation occurs below the ground level, where it enhances local heat release, this being an exothermic process (e.g., Chiodini et al. 2005). Here, the temporal resolution of continuous monitoring makes it possible to track and define any change in steam release.

A new eruptive cycle began during our monitoring period. Over a time interval of 1082 days (September 2009–September 2012), all changes in the soil heat flux were linked to variations in activity, distinguishing between pre-eruptive, eruptive, and post-eruptive regimes. Our monitoring method thus allowed us to identify a connection between monitored steaming zone on the NE rift of Etna and the eruptive conduits that fed the most recent volcanic cycle at the SE crater.

Such a simple monitoring method gives a measure of the secondary thermal effects exerted by magma dynamic. Tracking variation in these effects offers great monitoring potential. For example, an increase in SHS heat flux could suggest that the intensity of an eruption is going to increase and a decrease could herald cessation. If the monitoring station is located on a fractures network connected to the eruptive conducts, this system offers a dataset that can be used as a reference for evaluating magma flux conditions when the other observational data are discontinuous.

Thermal monitoring allowed us track over the energy released from a thermal anomaly along Etna’s NE rift zone between 2009 and 2012 and to relate variations in this flux to activity at the summit craters about 4-km distant. The analysis of soil temperature data allowed us to identify heat flux variations related to the continuous release of thermal energy and to compare with trends apparent in time series of observational data. We find that real-time, continuous, and operational monitoring of heat release reduces the risk of incorrect interpretations, interpolated by other discontinuous observations, allowing the heat flux response of the system to increase and diminish mass fluxes into the system to be tracked. Building on previous work from Japan (Matsushima et al. 2003; Sekioka and Yuhara 1974; Ohara and Okamoto 1980), Vulcano (Diliberto 2017; Cannata et al. 2012; Aubert et al. 2008), Vanuatu (Lardy and Tabbagh 1999), and Etna (Madonia et al. 2013; Aubert 1999; Tabbagh and Trezeguet 1987), we thus argue that the SHS heat flux method provides a new parameter to be considered for continuous, operational monitoring and for generation of deterministic models aimed at interpreting volcanic activity in terms of effects on hydrothermal fluid circulation.