In order to reconstruct Etna's dynamics during 2013, different kinds of data were used (see Figure 2): seismic, infrasonic, tilt, GPS, geochemical, and volcanological. In the next sections, all the analyses and results obtained from these data are described.
3.1 Seismic Data
At Mount Etna, the seismic permanent network run by Istituto Nazionale di Geofisica e Vulcanologia (INGV)–Osservatorio Etneo comprises 33 broadband and 12 short-period stations. Analysis of seismo-volcanic signals is performed on the recordings of the 19 stations closest to the summit area. These stations are equipped with broadband (40 s cutoff period), three-component Trillium seismometers (NanometricsTM) acquiring in real time at a sampling rate of 100 Hz (Figure 1b). It is worth noting that two of these stations (EBEL and ETFI; see yellow circles in Figure 1b) were destroyed during the paroxysmal episodes of 28 February and 11 November 2013, respectively.
We analyzed LP events and volcanic tremor recorded during 2013, as well as the seismic signal accompanying the 5 September BN explosion.
As for the former, we detected more than 200,000 events, whose peak-to-peak amplitudes and spectral features are shown in Figures 2b and 2c, respectively. The peak-to-peak amplitude was obtained from the seismic signal recorded by ECPN station (see Figure 1b), which is the reference station for LP events study at Etna [e.g., Cannata et al., 2009]. Concerning the spectral features, we calculated the normalized pseudospectrogram (examples of the application of pseudospectrograms can be found in Alparone et al.  and Spina et al. ), allowing to track the evolution of the LP spectral content, as follows: by the FFT algorithm, we calculated one spectrum for each event using a 10 s long window, starting 0.5 s before the onset of the LP event. Next, an averaging process was performed on the amplitude spectra of all the events falling in nonoverlapping 1 day windows, sliding through the investigated time interval. Finally, we normalized each spectrum by its maximum value and gathered the normalized averaged spectra as columns in a single matrix arranged in temporal order. This matrix was visualized as a pseudospectrogram with time in the x axis, frequency in the y axis, and the color scale showing spectral amplitude. From January to mid-February 2013, LP events were characterized by weak amplitudes (Figure 2b) and relatively broad spectral contents (~2–5 Hz; Figure 2c). During the first cycle of lava fountains of 2013, we observed an increase in the amplitudes of LP events. However, the most important change in terms of both amplitude and spectral content took place successively, during the 28 April to 5 September time interval. Indeed, a gradual amplitude increase occurred, ending on 5 September with a very sharp decrease, at the same time as the explosion at the BN. During the same time interval, we noted an important variation in LP spectral content: most LP events became monochromatic with frequency peaks <1.2 Hz (Figure 2c). Similarly to most of the LP events at Mount Etna, these LP events were not accompanied by acoustic trace. It is also possible to observe a gradual decrease of the frequency peak from ~1.0 Hz to ~0.6 Hz, taking place from May to the end of August. Moreover, this peculiar low-frequency feature sharply ended on 5 September, when the LP event spectral content became again higher and broader (3–6 Hz). The rapidity of the changes of the LP events, which took place at the same time as the 5 September BN explosion, is evident in the seismograms shown in Figure 3: indeed, the disappearance of the “signal spikes” after the BN explosion is due to the strong decrease of LP event amplitude (Figure 3a), while the changes in the LP waveforms are visible in the seismograms in Figure 3c (typical LP event waveform during 28 April to 5 September) and Figure 3d (typical LP event waveform after the explosion).
The LP events were also located by a grid-search method based on the joint computation of two different functions: (i) semblance, used to measure the similarity among signals recorded by two or more stations [Neidell and Taner, 1971], and (ii) R2, calculated on the basis of the spatial distribution of seismic amplitude (see Cannata et al.  for further details). LP event sources in 2013 were similarly located to those obtained in the previous years [e.g., Patanè et al., 2008; Cannata et al., 2009, 2013], that is below the VOR-BN craters at shallow depths (> ~2 km above sea level (asl); Figures 5a and 5b). Moreover, if we consider only LP events with low-frequency peaks (<1.2 Hz) taking place during 29 April to 5 September (Figures 5c and 5d), there are no significant differences with the locations obtained during all of 2013.
The time variations of volcanic tremor, in terms of amplitude and source locations, were also investigated (Figures 2d, 6, and 7). In particular, the time evolution of its amplitude was followed by the RMS amplitude of the seismic signal recorded by the vertical component of ESLN station, calculated on 10 s long moving windows filtered in the band 0.5–2.5 Hz (Figure 2d). Since the seismic RMS depends not only on volcanic tremor but also on amplitude transients (LP events, volcano-tectonic earthquakes, regional earthquakes, and so on), we calculated the 25° percentile on 8 h long moving windows of RMS to exclude the contributions of amplitude transients and highlight the variations exclusively related to volcanic tremor changes. The most important RMS peaks, coinciding with the lava fountains, were observed during February–April and October–December periods. If we focus on the 25° percentile trends during the period in between the two lava fountain cycles, it is possible to observe a gradual increase in the volcanic tremor amplitude (particularly evident from June), which slightly reduced on 5 September. This amplitude reduction is visible in Figure 7
- Domenico PatanèDirigente di Ricerca Istituto Nazionale di Geofisica e VulcanologiaVerified email at ingv.it
- Eugenio Priviteraistituto nazionale di geofisica e vulcanologiaVerified email at ingv.it
- Carmelo CassisiIstituto Nazionale di Geofisica e Vulcanologia - Sezione di Catania, Osservatorio EtneoVerified email at ingv.it
- Marco LiuzzoIstituto Nazionale di Geofisica e VulcanologiaVerified email at ingv.it
- Enzo BoschiProfessor of Solid Earth Geophysics, University of BolognaVerified email at unibo.it
- Giuseppe SalernoIstituto Nazionale di Geofisica e Vulcanologia, Osservatorio Etneo, ItalyVerified email at ingv.it
- Luciano ZuccarelloIstituto Nazionale di Geofisica e Vulcanologia - Sezione di CataniaVerified email at ingv.it
- Mario MattiaIstituto Nazionale di Geofisica e VulcanologiaVerified email at ingv.it
- Mauro ColtelliINGV Osservatorio Etneo, Catania, ItalyVerified email at ingv.it
- Tom D PeringUniversity of SheffieldVerified email at sheffield.ac.uk
- Marco ViccaroUniversità di Catania, Dipartimento di Scienze Biologiche Geologiche e AmbientaliVerified email at unict.it
- Sonia CalvariResearch Director, Istituto Nazionale di Geofisica e Vulcanologia - Osservatorio Etneo, ItalyVerified email at ct.ingv.it
- Rosa Anna CorsaroSenior Researcher Istituto Nazionale di Geofisica e Vulcanologia, Osservatorio EtneoVerified email at ingv.it
- Carmelo MonacoProfessore di Geologia Strutturale, Catania UniversityVerified email at unict.it
- Mike JamesLancaster UniversityVerified email at lancs.ac.uk