TOMO-ETNA experiment at Etna volcano: activities on land

In the present paper we describe the on-land ﬁeld operations integrated in the TOMO-ETNA experiment carried out in June-November 2014 at Mt. Etna volcano and surrounding areas. This terrestrial campaign consists in the deployment of 90 short-period portable three-component seismic stations, 17 broadband seismometers and the coordination with 133 permanent seismic station belonging to Italy’s Istituto Nazionale di Geoﬁsica e Vulcanologia (INGV). This temporary seismic network recorded active and passive seismic sources. Active seismic sources were generated by an array of air-guns mounted in the Spanish oceanographic vessel “Sarmiento de Gamboa” with a power capacity of up to 5200 cubic inches. In total more than 26,000 shots were ﬁred and more than 450 local and regional earthquakes were recorded. We describe the whole technical procedure followed to guarantee the success of this complex seismic experiment. We started with the description of the location of the potential safety places to deploy the portable network and the products derived from this search (a large document including full characterization of the sites, owners and indication of how to arrive to them). A full technical description of the seismometers and seismic sources is presented. We show how the portable seismic network was deployed, maintained and recovered in different stages. The large international collaboration of this experiment is reﬂected in the participation of more than 75 researchers, technicians and students from different institutions and countries in the on-land activities. The main objectives of the experiment were achieved with great success.


Introduction. The importance of an active seismic tomography in Etna volcano
Mt. Etna is one of the most fascinating natural laboratories for studying the deep structure of an active volcano with seismic tomography. Mt. Etna's eruptive dynamics are the result of a complex interaction between magma in its plumbing system, the regional (deep crust) tectonic regime, and local (shallow crust; volcanic edifice) structures, partially controlled by flank instability. Magma ascent driving conditions (e.g., structural setting, tectonic forces) are not yet completely understood. One of the main limitations in understanding the volcano's eruptive dynamics is the insufficient knowledge of its intermediate and deep crust. As indicated in Ibáñez et al. [2016, in this volume] in the last 25 years, several seismic and tomographic experiments have been focused on the study of Mt. Etna's structure using different techniques and data sets. The TOMO-ETNA experiment is an international and multi-institutional joint effort focused into providing answers to some of the open questions associated to the structure and dynamics of Mt. Etna. This experiment will take advantage of the use of both active and passive seismic sources and the combination of information of other seismic and geophysics measurements such as marine seismic reflection [Coltelli et al. 2016, in this volume;Firetto Carlino et al. 2016, in this volume] or magnetic profiles [Cavallaro et al. 2016, in this volume].
A tomography image involves a great amount of seismic data [Lees 2007] recorded by a dense seismic network, a requisite often difficult to achieve in a reasonable time interval due to an absence of seismicity or the lack of stable and dense seismic networks [e.g Chouet 1996Chouet , 2003Chouet and Matoza 2013;McNutt 2002McNutt , 2005Wassermann 2002;Zobin 2011]. This problem can be solved by planning active seismic experiments such as we will summarize in this section. Following Koulakov and Shapiro [2015], it is possible to use different seismic sources that they divided into three categories: human-made sources, earthquakes, and ambient seismic noise. Fully controlled artificial sources are expected to provide the highest accuracy of tomographic images. At the same time, their implementation on volcanoes is very sophisticated and expensive. Another shortcoming of artificial sources is that they cannot be produced at depth. In regions with welldistributed volcano-tectonic seismicity, earthquakes may give a distribution of sources plausible for highresolution seismic tomography. However, many volcanoes do not generate relevant seismicity. Human-made seismic sources include explosions, vibrators, air-gun shots, and weight (or mass) drops and are usually called active sources. There are some onshore experiments which use chemical explosions of sufficiently large magnitude to generate seismic signals propagating to distances of dozens kilometers and depths of a few kilometers. The problem of such explosions is that they are fairly expensive and thus their number is strongly limited. Furthermore, chemical explosions are not ecologically sustainable and they are prohibited in most volcanic areas [Koulakov and Shapiro 2015]. An alternative method of active-source onshore generation might be using vibrators which are more ecologically safe compared to explosions. However, for the purposes of volcano tomography, the vibrators should be sufficiently powerful and heavy to enable the required propagation of seismic rays. In reality, the transportation of powerful vibrators appears to be not possible in most volcanic areas; therefore, they are not widely used in practice. Again following Koulakov and Shaphiro [2015], in case a volcano is located close to the seashore, it is possible to use underwater air-guns as sources of seismic signal. The advantage is that air-gun shots are rather cheap and sufficiently powerful. They can be frequently generated along the ship cruise that gives a huge data base with a clear technology of processing. The problem is that these sources are limited by the shore line and not suitable for most volcanoes. The major advantage of the active sources is that their parameters are known. The problem is that they are lo-cated close to the surface, and seismic signals propagate along the nearly horizontal ray paths. Another problem, which is mostly observed for the air-gun shots, is that they generate very weak S waves which cannot be picked and used for tomography studies.
In the last 20 years the number of dedicated experiments using active seismic sources has been increasing. In order to give a real dimension of the nature of the TOMO-ETNA experiment we will present, in Table 1, a short revision of some of the most important experiment carried out in active volcanoes around the word. In Figure 1 we plot the position of the volcanoes described in the Table 1.
The TOMO-ETNA active seismic experiment is probably one of the most complex experiments performed in an active volcano to obtain a seismic tomography. Ibáñez et al. [2016, in this volume] provide a full description of the objectives, plans and involved research projects associated to this experiment. In order to solve some of the problems associated to the exclusive use of active sources this experiment was designed using both active and passive seismic sources. Passive sources (earthquakes) were selected for the period in which the active seismic survey was performed ( June-July 2014) in order to produce a real snapshot of the Mt. Etna structure and to avoid potential influences of structural evolution with time, following the observations done previously by Barberi et al. [2016, in this Zollo et al. [1996Zollo et al. [ , 1998Zollo et al. [ , 2002; Gasparini [1998] volume]. In the present manuscript we describe all activities developed to carry out the necessary activities to work on land, from the search of sites to the recovery of equipment and data.

On-land portable short-period seismic stations
On-land portable seismic stations were provided by the Geophysical Instrument Pool Potsdam (GIPP), Germany [Ibáñez et al. 2016, in this volume]. They are composed of a DATA-CUBE3 recorder and triaxial seismometers PE-6/B 4.5 Hz or Mark L-4C-3D. In Figure 2 we show pictures of these types of seismic stations. The data logger is a three-channel recorder (Figure 2a) with sampling rate of 100 samples per second. Its main characteristics are a selectable gain ×1 to ×32 full scale input of 4.1 Vpp and gain 1 and sample rate up to 400 sps (3 channels), and an effective resolution typically of 21.5 bit and 100 sps. Tables 2 and 3 show some important parameters of the acquisition systems and the seismic sensors such as the sensitivity and clip values. The GPS of the station is integrated on the top of the Cube data logger and has an accuracy of <10ms over 20 days without processing or <0.01ms with processing. These stations were powered with 2D cells batteries with an average duration of 14 days. The sensor is composed of one vertical component SM-6/U-B 4.5 Hz geophone and two horizontal component SM-6/H-B 4.5Hz geophones ( Figure 2b). The flat response of the sensor ranges from 4.5 Hz to > 150 Hz with a damping of 0.56. Additionally, some Mark L-4C-3D sensors were deployed with the Cube recorders ( Figure 2c). These geophones are three-component and have a response from 1 Hz to >100 Hz with a damping of 0.7.
For TOMO-ETNA experiment we deployed 80 Cube stations with 78 3-D PE-6/B geophones and 2 Mark L-4C-3D geophones, some of them were transferred from one location to another during the experiment in order to maximize the network coverage.  Table 3. Main characteristic of the seismic sensors used.

On-land broadband portable seismic stations
On-land broadband (BB) portable seismic stations were provided by the GIPP, Germany , in this volume]. They are composed of an Earth-Data PR6-24 recorder and Nanometrics Trillium Compact seismometer ( Figure 3). The PR6-24 is a data logger for stand-alone use with three-channel digitizer with a 140dB dynamic range at 100 sps of sampling rate. The seismometer is a Trillium Compact, a broadband three-component sensor with a flat response that ranges from 120 sec to 100 Hz. Its three symmetric triaxial sensors are resonance-free up 200 Hz. These stations were powered with solar panels connected to a 75Ah 12 V batteries.
For TOMO-ETNA experiment we deployed 17 broadband stations that operated from June 20 to October 28, 2014. They registered not only the air-gun shots from the active experiment but natural seismicity occurred during the 4-month period, allowing us to increase and enhance the passive database [Barberi et al. 2016, in this volume].
In Figure 4 we plot the response curve for the Trillium Compact (+EarthData), the Mark L-4C (+Cube) and the PE6/B (+Cube) seismometers.

Spanish OBSs
The Unidad Tecnológica Marina (CSIC-UTM, Barcelona, Spain) provided 15 ocean bottom seismic stations (OBSs). Their technical name is LC SP 4×4 IGPP-SIO-UCSD OBS, and they include a HighTech HTI-90-U hydrophone. The geophone is a short-period Sercel L28 model, with natural frequency of 4.5 Hz and a sample rate of 200 sps. Data logger and electronics are BART ORE Offshore software sampling at 200 sps ( Figure 5). The OBSs clock is Seascan TCXO (0.02-0.05 ppm), which must be synchronized with high-precision on-board GPS before deployment and recovery phases, in order to correct time shifts during the recording period. Usual time drift of this clock for 20 days is around 10 msec. These stations were powered with 12 V batteries with autonomy of 2 months. IBÁÑEZ ET AL. 6

Italian OBSs
The Observatory of Gibilmanna, near Palermo, run by Italy's Istituto Nazionale di Geofisica e Vulcanologia (INGV), provided two types of OBS stations to be deployed: i) short-period SM/6 geophone registering within a frequency range of 4.5-140 Hz and with 250 sps; and ii) broadband Güralp CMG40T-OBS sensors that registered between 60 s and 100 Hz with a sample rate of 100 sps. In Figure 6 we show a picture of one of this OBS station. A total of 10 short-period OBSs and 2 broadband OBSs were used in this experiment. All seismic stations were powered with 12 V batteries with autonomy of 1 month for the short-period and 6 months for the BB.

Array Pozzo Pitarrone
Two broadband seismic stations of the permanent monitoring network are installed at Pozzo Pitarrone [Zuccarello et al. 2016, in this volume], one at surface (EPIT) and one at the well bottom (EPIP), at depth of about 130 meters. A temporary seismic array was deployed in the area around Pozzo Pitarrone between May and October 2014. The array consisted of 8 three component broadband stations arranged in two roughly concentric rings of radii 100 m and 200 m centered at EPIT station. The seismometers used for the array were Guralp CMG-40T 60 s (http://www.guralp.com) and Nanometrics Trillium 120 s compact (http://www. nanometrics.ca), and signals were acquired by Nanometrics Taurus data loggers recording at 200 sps, with 24 bits dynamic range.

Permanent network
We have used also the data recorded by the INGV permanent network operated by the Osservatorio Etneo (the INGV Etna Observatory; Figure 7). Seismic stations of this network are broadband three-component Nanometrics Trillium seismometers. These sensors have a period of 40 s, a flat frequency response from 0.025 Hz and a sensitivity of 1553 V/M/S. They register with a sample frequency of 100 Hz by a Nanometrics Trident digital system (24 bits). For TOMO-ETNA experiment we used the data registered in 71 seismic stations. The dataset is provided in DMX format, which is a simplified type of SUDS format. In this case, careful data processing was needed to transform DMX to a SAC format.

Air-gun signals
Active seismic sources were provided by the Spanish oceanographic research vessel (hereafter, R/V) TOMO-ETNA EXPERIMENT AT ETNA VOLCANO   "Sarmiento de Gamboa" and consisted of 16 air-guns deployed along a two-line array 10 m below sea level ( Figure 8). The volume of the guns was of: 520×4, 380×4, 250×4, 150×4 cu.in.; having a total volume of 5200 cu.in. They were spaced 1 m from each other, within the line, and 9 m between different lines. For the TOMO-ETNA refraction experiment the air-guns shot every 90 s, while for the reflection experiment was of 20 s with a 4340 cu.in. of total volume.
The signal associated to this bubble is simulated using a software package named GUNDALF (Figure 9). Before the onset of the experiment, many tests were performed by the Unidad de Tecnología Marina (UTM-CSIC, Spain) to obtain the most efficient distribution and power of the guns. The tests carried out permitted to adjust several parameters for an optimal configuration of the guns that lead to the most efficient output signature. The chosen signature has a dominant frequency of 20 Hz and a total acoustic energy of 479,939.9 J and an acoustic efficiency of 40.8% (Figure 9). In Figure 10 we show a picture of the real bubble generated by the air-guns mounted in the Spanish R/V "Sarmiento de Gamboa".

Earthquakes
A total of 452 earthquakes were selected from the INGV database for the period June 1st to November 30, 2014. The preliminary filter was done on the base that they must be recorded in at least three seismic stations from the permanent INGV network, and inside the re-gion selected for this study. In Figure 11 we plot the epicenters of the selected seismicity. The magnitude (Mw) of these events ranged from 0.3 to 3.8, and the focal depths were from 1.5 km above sea level (Etna edifice) to 215 km below sea level (subduction structure below Aeolian Islands). The selected period of time was chosen in order to produce a seismic tomography as a snapshot of the structure of the region. As Barberi et al. [2016, in this volume] evidenced, the continuous magmatic dynamic of the volcanic structure produce significant changes in the physical properties of the volcanic region, and therefore changes in the velocity patterns. If a larger time interval for the earthquakes was selected, we could introduce a bias in the final picture of the structure due to potential changes of the velocity values, mainly in the Mt. Etna edifice and roots.
In Barberi et al. [2016, in this volume] there is an exhaustive description of the seismic and volcanic activities of the region and in Mt. Etna during the period of the experiment.

Search of the sites
The on-land phase started with the search of the potential places to locate the portable seismic stations. This is a very important and sensitive step since a careful selection of the site will permit to deploy efficiently and safely the seismometers. This search was done imposing a series of quality criteria to place a seismic sensor: (a) low ambient/cultural/volcanic noise level (b) protected and safety sites (c) easy to arrive and maintain (d) consolidated host rock.
Initially, an area of 5500 km 2 was selected as the region to deploy the stations. The size was estimated taking into account the main objectives of the project: to produce a seismic tomography of Mt. Etna volcano and surrounding areas in order to place this volcano in a regional framework. This zone comprises the eastern coast of Sicily and Mt. Etna volcanic edifice (see Figure  1 of Ibáñez et al. 2016, in this volume]. Although the area under interest comprises part of eastern Sicily Island, the Aeolian Islands and part of the south of the Italian Peninsula (Calabria), since in both Aeolian Islands and Calabria the permanent INGV seismic network has several stations, we centered our efforts exclusively in Sicily Island. The first step was to place on a map a set of 200 theoretical locations for the portable seismic stations according to the above quality requirements. Once this preliminary map was defined, from the beginning of April to the end of May of 2014 a team led by A. Díaz-Moreno from the UGR and O. Cocina from the INGV, Section of Catania (hereinafter: INGV-Catania), with the support of several researchers of the INGV-Catania, visited each of these locations, in some cases more than once. Finally 150 potential locations were selected covering almost homogeneously the region of interest. The work of this team included to stablish contact with the private owners or the public authorities responsible of the sites and asking for permission for the temporal deployment of the seismic stations. Official documents were provided by the INGV in order to have an additional support in this negotiation. Preferentially they selected public properties such as cemeteries, parks or schools (the planned shooting period was during a non-teaching season). If no public properties were identified, then private areas with the possibility of restricted access were selected such as farms, rural hotels, and golf resorts. In the Mt. Etna region, the special protected area due to the Natural Park declaration facilitates this search and location of sites. TOMO-ETNA EXPERIMENT AT ETNA VOLCANO  The Natural Park facilities were also used for several stations placed in the Peloritani Mountains.
In several times, the researchers had to convince both authorities and private owners of the importance of the experiment. Many private owners were concerned about possible health issues regarding the installation of the stations (the most repeated question done for many owners was if the seismic stations had any level of radioactivity or how noisy they could be, or the possible consequences on their properties deriving from the 'measurements' of earthquakes). The search teams were always composed by at least one Italian researcher from the INGV due to of the importance of having a local and institutional representative.
The BB stations needed additional requirements such as a place to install solar panels and consolidate and cover basement since they will be operating for longer time. Furthermore, their objective was to record natural seismicity generated by Mt. Etna volcano, and therefore they must be located close to the volcano edifice. Therefore most of the selected places for the portable BB stations were placed already used for the allocation of other geophysical or geochemical instruments.
At the end, mostly of the selected places were public sites. As a curiosity we can mention that more than 25 cemeteries were used to install seismic stations. Indeed, no robbery or damage of any instruments had been reported during the experiment. At the end of the experiment the only risk that the seismic stations were exposed to were forest-fires and the disturbance from farm animals. In any case we were lucky and no damage was reported. As an example, near the village of Catenanuova a fire was declared. Flames stopped just few meters from the station as illustrated in Figure 12.
In total around 10,000 km were driven by the different teams during the search period. Once a site was identified as suitable, the search team prepared a document containing information on the location, details  about driving instructions, owners' contacts, photos, etc. In Figure 13 we report an example of this type of document. All these documents were compiled in an electronic and printed format under the coordination of A. Díaz-Moreno and with the collaboration of Italian and Spanish researchers. The aim of this book was to allow to the deployment teams to arrive promptly to the locations. Taking into account that many of the researchers involved had never visited the place, the information should be very accurate and detailed. It is noteworthy that this information was a complete success since all the researchers were able to reach the locations without major problems, leading to a much more efficient work. At the present it is planned to print the complete information (more than 250 pages of data) in a type of open file report in order to be accessible to the rest of the scientific community working at Mt. Etna volcano. Due to the large extension of the region under study, once the best sites were identified the whole region was divided in four sectors with an almost same number of seismic stations contained in each one. In each sector a city or village was selected and placed to accommodate the researchers responsible of the deployment, maintenance and recovery of the seismic stations. The four subteams were located in the city of Catania and in the towns of Randazzo, Milo and Villafranca.

Deployment
In June 16, 2014, the TOMO-ETNA active seismic experiment, officially started with a preparation of the seismometers to be deployed at the Etna volcano and surrounding area. The first activity carried out in this new phase of the experiment was the revision of the seismic stations provided by the GIPP-GFZ institute, 80 short-period and 20 broadband stations. All equipment were stored in the INGV headquarters in Nicolosi and it was necessary to transport a large amount of material from Catania to this place. In Figure 14 we illustrate this process of transportation and storage of the equipment.
On June 17 a training session for the use of the equipment was programmed for a selected group of Italian and Spanish researchers (deployment team leaders) that will be responsible for a process of further training to the other researchers. This first training was performed by Prof. B.-G. Lühr from the GFZ of Potsdam (Germany). In Figure 15 we illustrate this process. On June 18 and 19 this team of leaders deployed some of the seismic stations in some sites near the Etna volcano under the supervision of Prof. Lühr in order to check the right operation procedure and to test potential mistakes, faults or other minor problems. On July 20 the whole TOMO-ETNA team (a great part of the non-Italian researchers arrived on June 19) participates in a new training session in Nicolosi and received the seismic stations that each subteam must deploy. As a part of the training process each TOMO-ETNA EXPERIMENT AT ETNA VOLCANO Figure 13. Information slides of one site selected for station location. In this case it is the Cemetery of Regalbuto town. From all potential locations described in the book, we selected the best 100 places following the criteria described in Section 4.1 and geometrical constraints to obtain a dense regular grid for the seismic network.
team received the information of the pre-selected sites to deploy the seismic stations. This information was presented in the document described in Section 4.1.
For each seismic station we require the compilation of a file with technical information associated to the instruments and deployment procedure. In Figure  16 we show an example of this sheet.
Each of the four working groups was equipped with: -a station wagon and a van -a smartphone with Italian card, including internet connection and a GPS application -four halogen headlamps with batteries -a toolbox with various kinds of screwdrivers, pliers, wrench, scissors, electrical duct tape, duct tape, gloves, compass, level and a tester -two picks, two shovel and two hoes -a fluorescent jacket for each person with identification of the project, INGV and UGR  -two back-packs of 50 liters of capacity to transport sensors and batteries -their correspondent number of data logger, seismometers and batteries, including batteries of replacement for the data logger -a folder with the information of their respective stations (instruction of use, replacement and maintenance), sites and routes to arrive to them -a clear identification sheet stating that the car and the persons in it belong to the INGV-Catania and to the MED-SUV project and TOMO-ETNA experiment.
Additionally we were supported by different INGV technicians from Catania, Rome, Naples and Bologna. These technicians were in charge of the deployment of the BB stations.
In June 23 the deployment started. The workingplan for the deployment was conceived in order to optimize the different assigned routes and to deploy the seismic stations as fast and safe as possible. In Figure 17 we plot the initial planned sites distributed according the different working teams and we mark, as an example, the routes followed by the different teams in one day of work. Additionally, in Figure 18 we illustrate the installation process. As a summary of the number of seismic stations deployed: (a) Villafranca group: 14 short-period seismic stations (b) Randazzo group: 18 short-period seismic stations (c) Milo (Etna) group: 12 short-period seismic stations (d) GPS technicians: 8 broadband seismic stations (e) seismiological technicians: 6 broadband seismic stations (f ) Catania group: 4 broadband and 36 short-period seismic stations.
In June 26 the 95% of the seismic stations had been deployed in their different sites according with the preestablish selection. The final map with the sites of the whole seismic network deployed for the TOMO-ETNA experiment, including the permanent seismic network of the INGV, is presented in Figure 18.

Maintenance
After the phase of installation of the seismic stations, each group developed activities of maintenance and control. These activities were focused mainly in controlling the proper operation of the seismic stations and in changing the batteries every 7 days to ensure data acquisition. To avoid potential vandalism on the seismic stations, every site was visited at least once every 2 or 3 days.
As routine procedure, every day the operation coordinator, Dr. J. Prudencio, contacted each headquarter to indicate the activities of the day, potential changes, recovery information and potential problems, including helping other teams reaching the sites.  During the second leg of the shooting process, after July 10, some of the short-period seismic stations were moved from their original planned sites to new one in order to cover additional sites and improve the spatial distribution. In total 23 short period-stations were moved. In this case the initial planned 98 portable station sites were increased to 121 sites.

Recovery
On July 24, 2014, the active phase of the seismic experiment TOMO-ETNA ended. At this date the whole portable short-period seismic network was recovered. The broadband seismic network remained operative until October 28, 2014. At the same time the R/V "Sarmiento de Gamboa" arrived to the Catania harbor after the recovery of 21 short-period OBSs. The rest were recovered by the R/V "Aegaeo" in November 25, 2014. The uninstallation process of all seismic stations was a success and all the recovered instruments were fully operating and with no damages. In total, more than 25,000 km were driven during the experiment. In Figure 17 we show the final position of the whole stations used in the TOMO-ETNA experiment.

Seismic data bases
The TOMO-ETNA dataset comprises 26,364 airgun shots (9705 refraction shots and 16,659 reflection shots) registered in 80 short-period, 17 broadband, 27 ocean bottom stations (OBSs) (15 provided by the Spanish team and 12 by the Italian team) and 70 broadbandshort-period (INGV permanent network) stations, summing 5,140,980 seismic signals. In Figure 19 we illustrate the route followed by the R/V "Sarmiento de Gamboa" during the whole shooting process. For additional information of the marine activities see Coltelli et al. [2016, in this volume]. For the refraction experiment we have a database of 1,892,475 seismic signals (including all stations). Additionally we recorded a total of 452 earthquakes in a dense seismic network. When the magnitude of the earthquake was larger than 3.0 it was recorded in more than 50 seismic stations. In Figure 20 we show an example of a regional earthquake recorded by this network. Red dots indicate the estimate P-wave onset determined by using an automatic algorithm described by Álvarez et al. [2013] and García et al. [2016, in this volume].
In general the quality of the signal is high. Due to the large area covered in this experiment, it was not possible to record the air-gun shots in the whole seismic network. In average the maximum distance shootstation to have good quality was of around 50 km. In Figure 21 we plot an example of 3 hours of shooting period recorded by a single station. Red starts indicate, as in the case of the earthquakes, the estimated P-waves onsets of every explosion calculated automatically. In order to show the quality of the recorded signals we plot an example of a single shot recorded in four types of seismometers (short-period, broadband, permanent and OBSs) at different distances ( Figure 22). It is remarkable how the signal has been filtered by the medium. As indicated in the characteristics of the active source, the theoretical source spectra are wide, between 4 and 40 Hz. However the spectrograms of the recorded signals show a peak of amplitude, between 6 and 15 Hz. This observation was used to select a filter between 4 and 12 Hz for the signal analysis in the tomographic inversion [Díaz-Moreno et al. 2016, in this volume].

Database managing
The first step was the unification of the data formats, because all the seismic data are recorded in a dif- ferent formats depending on the station features (shortperiod, broadband, OBS, etc.). This procedure is not as easy as it may seem, since it is a very sensitive procedure. In the following paragraphs the different original formats and their conversion are detailed. Our preferred format to work with is .mat (matlab) so that we can process the data in a less time-consuming way.
We analyzed separately both earthquake and shots data. For the shot database we started with the Cube stations. These stations save the data in binary daily files that should be then converted to an appropriate format. The GFZ Potsdam institute provides useful algorithms to convert Cube format to MiniSeed format. During the recorded period, many of the 80 stations deployed lost the GPS signal sometimes, leading to more than one daily file. We found out that the registered data had some gaps, meaning that for certain periods of time the station did not register. The most probable cause of this is the loss of GPS signal. Fortunately, MiniSeed files are written with an absolute scale of time, thus, allowing us to identified and exclude these gaps. Another step we performed was to convert MiniSeed into SAC in order to choose the more suitable type file to work with. On the other hand SAC files have the time information not in absolute values but in a single time vector from 1 to the end of samples. Therefore, gaps could not be well identified using this type of files. For this reason we decided to work directly with the MiniSeed data. To reduce computation time we transformed the data into .mat format taking only the pieces of signal we needed. Indeed, we cut the signal 30 s before that data and 60 s after each shot. When we find a gap in the registration, the shots occurred in that period were not taken into account for that station.
Once the Cube database is organized in the new format, we plotted all shots registered at one station during one day using Matlab tools. This way of representation allows us view the movement of the vessel and to estimate the acoustic wave (1.5 km/s) registered on the stations near the coast. Indeed, controlling the arrival time of the acoustic wave is a simple and very effective procedure to check that the GPS data times are correct, and that there is synchrony between the GPS data time of the air-guns and the stations.
The data extracted from the broadband stations are in MiniSeed format; therefore not further conversion is needed, only into .mat files. In this case we followed the same procedure as for the Cubes, we took the time data of each explosion to cut the MiniSeed file 30 s before it and 60 s after it and save it in Matlab files. The only problem we addressed in this step was that the coordinates of the broadband stations were daily calculated and differ from one day to another in some decimals. In order to have a final station coordinates we calculated the mean of all these daily GPS coordinates.
The INGV permanent stations provide data formatted in .DMX files, which result to be a compressed type of .SUDS files. This type of file is organized in a TOMO-ETNA EXPERIMENT AT ETNA VOLCANO Figure 19. The whole route followed by the research vessel "Sarmiento de Gamboa" during the shooting period.
completely different way as the previously mentioned MiniSeed or SAC. They save registration of 1min for each station. In this case, the INGV research team provided the necessary tools to convert them to a user-friendly SAC format. In any case we may address the same problems as with the Cube data; if any gaps are present we will not be able to detect them with the SAC format. Nonetheless, we converted them into .mat files following the previously discussed procedure and fortunately, we found no gaps in the registration of these stations.
The Spanish OBS data are provided in RAW binary format. The tools for conversion permit to obtain an output format known as PascalSegy. This data files were, then, organized in the common .mat files for their analysis.
The Italian OBS data available at the time of this report are only those recorded by the broadband OBSs. These data are saved in SAC files and therefore easy to handle. The procedure was the same as for the rest of data, obtaining .mat files as the final format.

Final remarks
As reported in the present manuscript, probably the TOMO-ETNA experiment is the most complex experiment focused to obtain information on the inner structure of an active volcano and surrounding areas using both active and passive seismic sources. The efficiency and coordination in the performed work has allowed deploying in a short time a large number of seismic stations. All of them were operative along the scheduled time, safely recovered, and even if the island if Sicily is a very populated region, no seismic station was damaged. The first step in the success of this deployment was based on an efficient search of potential sites a few months before the arrival of the working team, including complete and detailed information on the characteristics of the sites, owners, contacts and easy indications of how to arrive to them. The final document generated in this search is very valuable information that could be used in the future for new experiments. The second factor contributing to the success was the division into several teams responsible of the deployment and the maintenance of the seismic stations distributed in different sectors covering the whole region under study. These teams were autonomous and, following the indications of the provided documents, were able to design an efficient daily working program, coordinated by a single person who was responsible for any final decision. It is remarkable the high quality, portability and feasibility of the used instruments, in particular the Cube shortperiod seismic stations are ideal to be used in this kind of experiments. The quality of the recorded data predicts final results that could have a high impact in the scientific community.
Acknowledgements. This paper has been partially funded by the following research projects: the European project MED-SUV funded by the European Union's Seventh Framework Program for research, technological development and demonstration under grant agreement No. 308665; the Spanish COCSABO project (COC-DI-2011-08); the European project EUROFLEETS2 (Seventh Framework Programme, grant agreement No. 312762) through transnational access to the research vessels "Sarmiento de Gamboa", operated by CSIC (Spain), and "Aegaeo", by HCMR (Greece); the Geophysical Instrument Pool Potsdam (GIPP) from GFZ (Potsdam) with the project (Seismic TOMOgraphy of ETNA volcano and Eolian Islands, Italy, using active and passive seismic data). We would like to thank the following supporting institutions: Dipartimento Regionale della Protezione Civile, Regione Siciliana; Dipartimento Azienda Regionale Foreste Demaniali, Ufficio Provinciale di Catania; Ente Parco dell'Etna; Unidad de Tecnología Marina -CSIC in Barcelona (Spain); Stato Maggiore Marina (Italian Navy General Staff ), CINCNAV (Command in Chief of the Fleet) and Marisicilia (Navy Command of Sicily); Coastal Guard of Messina and Riposto; to obtain support and navigation permissions for the oceanographic cruises: Spanish Foreign Office and Italian Foreign Office. This paper has been partially supported by the Spanish projects TEC2015-68752-R (MINECO/FEDER), KNOWAVES and CGL2015-67130-C2-2. We would like to thank all private and public owners of the sites selected to deploy seismic stations for their kind and unselfish disposal to use their properties. This manuscript has been largely improved by the insightful comments of Dr. Simona Petrosino and an anonymous reviewer and by the editor José Morales. TOMO-ETNA EXPERIMENT AT ETNA VOLCANO