The first ultra-high resolution Digital Terrain Model of the shallow-water sector around Lipari Island (Aeolian Islands, Italy)

Very high resolution bathymetric map obtained through multibeam echo-sounders data are crucial to generate accurate Digital Terrain Models from which the morphological setting of active volcanic areas can be analyzed in detail. Here we show and discuss the main results from the ﬁrst multibeam bathymetric survey performed in shallow-waters around the island of Lipari, the largest and the most densely populated of the Aeolian Islands (southern Italy). Data have been collected in the depth range of 0.1-150 m and complete the already existent high-resolution multibeam bathymetry realized between 100 and 1300 m water depth. The new ultra-high resolution bathymetric maps at 0.1-0.5 m provide new insights on the shallow seaﬂoor of Lipari, allowing to detail a large spectrum of volcanic, erosive-depositional and anthropic features. Moreover, the presented data allow outlining the recent morphological evolution of the shallow coastal sector of this active volcanic island, indicating the presence of potential geo-hazard factors in shallow waters.


Introduction
Since the beginning of 1930s, single-beam echosounders were used to produce early remote-sensing based bathymetric maps. In the last decades, seafloor mapping systems have undergone a technological revolution, especially through the development of multibeam echo-sounders (MBES). While the use of the former single beam sensors remains mainly restricted to recreational applications, multibeam systems allowed to extensively map continental margins, greatly enhancing the knowledge on geological and oceanographic processes that are sculpturing the seafloor [Augustin et al. 1994, Bourillet et al. 1996]. The main reason for such improvements in seafloor mapping is that MBES, with respect to single beam echo-sounders, are equipped with hundreds of narrow adjacent beams arranged in a fan-like swath of typically 90 to 170 degrees across, providing very high resolution measurements up to few centimeters. MBES mapping also benefited from GNSS satellite data to precisely positioning the vessel during surveys, also supported by on board integrated motion-sensors used to estimate and correct continuously the vessel motion (roll, pitch and yaw) during the survey. The GPS satellites of the Global Navigation Satellite System (GNSS) provides the precise position of the vessel and the MBES sensor during navigation even up to a few cm of accuracy, then collecting the soundings data in the WGS84 geodetic reference system. Additional sensors are used to determine the speed of sound in the water column along the routes, to correct for the refraction effect of the sound waves in response to changes in water density due to temperature, conductivity and pressure from shallow to deep waters. Finally, powerful computer systems equipped with advanced computer graphics support navigation data and the ongoing surveys in real time. The post-processing of large amount of MBES data (in the order of 10 5 -10 7 of soundings) has the final goal to produce Digital Terrain Models (DTM) of the investigated areas at very high resolution, giving unprecedented detailed 3-D views of the seafloor. Nowadays MBES technique represents the most significant advance in the field of seafloor and continental water basins mapping, becoming a crucial tool for marine geomorphological studies [Locat and Lee 2002, Morgan et al. 2003, Esposito et al. 2006 and references therein, Anzidei et al. 2008, Bosman et al. 2009, Blondel 2012, Romagnoli et al. 2012, Romagnoli et al. 2013a].
With the aim to produce a very high resolution DTM of the shallow-water sector of the active volcanic island of Lipari (Aeolian Islands, Italy), an ultra-high resolution multibeam survey was carried out on September 2014. The aim of the paper is to present the technical details of the survey and briefly depict the first results arising from the ultra-high resolution multibeam bathymetry, evidencing possible implications on recent volcanism, tectonics and coastal hazard at Lipari Island.

The investigated area: tectonic and volcanic framework
Lipari Island is the largest and the most densely populated among the Aeolian Islands (southern Tyrrhenian Sea; Figure 1 inset). It displays an area of about 38 km 2 and represents the culmination of a broad largely submarine volcanic edifice ( Figure 1) belonging to the Lipari-Vulcano-Salina volcanic belt [Casalbore et al. 2014c]. The eruptive history of Lipari spans between ≈267 ka and medieval ages (AD 776-1220) and can be divided into nine epochs of activity interrupted by dormant periods, volcano-tectonic phases and episodes of terrace formation during the Last Interglacial [Calanchi et al. 2002, Forni et al. 2013. Active tectonics along NNW-SSE and subordinate N-S and E-W fault systems controlled several volcanic edifices that were active through time. Lipari, together with Vulcano and Salina Islands, is located at the northern end of a major NNW-SSE trending right-lateral strike-slip fault system, known as "Aeolian-Tindari-Letojanni". The latter has been interpreted as a tectonically active lithospheric discontinuity extending from the Aeolian Islands to the Ionian coast of Sicily [Serpelloni et al. 2011, Ventura 2013, Barreca et al. 2014. It is worth noting that Lipari Island is only the tip of a large volcanic edifice that rises about 1300 m above the seafloor, reaching a peak of about 602 m above sea level at Mt. Chirica. The submarine portion accounts for about the 80% of the entire areal of the edifice, so that several marine surveys were carried out in the last decade to map and characterize the main volcanic and erosive-depositional features offshore Lipari (Figure 1) [Romagnoli et al. 2013b, Casalbore et al. 2014a. Nevertheless, the use of large Research Vessels (R/V Urania and Thetis belonging to the Consiglio Nazionale delle Ricerche) and the necessity of ensuring safety condition during the navigation, prevented from performing the surveying in shallow-water areas (<−40/−50 m, white area in Figure 1), despite the areas close to the coastline represent the most critical for marine geohazards, especially in geologically-active setting [Chiocci et al. 2008, Bosman et al. 2014].

The Multibeam bathymetric survey
The multibeam bathymetric survey covered an area of 17 km 2 and was performed at the end of September 2014. Data were collected all around the coast of Lipari Island mostly in the depth range between about 0.5 and 150 m, using a 7 m long boat named BigOne belonging to the Istituto Nazionale di Geofisica e Vulcanologia (henceforth, INGV; Figure 2a). The vessel was equipped with a pole-mount at whose bottom end was installed a ultra-high resolution Teledyne RESON SeaBat 7125 SV2 multibeam system belonging to the Istituto di Geologia Ambientale e Geoingegneria of the National Research Council (Figure 2a). This system works at the frequency of 400 kHz emitting up to 512 beams across a 140°/165° wide swath; each beam has a width of 1° × 0. 5° (www.iho.int;International Hydrographic Organization [2005]). Besides the multibeam transducer, the system architecture ( Figure 2b)  Casalbore et al. [2014c]). The white zone shows the area covered by the 2014 ultrahigh resolution multibeam surveys discussed in this paper, where shallow-water bathymetric data were lacking from previous surveys.
includes: i) a RTK-DGPS positioning system, ii) a sound velocity profiler, iii) a sound speed and temperature sensor used to update in real-time the sound velocity values close to the flat face of the multibeam transducer, iv) a portable tide gauge station, v) a main control center workstation to synchronize, process and store all the raw data files.
In detail, the vessel positioning was supplied in real-time by an Applanix Position and Attitude System (POS/MV 320 V5) using RTK corrections received by a temporary GPS master base-station located on land at Pignataro harbor, through a high frequency link at 1 Hz rate both in transmission and acquisition (blue triangle in Figure 3a). Raw GPS data were recorded both in the master base station at Pignataro (Trimble SPS receiver equipped with Zephir GPS antenna) and in the rover station (two Trimble antennas) mounted onboard the vessel for Post-Processing Kinematics (PPK) corrections. Horizontal and vertical positional accuracy of this system is typically of +/−8 mm and ±15 cm, respectively. Attitude (pitch, roll, yaw, and heave) data were recorded at 100 Hz by the Inertial Motion Unit (IMU) with an average pitch, roll, and yaw with an accuracy of ±0.03°, whereas heave accuracy was main-MULTIBEAM BATHYMETRY OF LIPARI ISLAND  tained at ±5% or 5 cm. Lever arm offsets of each instrument (multibeam transducer, GPS antennas) relative to the Inertial Measurement Unit (IMU) were measured at the beginning of the survey, as they serve to reference the multibeam data to true position and to properly apply motion compensation corrections. In order to minimize the offset estimation between the master GPS antenna, IMU sensor Reference Point and MBES, the devices were placed in an ad-hoc designed box and pole system. Survey track lines (green lines in Figure 3a) were mostly run parallel to the isobaths and overlapping at 20% to guarantee the full coverage of the seafloor in the bathymetric range encompassed between the coastline (or at the minimum distance for safety of navigation) and the deeper multibeam dataset collected in the framework of the MaGIC Project (Figure 1; www.mag icproject.it). Patch tests (i.e. ad-hoc sounding lines acquired for the calibration of the multibeam sensor) in areas close to the survey zone were daily acquired on flat bottom and steep target between −10 and −60 m: the tests were realized to compute roll, pitch, yaw angles and time delay of the MBES and IMU sensors.
Eighteen sound speed profiles were collected during surveys around Lipari Island using a Valeport SVP to correct for any variation in sound velocity due to temperature and salinity changes throughout the water column at 0.1 m depth intervals (Figures 3a and 3b). Moreover, real-time sound velocity close to the transducer was provided by a Valeport Mini-SVS mounted on its port side, since any error in this value would introduce an angular error both in the beam angle and the ray-tracing.
Tidal correction during surveys was performed using a portable tide gauge station (Tidemaster Valeport with accuracy ±0.1% full scale) that was installed along the pier of Pignataro harbour (black triangle in Figure 3a) at 5 m water depth. Before the installation, the sensor has been calibrated on site based on the local oceanographic conditions. Tidal data were collected with a sampling rate of 4 minutes and thirty seconds of burst interval, successively compared and levelled to the averaged value recorded at the Ginostra tide gauge station of the Italian National Tidal Network (see www.mareografico.it for data and products), located in the nearby island of Stromboli (Figure 4).

GPS data processing
To define the geodetic reference frame of the collected data, we used data from the Global Positioning System (GPS) satellites analyzed in the Post-Processing Kinematic (PPK) mode. GPS data processing was performed in two steps: in the first we estimated the precise position of the temporary base station with respect to the GPS station LOSV (red triangle in Figure 3a) (LOSV position is: Lat 38°26´44.53˝; Lon 14°56´53.36˝; Height 273.30 m). This station, which is the nearest to the investigated area, is belonging to the RING national GPS network managed by the INGV [Avallone et al. 2010]. In the second step, we estimated the RTK-DGPS kinematic positions of the rover GPS placed on board the vessel reprocessing the RTK solution using the corrections provided by LOSV station .
Data have been post-processed using POSPac MMS Software to reduce horizontal and vertical positioning uncertainties, based on the satellite constellation orbits, clock corrections, atmospheric delays coupled with attitude data (L1+L2, ephemeris precise). Postprocessed kinematic techniques were used to generate a Smoothed Best Estimate of Trajectory (SBET) file. SBET file is the output of a post-processing solution that tightly integrates the orientation and position data of the IMU and GPS sensors through statistical filtering. SBET solution includes rotational motion around the three axes, as well as heave due to surface waves and GPS tidal variation over the survey period, that were all tied to LOSV station. Finally, the 3D positions of the GPS rover placed on board of the vessel, have been estimated at a few centimeters level.

Multibeam data processing and DTM generation
Multibeam data processing was performed using Caris Hips and Sips 8.1 [CARIS 2000] encompassing the following steps ( Figure 5): a) conversion and import of multibeam raw data files (s7k), b) application of tide data leveled to the Ginostra station, c) importing and replacing of the new Smoothed Best Estimated Trajectory on the raw multibeam data, d) ray-tracing by sound speed profiling with interpolation by time and/or distance between velocity depth-profiles, e) matching of multibeam lines by patch test on specific targets performed every day, f ) application of statistical and geometrical (angle and distance) filters for each swath to remove coherent/incoherent noise, g) the manual deletion of spikes due to single fake soundings during editing. Some 600 × 10 6 processed soundings were merged and gridded (e.g. using a weighted averaging algorithm; Ware et al. [1991]) for the generation of DTMs at different resolutions: cell size, varying from 0.1 m in very shallow water (down to −40 m) to 0.5 m at greater depths. The final DTM in this work was realized with cell-size of 0.5 m ( Figure 6). The sub-aerial DTM is ob-MULTIBEAM BATHYMETRY OF LIPARI ISLAND  tained from LIDAR data, provided by the Ministero dell'Ambiente.
Moreover, the multibeam backscatter signal was processed through SIPS SST (Side Scan Tools), applying radiometric and geometric corrections to the data, including slant-range corrections based on the available bathymetry, as well as corrections for beam pattern effects and time and angle varying gain. SIPS SST also calculates the Time Variable Gain (TVG), despeckle and gain normalization ( Figure 5). Multibeam backscatter mosaic was realized with a pixel size of 0.1 m.

The morpho-bathymetric map and preliminary analysis
The preliminary analysis of the new morpho-bathymetric data reveals the presence of several unknown geomorphic features in the shallow-water sectors around Lipari Island. These are related to volcanic, erosive-depositional and biological processes as well as to the anthropogenic interference. The most remarkable features are a number of canyons, whose headwall extend up to −5 m of depth and a few tens of meters far from the coasts (Figure 7). These features are mainly developed in the southern and eastern coast of the island, where the insular shelf is lacking (Figure 1; Romagnoli et al. [2013a], Casalbore et al. [2014c]). In contrast, the western part is mainly characterized by flat and relatively smooth seafloor surfaces, representing the top of four order of prograding depositional bodies interpreted as Submarine Depositional Terraces (SDT in Figure 8) by Chiocci and Romagnoli [2004]. SDT are characterized by a smooth seafloor, except for near-shore areas where blocky facies were often identified at −20 m (Figure 8). The blocky facies represent the reworking of coarsegrained material derived from the erosion of submarine lava flows and rock-falls processes that affected the overlying coastal cliff. Sometimes, relict volcanic features crop out from the surrounding seafloor, rarely emerging from the sea as in the case of Pietra del Bagno islet (Figure 8). Large areas of SDT also show smallscale roughness of the seabed, due to the alternation of blocks with metric/sub-metric size, sandy areas and widespread Posidonia oceanica meadows down to −40 m, similarly to that observed at the nearby Stromboli Volcano. In detail, Posidonia meadows are easily recognizable on the data due to their peculiar morpho-acoustic facies and backscatter pattern ( Finally, it is remarkable the relief of an archaeological structure, likely a roman age pier located in the harbor of Lipari Marina Grande. This structure is about 150 m long and 20 m across and its top, largely buried by a thick layer of young sediments, is submerged up to a depth of about −11 m, inferring higher rates of land subsidence and relative sea level rise for this area with respect to other regions of the Mediterranean coasts [Pirazzoli et al. 1996, Morhange et al. 2001, Lambeck et al. 2004a,b, Lambeck and Purcell 2005, Anzidei et al. 2014b] and of the Aeolian Islands [Tallarico et al. 2003, Anzidei et al. 2014a]. This pier is currently under investigation by the archaeologists of the Soprintendenza del Mare della Sicilia that revealed part of the valuable constructional features of this site lying very close to the modern pier currently used by local navigation companies (http://www.regione.sici lia.it/beniculturali/archeologiasottomarina/).
The outer mole of Pignataro harbor is placed across the edge of the high-stand depositional terrace. Its end is laying on a steep slope at 25°-29° (Figures 7  and 9c). Widespread fractures observed along this structure can be related to seafloor instabilities that affected the foundations since the time of its construction.

Discussion
The new ultra-high resolution multibeam survey provided the first 3D detailed morpho-bathymetric map of the coastal seafloor of Lipari Island (i.e. between −0.1 and −150 m) at a resolution better than 0.5 m ( Figure 6). Moreover, these data allowed to complete the previous seafloor mapping of the deeper submarine portion of Lipari edifice performed in the framework of the MAGIC project (http://www.magicproject.it).
Although the volcanological and geomorphological interpretations are not the specific goals of this paper, which aims to present the bathymetric surveys and the first very high resolution map of the shallowwater sectors of Lipari Island, the collected data provide information on the geometry and characteristics of morphological features related to volcanic activity, sea-level changes, sliding or rock-fall processes as well as of submerged archaeological structures. The first new evidences arising from the very high resolution multibeam bathymetry can be summarized as follows: -the set of four order of STDs lying between -10 and −140 m characterizing the insular shelf of Lipari [Chiocci and Romagnoli 2004] appears well preserved along the western side of the island. Conversely, on the eastern side the terraces are absent or cut by submarine canyons that run up to the current shoreline; -slide scars or rock-fall blocks are very common offshore the island, especially along the steepest coasts or at the outer edges of the submarine marine depositional terraces; -despite the recent volcanic activity, data did not reveal any relevant submarine gas exhalative centers. This is in contrast with the nearby island of Panarea and Vulcano that show several active exhalative centers even with explosive features in the shallow seafloor [Esposito et al. 2006 andreferences therein, Monecke et al. 2012]; -in the area of Marina Grande, which is the location of the main commercial harbor of the island, the remnants and the shape of a long pier, probably of roman age, may correspond to the entrance of the ancient harbor of Lipari (http://www.regione.sicilia.it/ beniculturali/archeologiasottomarina/); -the occurrence of several active canyons heads very close to the coastline is relevant for hazard assessment of Lipari Island.
Therefore the new bathymetry opens questions on the recent geological evolution of this island suggesting further investigations through the integration of MULTIBEAM BATHYMETRY OF LIPARI ISLAND different geological and geophysical studies. Mainly, if the subsiding behavior of Lipari inferred from instrumental geodetic [Mattia et al. 2008, Barreca et al. 2014 and historical data (Calanchi et al. [2002], Mazza [2013]; http://www.regione.sicilia.it/beniculturali/archeolo giasottomarina/) can be addressed to the interaction between tectonic structures, volcanic activity and retrogressive erosive processes mining the stability of the subaerial flanks. Moreover, the proximity of active canyon heads at the coast together with steep slope gradients may favor the onset of retrogressive submarine slides, as reported in other active canyons [Assier- Rzadkieaicz et al. 2000]. The latter may cause a loss of stability of the coastline and of the near offshore which can become the place of potential gener-ation for local tsunamis. In this regard, the bathymetric data can represent a reference base for future repeated bathymetric surveys in order to assess eventual changes in the morphology of the canyon heads related to slope failures. Although historical landslide-generated tsunamis are not reported at Lipari, tsunamigenic slides recently occurred at the head of active canyons close to the coastlines, such as for the Gioia Tauro Canyon in 1977 [Colantoni et al. 1992 [Mastronuzzi et al. 2013]. Similar events also occurred in the Aeolian Islands during the 2002-2003 eruption of Stromboli volcano [Maramai et al. 2005, Chiocci et al. 2008. Regional tsunamis may also affect the coast of Lipari, such as in the case of the 2003, M = 6.9 Boumerdès earthquake in northern Algeria, that struck most of the coasts of the central Mediterranean, including Italy, causing damages and casualties [Vecchio et al. 2014, and references therein]. In this context, the presence of submarine canyons with high slopes near the coastline is a critical condition for the propagation of tsunami waves approaching the coast [Weiss 2008].
At Lipari, the steepest seafloor areas are close the town of Lipari with its industrial and touristic activities and coastal installations, therefore this area is particularly sensitive for potential geo-hazard. Further analysis of the morpho-bathymetric maps, in combination with additional marine (seismic, seafloor sampling, etc.) and aerial surveys are needed to better constraint the recent evolution of the island as well as to assess coastal dynamics and related hazard. In particular, the realization of a very high resolution terrestrial DTM, like for other active Italian volcanoes [Achilli et al. 1998, Baldi et al. 2006, Baiocchi et al. 2007, Pesci et al. 2007] and its integration with the present multibeam bathymetry will represent a further step for a comprehensive analysis of coastal hazards, including tsunami modeling for near and far-field sources and coastal slides, similarly to that was realized for the nearby Panarea Island [Fabris et al. 2010] or in continental volcanoes [Baiocchi et al. 2007].

Conclusion
The ultra-high resolution multibeam swath-bathymetry at 0.1-0.5 m performed in the shallow waters surrounding the coast of Lipari Island, allowed to complete the already existent DTMs for this area. Data revealed the presence of morphological structures which represent a potential marine geohazard for this active volcanic island. Particularly, the possible interaction between tectonic structures, volcanic activity and retrogressive erosive processes can threaten the stability of the submarine flanks of the Lipari edifice, with consequent natural hazards. Therefore, due to the presence of industrial installations and touristic activities, this area deserves the same monitoring and hazard assessment effort of any active volcano-tectonic region within an urbanized area.