The high resolution bathymetric map of the exhalative area of Panarea (Aeolian Islands, Italy)

On November 3, 2002 a shallow submarine gas eruption occurred in an area of 2.3 km east of Panarea (Aeolian volcanic arc, Southern Thyrrenian Sea, Italy). The exhalative area, surrounded by the islets of Dattilo, Panarelli, Lisca Bianca, Bottaro and Lisca Nera, has been known since historical times for the hydrothermal activity related to the Panarea volcanic complex. Due to the exceptional characteristics of the phenomenon, different geological, geochemical, geophysical and studies were carried out in this still poorly known volcanic area. A particular effort was devoted to producing a high resolution bathymetric map that also aimed to estimate the amount and location of the active exhalative centers and their variations in space and time. Data were obtained by three RTK multibeam surveys performed between December 2002 and December 2003. Here we show and discuss the technical details of the bathymetric surveys, the bathymetric map at 0.5 m resolution, and the accurate location of the 606 main exhalative centres active during the 2002-2003 crisis. The bathymetric data and the maps show two prevailing principal NE-SW and NW-SE alignments that match the spatial distribution of the exhalation centres. The accurate positioning at submeter accuracy of the gas vents is useful in the monitoring activity and to study their temporal and spatial variability. Mailing address: Dr. Marco Anzidei, Istituto Nazionale di Geofisica e Vulcanologia, Via di Vigna Murata 605, 00143 Roma, Italy, e-mail: anzidei@ingv.it

Panarea is the smallest among the Aeolian Islands, representing the subaerial portion of a submarine stratovolcano about 2000 m high and 20 km wide (Gabbianelli et al., 1993;Gamberi et al., 1997).East of Panarea, the islets of Basiluzzo, Dattilo, Panarelli, Lisca Bianca, Bottaro, Lisca Nera and Le Formiche (hereafter referred as the 'Islets') form an archipelago that The high resolution bathymetric map of the exhalative area of Panarea (Aeolian Islands, Italy) emerges from the eastern flank of the volcano (fig. 1b).Panarea is made up of dacite to rhyolite lava domes, dated up to 149 ± 5 kyr (Calanchi et al., 1999), interbedded with pyroclastic deposits.The island is topped by the pyroclastic deposits of Punta del Torrione formation, dated between 42 kyr and 11 kyr, which also outcrop at Salina, Lipari, Filicudi, Vulcano and along the northern coast of Calabria and Sicily (Lucchi et al., 2003).
At the beginning of November 2002, during the eruptions of Etna and Stromboli volcanoes, a submarine gas eruption started in a shallow area up to 30 m water depth and 2.3 km 2 of surface, bordered by the Islets.The event occurred  suddenly, without significant seismicity (Saccarotti et al., 2003) and reached an intensity level never observed before during the last century (SGA, 1996).The gas output was estimated to be 10 9 l/day, two orders of magnitude higher than that previously measured (Caliro et al., 2004).The most active exhalation center (EC-1, fig.7b) was located close to the SW wall of the islet of Bottaro, with gas flowing vigorously up to the water surface.To provide the present day locations of the submarine gas Exhalation Centres (EC) with respect to previous studies (Gabbianelli et al., 1990;Anzidei, 2000) and a first high resolution morphobathymetric map of the exhalation centres, bathymetric surveys were planned and performed during the maximum activity.Surveys were carried out in the early days of December 2002, with partial repetitions in selected areas during July 2003 and December 2003.

Bathymetric surveys
A multibeam high resolution survey that covered an area of about 9 km 2 around the Islets was performed for the first time in December 8-12, 2002, using the Alsea boat of Coastal Consulting and Exploration Company (fig.1c), equipped with an ultra high resolution Reson Seabat 8125 multibeam (240 beams, 120°sector coverage, 455 kHz) (table I).
Before starting the surveys, a check of the health of the GPS/RTK data link was performed  1a), which belongs to the Tyrrhenian Geodetic Network (TyrGeoNet), whose 3D coordinates are known at a few mm level (Anzidei et al., 1995;Serpelloni et al., 2005), (table II).
The decimetric positioning of the vessel was computed by GPS technique in RTK mode.Real time coordinates were obtained by installing an Ashtech Aquarius reference station located on the GPS station PANA and transmit-  ting the differential corrections by a High Frequency link at 1 Hz rate to the mobile Aquarius receiver which was placed on board of the vessel.In addition to this, a Sg-Brown Meridian Surveyor gyrocompass, a Tss DM505 MRU and a Fugro Omnistar Differential GPS, were interfaced to the Reson PDS2000 Navigation software (www.reson.com)for data acquisition (multibeam and positioning), control, calibration and pre-processing.A SBE 37-SI Microcat CTD probe was located at the sonar head and interfaced to the sonar processor, so providing the real time speed of sound data for the beam forming, while the Navitronics SVP15 and a SeaBird CTD probes were also used for profiling the speed of sound along the water column (fig.3).Additional details on the instrumentation are reported in table I. Navigation routes were planned with the aim to achieve the full coverage of the sea bottom (fig.4), with at least 20-30% overlap of the nearby swaths.The Reson PDS2000 software was able to show in real time to the operator and the pilot the ongoing multibeam and Digital Terrain Model (DTM) and the positioning information that were used for guidance.At the beginning of each survey, a full set of multibeam calibration lines were acquired, on flat bottoms and steep targets at 30-70 m water depth.The roll, pitch and yaw correction angles were then used to correct the installation geometries.Calibration parameters were taken into account during data analysis to correct the observations.

Data analysis, bathymetric map and exhalation centres
Data analysis was performed by the Computer Aided Resource Information System-Hydrographic Information Processing System (CARIS-HIPS) Pro v. 5.2 software (CARIS, , specifically designed to process multibeam data under Windows NT ® and capable to manage images in a mosaic of the seafloor and produce raster and analytical maps.The data processing sequence was as follows: 1. System calibration and multibeam data reprocessing.
2. Data quality check: only 15% of the data were discarded due to a low signal to noise ratio.
3. Sea level correction using the tide gauge and RTK data (fig.5a,b).
4. High and medium frequency spike removal, keeping intact the signatures produced by the uprising gas bubbles from the seafloor (fig.6a-d).

Production of high accuracy Marine Digital Elevation Models (MDEM) (table III).
Through the repeated computations, the signal to noise ratio was greatly improved, allowing a precise discrimination of the significant signal and the subsequent identification and location of EC (table IV).
A total of 606 EC were detected (Appendix) and classified by intensity and significance level (fig.7b).The latter criteria were expressed by the number of soundings that identified each EC from multiple overlapping swaths (0-1-2 levels of table IV) and from their height above the seafloor (A-B-C levels of table IV).
The exhalation centres were further divided into three classes (table IV): A) High activity centres, with bubbles columns interesting the entire water column up to the sea surface (figs.6d and 8).
B) Medium activity centres, with bubble columns extending for 3 to 4 m from the bottom but not reaching the sea surface (fig.8).
C) Moderate activity centres, with bubble columns extending up to 2 m from the ground floor, not reaching the sea surface (fig.6a).
The classified EC were mapped over the MDEM, allowing a view of the active exhalation centres for the whole area during the time span 2002-2003 (fig. 7a), and at December 2002, during the maximum intensity of the crisis (fig.7b).Bathymetric surveys were partially repeated during July and December 2003 (fig.4), focusing on the most active area around Lisca Bianca, Bottaro and EC-1 (fig.7b).
To produce the MDEM at 0.5 m average pixel size, of 9 million of 3D punctual data in the UTM33-WGS84 coordinate system were used.
The MDEM for each survey data set (December 2002, July and December 2003), was analysed to reduce any positioning offset or error.The difference grids display pixels whose values are the difference of depth at two survey epochs.The residuals between the three MDEMs were estimated, as reported in table III.By the average standard deviations and excluding values >1 m, supposed to be related to the active exhalation centres, the statistical difference between the surveys for all the investigated area was estimated within 14 cm.    the EC-1, we tentatively estimated that the explosive opening of the crater produced an extraction at 3852 m 3 of rock during the maximum of the crisis.One year after its opening, the crater was partially filled with approximately 221 m 3 of debris due to rock falls of the inner flank of the crater the upwelling gas and water flows.

Discussion and conclusions
The high resolution multibeam technique provided the first 3D detailed bathymetric map of the seafloor at 0.5 m average pixel resolution and the location and distribution of the exhalation centres, during the crisis that has affected the Panarea area since November 2002 (fig.10a,b).These data are useful to improve and support the geological, volcanological, geochemical, geophysical research and monitoring of this still poorly known volcanic area.Moreover, they represent the first accurate record of the last submarine exhalation crisis which occurred at Panarea.
The MDEM (fig.10a,b) shows that the area of Panarea Archipelago represents a positive geomorphic feature which is defined by the the coalescence of individual volcanic edifices.This is an asymmetric structure sloping at high angle in its southeastern flank while the others flanks display smoother slopes.The area within the Islets is characterised by a shallow seafloor platform, bewteen 0 and -30 m, gently slooping to NW between Panarelli and Secca dei Panarelli (fig.10a,b).Lisca Bianca, Bottaro and Lisca Nera islets are NE-trending coalescent structures.The Dattilo structure shows an elongated tongue-like marine abrasion platform, SE trending, for 0.8 km at depths between -5 and -8 m.
The area within the Islets is dotted by hundreds of circular or horseshoe-shaped depressions up to several metres deep and wide, largely distributed especially surrounding Dattilo, Lisca Nera and Bottaro.The high frequency relief of coarse textured morphology, at metre to decimetre size pinnacles and troughs, is well developed between Lisca Bianca, Bottaro and Secca dei Panarelli.In the area between Panarelli and Secca dei Panarelli the map reveals large lava flow surfaces, smooth and lobed.
The deepest sector dislays evident morphological structures such as those located in the northwest area, NE-NNE-NNW-and E-trending lineaments (fig.10a: A,B).The southwest sector displays a sub-elliptical shape wide depression probably due gravitative movements (C in fig.10a).
The shaded relief (fig.10a,b) does not show a caldera-like structure as suggested in Gabianelli et al. (1990) and Caliro et al. (2004).
The high quality of the surveys disclosed 606 exhalation centres (Apppendix), which were classified on the basis of their different size and activity level, as estimated from the interferences on the multibeam sensor.The location of the gas vents, mapped on the bathymetric maps, revealed the existence of preferential EC aligments NE-SW and NW-SE trending (fig.10a,b) and along which the gas eruption took place.These alignments match those measured from the fractures, the alunite veins and the fossil gas-pipes exposed on the lava of the seafloor and along the Islets' cliffs (Anzidei et al., 2003(Anzidei et al., , 2004a,b;,b;Esposito et al., 2005).Therefore we suggest that the NE-SW and NW-SE systems are the main pathways for the upwelling of hydrothermal fluids.
The comparison between the gas bubble alignments and those inferred by morphologycal analysis, suggests that presently the gas path is mainly driven by an active extension along the NE-and NW-trending fracture systems.The trends of the fracture systems are in agreement with the main regional tectonic structures (De Astis et al., 2003).
The maximum number of exhalation centres was located in a limited extended zone, west of Bottaro and Lisca Bianca islets, whereas clusters of minor centres were sparse in the area.The largest and most active exhalation centre EC-1 was located a few tenths of metres south of Bottaro (figs. 7b and 11).The ellipsoidal crater rim spans between -8 m and -15 m.Its main axis, NW-oriented, is 40 m long and the minor axis is 25 m long.The hundreds of depression features identified on the seafloor (fig.10a,b) can be related to fossil exhalation cen-tres suggesting that the area has been the site of past episodes of gas eruptions similar to that of November 2002 (Anzidei et al., 2003(Anzidei et al., , 2004 a,b; a,b;Esposito et al., 2005).
The differential MDEM obtained from multibeam data in the time span December 2002-December 2003, estimated 3852 m 3 of extracted rock during the gas explosion; 221 m 3 were recovered after 12 months, due to a continuous but decreasing gas flux from the seafloor (fig.12).As far as the tectonic and vol-canic interpretations are not the object of this paper, which aims to shows bathymetric survey data and results applied to the submarine part of Panarea volcano, we think that a change in the regional and local strain fields in this area may have produced a reactivation of previous fractures or the opening of new ones, causing a subsequent gas exhalation during the 2002-2003 Panarea crisis, as more extensively described in Caliro et al. (2004), Caracausi et al. (2005) and Esposito et al. (2005).

Appendix.
Coordinates of the 606 exhalation centres (UTM 33 coordinates).Their depth reported to the mean sea level during surveys, classification and epoch of detection, are also reported.

Fig
Fig.1a-c.a) Regional setting of Panarea Island; b) Panarea Archipelago; c) the ALSEA vessel used during bathymetric surveys.In the map are also displayed the locations of the tide gauge, the TyrGeoNet GPS station and the area investigated by the bathymetric surveys.

Fig. 2 .
Fig. 2. Sketch (not in scale) of the Panarea Pier with the location of the tide gauge and the geodetic benchmark used to define its coordinates with respect to the TyrGeoNet GPS station of Panarea.

Fig. 4 .
Fig. 4. Course over the ground performed by ALSEA vessel during bathymetric surveys.The 30% overlapping between nearby lines guarantee the full coverage of the area.

Fig. 3 .
Fig. 3. Sound speed profile.Data were used to calibrate the multibeam system for sound speed velocity in the water to determine depths at 1 cm formal accuracy.

Fig
Fig. 5a,b.a) Example of tide gauge recordings collected at 5 min sampling rate at Panarea Pier in the time span December 19-21, 2003 during bathymetric surveys.Tide amplitudes are <0.4 m, typical of the Central Mediterranean Sea.b) Comparison between tide data recorded by the tide gauge located at Panarea (grey line) and those registered by the GPS/RTK system located on board of the vessel (black line) during bathymetric surveys.The GPS/RTK tide data have been collected at 1 min sampling rate but averaged at 10 min to remove the high frequency noise.The two data sets are in good agreement.Tidal values were used to correct depth measurements, that were reduced to a mean sea level.

Fig.
Fig.6a-d.a-c) Multibeam images of some typical exhalation centres (vertical exaggeration 4X), showing uprising gas bubbles from the seafloor and the related small size craters produced by the exhalation activity; d) underwater picture of exhalation centres west of Bottaro Island.Gas bubbles are reaching the sea surface (0A class).

Fig. 7a ,Fig. 8 .b
Fig. 7a,b.a) Bathymetry of Panarea Archipelago at 2.0 m countour level.The red dots are the A and B classes of exhalation centres, respectively, detected in the time span December 2002-December 2003.Blue lines are the cross sections aal and bbl; b) location of the main exhalation centres active during December 2002, based on their classification (Appendix).In the map are shown the EC-1 (latitude 383818.8N;longitude 150634.0E,depth 13.3 m), EC-8 (latitude 383819.9N,longitude 150637.1E,depth 8.7 m), which were the largest exhalation centres and the EC named black smoke (latitude 383815.1N,longitude 150617.4E,depth 20.8 m).The latter is a 175 m wide and 5 m deep subcircular depression, interested by moderate dark exhalation, surrounded by lava outcrops that show deposition of alunite, sulfur and sulfates.

Fig. 11 .
Fig. 11.Digital Terrain Model of the EC-1.The cross section profiles obtained by multibeam data displayed a crater about 40 m long, 15 m wide and 9 m deep.The spikes located in the deepest part of the crater are the uprising gas bubbles.

Fig. 12 .
Fig. 12. Topographic changes of the groundfloor at the EC-1, after one year from its opening.The differential MDEM obtained from multibeam data in the time span December 2002-December 2003, estimated 3852 m 3 of extracted rock during the gas explosion; 221 m 3 were recovered after 12 months, due to a continuous but decreasing gas flux from the seafloor.Positive values are negative groundfloor changes (blue) due to slides inside the crater; negative values (red) are positive groundfloor changes, due to the progressive filling of the bottom of the crater.

Table I .
Instrumentation used during the bathymetric surveys.

Table III .
Accuracy of the MDEMs.

Table IV .
Classification of the exhalation centres.