Somma Vesuvius volcano : ground deformations from CGPS observations ( 2001-2012 )

This paper is a contribution to the evaluation of ground deformations at Somma-Vesuvius volcano by means GPS measurements from 2001 to 2012. In this study we use a dataset from nine continuous GPS stations of the Neapolitan Volcanoes Continuous GPS network (NeVoCGPS), which covers the Neapolitan volcanic area, and is operated by the Istituto Nazionale di Geofisica e Vulcanologia. The GPS data processing is performed by the Bernese software v. 5.0. The results of the data processing show that the dynamics of the Somma-Vesuvio volcano, between 2001 and 2012, is characterized by a general subsidence, with maximum values on the Gran Cono at BKNO (−11.7 ± 0.65 mm/year) and BKE1 (−4.92 ± 0.36 mm/year) stations. The subsidence decrease from the crater down to the coast and the horizontal displacements are concentrated in Gran Cono area, the youngest part of the volcano. The parameters of the principal strain components indicate that Somma-Vesuvius is affected by a predominant contraction phase, which is concentrated in the areas with the greatest altitudes.


Introduction
The Somma-Vesuvius is a strato-volcano located 15 km east of Naples (Figure 1).It is about 1,200 meters high and covers a surface area of 165 km 2 [Principe et al. 1987].About 2 million people live in the surroundings of the volcano: hence the Somma-Vesuvius may be considered a relatively small volcano with an associated high level of risk [Auger et al. 2001].This is the main reason to demand for the improvement in knowledge regarding volcano structures and dynamics in recent years [e.g.Zollo et al. 1996a, 1996b, 1998, Bruno and Rapolla 1999, Cubellis et al. 2001, Del Pezzo et al. 2006, Berrino and Camacho 2008].In particular, TO-MOVES [Gasparini et al. 1998] and BROADVES [De Gori et al. 2001] projects have given a strong stimulus towards structural studies, considerably changing our ideas about the internal structure of volcano [De Natale et al. 2006].In summary, the results of TOMOVES project has shown the following evidences: a) a highvelocity zone which extends vertically from about 400 m below the crater down to at least 3000 m and a shallow 300-500 m thick low-velocity layer which covers the edifice [Zollo et al. 1996a[Zollo et al. , 1996b]]; b) a detailed image of the top of Mesozoic carbonate rocks forming the basement of the volcanic area.[Zollo et al. 2002]; c) an extended low-velocity layer at about 8-kilometer depth, interpreted as a sill with magma interspersed in a solid matrix [Auger et al. 2001, Capuano et al. 2003].Moreover, the seismic tomography at a regional scale (BROADVES project) also put in evidence a low velocity anomaly under Somma Vesuvius volcano, between 15 and 35 km of depth [De Gori et al. 2001, De Natale et al. 2006].
Since the last eruption in 1944, the Somma-Vesuvius system has been in a quiescent phase.Its current activity is characterized by hundreds of earthquakes per year with low energy, low temperature fumaroles at the summit crater and subsidence.The largest earthquake (MD = 3.6) was recorded on October 1999 [Del Pezzo et al. 2004, De Natale et al. 2004, Galluzzo et al. 2004, Cubellis et al. 2007].
De Natale et al. [2000] have interpreted the seismicity in this quiescent period as mainly due to the gravitational loading of the volcanic edifice, coupled to a particular substructure in which central conduits, filled by quenched magma, are characterized by a marked rigidity contrast, focusing gravitational and any local and regional stresses.The gravitational loading (not only on the volcano) has been postulated by many authors [Prandtl 1921, Hencky 1923, Terada 1929, Borgia et al. 1992, Mc Guire et al. 1997], including flank instability and volcano spreading [Borgia et al. 2000a, 2000b, Lanari et al. 2002, Borgia et al. 2005, Guglielmino et al. 2011].
In this paper, we evaluate ground deformations at Somma-Vesuvius volcano, from 2001 until 2012, by

Article history
Received November 2012; accepted June 2013.
Special Issue: Vesuvius monitoring and knowledge using GPS data.From a ground deformation viewpoint, we show that the dynamics of volcano is characterized by diffuse subsidence and by planimetric movements concentrated only in the high elevation zone of Somma-Vesuvius.

Somma-Vesuvius GPS network and collected data
The first discontinuous GPS measurements at Somma-Vesuvius were carried out in 1992 [Pingue et al. 1998], but continuous observations began only four years later.Indeed, the first permanent GPS station was installed at Somma-Vesuvius in February 1996 and it recorded GPS data just for one week [Esposito et al. 2001].However, only since 2001 the CGPS network had a stable configuration and all stations have been equipped with choke ring antennas (TERZ since 2003).Today, the Somma-Vesuvio GPS network is a part of a continuous GPS network (Figure 1), called NeVoCGPS (Neapolitan Volcanoes Continuous GPS) aimed monitoring of the entire Neapolitan volcanic area.It consists of 29 permanent stations equipped with dual-frequency receivers and choke ring antennas.Every day, 24 hours of observed dual-frequency GPS data, with 30 seconds sampling interval and 15 degrees cut-off angle, are downloaded from each site to the Osservatorio Vesuviano (Istituto Nazionale di Geofisica e Vulcanologia, Naples section), by the Internet, Wi-Fi and commercial telephone lines, depending on the communications infrastructure available at each station.
The data quality must be quickly and accurately tested before processing and storage in a database.The Osservatorio Vesuviano uses an in-house automatic application (SETA) for a quality check.It makes use of freeware programs, e.g.TEQC software [Estey and Meertens 1999], to translate from the binary receiver format to the standard RINEX format, perform a quality check of the data and zip the data files in an archive.All the procedures for managing the remote stations, downloading raw data, creating RINEX files and data quality control take place automatically.
In this study, we use continuous GPS measurements from nine sites of the NeVoCGPS network.The oldest stations (ONPI, PRET, and SANA) have been active for eleven years and the newest (BKNO) for 16 months.Table 1 shows the time span of the data sets and the equipment used in each station.

GPS data processing and time series
Data processing is performed by the Bernese Processing Engine (BPE) of the Bernese GPS software v. 5.0 [Dach et al. 2007] in double difference mode.The elevation cut off is set at 15° and the IGS absolute phase center variations (APCVs) for the satellite and receiver antennas are applied.The data are automatically processed on a daily basis using the Ultra-rapid International GNSS Service (IGS) products [Dow et al. 2009].When the IGS final orbits and Earth orientation parameters (EOPs) become available, the data are reprocessed on a weekly basis.
Independent baselines are selected taking into account the criterion of maximum common observations [Dach et al. 2007].The ambiguity resolution is based on the Quasi Ionosphere Free (QIF) analysis.Daily coordinates of the stations are estimated, together with the troposphere, in the final ionosphere free L3 solution.The dry part of the troposphere is modeled using the dry-Niell a-priori model and estimating the troposphere zenith delay (TZD) parameters every hour at each site using the wet-Niell mapping function.The geodetic datum is realized by three no-net-translation conditions imposed on a set of six IGS08 reference stations (minimum constraint solution), which are included in the processing.Table 2 shows the difference between the SOMMA-VESUVIUS: GROUND DEFORMATIONS FROM CGPS (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012) [Williams 2008].Due to the short time series (only 1.5 years), the flicker noise for BKNO station is not calculated.
IGS08 coordinates and estimated value for reference stations.Residuals demonstrate the consistency of estimated coordinates with IGS08 at level of 2.3 mm in total RMS from Helmert transformation.Velocity differences are very small (below 0.5 mm/yr) for all stations.Daily and weekly GPS time series of the north, east and up components are calculated in the global IGS08 reference frame.The Neapolitan volcanic area is a part of the Eurasia plate, which has, in terms of GPS horizontal component, a velocity about 2.5 mm/yr to the NE.Therefore, to eliminate the global tectonic motion and highlight the volcanic deformation, the planimetric velocities of GPS permanent stations are calculated with respect to a local reference station (ENAV; Figure 1) located on the limestones of Sorrento Peninsula, outside the volcanic area, and with long time span of measurements and constant horizontal velocities.We subtract ENAV velocity (16.5 mm/yr and 21.3 mm/yr, in the north and east component respectively) from the GPS planimetric time series for every station [De Martino et al. 2011].
Afterwards, the time series are analyzed to determine noise properties, linear velocities and periodic signals using Maximum Likelihood Estimation (MLE).This technique allows simultaneous estimations of the noise properties structure together with the parameters of a time-dependent model of the data.The quantities estimated by MLE analysis are linear trend, offsets, annual and semi-annual periodic signals [Williams 2003].The MLE analysis is performed using the CATS software [Williams 2008], and all the parameters are estimated with white plus flicker error models [Langbein and Johnson 1997, Mao et al. 1999, Williams 2003, Amiri-Simkooei et al. 2007].Their uncertainties can thus be considered to be realistic estimates based on the analysis of the noise at each station [Avallone et al. 2010].The analysis results are listed in Table 3 and shown in Figure 2.
Ground deformations can be considered as the superposition of a background signal, which is equal at all GPS stations, plus a local signal.For background deformation we intend a stationary level of deformation in which every transient can be considered indicative of changes in the studied area.
A time-series is stationary at the nth order if its statistical moment of order n is time independent.In our case, a deformation transient can easily be recognized because it significantly perturbs the average value of a stationary signal [Bottiglieri et al. 2007].
The stationarity has been tested using a two-tailed t-test at significance level t 95% for constant length running windows: the mean in each time window has been evaluated and then all the possible couples of means have been compared to another.A signal is considered stationary when the 98% of the averages can be considered equal to a 95% significance level [Bottiglieri et al. 2007].
As a result we find that the UP components exhibiting a deformation trend are those from the stations BKE1, BKNO, ONPI, OSVE, PRET, SANA, and TERZ stations.Instead, regarding horizontal components, only the stations located in the maximum vertical displacements area (BKNO, BKE1, OSVE) and PRET station on Mt.Somma show significant planimetric movements.

Discussion and conclusion
The GPS data presented in this paper show that the dynamics of the Somma-Vesuvio volcano, between 2001 and 2012, is characterized by a general subsidence (Table 3), with maximum values on the Gran Cono (BKNO −11.7 ± 0.65 mm/year and BKE1 −4.92 ± 0.36 mm/year).Down to the volcano foot, from Gran Cono to Mt. Somma, the subsidence velocity decreases until reaching the minimum values on the coast, where the station at a lowest altitude (AGR1) has a velocity (−0.28 ± 0.24 mm/year) comparable to the error.This is confirmed by data from the tide gauge network [Tammaro et al. 2007, Obrizzo et al. 2009] and from the leveling network [Pingue et al. 2013].Comparison between the velocity of the leveling benchmarks close to GPS stations and UP velocity of GPS sites is shown in Figure 5 and Table 4.
It is evident from Figure 3, which shows altitude of GPS stations versus subsidence velocity, and from Figure 4, which show the load (t*g*h) with respect to the volcano topography, that the GPS stations with highest velocities are those located at higher altitudes, on the Gran Cono.This is to be expected, since the stations at higher altitude are those with higher potential energy, and show that the vertical movements are also affected by the topography [Meo et al. 2008].Lanari et al. [2002], from SAR and leveling data (1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000), show that the Somma-Vesuvius volcano is subject to a deformation process characterized by a continuous subsidence in the Vesuvius area (Gran Cono).However, this analysis is restricted to the vertical component and only in subsectors where the coherence of the interferometric signal is preserved (Gran Cono area) or, at lower altitudes, that are not covered by vegetation.On Somma-Vesuvius volcano, the terrestrial geodetic networks, such as leveling and GPS (NeVoCGPS) networks, cover areas that are not measured by the SAR.
If we consider the altitude versus modulus of horizontal velocity (Figure 6), we note that as the altitude increases, planimetric velocity increases.Horizontal velocity below 300 meters is less than 1 mm/yr, only the GPS stations above 600 m (Gran Cono area) have velocity > 1 mm/yr.
The horizontal velocities of the GPS stations located at Mt. Somma (Figure 7) seem unaffected by the presence of the volcano.Indeed, the values of horizontal velocities are comparable to the error and vary from (0.24 ± 0.22) mm/yr to (0.37 ± 0.26) mm/yr, except PRET, which has a velocity of (−0.67 ± 0.08) mm/yr.
We have calculated the parameters of the principal strain components starting from displacements evaluated in the longer period of time (over 10 years) estimated by the velocity of the seven permanent GPS stations and considered constant for the whole period.Only BKNO station is not considered, due to its too short operating time (about 1.5 years).The main strain parameters refer to the barycenters of the seven triangles in which the area covered by the GPS stations was divided.The results obtained (Figure 8) indicate, as already in the past [Bonasia and Pingue 1981] that Somma-Vesuvius is affected by a predominant contraction phase, albeit small (less than 10 ppm).This contraction phase is more marked (Figure 8) in the areas with the greatest altitudes, according to the phenomenon of diffuse and modest subsidence observed in the Gran Cono area [Bonasia and Pingue 1981,    Table 4. GPS velocities and average velocity of the leveling benchmarks closest to GPS station.Pingue and Esposito 1988, Pingue et al. 1998, Lanari et al. 2002, Pingue et al. 2013].
We argue that a key to understanding the static deformation in Gran Cono area is the joint effect of compaction, sliding, including flank instability, and gravitational stress.As already suggested by Lanari et al. [2002], the discontinuity between the younger Vesuvius and the older Mt.Somma structure is not so well defined, but it is likely inward dipping with a normal fault-like sliding.This occurrence is also favored by the regional extensional tectonic regime in the area [Montone 1995].Subsidence around the Gran Cono also is in agreement with the hypothesis that local seismicity, clustered below the crater itself, is mainly driven by gravitational stress [De Natale et al. 2000, 2006, D'Auria et al. 2013].
Instead, static deformation (mainly subsidence) at Mt. Somma area is due to the contact, at a greater depth, between volcanic edifice and the sedimentary sequence below the volcano.It may also be induced by normal fault-like slip with a low angle dip, occurring along the contact zones between the volcanic rocks and the carbonatic basement due to the joint effect of the tensional regional stress regime and the volcano loading [Lanari et al. 2002, De Natale et al. 2006].

Figure 1 .
Figure 1.NeVoCGPS network.Red triangles indicate GPS stations of the network.In the insert box, the nine stations used in this study are shown.

Figure 2 .
Figure 2. Weekly position time series of AGR1 GPS station.North and east components are relative to ENAV station.Red line represents the sum of linear trend, annual and semi-annual periodic signals estimated with the CATS software.

Figure 3 .
Figure 3. Altitude of GPS stations versus subsidence velocity.

Figure 4 .
Figure 4. Gravitational load (t*g*h) as function of the volcano topography.

Figure 5 .
Figure 5.Time series of leveling benchmarks closest to GPS stations and comparison between benchmark and GPS velocities.Red line show linear fit of leveling data.The offset between each couple of time series is arbitrary.The Y scale is linear and spacing tick is of 2 cm.

Figure 6 .
Figure 6.Altitude of GPS stations versus horizontal velocity.

Figure 7 .
Figure 7.The horizontal GPS velocity field (relative to ENAV) and the 95 per cent confidence ellipse.

Figure 8 .
Figure 8. Principal strain axes for each of the seven triangles in which Vesuvius area has been divided.In the insert the dilatation values (in ppm) are also shown.

Table 1 .
Characteristics of the CGPS stations used in this paper and of the collected data.Station TERZ was equipped with choke ring antenna in 2003 and due radiofrequency interference first observation of OSVE started in 2005.

Table 2 .
Residuals from Helmert transformation between IGS08 and solution for the reference stations.

Table 3 .
Results of the time series analysis for Somma-Vesuvio GPS permanent stations.The analysis and noise characterization are performed with CATS software