Structural assessment of Mount Etna volcano from Permanent Scatterers analysis · 2018-06-08 · 1....

Article

Volume 12, Number 2

9 February 2011

Q02002, doi:10.1029/2010GC003213

ISSN: 1525‐2027

Structural assessment of Mount Etna volcanofrom Permanent Scatterers analysis

Alessandro Bonforte, Francesco Guglielmino, and Mauro ColtelliSezione di Catania, Istituto Nazionale di Geofisica e Vulcanologia, Piazza Roma 2, I‐95123 Catania,Italy ([email protected])

Alessandro FerrettiTele‐Rilevamento Europa, Via Vittoria Colonna 7, I‐20149 Milan, Italy

Giuseppe PuglisiSezione di Catania, Istituto Nazionale di Geofisica e Vulcanologia, Piazza Roma 2, I‐95123 Catania,Italy

[1] A study of the deformation pattern of Mount Etna volcano based on the results from the PermanentScatterers (PS) technique is reported. Ground motion data provided by the interferometric synthetic aper-ture radar (InSAR) PS technique from 1995 to 2000 are compared and validated by GPS data. An analysisof the ascending and descending line of sight (LOS) components of ground velocities has yielded detailedground deformation maps and cross sections. This analysis allows detection and constraint of discontinu-ities in the surface velocity field. LOS velocities have then been combined to calculate the vertical and hor-izontal (E‐W) ground velocities. A wide inflation of the edifice has been detected on the western andnorthern flanks (over an area of about 350 km2). A seaward motion of the eastern and southern flankshas also been measured. PS data allows the geometry and kinematics of the several blocks composingthe unstable flanks to be defined even in the highly urbanized areas, and their displacement rates have beenmeasured with millimeter precision. This analysis reveals the extension of some features beyond their fieldevidences and defines new important features. The results of this work depict a new comprehensive kine-matic model of the volcano highlighting the gravitational reorganization of the unbuttressed volcanic pileon its slippery clay basement on the southern flank, but an additional drag force due to a strong subsidenceof the continental margin facing the Etna volcano is necessary to explain the PS velocity field observed onthe eastern flank.

Components: 10,900 words, 6 figures.

Keywords: volcano tectonics; PSInSAR; remote sensing; flank instability; GPS; ground deformation.

Index Terms: 8485 Volcanology: Remote sensing of volcanoes (4337); 1240 Geodesy and Gravity: Satellite geodesy:results (6929, 7215, 7230, 7240); 8494 Volcanology: Instruments and techniques.

Received 5 May 2010; Revised 2 December 2010; Accepted 29 December 2010; Published 9 February 2011.

Bonforte, A., F. Guglielmino, M. Coltelli, A. Ferretti, and G. Puglisi (2011), Structural assessment of Mount Etna volcanofrom Permanent Scatterers analysis, Geochem. Geophys. Geosyst., 12, Q02002, doi:10.1029/2010GC003213.

Copyright 2011 by the American Geophysical Union 1 of 19

http://dx.doi.org/10.1029/2010GC003213

1. Introduction

[2] The improvements in geodetic techniques andtheir increasing use for studying and monitoring thedynamic of Mount Etna during the second half ofthe 1990s have provided a large amount of data forinvestigating the evolution of the deformationpattern of this volcano, inferring its magmatic andtectonic sources and, in a broad sense, the dynamicsof this volcano [Nunnari and Puglisi, 1994; Frogeret al., 2001; Puglisi et al., 2001; Bonforte andPuglisi, 2003, 2006; Houlié et al., 2006]. The useof SAR techniques and the extended use of GPSsurveys represent the main improvements in thesurveys techniques of this stimulating period forvolcano geodesy [Massonnet et al., 1995; Brioleet al., 1997; Lanari et al., 1998; Beauducel et al.,2000; Bonaccorso et al., 2006; Bonforte et al.,2007a, 2008; Puglisi et al., 2008]. Due to thesefavorable conditions, Mount Etna was used as a testsite for the application of the Permanent ScattererSAR technique (PS). The PS technique enables oneto overcome most of the limitations of classic Dif-ferential SAR Interferometry (DInSAR), such as theinability to define themovements of isolated coherentscatterers and to completely use all interferometricpairs, even those with very large baselines. More-over, the PS technique allows estimation and removalof the atmospheric phase delay that may cause mis-interpretation of the volcano deformation.

[3] In this paper, a detailed map of the grounddeformation ofMount Etna has been produced usingthe PS technique, which was applied to repeatedERS images taken from ascending and descendingsatellite passes. Particular care has been devotedto the estimation of the atmospheric profiles even onareas where just a few PSs were identified, thusobtaining a reliable ground motion estimation alongdifferent lines of sight (LOS). After an accuratecoregistration of the two ascending and descendingPS data sets, the E‐Wand the vertical components ofthe PS velocity could be determined.

[4] Thanks to the absence of flank eruptions from1993 to 2001, which resulted in an absence of localstrong anelastic deformations, the studies of MountEtna through this period focused on the dynamicsrelated to long‐living phenomena as produced eitherby deep volcanic or tectonic sources. The previousstudies based on geodetic and geological datahighlighted that deformation of the eastern andsoutheastern flanks of Mount Etna are quite dif-ferent to the radial pattern that one would expect ina central volcano like Mount Etna [see Bonforteet al., 2008, and references therein]. All data sets

considered here confirm that the movements onthese flanks are larger than expected from simplemagmatic sources and increase at lower altitudes.

2. Geological Setting and GeodeticMonitoring

[5] Mount Etna is a Quaternary polygenetic vol-cano located on the east coast of Sicily, rising about3300 m above sea level (Figure 1). The volcano lieson the Appennine‐Maghrebian chain to the northand on the Catania‐Gela foredeep to its southern-most part; it is characterized by a complex geody-namic framework characterized by a compressiveregional tectonics, active in northern Sicily along aca. N‐S trend and an extensional regime roughlytrending E‐W, affecting the eastern coast of Sicily[Bousquet and Lanzafame, 2004]. During the past400 years, the volcano has produced over sixty flankeruptions [Branca and Del Carlo, 2004], while thesummit craters were almost continuously active,with persistent degassing and frequent explosiveactivity at the vents on the crater floor. In particu-lar, the 1991–1993 flank eruption [Calvari et al.,1994] represented the most important flank erup-tion in the last 300 years, both in terms of duration(472 days) and volume of lava erupted (about 235 ×106 m3). After the end of this eruption, the volcanoshowed a gently degassing activity, then an almostcontinuous summit activity, increasing in intensityfrom July 1995 to June 2001. Between 1993 and2001, amagma recharge phase has beenwell imagedby ground deformation, that showed an almostcontinuous expansion of the volcano [e.g., Bonforteet al., 2008, and references therein]. Furthermore, animportant contrast in deformation patterns char-acterizes the eastern flank which shows a fairlycontinuous seaward motion, due to the interrela-tionship between gravity instability and magmaintrusion [e.g., Borgia et al., 1992; Rust and Neri,1996; Bonforte and Puglisi, 2003, 2006].

[6] Ground deformation monitoring began onMountEtna with installation of the first trilateration net-works, dedicated to measuring ground distancesvariations using Electronic Distance Meters (EDM)and theodolites during the 70s. Since then, earlyEDM networks were gradually extended and im-plemented in order to cover the most active areas[see Puglisi et al., 2004, and references therein].Ground deformation monitoring by GPS began onMount Etna in 1988, with annual GPS surveys andmore frequent observations during periods of greatervolcanic activity. The periodically measured dense

GeochemistryGeophysicsGeosystems G3G3 BONFORTE ET AL.: STRUCTURAL ASSESSMENT OF MOUNT ETNA 10.1029/2010GC003213

2 of 19

GPS network is continuously implemented andcurrently consists of more than seventy benchmarks.The Permanent Scatterers technique introduced inthis paper represents a further improvement in thegeodetic studies on Mount Etna, allowing the defi-nition of detailed ground deformation maps, typicalof the SAR‐based images, with millimeter precision,typical of the classical geodetic approaches.

3. Permanent Scatterer Methodology

[7] The Permanent Scatterers (PS) technique is analgorithm developed for processing data acquired

by space‐based Synthetic Aperture Radar (SAR);it represents a substantial improvement to classicdifferential interferometry (DInSAR) approaches[Curlander and McDonough, 1991]. The DInSARapproach is affected by several additional factorsthat add noise to the interferometric phase, such astemporal and geometrical decorrelation problems,topographic effects and orbit errors, as well asatmospheric artifacts [Zebker and Villasenor, 1992;Massonnet and Feigl, 1995; Zebker et al., 1997;Bonforte et al., 2001; Onn and Zebker, 2006].The PS approach [Ferretti et al., 2000, 2001;Colesanti et al., 2003], allows the main limitations

Figure 1. Structural sketch of Mount Etna reporting main faults known from geological surveys, villages, and theGPS stations and velocities used for PS data validation. RNE, NE rift; PF, Pernicana fault; RPN, Ripe della Nacafaults; SVF, S. Venerina fault; TFS, Timpe fault system; ACF, Acicatena fault; MTF, Mascalucia‐Tremestieri fault;TCF, Trecastagni fault; RF, Ragalna fault. Bottom left and upper right corners coordinates are in WGS84 Datum, bothin latitude and longitude degrees and in km for UTM projection, zone 33 north.


3 of 19

of traditional SAR interferometric technique to beovercome. PS analysis separates interferometricphase contributions on single selected targets, iso-lating the ground deformation component. Grounddeformation can be accurately monitored by iden-tifying permanent scatterers on the ground, whichare affected by very low decorrelation and then thatare active throughout the time span that satelliteobservations are available. Measurement pointschosen to be permanent scatterers were selectedbased on the values of the amplitude stability index,as well as the mean value of the coherence of thedifferential interferograms. In this way, each scat-terer is comparable to a natural geodetic benchmark,on which we can measure its motion, and the entireradar image can be considered a very dense networkthereby giving us a highly detailed time series ofdisplacements along the line of sight (LOS).

[8] All the available ERS1‐ERS2 images acquiredon Mount Etna from 1995 to 2000 were used in thiswork; in particular, we used both the ascending anddescending data sets. The descending data setconsists of 66 images acquired on the Track 222frame 2853 and spans from 25 April 1995 to 20December 2000. The ascending data set consists of38 images acquired on the Track 129 frame 747,spanning from 18 April 1995 to 4 October 2000.

[9] All interferograms have been generated using asingle master acquisition; no threshold has beenapplied either on temporal baseline or on normalbaseline values. For the ascending data set weselected as the master the image taken on 15 April1998; the image taken on 15 September 1999 wasselected as the master for the descending data set.Each data set provides a dense coverage of surfacevelocity observations along the line‐of‐sight direc-tion of the satellite. The SRTM (Shuttle RadarTopography Mission) [Farr et al., 2007] digitalelevation model was used for a first estimation ofthe local topography, although the precise elevationof each PS has been performed according to thePS technique [Ferretti et al., 2000, 2001]. Finally,it is important to point out that no prior informationwas used in InSAR data processing.

4. PS‐GPS Validation

[10] In order to validate the PS ascending anddescending velocities, we considered the deforma-tion measured at 35 GPS stations during surveyscarried out from July 1995 to July 2001. For eachGPS station, the cumulative 3‐D displacementswere calculated from all available measurements

through the entire period investigated by the PStechnique (Figure 1). Then the LOS components ofthe cumulative displacements were calculated forboth ascending and descending geometries, and themean LOS velocities, with their associated errors,were estimated by a linear regression. The resultingLOS velocities of GPS stations were finally com-pared with those of their nearest pixels, assuming acircular buffer, centered on the GPS stations, havinga radius of 100 m. The results of such comparisonfor both ascending and descending geometries arereported in Figure 2. The fit is good for theascending PS data set (Figure 2a); the correlation is0.7 (Figure 2c), the mean difference is about 6 mm/yrbetween PS and GPS velocities and the standarddeviation is 5.3 mm/yr. The fit between GPS and PSdescending velocities is a little bit worse (Figure 2b)with a correlation 0.4 (Figure 2d), a mean differ-ence of about 7 mm/yr and a standard deviation of6.8 mm. Considering that the errors of GPS posi-tioning, for the considered surveys, are in the orderof 3–4 mm, for the horizontal components, andabout twice that for the vertical component [Puglisiand Bonforte, 2004], we can affirm that the rela-tively large differences in comparing GPS and PSvelocities are mainly due to the uncertainties ofGPS measurements.

[11] Since there are no technical reasons to justifythe discrepancy between the ascending and des-cending comparisons (indeed, the original GPS dataare the same for both data sets, as well as the PSalgorithm), the highest misfit between the GPS andthe descending PS data set should be attributableonly to some characteristics of the deformationpattern of the volcano with respect to the SARlooking geometries. At the first order, the grounddeformations at Mount Etna are characterized bytwo patterns, as mentioned above: the deformationoriginated by the “deep” volcanic sources, mainlyon the western flank, and the subsidence andeastward motion of the eastern and southern flanks(Figure 1) [see Puglisi et al., 2004, and referencestherein]. As many recent studies suggest [e.g.,Bonforte et al., 2008, and references therein], thedeformation on thewestern flank is mainly producedby elongated pressure sources. In a such conditions,both vertical (uplift) and horizontal (westward)components of motion on the western flank pro-duce a strong shortening of the LOS distance forascending geometry (Figure 3a–3c), while the twocomponents act in opposite ways for the descend-ing geometry, resulting in lower LOS distancevariations when compared to the ascending data set(Figure 3b–3d). Similar reasoning explains the low


4 of 19

signal for the descending geometry we observe onthe eastern and southern flanks. Thus, the originof the differences in the comparison betweenascending and descending PS data and GPS is thepeculiar geometry of the SAR system with respectto the deformation pattern of Mount Etna volcanoduring the 1993–2001 time interval, which pro-duces an unfavorable signal‐to‐noise ratio in thedescending data set with respect to the ascendingone (Figure 3e).

5. PS Data Analysis and Description

[12] LOS velocity for each pixel has been codedaccording to a color scale and projected on aMount Etna map, in order to draw the ground LOSvelocity patterns for ascending and descendingviews (Figure 3). Color scales are based on theNatural Break algorithm described by Jenks andCaspall [1971]. This algorithm is widely usedwithin GIS packages and performs well especiallywhere large changes in value distributions occur. In

sections 5.1–5.3, the ascending and descendingPS data are analyzed in order to characterize theground deformation pattern.

5.1. Ascending View[13] The ascending data set (Figure 3a) detectedthe maximum approaching velocity on the uppernorthwestern flank of the volcano. Since the sensorlooks from west to east, with an incidence angle ofabout 23° (from zenith), this motion means that theground surface is rising and/or moving westward,in good agreement with the inflation detected forthe same period by GPS data [Puglisi and Bonforte,2004;Houlié et al., 2006]. The approaching velocityseems to decrease roughly uniformly in any direc-tion away from the maximum area, but no infor-mation is available on the summit area due to thepoor coherency of SAR interferograms in this area.Eastward of the summit area, the deformationpattern changes. The eastern flank, indeed, showsa diffuse movement away from the sensor. In thisflank there is no localized maximum, but the

Figure 2. PS‐GPS data validation. GPS station velocities (mm/yr) have been calculated and projected along(a) ascending and (b) descending views. These velocities are then compared to those measured by PS on the nearestpixel of each GPS station from both ascending (Figure 2a) and descending (Figure 2b) views. The fitness values for(c) ascending and (d) descending geometries are shown. Errors relevant to PS velocities are not visible since they areabout 0.6 mm/yr for the ascending data set and about 0.7 mm/yr for the descending one.


5 of 19

highest velocities are recorded in the Valle delBove area and diffusely along the coastline, with amaximum in its central part, decreasing towardhigher flanks of the volcano. This kind of defor-

mation pattern indicates that the ground is subsid-ing and/or moving eastward, as already describedby Bonforte and Puglisi [2003, 2006] and Puglisiand Bonforte [2004]. On its north side, the moving

Figure 3. (a) Ascending and (b) descending ground velocity maps (mm/yr); dashed lines represent the position ofthe cross section. Sketch showing the effects of ground motion on the ascending (c) and (d) descending geometries.(e) E‐W cross section of ascending (red) and descending (blue) measured LOS ground velocities (mm/yr).


6 of 19

area on the eastern flank is bounded by the Perni-cana fault, where this motion abruptly disappears.This pattern is clearly visible only at lower altitudes;at greater elevations the vegetation cover reducesthe coherency andmakes it difficult to define exactlythe limit of the moving area. On the southern part thevelocity decreases stepwise toward the south, acrossa few discontinuities. On the southernmost partof the image, on the volcano’s periphery, the PShighlights an ENE‐WSW trending area where themovement approaches the sensor.

[14] The E‐W cross section (Figure 3e) demon-strates the opposite movement directions of the twosides of the volcano, separated by a transition zonecoincident with the summit craters area. This crosssection illustrates evidence for the more uniformbehavior of the western with respect to the easternflank.

5.2. Descending View[15] The descending data set (Figure 3b) depicts aground deformation pattern somewhat different tothe ascending one, as expected from this view,which is looking downward from east to west. Theabsolute velocities on the western side of the vol-cano are generally lower than those observed onthe ascending data set. At lower elevations theyare moving away from the sensor (then westward)with a maximum (negative) velocity localized atlower altitude than that detected on the ascendingdata set. The northern side and the southernmostmargin of the volcano, conversely, show a move-ment approaching the sensor as observed in theascending data set. The entire eastern and southernflanks of the volcano show movements approachingthe sensor (different from the ascending look) con-firming that these flanks are moving eastward;however, also in this case the absolute velocitiesdetected with the descending geometry are lowerthan those measured by the ascending one. Fur-thermore, in this data set, the maximum velocity islocalized in the southern part of the coastline.

[16] The E‐W cross section (Figure 3e) in this casealso shows evidence of a more uniform behaviorof the western flank of the volcano, even if all thevelocity range is reduced with respect to theascending data set. In fact, all velocities are confinedbetween −5 and +10 mm/yr for the descendingview, much lower than those of the ascending one,ranging from −20 to +15 mm/yr. In the descendingview, the inversion from negative to positivevelocities occurs west of the summit craters.

5.3. Main Features of the LOS VelocityPatterns[17] Both ascending and descending ground velocitymaps (Figure 4) show particular features (hereaftercalled PSF, Permanent Scatterers Features) in theLOS velocity distribution. Below we describe thedifferent features, considering only their effects onthe ground deformation pattern, leaving aside theanalysis of their kinematics and role in the dynamicof the volcano, which is discussed in section 6.

[18] A first discontinuity (PSF‐1 in Figure 4) marks,with a roughly E‐W direction along the NE part ofthe volcano, the abrupt decay of ground velocity,bounding the deforming area to the south from thestable part to the north. Southward, a series ofNW‐SE discontinuities on the velocity fields arevisible, either on the ascending and descendingdata set. Here, each discontinuity (PSFs 2 to 8 inFigure 4) separates a higher velocity side on theeast from a lower one on the west. With thisarrangement, LOS velocities are progressively, butnot gradually, reduced until reaching zero after thesouthernmost discontinuity (PSF‐8 in Figure 4),from the northern urban area of Catania up to thelower SSW flank of the volcano. On the south-western side of the volcano, an arrangement ofthree discontinuities (PSFs named 9, 9a and 9b inFigure 4) seems to isolate a triangular wedge aroundthe village of Biancavilla (hereafter the “Biancavillawedge”), characterized by a different deformationpattern with respect to the surrounding areas. Finally,a NE‐SWdiscontinuity (PSF‐10 in Figure 4) crossesthe entire western flank of the volcano, causing adecay of LOS velocity from north to south; this lastdiscontinuity is not so evident in the maps becauseof the smaller effect on the color scale, but itis easily distinguishable on the cross sections. Inaddition to these linear discontinuities, some localeffects are clearly visible, affecting some particularareas on the volcano’s flanks. Some of these localanomalies are clearly related to the thermal con-traction and clast repacking of recent lava flows,perfectly shaping their extension. They are visiblein the Valle del Bove area (PSF‐a in Figure 4)and on the upper southern and northeastern flanksof the volcano (PSF‐b and PSF‐c in Figure 4). Amore local and circular anomaly (PSF‐d in Figure4) has been detected on the lower northwesternside of Mount Etna and is probably related to waterpumping from a well at that location. A wider andV‐shaped anomaly (PSF‐e in Figure 4) affectsthe lower southeastern sector; this strong anomalyseems to be confined to the south‐west by the PSF‐6


7 of 19

Figure 4. Ground LOS velocity (mm/yr) maps and cross sections for ascending (red) and descending (blue) views.See text for details.


8 of 19

and PSF‐7, showing a more gradual decay towardthe north. On the southernmost periphery of theimaged area, an ENE‐WSW elongated zone (PSF‐fin Figure 4), from Misterbianco town to thenorthern part of the city of Catania (hereafter the“Misterbianco ridge”), shows a clearly positiveanomaly of the velocity pattern; this anomaly isabruptly truncated by the PSF‐8 on its northeasterntip, while it shows a gradual decay on both north-western and southeastern parts. It is not possible toinvestigate a possible southwestern prolongation ofthis anomaly due to the low coherency in the PSdata sets.

[19] In order to better identify the main PSFs inthe ground deformation pattern defined by theascending and descending data sets, several crosssections have also been analyzed (Figure 4). Dif-ferent azimuths and positions were chosen, in orderto cover the entire volcano, intersecting all PSFsin different positions and with different angles.Furthermore, cross sections could help identifyother minor PSFs, defining jumps in the velocityfield that are not strong enough to produce effectson the color scale used in the map.

[20] Cross section A‐A′ intersects all the PSFs onthe SE side of the volcano. No ground motion isdetected from the southernmost part until PSF‐8 iscrossed. From here toward the northeast, groundvelocity starts to increase for ascending view. Thevelocity increase is not gradual but abrupt varia-tions occur along the PSFs 7, 6, 4 and 3, each oneproducing a jump of 3–5 mm/yr. These jumpsclearly also affect the descending data set. Betweentwo subsequent discontinuities, LOS velocities tendto increase for descending view (approaching) andto decrease for ascending one (moving away),showing similar positive gradients on both datasets. The last jump occurs along the PSF‐2 thatleads to the maximum velocity of 20 mm/yr on theascending data; descending data suggest the pres-ence of more parallel discontinuities, confining anarrow high velocity strip, visible also in the map(Figure 4).

[21] Cross section B‐B′ crosses the entire southernflank of the volcano with an E‐W direction. On thewesternmost part, it shows a progressive and gradualvelocity increase on both data sets. PSF‐9 producesa strong drop in the velocity values, especially onthe ascending data set. From here toward the east, astepwise behavior, similar to that revealed by theA‐A′ cross section, can be observed, locally dis-turbed by the anomaly “b” (on the 1983 lava flow),until it meets PSF‐6. Here, negative gradient starts

on both data sets until PSF‐4. From PSF‐4 to 3 thevelocity gradient is positive on both data sets, whileeast of PSF‐3 the behavior on the two data sets isopposite and controlled by the structures groupedas PSF‐2, as seen also by cross section A‐A′.

[22] Cross section C‐C′‐C″ runs around the volcanofrom the northwestern to the southern lower flanks.It crosses the PSFs affecting the ground velocitypattern on the western and southwestern sidesof the volcano. The PSF‐10 evidences a negativejump of about 3 mm/yr on the ascending data set inthe cross section, while having a minor effect on thedescending one. Southward, both ascending anddescending velocities remain at constant values.The PSF‐9a produces a dramatic change of defor-mation on the descending data set with a jump ofabout 8mm/yr (frommoving away to approaching thesensor), while having minor effects on the ascendingone. From PSF‐9a to 9 across the Biancavilla wedge,velocities decrease to zero on both data sets. Fromhere toward the south‐eastward a similar patternaffects both ascending and descending data sets,clearly drawing the maximum values just northof Catania before decreasing again to zero towardthe coastline.

[23] Cross section D‐D′ evidences the abrupt end,on PSF‐8, of the positive anomaly on the south-ernmost periphery of Mount Etna. PSF‐7 producesa negative jump on both data sets, and the negativeanomaly PSF‐e is clear northward of this discon-tinuity. From here northward, LOS ground veloci-ties show a similar pattern on both data sets, showingother minor jumps related to PSF‐4 and PSF‐3 thatare intersected with low angles by this cross sec-tion. North of PSF‐3, anomalies PSF‐a and PSF‐care detected on 1991–1993 and 1971 lava flows,respectively. On the northernmost part of this pro-file, PSF‐1 clearly closes the moving area, abruptlyreducing both ascending and descending LOSvelocities to zero; in particular, ascending data showa sudden jump of about 10 mm/yr.

[24] The cross section E‐E′, similar to C‐C′‐C″,shows the uniform and gradual gradient of defor-mation on the western side of the volcano, disturbedonly by the very local PSF‐d anomaly, related to awater well near Bronte village. This smooth patternis only perturbed by the PSF‐10. A few kilometerssouthward, we observe a transitional area wherethe velocities change pattern until PSF‐9b, whichcauses an evident inversion of the velocity valueson the descending dataset, passing from negativeto positive velocities in a few hundred meters.From here south‐eastward, ascending and descend-


9 of 19

ing datasets show similar patterns. PSF‐6 is crossedwith a very low angle and doesn’t show the jump invelocities as seen in cross section A‐A′. Eastwardof the PSF‐e anomaly, velocities show an inversionof the gradient on both datasets, locally interruptedby PSF‐5, which causes a jump of more than5 mm/yr on the ascending LOS component. At theeastern end of profile, the ascending velocity showsa very strong positive gradient.

[25] Cross section F‐F′ traverses the entire easternflank of Mount Etna along the coastline. At thesouthern end of this profile, the ground velocitiesare close to zero, and slightly increase northward.PSF‐8 generates a little jump (about 2 mm/yr) inthe increasing trend of the ascending data set, whilethe PSF‐7 produces a more significant changein the ground motion pattern (a jump of about a5 mm/yr on the ascending LOS velocities and a8 mm/yr in the descending ones). Furthermore,from here northward, an opposite trend affects bothascending and descending data sets. Other PSFs areintersected with low angles by this cross section;nevertheless, jumps of 6 mm/yr at PSF‐5 and of8 mm/yr at PSF‐2 are still visible on the ascendingdata set. Between PSF‐2 and PSF‐1, ascendingLOS velocities show the highest values measured,which gradually increase approaching PSF‐1. Bothascending and descending LOS velocities return tovalues close to zero just north of PSF‐1, as observedon cross section D‐D′ at higher altitude but with asmoother decay.

6. Combination of Ascendingand Descending Data Sets

[26] The combination of the ascending and des-cending data sets allows discrimination between thevertical and horizontal (in E‐W direction) compo-nents of the movements. Considering that the twodata sets are obtained from two nearly oppositeazimuth angles, the horizontal (in E‐W direction)movement acts in contrasting ways for the twopoints of view, while the vertical one produces thesame effect on both data sets.

[27] The LOS distance variation is given by

dLOS ¼ UN sin�þ UE cos�ð Þ sin�þ UV cos� ð1Þ

Here UN, UE and UV are the N‐S, E‐W and verticalcomponents of ground motion, respectively; F isthe azimuth angle, which conventionally we mayassume positive for descending and negative forascending orbits and l is the incidence angle. At

Mount Etna, sinF ≈ 0 (0.19) and cosF ≈ ± 1 (±0.98);then

dLOS ffi �UE sin�þ UV cos� ð2Þ

where + is for descending passes and – for ascendingones.

[28] The mean value of the unit versor was used forthe estimation of horizontal and vertical compo-nents, since its variation within the area of interestcan be considered negligible. Variations of theincidence angle, in fact, are lower than 4 degrees.Considering a LOS displacement of 1 cm, themaximum error introduced by assuming the meanangle is always lower than 1 mm.

[29] From equation (2) is evident that it is impos-sible to discriminate the horizontal N‐S componentof ground motion that, indeed, has a negligiblecomponent along the LOS. North‐south displace-ment values, though present, cannot really com-promise the estimation of vertical and east‐westcomponents. In particular, east‐west displacementcomponents are almost unaffected by possiblenorth‐south motion. Even considering a purelyhorizontal displacement of 1 cm, in north‐southdirection, we would erroneously estimate a verticalcomponent of displacement of just 0.8 mm andalmost no east‐west displacement (<0.1 mm).

[30] From (2) it follows that

UE ffi 1

2

dLOS descð Þ � dLOS ascð Þsin�

ð3Þ

UV ffi 1

2

dLOS ascð Þ þ dLOS descð Þcos�

ð4Þ

The cost of this approach is that the number ofpixels we can use to estimate the two componentsof the movement is reduced with respect to theamount of pixels available in each data set. This isbecause it is necessary that each pixel has goodcoherency both on ascending and descending datasets.

[31] The result of the combination of ascending anddescending data sets is reported in Figure 5 (firstand second panels), where east‐west and verticalvelocities are plotted, respectively. The analysis ofthese maps allows description of the main char-acteristics of the movement components and,whenever possible, characterizing the PSFs tracedon ascending and descending LOS velocity datasets that isolate domains affected by uniform orcontinuous gradient of the velocities. To detail thekinematics of these discontinuities, the same cross


10 of 19

sections previously described are analyzed also toinspect the data sets for vertical and horizontalvelocities as reported in Figure 5.

6.1. East‐West Velocities[32] Figure 5 shows a diffuse westward movement(negative velocities) of the whole western flank ofthe volcano, with a maximum velocity located atmiddle elevation. The westward velocity progres-sively decreases in all directions outward fromthis maximum with a fairly uniform gradient. Thehorizontal velocities are positive (eastward) on theeastern and southern flanks and the highest veloci-ties are located along the coastline, between theeastward prolongation of the Pernicana fault, asdeduced by Bonforte and Puglisi [2006], and theAcicastello village (Figure 5). On the eastern flank,the decay of the eastward velocity is abruptacross the NE rift–Pernicana fault system (PSF‐1 inFigure 5), while southward the decay is stepwise.

6.2. Vertical Velocities[33] Figure 5 shows a wide uplift (positive values)on the upper part and western flank of the volcano,with a maximum upward velocity located on itsnorthwestern side. Similarly to the horizontal west-ward motion of this side of the volcano, the upliftgradually decays in all directions with a fairlyuniform gradient. Unlike the horizontal velocity,however, the entire upper part of the volcano isinvolved in the uplift, down to the middle slopes onthe southern flank. At lower altitudes, the easternand southeastern flanks are characterized by nega-tive, or slightly positive, values of vertical variation.Along the coast, we measured the highest subsi-dence rates in an area characterized by high eastwardvelocity; as in the case of the E‐W velocities, thesubsidence is bounded northward by the abruptdecay in association with the PSF‐1 while, unlikethe E‐W velocity, southward the highest subsidencestops near the location of Stazzo village (northwardPSF‐3; Figure 5). Inland, on the southeastern flank,we measured other subsiding areas at mediumelevations (300–800 m).

[34] The map of the vertical velocity highlights theobservation that the movement of the PSF‐f anom-aly is mainly uplift. The area with highest velocitybegins in the area of Misterbianco town, but someevidence is also visible farther westward around ofPaternò town, and further involves the northern andwestern suburbs of the town of Catania, with valuescomparable with those measured on the upper partof the volcano. Figure 5 shows how the movement

of the PSF‐d and PSF‐e anomalies is exclusivelyvertical, both showing subsidence.

[35] Other minor local vertical features are thesubsidence located on the upper southern flank andinside the Valle del Bove. The shapes of these localfeatures are irregular and coincident with the recentlava flows emplaced during the years before theperiod investigated by PS analysis (1983, 1985,1986, 1989, 1991–1993).

6.3. Main Features of the GroundDeformation Pattern[36] The analysis of the vertical and E‐W velocitymaps allows definition of the general pattern ofground deformation and of the kinematics acrossthe PSFs. The whole northwestern side of the vol-cano shows a general uplift and westward move-ments. A broader uplift has been detected on thesouthern flank with respect to the southwesternflank. The entire eastern flank shows an eastwardmotion accompanied by a wide subsidence. On thesouthern half of the seaward moving sector, E‐Wvelocities gradually increase from the upper to thelower flanks without abrupt changes, unlike in thevertical velocity field. On the southeastern part of thevolcano, vertical motion pattern is strongly shapedby theNW‐SE trending PSFs; this is very evident forPSFs 2, 4, 6 and 7 whose eastern sides always showstronger subsidencewith respect to the western ones.

[37] The same cross sections previously describedto investigate the LOS velocity data sets, are hereanalyzed in order to investigate the horizontal andvertical kinematics of the volcano’s flanks.

[38] Cross section A‐A′ confirms that the east-ward component of velocity progressively increasestoward the coast. The E‐W velocity rapidlyincreases north of PSF‐8, reaching 15 mm/yr north-east of PSF‐7; this velocity remains fairly constantover the entire SE flank, jumping to more than20 mm/yr northeast of PSF‐3. All PSFs show a ver-tical component of motion, with the general trend ofvertical velocity showing an increasing subsidencefrom SW to NE. Each PSF marks an abrupt increaseof downward motion on its NE side. On the contrary,after each jump, the subsidence tends locally todecrease from SW to NE, thus revealing a SW‐wardtilt of each block bounded by two PSFs. Northeastof PSF‐3, both the subsidence and the eastwardvelocities reach their maximum values.

[39] Cross section B‐B′ shows the stability of theBiancavilla wedge, which is affected only by aslight uplift on its upper part. East of PSF‐9, the


11 of 19

Figure 5. Ground velocity (mm/yr) maps and cross sections for vertical (red) and E‐W (blue) components. See textfor details.


12 of 19

eastward motion of the southern flank starts andgradually increases toward the east, reaching a stablevalue of 15 mm/yr from west of PSF‐4 to PSF‐3.From PSF‐3 eastward, a continuous increase ofeastward velocity is measured until the coastline (asin the A‐A′ cross section). The other PSFs pri-marily control the vertical motion distribution.

[40] From cross section C‐C′‐C″ the effect of theinflation on the western flank is evident with ageneral uplift and westward motion. The PSF‐10,located on a low coherent area, produces a smalljump of about 2 mm/yr on the vertical velocity,defining a zone characterized by minor upliftbetween PSF‐10 and PSF‐9a over the town ofAdrano. The inflation stops south of PSF‐9a, wherethe stability of the triangular Biancavilla wedge isconfirmed, although few data are available in thisarea. The absence of significant ground motioncontinues SE of the lower part of PSF‐9, untilcrossing the PSF‐f where the horizontal and verticalvelocities define it as an actively growing anticline,uplifting at a rate of about 10 mm/yr.

[41] The southern end of cross section D‐D′ startswith the uplift of the southern flank of the anticline,which is abruptly stopped by the PSF‐8. From herethe vertical velocities invert and decrease, passingacross the PSF‐7, until an absolute minimum isreached at the center of PSF‐e. Northward, the ver-tical motion shows a slight uplift until the northernend of the section, probably due to the effect of thegeneral inflation of the volcano. The eastwardmotion is instead affected by a gradual increasefrom the southern end, even across the PSF‐8,which stops at the PSF‐7, where we observe a jumpof about 10 mm/yr. From here to the north, theeastward motion shows a constant velocity at about15 mm/yr increasing, again north of PSF‐3 andreaching a maximum value of 25 mm/yr, constantfor more than 5 km south of PSF‐1. At PSF‐1, asudden drop of 20 mm/yr clearly defines the end ofthe eastward moving block.

[42] The inflation of the volcano is well expressedin section E‐E′ where the maximum westwardvelocity is located near the PSF‐10. From PSF‐9bwestward to PSF‐10 the velocity rapidly decreaseswhile vertical velocities are rather disturbed, defin-ing a narrow area characterized by a minor uplift(down to zero just west of PSF‐9b). This area isnarrower than that detected by the profile C‐C′‐C″and defines a NE‐SW elongated strip above thetown of Adrano, characterized by relative subsi-dence with respect to the surroundings, hereafternamed the “Adrano graben”. The area surrounding

the PSF‐9b closes the inflation by inverting thehorizontal motion from westward to eastward andproducing a jump on the uplift pattern. South of thePSF‐9b, positive vertical velocities resume, con-sistent with the wider uplift of the upper southernsector of the volcano. From PSF‐9b SE‐ward,vertical velocity shows a decreasing trend, reachinga maximum in subsidence east of PSF‐6 (anomalyPSF‐e) From this minimum to the coastline, thevertical behavior evidences the westward tilt of theblocks as revealed also by cross section A‐A′ athigher altitude. Once again, the highest velocitiesare measured along the coast.

[43] Along the coastline, cross section F‐F′ revealshow the PSF‐8 produces only a little increase ofabout 5 mm/yr of the eastward velocity, while themost significant jump is produced by PSF‐7, with asudden increase of eastward motion of almost15 mm/yr. North of PSF‐7, the E‐W velocity showsconstant values of about 15–20 mm/yr, which fur-ther increase north of PSF‐3, up to 25–30 mm/yr.Unlike cross section D‐D′, at this altitude the E‐Wvelocity starts to decrease a few km south of PSF‐1,producing a drop of 10 mm/yr. Vertical velocitypatterns also confirm the tilting of the individualblocks decoupled by subsequent PSFs. PSF‐2 showsa jump of about 10 mm/yr, revealing the highestrate of subsidence on the easternmost periphery ofthe volcano.

7. Discussion

[44] The ground motion maps resulting from the PSanalysis and combination reveal that the deformationofMount Etna is characterized by twomain domains.The first domain involves the entire western andnorthern flanks of the volcano and its summit area,showing roughly radial deformation and upliftpatterns (Figure 6). These deformation patterns con-firm the fairly continuous inflation alreadymeasuredby GPS surveys between the 1991–1993 and 2001eruptions and visible also fromGPS velocity vectorson the western and northern side of the volcano,reported in Figure 1. During this period, volcanicactivity evolved from quiescence, during the twoyears after the 1991–1993 eruption to the strombo-lian activity that began in late 1995 at the NEC,involving all summit craters in the following years,with continuously increasing intensity. Several pres-surizing sources at depths ranging from 3 to 9 kmand some shallower intrusions have been modeledby different geodetic techniques beneath the vol-cano by considering different time windows from1993 to 2001 [Puglisi and Bonforte, 2004; Bonforte


13 of 19

et al., 2008; Puglisi et al., 2008]. The long‐termground deformation pattern on this part of MountEtna, as measured by PS technique over the whole1995–2001 period, comprises the effects of all theabovementioned deep and shallow sources activebeneath Mount Etna during the inflating phase;the measured pattern is fairly spatially continuouson the northern and western sides of the volcano.Between the PSF‐10 and the PSF‐9a‐9b (see sec-tions C‐C′‐C″ and E‐E′ in Figures 4 and 5), wehighlighted a wide area where the velocity patternschange irregularly, although tectonic features cor-responding or similar either to PSF‐10 and/or thePSF‐9a‐9b are not recognizable here in the field.Mattia et al. [2007] studied the deformations mea-

sured by an EDM network located just north-ward of the PSF‐10 and proposed the presence ofa NE‐SW buried fault, which should producethe right‐lateral shear strain measured by the EDMsurveys. PS analysis does not confirm the existenceof such a fault and suggests instead that the strainedarea extends southward with respect to the EDMnetwork, maintaining the same NE‐SW trend.

[45] The second domain, well identified on the east-ern and southern flanks of the volcano, is character-ized by the general eastward and downward motion,as already measured by GPS surveys [Bonforte andPuglisi, 2006]. Velocity vectors reported in Figure 1highlight how the entire eastern flank is affected by

Figure 6. Main kinematic domains of Mount Etna volcano.


14 of 19

an ESE‐ward motion, at a rate of about 1 to 3 cm/yrduring the 1995–2001 period. The detailed PSanalysis performed in sections 5.3 and 6.3 showsthat this domain is divided into several blocksshowing slight different velocities (Figure 6), inagreement with the analysis of the dense GPS net-work installed here [Bonforte and Puglisi, 2006;Bonforte et al., 2008]. The PS velocity maps nowallow imaging the discontinuities (PSFs) in theground deformation field, where the relative motionof the blocks occurs, giving the opportunity todefine the complex structural framework affectingthe southern and southeastern lower flanks ofMountEtna. Most of the PSFs identified in section 5.3correspond to known faults. In a few cases, somePSFs are more extensive than fault segmentsidentified by classical geological surveys, whilesome other PSFs on the southern flank were neveridentified by previous field survey studies. Onlysome long‐period DInSAR interferograms havepreviously imaged some features affecting theground displacement pattern on this flank of thevolcano, such as the PSF‐6, 7, 8, part of the PSF‐3,and the anticline [Borgia et al., 2000; Froger et al.,2001].

[46] Below, we discuss the kinematic and the roleof the features identified in sections 5.3 and 6.3,in the framework of the dynamic of the volcano.On the northeastern side of the volcano, PSF‐1corresponds to the NE rift‐Pernicana fault system,which abruptly marks the northern border of theseaward‐moving area. The trace of this fault is wellestablished from previous geodetic and geologicalsurveys [Azzaro et al., 1998; Neri et al., 2004;Palano et al., 2006; Bonforte et al., 2007b]. PSanalysis allows to trace the path of the fault from itsconnection to the NE rift down to the coastline. Theposition of the fault at lower elevations correspondsto that deduced previously from GPS data [Bonforteand Puglisi, 2006] and subsequently confirmed bysurface fracturing during the 2002–2003 eruption[Neri et al., 2004]. The strong eastward motionabruptly disappears across the Pernicana fault, pro-ducing a left‐lateral transcurrent movement ofabout 25 mm/yr. The eastward motion also changesacross the NE rift, thus producing a significantextension that could promote magmatic drainagefrom the central conduit by partly passive intru-sions as reported by Bonforte et al. [2007c]. Anotherinteresting feature is the decay of the eastwardmovement from higher to lower altitudes as shownby comparing the cross sections D‐D′ and F‐F′(Figure 5).

[47] On the southern flank, PSF‐8 represents thesouthernmost border of the seaward moving sector.It does not correspond to any known geologicalstructure or fault visible on the field. Its effect is mostevident where it truncates the Misterbianco ridge.The PSF‐8, indeed, cuts and closes the growingMisterbianco ridge which upraises the northern partof Catania at a rate of about 5 mm/yr (Figure 5 andcross section D‐D′). Even if PSF‐8 represents thesouthernmost border of the mobile flank, however,the main decay of deformation occurs at PSF‐7.This discontinuity, indeed, shows both horizontaland vertical movements. With respect to the PSF‐8,this feature shows an increasing eastward move-ment from a few mm/yr on its westernmost part to15 mm/yr near the coastline (Figure 5 and crosssections A‐A′, B‐B′, D‐D′ and F‐F′). The verticalmovement progressively decreases from the west(4 mm/yr) to east (0 mm/yr).

[48] From the E‐W velocity map (Figure 5, firstpanel), PSF‐8 appears to continue northwestward,with minor evidence, toward PSF‐9, which coin-cides with the Ragalna fault. The PS data, however,allow us to improve the definition of the actualextent of the Ragalna fault which is indeed com-posed by a complex arrangement of at least threemain faults (PSF‐9, 9a and 9b) dissecting theSW flank of the volcano. The known Ragalna faultsegment [Neri et al., 2007], trending N‐S, coin-cides with the PSF‐9 and represents the south-western boundary of the eastward moving sector,separating a stable block on its western side fromthe mobile southern sector on its eastern side(Figure 5, cross section B‐B′). Westward of PSF‐9the Biancavilla wedge (Figure 6) is isolated alsofrom the inflating western flank of the volcano,representing a sort of locked triangular sector onthe southwestern periphery of the volcano, betweenthe inflating and sliding flanks of Mount Etna.

[49] In general, all the features discussed aboveproduce a structural arrangement of Mount Etnawith an SSE‐ward spreading sector on the southernside and a westward inflating one on the westernside. This geometry yield a very similar behavior tothat reproduced by analog models for spreadingedifices [Merle and Borgia, 1996]. On the northernand western sectors of Mount Etna, the higher ele-vation rocky substratum below the volcano pre-cludes or minimizes any radial spreading of theedifice, while the entire eastern and largely un-buttressed sector of the volcano is affected in ageneral seaward‐directed slide. On the southernand southwestern sides of the Etna edifice, nearly


15 of 19

flat lying sedimentary basement underlies the vol-cano at low elevation. The clayey sediments fromthe Catania foredeep may act as a low‐viscositylayer beneath the main part of the volcanic edificeand the lava flows that are on top of the sediments.The southern and southwestern flanks of MountEtna are not buttressed against solid bedrock likethe northern and western flanks and lie on top ofplastic clay‐rich sediments. These observations,coupled with our PS data, are consistent with thegeometric and kinematic conditions assumed byMerle and Borgia [1996] in their models of grav-itationally driven spreading of volcanic edifices. Insuch conditions the edifice shows a roughly radialspreading of the volcanic pile over the plastic sedi-mentary basement under the effect of gravity andprobably induced or, at least, favored by the inflationof the edifice. The concentric stretching due to thespreading, as stated by [Merle and Borgia, 1996],induces the formation of the radial Adrano grabendefined by PSFs 9a–9b and 10 and the triangularstable Biancavilla wedge (Figure 6). The conse-quent radial shortening on the periphery of thevolcano has dragged and pushed the underlyingsediments causing the uplift of the U‐shapedMisterbianco ridge (Figure 6), which can be inter-preted as a concentric ridge, as defined by Merleand Borgia [1996]. The active anticline follows apreexisting Plio‐Plesitocene E‐W anticline, andbounds the volcano roughly from Paternò to AciTrezza villages, where it is abruptly truncated bythe transcurrent movement along PSF‐8.

[50] The Misterbianco ridge (PSF‐f) also providesan indirect evidence of the southward movementsof the southern flank, which cannot be measured bySAR Interferometry. This kinematic arrangementon the southern slope confirms the hypothesis ofthe southward moving triangular slice on the uppersouthern flank ofMount Etna, proposed by Bonforteet al. [2009] from analysis of GPS surveys. Ac-cording to this interpretation, the anticline is rootedat shallow depth and its movement is more relatedto the gravitational loading of the flank of thevolcano rather than to the inflation of deep mag-matic sources. This interpretation is consistent withthe very low levels and shallow focii of seismicityon the southern flank of the volcano.

[51] PSF‐6 (Figure 5) follows the path of the Tre-castagni fault, but it extends beyond the mappedlimits of the fault, from lower altitude (about300 m) up to medium‐high flank (about 1500 m).At higher elevations, this feature corresponds to thearea affected by huge ground fracturing during the1989 eruption and significant ground deformation

during the 2001 eruption [Bonforte et al., 2004].This fault, like PSFs 4 and 5 and the higher part ofPSF‐7, shows a primarily vertical displacement,producing a downthrow of the eastern side at a rateof about 4 mm/yr. PSF‐7, 6 and 4 do not affectsignificantly the eastward velocity patterns (Figure 5,cross sections A‐A′ and E‐E′). PSF‐5, which cor-responds to the Acicatena fault, conversely shows amore complex path and wider extension than thosedetectable by geological surveys. It shows bothsignificant vertical and horizontal displacements,with rates of about 5 mm/yr (Figure 5, cross sectionsE‐E′ and F‐F′) and bounds on the west side theAcireale area, affected by very high eastwardvelocity (20–25 mm/yr).

[52] PSF‐3 delimits southward the other sectorcharacterized by the highest eastward velocity androughly corresponds to the S. Tecla‐S. Venerinafaults alignment. This sector, involving the entireNE flank of the volcano from the NE rift to thecoast, moves eastward between the Pernicana fault(PSF‐1) and the S. Tecla‐S. Venerina faults (PSF‐3)at a rate higher than 20 mm/yr (Figure 5, crosssection D‐D′), reaching a maximum velocity ofabout 30 mm/yr at lowest altitude (Figure 5, crosssection F‐F′). These rates are in very good agree-ment to the displacement rates measured by localGPS network installed across the Pernicana fault[Bonforte et al., 2007a; Palano et al., 2006].

[53] Along the northern Ionian coast PS groundvelocity maps record the larger eastward anddownward motion on the entire volcano easternflank, confirming a feature previously highlightedby GPS data [Bonforte and Puglisi, 2006]. Thistriangular area named the “Giarre wedge” (Figure 6)is well delimited by PSF‐2 on the south and lessclearly delimited by the decreasing vertical velocityon the north. It fits well the costal triangular areashowing strong subsidence between September 1997and August 1998 as observed in GPS data [seeBonforte and Puglisi, 2006, Figure 3a].

8. Conclusions

[54] PS analysis described here clearly shows,with a very high spatial resolution, the main fea-tures of the ground deformation pattern affectingMount Etna volcano during a noneruptive period,from 1995 to 2000. Before using the results of thePS analysis for defining the structural assessmentof the volcano, we verified the good agreementbetween these data and the velocities measured byGPS in 35 GPS stations surveyed throughout the


16 of 19

considered period. PS data allow the main struc-tural features of the entire volcano to be traced withhigh spatial resolution and their kinematics for theperiod from 1995 to 2000 to be analyzed in orderto investigate the dynamics driving the grounddeformation of the volcano. Two first‐order pat-terns of ground motion were detected, each beingrelated to two different large‐scale deformationpatterns affecting opposing sectors of the volcano.

[55] The first ground deformation feature is anevident inflation of the northern and western flanksof the edifice (Figure 6). This is related to themagma accumulation at depth and the consequentpressurization of the entire feeding system of thevolcano, from depth and intermediate storage zonesup to the central conduits. This uniform and diffusedeformation is mainly visible in the PS images onthe northern and western slopes of the volcano,because this region of the edifice is little affectedby structural discontinuities that drive the strainpropagation; indeed, these flanks lie over the tec-tonic units of the orogenic chain and are thus notdisturbed by any gravitational spreading process.

[56] The second ground deformation feature is theeastward and downward motion of the southernand eastern flanks of the volcano. This kind ofmotion reveals an independent behavior of the sea‐facing and foredeep‐facing slopes of Mount Etna,which is not related to its magmatic and eruptivedynamics. At a wider scale, horizontal seaward‐directed velocities progressively increase fromhigher to lower elevations, while subsidence de-creases, consistent with a wide, first‐order rotationaldecollement. At a more local scale, it is evident howthe motion of the seaward moving flanks is verycomplex since it is segmented into a numberblocks, each of which show distinct velocities andkinematics (Figure 6).

[57] The Giarre wedge is the continuation on theIonian sea of the NE block (Figure 6) showing ahorizontal velocity that slightly increases towardthe coast and a vertical velocity that becomesstrongly negative on the coast. The NE block doesnot show any significant discontinuity in its east-ward (seaward) movements, whereas to the souththe horizontal and vertical velocity maps (Figure 6)show a complex pattern on the volcano flankmarked by discontinuities from PSF‐3 to PSF‐8 thatpoint out a change in seaward movements. Both themedium east and SE blocks (Figure 6) have radialdiscontinuities with tilt component toward the NE,evident from section A‐A′ (Figure 5), indicating arotation of individual blocks following the main

seaward movement of the main NE block. Finally,this behavior vanishes on S and SW flanks passinginto the outward‐directed deformation on the Bian-cavilla wedge and Adrano graben (Figure 6) andthen to the inflation affecting the western andnorthern flanks.

[58] At a glance the main components of the Eand SE flank of the volcano are a radial seawardmovement and a strong subsidence of the Ioniancoast. The kinematic model of these movements isa tectonic lowering of the Giarre wedge that dragsthe blocks toward the sea; it drags directly the NEblock and, through slippery steps, that are probablyrelated to the basement discontinuities, the mediumeast and SE blocks. This complex kinematics canbe explained as a tectonic inducement to slip by astrong subsidence area localized below sea level infront of Riposto village.

[59] The sliding mechanism changes abruptly to thesouth where the southern flank pushes against theMisterbianco ridge (Figure 6), a structural barrier(ancient anticline) that stops the movement of out-ward sliding of the volcano by gravity. This featurecould in fact be related to the drag and compressioneffects of the southward motion of the southernflank of Mount Etna on the plastic sedimentaryclays lying beneath the southern base of the volcano.Only here, in fact, does the presence of such aplastic basement create the ideal conditions ofanalog models of radially spreading volcanoes,allowing also the formation of a triangular horst(Biancavilla wedge) bordered by a sector graben(Adrano graben) on the southwestern side of thevolcano. The anticline is thus rooted at shallowdepth and related to the gravitational loading ofthe flank of the volcano.

[60] On the southern flank the observed movementsare attributable only to the gravitational reorgani-zation of the volcanic pile on its slippery claybasement, as proposed byMerle and Borgia [1996].In contrast, on the eastern flank PS analysis allowsus to propose that an additional drag force, dueto a strong subsidence of the continental marginfacing the Etna volcano, is necessary to arrange theobserved PS velocity field.

[61] Because of the spatial continuity and high res-olution of the PS data we are able to accuratelydefine all different blocks composing the unstablesectors of the volcano, the discontinuities boundingeach block and the precise kinematics of eachdiscontinuity. This kind of information provided byPS technique represents a major improvement inthe knowledge of the structural setting of Mount


17 of 19

Etna, since it allowed tracing the real extension ofsome faults (partially known from their morpholog-ical evidence) and discovery of some new structures,even in urban areas, which were not previouslydetected by geological surveys.

Acknowledgments

[62] We are grateful to Claudio Prati and Fabio Rocca for pro-moting this work and the fruitful discussions. We thank alsothe INGV ground deformation staff for the support in the fieldsurveys. We wish to thank Richard Herd and two others anon-ymous reviewers for their important suggestions for improvingthe early versions of the paper. Thanks are due to the directorof the Catania section of INGV, A. Bonaccorso, for supportingthis work.

References

Azzaro, R., S. Branca, S. Giammanco, S. Gurrieri, R. Rosa,and M. Valenza (1998), New evidence for the form andextent of the Pernicana Fault System (Mount Etna) fromstructural and soil‐gas surveying, J. Volcanol. Geotherm.Res., 84, 143–152, doi:10.1016/S0377-0273(98)00036-5.

Beauducel, F., P. Briole, and J. L. Froger (2000), Volcano‐wide fringes in ERS synthetic aperture radar interferogramsof Etna (1992–1998), Deformation or tropospheric effect?,J. Geophys. Res., 105, 16,391–16,402, doi:10.1029/2000JB900095.

Bonaccorso, A., A. Bonforte, F. Guglielmino, M. Palano, andG. Puglisi (2006), Composite ground deformation patternforerunning the 2004–2005Mount Etna eruption, J. Geophys.Res., 111, B12207, doi:10.1029/2005JB004206.

Bonforte, A., and G. Puglisi (2003), Magma uprising and flankdynamics on Mount Etna volcano, studied using GPS data(1994–1995), J. Geophys. Res., 108(B3), 2153, doi:10.1029/2002JB001845.

Bonforte, A., and G. Puglisi (2006), Dynamics of the easternflank of Mount Etna volcano (Italy) investigated by a denseGPS network, J. Volcanol. Geotherm. Res., 153, 357–369,doi:10.1016/j.jvolgeores.2005.12.005.

Bonforte, A., A. Ferretti, C. Prati, G. Puglisi, and F. Rocca(2001), Calibration of atmospheric effects on SAR inter-ferograms by GPS and local atmosphere models: First results,J. Atmos. Sol. Terr. Phys., 63, 1343–1357, doi:10.1016/S1364-6826(00)00252-2.

Bonforte, A., F. Guglielmino, M. Palano, and G. Puglisi(2004), A syn‐eruptive ground deformation episode measuredby GPS, during the 2001 eruption on the upper southern flankof Mt Etna, Bull. Volcanol., 66, 336–341, doi:10.1007/s00445-003-0314-x.

Bonforte, A., S. Gambino, F. Guglielmino, F. Obrizzo,M. Palano, and G. Puglisi (2007a), Ground deformationmodeling of flank dynamics prior to the 2002 eruption ofMount Etna, Bull. Volcanol., 69, 757–768, doi:10.1007/s00445-006-0106-1.

Bonforte, A., S. Branca, and M. Palano (2007b), Geometricand kinematic variations along the active Pernicana fault:Implication for the dynamics of Mount Etna NE flank (Italy),J. Volcanol. Geotherm. Res., 160, 210–222, doi:10.1016/j.jvolgeores.2006.08.009.

Bonforte, A., D. Carbone, F. Greco, and M. Palano (2007c),Intrusive mechanism of the 2002 NE‐rift eruption at Mt Etna(Italy) modelled using GPS and gravity data, Geophys. J. Int.,169, 339–347, doi:10.1111/j.1365-246X.2006.03249.x.

Bonforte, A., A. Bonaccorso, F. Guglielmino, M. Palano, andG. Puglisi (2008), Feeding system and magma storagebeneath Mount Etna as revealed by recent inflation/deflationcycles, J. Geophys. Res., 113, B05406, doi:10.1029/2007JB005334.

Bonforte, A., S. Gambino, and M. Neri (2009), Intrusion ofeccentric dikes: The case of the 2001 eruption and its rolein the dynamics of Mount Etna volcano, Tectonophysics,471, 78–86, doi:10.1016/j.tecto.2008.09.028.

Borgia, A., L. Ferrari, and G. Pasquarè (1992), Importance ofgravitational spreading in the tectonic and volcanic evolutionof Mount Etna, Nature, 357, 231–235, doi:10.1038/357231a0.

Borgia, A., R. Lanari, E. Sansosti, M. Tesauro, P. Berardino,G. Fornaro, M. Neri, and J. B. Murray (2000), Activelygrowing anticlines beneath Catania from the distal motionof Mount Etna’s decollement measured by SAR interferom-etry and GPS, Geophys. Res. Lett., 27, 3409–3412,doi:10.1029/1999GL008475.

Bousquet, J. C., and G. Lanzafame (2004), The tectonics andgeodynamics of Mount Etna: Synthesis and interpretationof geological and geophysical data, inEtnaVolcano Laboratory,Geophys. Monogr. Ser, vol. 143, edited by A. Bonaccorso et al.,pp. 29–47, AGU, Washington, D. C.

Branca, S., and P. Del Carlo (2004), Eruptions of Mt Etnaduring the past 3,200 years: A revised compilation integratingthe historical and stratigraphic records, in Etna: VolcanoLaboratory, Geophys. Monogr. Ser, vol. 143, edited byA. Bonaccorso et al., pp. 1–27, AGU, Washington, D. C.

Briole, P., D. Massonnet, and C. Delacourt (1997), Post‐eruptive deformation associated with the 1986–87 and 1989lava flows of Etna detected by radar interferometry, Geophys.Res. Lett., 24, 37–40, doi:10.1029/96GL03705.

Calvari, S., M. Coltelli, M. Neri, M. Pompilio, and V. Scribano(1994), The 1991–93 Etna eruption: Chronology and lavaflow field evolution, Acta Vulcanol., 4, 1–14.

Colesanti, C., A. Ferretti, C. Prati, and F. Rocca (2003), Mon-itoring landslides and tectonic motions with the PermanentScatterers Technique, Eng. Geol. Amsterdam, 68, 3–14,doi:10.1016/S0013-7952(02)00195-3.

Curlander, J. C., and R. N.McDonough (1991), Synthetic Aper-ture Radar: Systems and Signal Processing, John Wiley,New York.

Farr, T. G., et al. (2007), The Shuttle Radar TopographyMission,Rev. Geophys., 45, RG2004, doi:10.1029/2005RG000183.

Ferretti, A., C. Prati, and F. Rocca (2000), Nonlinear subsi-dence rate estimation using permanent scatterers in differen-tial SAR interferometry, IEEE Trans. Geosci. Remote Sens.,38, 2202–2212, doi:10.1109/36.868878.

Ferretti, A., C. Prati, and F. Rocca (2001), Permanent scat-terers in SAR interferometry, IEEE Trans. Geosci. RemoteSens., 39, 8–20, doi:10.1109/36.898661.

Froger, J. L., O. Merle, and P. Briole (2001), Active spreadingand regional extension at Mount Etna imaged by SAR interfer-ometry, Earth Planet. Sci. Lett., 187, 245–258, doi:10.1016/S0012-821X(01)00290-4.

Houlié, N., P. Briole, A. Bonforte, and G. Puglisi (2006),Large scale ground deformation of Etna observed by GPSbetween 1994 and 2001, Geophys. Res. Lett., 33, L02309,doi:10.1029/2005GL024414.


18 of 19

Jenks, G. F., and F. C. Caspall (1971), Error on choroplethicmaps: Definition, measurement, reductions, Ann. Assoc. Am.Geogr., 61, 217–244, doi:10.1111/j.1467-8306.1971.tb00779.x.

Lanari, R., P. Lundgren, and E. Sansosti (1998), Dynamicdeformation of Etna volcano observed by satellite radarinterferometry, Geophys. Res. Lett., 25, 1541–1544,doi:10.1029/98GL00642.

Massonnet, D., and K. L. Feigl (1995), Discrimination ofgeophysical phenomena in satellite radar interferograms,Geophys. Res. Lett., 22, 1537–1540, doi:10.1029/95GL00711.

Massonnet, D., P. Briole, and A. Arnaud (1995), Deflation ofMount Etna monitored by spaceborne radar interferometry,Nature, 375, 567–570, doi:10.1038/375567a0.

Mattia, M., D. Patane`, M. Aloisi, and M. Amore (2007), Fault-ing on the western flank of Mt Etna and magma intrusions inthe shallow crust, Terra Nova, 19, 89–94, doi:10.1111/j.1365-3121.2006.00724.x.

Merle, O., and A. Borgia (1996), Scaled experiments ofvolcanic spreading, J. Geophys. Res., 101, 13,805–13,817,doi:10.1029/95JB03736.

Neri, M., V. Acocella, and B. Behncke (2004), The role of thePernicana Fault System in the spreading of Mount Etna(Italy) during the 2002–2003 eruption, Bull. Volcanol., 66,417–430, doi:10.1007/s00445-003-0322-x.

Neri, M., F. Guglielmino, and D. Rust (2007), Flank instabilityon Mount Etna: Radon, radar interferometry and geodeticdata from the southwestern boundary of the unstable sector,J. Geophys. Res., 112, B04410, doi:10.1029/2006JB004756.

Nunnari, G., and G. Puglisi (1994), The Global PositioningSystem as a useful technique for measuring ground deforma-tions in volcanic areas, J. Volcanol. Geotherm. Res., 61,267–280, doi:10.1016/0377-0273(94)90008-6.

Onn, F., and H. A. Zebker (2006), Correction for interferomet-ric synthetic aperture radar atmospheric phase artifacts usingtime series of zenith wet delay observations from a GPS

network, J. Geophys. Res., 111, B09102, doi:10.1029/2005JB004012.

Palano, M., et al. (2006), Kinematics and strain analyses of theeastern segment of the Pernicana Fault (Mount Etna, Italy)derived from geodetic techniques (1997–2005),Ann. Geophys.,49, 1105–1117.

Puglisi, G., and A. Bonforte (2004), Dynamics of MountEtna Volcano inferred from static and kinematic GPS mea-surements, J. Geophys. Res., 109, B11404, doi:10.1029/2003JB002878.

Puglisi, G., A. Bonforte, and S. R. Maugeri (2001), Grounddeformation patterns on Mount Etna, 1992 to 1994, inferredfrom GPS data, Bull. Volcanol., 62, 371–384, doi:10.1007/s004450000112.

Puglisi, G., P. Briole, and A. Bonforte (2004), Twelve years ofground deformation studies on Mount Etna volcano based onGPS surveys, inEtna Volcano Laboratory, Geophys.Monogr.Ser, vol. 143, edited by A. Bonaccorso et al., pp. 321–341,AGU, Washington, D. C.

Puglisi, G., A. Bonforte, A. Ferretti, F. Guglielmino, M. Palano,and C. Prati (2008), Dynamics of Mount Etna before, duringand after the July August 2001 eruption inferred from GPSand differential synthetic aperture radar interferometry data,J. Geophys. Res., 113, B06405, doi:10.1029/2006JB004811.

Rust, D., and M. Neri (1996), The boundaries of large‐scale col-lapse on the flanks of Mount Etna, Sicily, in Volcano Instabilityon the Earth and Other Planets, edited byW. J. McGuire, A. P.Jones, and J. Neuberg, Geol. Soc. Spec. Publ., 110, 193–208.

Zebker, H. A., and J. Villasenor (1992), Decorrelation in inter-ferometric radar echoes, IEEE Trans. Geosci. Remote Sens.,30, 950–959, doi:10.1109/36.175330.

Zebker, H. A., P. A. Rosen, and S. Hensley (1997), Atmo-spheric effects in interferometric synthetic aperture radar sur-face deformation and topographic maps, J. Geophys. Res.,102, 7547–7563, doi:10.1029/96JB03804.


19 of 19

Structural assessment of Mount Etna volcano from Permanent Scatterers analysis · 2018-06-08 · 1....

Documents

Transcript of Structural assessment of Mount Etna volcano from Permanent Scatterers analysis · 2018-06-08 · 1....