This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy Journal of Volcanology and Geothermal Research 182 (2009) 172–181 Contents lists available at ScienceDirect Journal of Volcanology and Geothermal Research j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / j v o l g e o r e s Insight on recent Stromboli eruption inferred from terrestrial and satellite ground deformation measurements A. Bonaccorso, A. Bonforte, S. Gambino, M. Mattia, F. Guglielmino, G. Puglisi ⁎, E. Boschi Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Catania, P.za Roma, 2, 95123 Catania, Italy a r t i c l e i n f o Article history: Received 4 July 2008 Accepted 3 January 2009 Available online 14 January 2009 Keywords: Stromboli ground deformations source modelling flank instability a b s t r a c t The multi-parametric permanent system (tilt and GPS networks, robotized geodetic station) for monitoring ground deformation at Stromboli volcano was set up in the 1990s and later greatly improved during the effusive event of 2002–2003. Unlike other volcanoes, e.g. Mt. Etna, the magnitude of ground deformation signals of Stromboli is very small and through the entire period of operation of the monitoring system, only two major episodes of deformation, in 1994–1995 and 2000, which did not lead to an eruption but rather pure intrusion, were measured. Similarly to the 2002–2003 eruption, no important deformations were detected in the months before the 2007 eruption. However, unlike the 2002–2003 eruption, GPS and tilt stations recorded a continuous deflation during the entire 2007 eruption, which allowed us to infer a vertical elongated prolate ellipsoidal source, centered below the summit craters at depth of about 2.8 km b.s.l. Due to its geometry and position, this source simulates an elongated plumbing system connecting the deeper LP magma storage (depth from 5 to 10 km) with the HP shallower storage (0.8–3 km), both previously identified by petrologic and geochemical studies. This result represents the first contribution of geophysics to the definition of the plumbing system of Stromboli at intermediate depth. Finally, no deformation due to the plumbing system was measured for a long time after the end of the eruption. Meanwhile, the new terrestrial geodetic monitoring system installed within the Sciara del Fuoco, on the lava fan formed during the eruption, indicated that during the first months after the end of the eruption the ground velocity progressively decreased in time, suggesting that part of the deformation was due to the thermal contraction of the lava flow. © 2009 Elsevier B.V. All rights reserved. 1. Introduction Stromboli volcano is the northernmost island of the Aeolian Archipelagos. The volcano is about 2500 m high above the sea floor, but only the topmost 910 m of this volcano emerges from the sea, forming the island whose diameter ranges from 2.4 to 5 km. Its continuous volcanic activity is characterized by intermittent explosions at one or more of the summit craters, at intervals in the order of minutes, emitting pyroclastics at heights of tens or hundred meters (Strombolian activity). Occasionally, major explosions or paroxysms occur with frequencies of a couple of events per year (major explosions) or per century (paroxysms) (Barberi et al., 1993). Sometimes, at intervals in the order of years or tens of years, lava flows generated from the summit craters or fissures opened at their base flow into the Sciara del Fuoco (SdF), a large depression that marks the northwestern flank of the edifice. The SdF was produced by a huge sector collapse, in the order of 1 cubic km that very likely originated in the Holocene (Tibaldi, 2001). This sector collapse was the last of four episodes which affected the northeastern flank of the volcano during ⁎ Corresponding author. Tel.: +39 095 7165800; fax: +39 095 435801. E-mail address: puglisi-g@ct.ingv.it (G. Puglisi). 0377-0273/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jvolgeores.2009.01.007 the last 100 Ky, with volumes involved in collapses decreasing with age (Tibaldi, 2001). The sliding of the SdF is continuous, with velocity in the order of 1–2 mm/day, as indicated by recent terrestrial geodetic continuous measurements (Puglisi et al., 2005; Nunnari et al., 2008). Usually, lava flows originating from the craters or fissure systems reach the sea, aided by the very steep flank of the SdF; the last effusive eruptions occurred in 1985–1986, 2002–2003 and 2007. The last two eruptions were the occasions for volcanologists to significantly advance their knowledge of this volcano, due to the great efforts toward improving the monitoring systems both in number of stations and techniques and in quality of data sets. The 2007 eruption started on the morning of 27 February, with the opening of a fracture system within and at the base of the NE Crater. The images of the video monitoring system of the Istituto Nazionale di Geofisica e Vulcanologia (INGV) indicates that the lava emission within the summit craters had already started at about 12:00 GMT and that at about 13:00 a new vent opened at the northeastern base of the NE crater (~ 650 m a.s.l.) (Calvari et al., 2008). During the first half-day, the fracture system evolved down to the SdF, until at about 400 m of elevation, where an effusive vent opened at 18:30 GMT (Neri et al., 2008), so that in the afternoon the lava flows reached the sea. This vent remained almost continuously active until the end of the Author's personal copy A. Bonaccorso et al. / Journal of Volcanology and Geothermal Research 182 (2009) 172–181 173 Fig. 1. Map of the multi-parametric ground deformation monitoring system installed at Stromboli island. eruption, on 2 April, except from 8 to 9 March, when a second vent opened a few hundred meters south of the former, remaining active for about 24 h. The lava flows overlapped and formed a fan structure along the northern scarp of the SdF (Fig. 1a). Throughout the eruption, the usual explosive activity at the summit craters stopped. Only on March 15th did a sudden major explosion occur whose products reached the lower altitudes of the flanks of the volcano. Unlike the final phases of the 2002–2003 eruption, when the explosive activity slowly resumed at the summit craters during the last weeks of the eruption, in 2007 the craters remained inactive for several months after the end of the eruption, and gradually renewed the ordinary strombolian activity during the summer. In this paper, we study the different phases of the 2007 eruption by analyzing the ground deformation data sets that the multiparametric monitoring system on Stromboli acquired before, during and after the volcanic event. The deformation data provided useful information to monitor the ongoing phenomena and to model the shallow-intermediate storage which caused the deflation, the first one measured at Stromboli, associated with the effusive eruption. 2. Geodetic networks The monitoring of ground deformations at Stromboli volcano is currently performed by a multi-parametric system with the twofold aim of measuring the deformations due to magma movements within the plumbing system and movements of the SdF. The first aim is achieved by performing continuous GPS and tilt measurements and the latter by carrying out continuous terrestrial geodetic measurements. 2.1. Tilt network Tilt has been continuously monitored at Stromboli since 1992 (Bonaccorso, 1998). Two tilt stations were installed at Punta Labronzo (SPLB) and Timpone del Fuoco (STDF), beyond the southern and northern rim of the SdF, respectively (Fig. 1). A third station was installed at Punta Lena (SPLN) but this installation has never been completely stabilized, in the sense that the data show an anomalous continuous drift that prevents their use. The sensors are biaxial shallow boreholes with 0.1 μrad of precision (manufacturer's estimate), installed at 2.5–3.5 m of depth. The two axes are respectively oriented towards the crater area (radial component) and towards a direction 90° counter-clockwise (tangential component). Air temperature and ground temperatures, at different depths in the hole, are also recorded in order to filter thermoelastic effects on the signals. Sampling rate is 1/30 min and data are transmitted in realtime time to Catania. Following instrumental damage, PLB station was reinstalled in May 2005. 2.2. GPS network The GPS permanent network was installed in June 1997. The remote stations (SPLN, SPLB, STDF) are located at the same sites as the tilt stations (Fig. 1). In 1998, a fourth GPS station (SCPS) was added to the network so that the final geometry of a polygon, completely surrounding the cone of the volcano, was reached. In February 2003, during the 2002–2003 eruption, four 1 Hz GPS permanent stations were installed in order to monitor the deformation of the highest part of the SdF and the area around the craters. Three of these new stations were located in the summit area of the volcano (SDIC; SSBA; SCRA) and the other one near the geophysical Observatory of San Vincenzo (SVIN) used as reference for the high-rate real-time data processing (Mattia et al., 2004; Puglisi et al., 2005). This network (named SCIARADAT) was declared operational on 12 February 2003. Within a few days (20 February 2003) SSBA station was destroyed by a lava flow; SDIC and SCRA stations were destroyed by the paroxystic explosion occurring at the summit crater on 5 April 2003. Today, the GPS network consists of 5 stations (SPLN, SPLB, STDF, SCPS and SVIN) and its data are processed both daily by GAMIT-GLOBK software and in Author's personal copy 174 A. Bonaccorso et al. / Journal of Volcanology and Geothermal Research 182 (2009) 172–181 Fig. 2. Detail of the THEODOROS system for monitoring the Sciara del Fuoco movements; in italics the benchmarks destroyed by the 2007 lava flow; the new benchmarks installed on the lava fan are also reported. The dotted lines indicate the margins of the aerial photo (on the left), where the 2007 lava flows and the lava fan are visible. real time by means of the Geodetics RTD software, in this case by considering SVIN as reference station (King and Bock, 2000; Herring, 2000). 2.3. Terrestrial geodesy At the end of the 2002–2003 eruption, a terrestrial monitoring system was installed to routinely measure the motion of benchmarks installed inside the SdF (Puglisi et al., 2005). This system, named THEODOROS, is based on a remotely controlled robotized Total Station installed near Punta Labronzo, on the northern border of the SdF (Figs. 1 and 2). The monitoring network is composed of three groups of benchmarks: (1) the reference system; (2) the control group and (3) the Sciara group. The first two groups all consist of benchmarks installed outside the SdF (i.e. in the stable part of the volcano with respect to SdF) ensuring congruent measurements in time; the first allowing to calculate the coordinates of the Total Station, the orientation of the reference system and the atmospheric refractivity, while the second verifies the stability of the system. The third group consists of benchmarks installed inside the SdF to monitor its movements. The measurement cycles comprise an appropriate combination and scheduling of a series of measurements of the different groups: (1) measurements of the reference system, every half an hour, to assess the positioning and orientation of the system; (2) measurements of the same group of benchmarks, every hour, to estimate the atmospheric refractivity; (3) measurements of the control points, every 20 min; and (4) measurements on the Sciara group, every 10 min. Each series consists of left/right measurements. Data are continuously transmitted to the S. Vincenzo Observatory, where the control of THEODOROS system is installed, and from there to INGV in Catania. 3. Ground deformations before the 2007 eruption During the years before the 2007 eruption, the ground deformation monitoring system measured at least two major episodes of deformation (Fig. 3), without associated eruptions, related to intrusive episodes at shallow levels (1–2 km. b.s.l.) in 1994–1995 (Bonaccorso, 1998) and in 2000 (Mattia et al., 2008). On 5 April 2003, a further episode was measured surrounding the summit area by the high rate GPS network, which revealed the response to the action of the shallow final path during a violent explosive paroxysm (Mattia et al., 2004). Fig. 3. SPLB tilt signals recorded from 1993 to 26 February 2007. The major episodes of tilt changes have followed the main seismic strain release episodes occurring in this period. Author's personal copy A. Bonaccorso et al. / Journal of Volcanology and Geothermal Research 182 (2009) 172–181 175 Fig. 4. Example of a detrended GPS time series (STDF station). This image clearly show the stability of the benchmark and that large deformations related to magmatic or tectonic activity hasn't ever been observed since 1997. The studies of the 1994–1995 and 2000 cases highlighted that the intrusion propagated along a NE–SW trending zone of weakness that dissects the summit of the volcano. This information fits the structural framework of the island (Pasquarè et al., 1993; Tibaldi, 1996) and confirms that the regional stress field is very influential along this structural direction. The 1994–1995 and 2000 intrusions inferred by tilt signals occurred months after the 1994 and 1999 swarms (Fig. 3), and were interpreted as slow magma injection processes, lasting about 3 and 2 months respectively, along this weak structure triggered by the previous seismicity (Bonaccorso, 1998). It is noteworthy that, excepting the cases mentioned above (1994– 1995 and 2000 tilt changes, summit crater area deformations during the 5 April 2003 paroxysm), the ground deformation signals recorded by the GPS and tilt networks at Stromboli were usually of small magnitude and did not cumulate significant variations (Fig. 4). Therefore, for the last 2002–2003 and 2007 eruptions, the ground deformation networks did not detect significant deformation changes of the overall volcano due to possible inflation preceding these eruptions, except a small uplift toward SW of the tilt at SPLB (about 7 μrad), cumulated between August and December 2006, detectable only after an accurate filtering of the thermal noise (Fig. 5). The results of the terrestrial geodetic monitoring during 2003–2007 period show that ground movements affected only the northern sector of the SdF involved in the December 30th 2002 landslides. Although the cumulative deformations measured in the SdF after the end of 2002–2003 eruption reached up to several meters in a few benchmarks, up to January 2007, these movements are relatively slow, with velocities measured at all benchmarks diminishing throughout the 4 years after the end of the 2003 eruption. (Fig. 6). Furthermore, the ground deformations within this area were not uniform, showing a decreasing rate from the upper (up to 1 cm/ day) to the lower part (few mm/day) of the slope (Fig. 6). Conversely, the central part of the SdF did not show significant ground deformation. All these data suggest that the deformations measured from 2003 to 2007 were related to the continuous settlement of the SdF after the events of December 2002 that affected this flank of the volcano. Unfortunately, technical problems affected the Total Station after 19 January 2007 and no ground deformation data is available for the SdF just before the onset of the 2007 eruption. 4. Ground deformations at the onset and accompanying the 2007 eruption On 27 February 2007, the eruption onset corresponded to a small but abrupt change in both GPS and tilt signals. In particular, in the early morning (from 8:30 to 13:00; Fig. 7), the SPLB tilt station recorded the beginning of a slow decrease in radial component accompanied by fluctuations, visible also on tangential component, probably related to the local response to explosions and seismic activity. Starting from 13:30 to 14:30 (i.e. just after the opening of the vent at 650 m) both SPLB components began to show clear trends that continued in the following days of the eruption, up to the end of March, cumulating 2.4 μrad of a down tilt vector, SW oriented (Fig. 8). Author's personal copy 176 A. Bonaccorso et al. / Journal of Volcanology and Geothermal Research 182 (2009) 172–181 Fig. 5. Tilt components recorded at PLB station from its re-installation (May 2005) to 2007 February, 26. Signals have been filtered by a linear correlation with the ground temperature and highlight the June–August 2006 transient anomaly (see also Fig. 3) and a SW uplift cumulated between August and December 2006. We did not use data from STDF tilt station because it was affected by electronic noise that made estimating its changes uncertain. The beginning of the eruption also influences the time series of the slope distance measurements between GPS stations, although shifted by about 1 day with respect to the eruption onset, due to the GPS sessions used to measure the slope distances which is processed daily over a 24 h window and produces a “lag” in the signal representation. Indeed, the comparison between the solutions of 27 and 28 February, show a contraction of some slope distances (Fig. 7), which continued throughout the eruption. One of the peculiarities of the ground deformation signals associated to the 2007 eruption was the continuous deflation we observed during the eruption, never measured in the past eruptions, whose measurable effects were the down tilt vector recorded at SPLB and the contractions of GPS network (Fig. 8). Making the most of this unique data set, we inferred the presence of a deflationary ground deformation source. Bonaccorso et al. (2008) modelled this source as a prolate ellipsoid, centered below summit craters at depth of about 2.8 km b.s.l. Despite the small intensity of the recorded signals, compared with similar signals measured on other volcanoes (e.g. Mt. Etna), the modelling fitted the data set well (misfit of 0.1 μrad for the tilt and 1.5 mm and 3.8 mm for horizontal and vertical GPS components, respectively) excluding the presence of a near spherical magmatic reservoir in the last kilometers below the volcano. The ratios between its semi-axes and its Euler angles indicate that this source simulates an elongated plumbing system, with its vertical axis about ten times longer than the other two axes (Bonaccorso et al., 2008) (Fig. 9). With respect to the monitoring of the SdF movements, the 2007 eruption caused a dramatic change in the operations of THEODOROS. The 2007 eruption lava flows destroyed all benchmarks installed on the northern part of the SdF, leaving only those on its central part. Unfortunately, these remaining benchmarks were difficult to measure automatically before the 2007 eruption, because their prisms deteriorated from 2003 until 2006, and the distance from the Total Stations is at the limit of the range in normal conditions. Furthermore, the heat rising from the active lava flows and the steam column between station and benchmarks made the automatic measurements even more difficult. For these reasons, we were forced to discard automatic measurements and to perform manual measurements during the eruption, and in particular on 9 March, to verify the stability of the central part of the SdF. On 9 March, the geodetic terrestrial network was also improved by installing a new reflector on the pillar set up in 2003 at the Pizzo area (Puglisi et al., 2005) in order to check the stability of the area surrounding the craters; a further reflector was installed on the upper eastern border of the SdF (BASTI, Fig. 2), to better monitor the overall stability of this flank of the volcano. All the measurements carried out on these benchmarks revealed the stability of both the central part of the SdF and the summit and surrounding areas, even during the opening of the vent on 9 March. 5. Ground deformations after the 2007 eruption After the end of the 2007 eruption both GPS and tilt networks did not measure significant ground deformations for several months, at least until the resumption of the strombolian activity at the summit craters, at the beginning of June 2007 (Marotta et al., 2008). Unfortunately, the SPLB tilt station failed on 13 April, so our information is based only on the GPS measurements, which did not show significant changes. The 2007 eruption produced a lava fan at the base of the SdF, due to the rapid cooling of the lava flows when entering the sea; this phenomenon produced a continuous overlapping of several flows during the eruption, building a thick lava body emplaced on a very steep slope, implying a hazard due to the possible fast sliding of this fan into the sea. In order to monitor the stability of this lava fan, the terrestrial geodetic network, was implemented on 6 April by installing 5 reflectors along a profile crossing the lava body, approximately over the old coastline. Later on, in June 2007, 4 further reflectors were installed at higher and lower altitude with respect to the previous profile, to have further information on the overall deformation of the lava body. Measurements were rather noisy during the first months, but a better definition of the reference system, implemented also by installing a new benchmark (SENT, Fig. 2), strongly improved the quality of the data. Furthermore, after the installation of “SENT,” the benchmark “400” was no longer used as a reference but to improve the monitoring of the eastern border of the SdF, together with “BASTI” benchmark. The position of the 9 benchmarks over the lava fan allows the areal distribution of the deformation to be drawn. The measurements carried out every 10 min follow with high temporal detail their motion (Fig. 10). The data collected since the end of the eruption highlighted a significant downslope motion of the entire lava fan, decreasing from the South to the North, where the body is buttressed by the northern wall of SdF. The ground velocity, especially the vertical component, was initially very high but progressively decreased in time, and it is still decreasing albeit with a lower rate. This evolution of the phenomenon and the initial main vertical deformation suggest that, Fig. 6. Daily averages of the North (A), East (B) and Height (C) components relevant to the benchmarks within the SdF, belonging to the THEODOROS system, from November 2004 to January 2007. Author's personal copy A. Bonaccorso et al. / Journal of Volcanology and Geothermal Research 182 (2009) 172–181 177 Author's personal copy 178 A. Bonaccorso et al. / Journal of Volcanology and Geothermal Research 182 (2009) 172–181 Fig. 7. A) Tilt components recorded at PLB station during the first months of the 2007. Signals are filtered by a linear correlation with the ground temperature and highlight a SW deflation trend after February 27. A thermo-elastic residual effect is present on the tangential component. The inset reports the raw signals recorded during the eruption onset; sampling rate is 1/30 min; vertical lines indicate the 07:30–13:30 GMT time window. B) Distance changes recorded at the SPLN-STDF and SPLB-STDF baselines of the GPS permanent network. at least during the very first months after the end of the eruption, part of the deformation was due to the thermal contraction of the thick lava body. 6. Discussion and conclusive remarks Since the installation of the first permanent monitoring network at Stromboli, on 1992 the magnitude of ground deformation signals of Stromboli is very small. The two episodes of significant tilt changes recorded in 1994–1995 (Bonaccorso, 1998) and in 2000 (Mattia et al., 2008), were related to intrusive episodes along the NE–SW main structural trend at shallow levels (1–1.5 km. b.s.l.) during periods without effusive eruptions. In addition, the beginning of the 2002– 2003 eruption was not accompanied by significant ground deformation changes, and the only event of marked deformation recorded during the effusive eruption occurred on April 5, 2003 in concomitance of the paroxystic explosion. During that event, the powerful rise of magma deformed the final paths of the summit conduits and in the summit area provoked large displacements measured by the con- tinuous high rate GPS network (Mattia et al., 2004). After 2002–2003, significant movements were recorded only in the SdF related to the slow sliding of this area. During the 2007 eruption, the ground deformation measurements, whose data were collected by using both terrestrial (total station and tilt) and satellite (GPS) techniques, played a fundamental role in both monitoring different volcano aspects and, for the first time, modelling the source acting during the effusive eruption. With reference to the main information of the monitoring, the measurements with the total station provided useful monitoring of the SdF sliding, highlighting that the movements were confined only in the northern part of the depression without significant displacements in the central part, and that the western upper crater rim (Pizzo Sopra La Fossa) was stable. The THEODOROS system also monitored the stability of the lava fan produced by cooling flows in the SdF at the sea level height. The value of the multi-parametric deformation monitoring system was particularly appreciated during the critical episodes occurring on Author's personal copy A. Bonaccorso et al. / Journal of Volcanology and Geothermal Research 182 (2009) 172–181 179 Fig. 8. Recorded and expected GPS vectors and tilt. The expected values were obtained by using a vertically elongated depressurizing ellipsoid centered 2.8 km below summit craters (Bonaccorso et al., 2008). The GPS vectors are obtained by fixing SVIN as reference station. The star indicates the projection onto the surface of the model. 9 March when the terrestrial radar measurements recorded a significant acceleration of the upper SdF towards the sea (Casagli et al., 2009), while the lava output rate at the vent at 400 m of altitude was decaying. This information suggested the possibility of both a significant intrusion crossing the northern wall of the SdF and/ or destabilizing the SdF at whole; the consequence of these hypotheses for the civil protection activities was to be severe, including also the possible permanent evacuation of the island. The data provided by the deformation monitoring system excluded such hypotheses. The lack of appreciable changes at the SPLB tilt signals, indeed, conflicts with the presence of an intruding dike below the northern wall of the SdF, while the results of terrestrial geodetic measurements, indicating that the central part of the SdF was not affected by sliding acceleration, excluded the possibility of forming a main portion of landslide involving also the central part of the SdF. Taken together, these measurements would indicate that the deformation phenomenon was localized only in the upper part of the SdF, where indeed in the afternoon of March, 9 a new vent opened and emitted lava for a few hours. Besides the monitoring, which provided information on the deformation state of the volcano in different areas such as summit craters, SdF and lava fan, the main aspect was that tilt and GPS data for the first time shed light on a deflation pattern which accompanied the first days of the eruption and allowed to model the source geometry and position. Therefore, tilt and GPS changes recorded in the initial phase of the 2007 eruption represent the first significant movements recorded during an effusive phase, and for the first time, describe a clear deflating process which was modelled to infer the position and geometry of the source (Bonaccorso et al., 2008). This elongated source, located under the crater area at a 2.8 km depth (Fig. 9), would represent the intermediate storage connecting the deeper LP magma storage (depth from 5 to 10 km), producing Low Porphyritic (LP) magma erupted during the paroxysm as inferred by petrological studies (Bertagnini et al., 2003; Francalanci et al., 2005, Di Carlo et al., 2006), with the HP shallower storage (0.8–3 km) i.e. the part where gas slugs expand along the final conduit (Burton et al., 2007) feeding the ordinary strombolian activity and the recent effusive eruptions with High Porphyritic (HP) magmas. Many composite volcanoes are fed with magma through double magma chambers, the shallow chamber normally being much smaller than the deep source reservoir (Gudmundsson, 2006). In our case, the modelled source would represent the connection between these two sources and suggests that Stromboli volcano is characterized by a relatively near free pathway for magma ascent from the deep to the shallow source. Therefore the contribution of the deformation modelling improved the information on the shallow-intermediate Fig. 9. Three-dimensional sketch model of the vertical elongated plumbing system beneath Stromboli inferred from the deflation recorded during the initial phase of the 2007 eruption by modeling GPS and tilt data. Author's personal copy 180 A. Bonaccorso et al. / Journal of Volcanology and Geothermal Research 182 (2009) 172–181 Fig. 10. Configuration of the new network of benchmarks installed on the 2007 lava fan and example of deformations measured at benchmarks located at its extremities (i.e., close to the wall and toward the center of the SdF). The plotted deformations are the daily average variations of the three components (North, East and Height) with respect to the first measurement (6 April 2007). plumbing system. The origin of the depressurization of the modelled source, which provoked the observed deflation, is the rapid and strong magmastatic pressure decay due to the lava effusion that is coherent with the very high effusion rate of the first days of the eruption (Calvari et al., 2008). The tilt recorded at PLB would confirm this aspect since its variation started simultaneously to the opening of the fissures; this implies that the depressurization from the summit craters down to the deep source (that producing the measured deflation) was very fast. At Stromboli seismicity is usually very shallow (b1 km) and associated with the ordinary eruptive-explosive activity. However, the seismicity would suggest that the eruption was preceded by a refilling of the deep plumbing system as indicated by a small seismic swarm of 5 volcano–tectonic events (two of these with M N 3) occurring on the island of Stromboli between 10 April and 22 May 2006. These events, having hypocentre locations clustered below the volcano edifice at a depth of about 5–6 km (D'Auria et al., 2007), occurred during a period of increased explosive activity and were followed by two major explosions on May 22 and on 16 June. An interesting point is that the deformation did not detect significant changes in the period before the eruption. This could due to two main reasons. First, the recharge would produce small changes (inflation) that in the middle term are difficult to detect, while the same magnitude changes are readily observed if occurring in a short period (i.e., rapid deflation). Secondly, the emerged part of the volcano is much smaller than the under sea part, where deformation cannot be measured, and the deeper pressurizing storage would not produce a detectable deformation in the small upper portion of the volcano edifice. In conclusion, the ground deformation multi-parametric monitoring proved a powerful tool to monitor different volcanic activities (fracture propagation, sliding of unstable sectors, and deflation of the overall volcano) and contributed to a correct evaluation of the ongoing phenomena. Moreover, the modelling of the deflation, observed for the first time at Stromboli, allowed inferring the position and geometry of an elongated depressurizing source which improved the understanding of the shallow-intermediate plumbing system. Acknowledgments We thank A. Gudmundsson and the other anonymous referee for their reviews that led to substantial improvements of the manuscript. We are indebted to the technical staff of Ground Deformation Team of INGV-CT, who ensures the regular working of the monitoring networks; in particular we would like to thank O. Campisi, M. Cantarero, A. Ferro, B. Puglisi and M. Rossi. The authors are also grateful to P.G. Scarlato, I. Aquino and the many other INGV colleagues and researchers of other institutions involved in the Stromboli emergency for the stimulating professional discussions and personal support during the tasking days of the eruption. We also thank S. Conway for correcting and improving the English of this paper. Author's personal copy A. Bonaccorso et al. / Journal of Volcanology and Geothermal Research 182 (2009) 172–181 The authors are indebted to the National Department of Civil Protection (DPC) team working at the Advanced Operative Centre at S. Vincenzo Observatory for their support in managing the complex monitoring system during the eruption. Special thanks are also due to M. Zaia (Zazà) and the other Volcanological Guides, the “Guardia di Finanza” staff and the helicopter pilots and technicians who aided the difficult fieldwork during the eruption and in particular the installation of the new reflectors on the lava fan. References Barberi, F., Rosi, M., Sodi, A., 1993. Volcanic hazard assessment at Stromboli based on review of historical data. Acta Vulcanol. 3, 173–187. Bertagnini, A., Métrich, N., Landi, P., Rosi, M., 2003. Stromboli volcano (Aeolian Archipelago, Italy): an open window on the deep-feeding system of a steady state basaltic volcano. J. Geophys. Res. 108 (B7), 2336. doi:10.1029/2002JB002146. Bonaccorso, A., 1998. Evidence of a dyke-sheet intrusion at Stromboli volcano inferred through continuous tilt. Geophys. Res. Lett. 25 (22), 4225. Bonaccorso, A., Gambino, S., Guglielmino, F., Mattia, M., Puglisi, G., Boschi, E., 2008. Stromboli 2007 eruption: deflation modeling to infer shallow-intermediate plumbing system. Geophys. Res. Lett. 35. doi:10.1029/2007GL032921. Burton, M.R., Allard, P., Muré, F., La Spina, A., 2007. Magmatic gas composition reveals the source depth of slug-driven strombolian explosive activity. Science 317, 227–230. Calvari, S., Lodato, L., Steffke, A., Cristaldi, A., Harris, A.J.L., Spampinato, L., Boschi, E., The 2007 Stromboli flank eruption: chronology of the events, and effusion rate measurements from thermal images and satellite data. Submitted for publication to G3, December 2008. Casagli, N., Tibaldi, A., Merri, A., Del Ventisette, C., Apuani, T., Guerri, L., Fortuny-Guasch, J., Tarchi, D., 2009. Deformation of Stromboli Volcano (Italy) during the 2007 eruption revealed by radar interferometry, numerical modelling and structural geological field data. J. Volcanol. Geotherm. Res. 182, 182–200. doi:10.1016/j.jvolgeores.2009.01.002. D ' Auria, L., Giudicepietro, F., Martini, M., Orazi, M., 2007. The April–May 2006 volcano– tectonic events at Stromboli volcano (Southern Italy) and their relation with the magmatic system. Earth-prints Repos. (http://hdl.handle.net/2122/1506). 181 Di Carlo, I., Pichavant, M., Rotolo, S.G., Scaillet, B., 2006. Experimental crystallization of a High-K arc basalt: the golden pumice, Stromboli volcano (Italy). J. Petrol. 1–27. doi:10.1093/petrology/eg1011. Francalanci, L., Davies, G.R., Lustenhouwer, W., Tommasini, S., Mason, P.R.D., Conticelli, S., 2005. Intra-Grain Sr isotope evidence for crystal recycling and multiple magma reservoirs in the recent activity of Stromboli volcano, Southern Italy. J. Petrol. 46 (10), 1997–2021. doi:10.1093/petrology/egi045. Gudmundsson, A., 2006. How local stresses control magma-chamber ruptures, dyke injections and eruptions in composite volcanoes. Earth Sci. Rev. 79, 1–31. Herring, T.A., 2000. GLOBK: Global Kalaman Filter VLBI and GPS Analysis Program Version 4.10. Mass. Inst. Of Technol., Cambridge, Mass. King, R.W., Bock, Y., 2000. Documentation for GAMIT GPS Analysis Software Version 10.01. Massachusetts Institute of Technology and Scripts Institution of Oceanography, Cambridge Mass. Marotta, E., Calvari, S., Cristaldi, A., Di Vito, M., Lodato, L., Orsi, G. Modalities of explosive activity resumption at the summit craters of Stromboli volcano after the end of the 2007 flank eruption. Submitted for publication to G3, December 2008. Mattia, M., Rossi, M., Guglielmino, F., Aloisi, M., Bock, Y., 2004. The shallow plumbing system of Stromboli Island as imaged from 1 Hz instantaneous GPS positions. Geophys. Res. Lett. 31, L24610. doi:10.1029/2004GL021281. Mattia, M., Aloisi, M., Di Grazia, G., Gambino, S., Palano, M., Bruno, V., 2008. Geophysical investigations of the plumbing system of Stromboli volcano (Aeolian Island, Italy). J. Volcanol. Geotherm. Res. 176 (4), 529–540. doi:10.1016/j.jvolgeres.2008.04.022.. Neri, M., Lanzafame, G., Acocella, V., 2008. Dyke emplacement and related hazard in volcanoes with sector collapse: the 2007 Stromboli (Italy) eruption. J. Geol. Soc. 165, 883–886. doi:10.1144/0016-76492008-002. Nunnari, G., Spata, A., Puglisi, G., 2008. A warning system for Stromboli volcano based on statistical analysis. In: Díaz, J.I., Fernandez, J., Camacho, A. (Eds.), Earth Sciences and Mathematics. Pure Appl. Geoph., vol. 165, p. 8. doi:10.1007/s00024-008-0392-y. Pasquarè, G., Francalanci, L., Garduno, V.H., Tibaldi, A., 1993. Structure and geologic evolution of the Stromboli volcano, Aeolian Islands, Italy. Acta Vulcanol. 3, 79–89. Puglisi, G., Bonaccorso, A., Mattia, M., Aloisi, M., Bonforte, A., Campisi, O., Cantarero, M., Falzone, G., Puglisi, B., Rossi, M., 2005. New integrated geodetic monitoring system at Stromboli volcano (Italy). Eng. Geol. 79, 13–31. Tibaldi, A., 1996. Mutual influence of dyking and collapses at Stromboli volcano, Aeolian Arc, Italy. Volcano instability (Spec. Publ.) Geol. Soc. Lond. 110, 55–63. Tibaldi, A., 2001. Multiple sector collapses at Stromboli volcano, Italy: how they work. Bull. Volcanol. 63, 112–125. doi:10.1007/s004450100129.