GEOPHYSICAL RESEARCH LETTERS, VOL. 30, NO. 23, 2214, doi:10.1029/2003GL018896, 2003 Etna 2002 eruption imaged from continuous tilt and GPS data M. Aloisi, A. Bonaccorso, S. Gambino, M. Mattia, and G. Puglisi Istituto Nazionale di Geofisica e Vulcanologia Sezione di Catania, Piazza Roma, Italy Received 22 October 2003; accepted 12 November 2003; published 10 December 2003. [1] On the night of October 26, 2002, intense explosive activity and lava effusion began suddenly on the southern flank of Mt. Etna at an altitude of 2750 m. During the 27 and 28 October, a long field of eruptive fractures propagated radially along the north-eastern flank of the volcano. Ground deformation changes recorded between 26 and 27 October from GPS and tilt data collected at the permanent geodetic network of Mt. Etna, were modeled to infer the positions and dimensions of the two dikes. The observed deformation pattern was consistent with a response of the edifice to a composite mechanism consisting of a vertical uprising dike in the upper Southern flank and a lateral intrusion propagating along the north-eastern sector. The first dike, which triggered the eruption, crossed the volcano edifice in a few hours and was located in the same area as the 2001 eruption, while the second lateral dike, which crossed the NE flank, was the primary cause of the recorded INDEX TERMS: 1204 Geodesy and deformation pattern. Gravity: Control surveys; 5104 Physical Properties of Rocks: Fracture and flow; 8414 Volcanology: Eruption mechanisms; 8419 Volcanology: Eruption monitoring (7280). Citation: Aloisi, M., A. Bonaccorso, S. Gambino, M. Mattia, and G. Puglisi, Etna 2002 eruption imaged from continuous tilt and GPS data, Geophys. Res. Lett., 30(23), 2214, doi:10.1029/2003GL018896, 2003. 1. Introduction [2] The eruptions of Etna during the last 30 years are mainly related to two primary regional NNW and NE trending structural trends. On the flanks of the volcano these trends are characterized by eruptive fissures and parasitic pyroclastic cones that have formed over the past thousand years and they form the so-called S and NE rifts (Figure 1). [3] After the huge lava output (ca. 240  106 m3) of the 1991 – 93 eruption, an intense recharge period characterized Mt. Etna from 1994 to 2001 [Puglisi et al., 2003]. Throughout this period, the volcano was characterized by frequent explosive activity at the central craters, summit lava flows and strong seismic swarms (January 1998 and April 2001) on the western flank. On July 17, 2001 a lateral eruption occurred on the southern flank [INGV-CT, 2001]. The dike emplacement occurred over a period of five days; the ground deformation changes recorded by the permanent tilt and GPS networks infer a nearly N-S oriented vertical tensile dislocation, which crossed the volcano edifice on the high southern flank in coincidence with the seismicity zone [Bonaccorso et al., 2002]. During the dike emplacement a 7-km-long field of fissures, striking N-S, opened on the southern flank from 2750 m Copyright 2003 by the American Geophysical Union. 0094-8276/03/2003GL018896$05.00 SDE to 2000 m. The main lava flow came out from the lowest section of the fracture and ceased on August 10, while an explosive cone formed close to the upper limit of the fracture at 2600 m during the course of the eruption. After 15 months, on the night of 26 October 2002, an intrusion suddenly occurred in the same area as the 2001 intrusion. This time the intrusion propagated much faster than 2001, and occurred in only a few hours as tracked by the associated seismicity and the tilt changes. An explosive cone formed on the high southern flank at 2750 m, close to the explosive cone of the 2001 eruption. The southern dike intrusion produced only a limited fracture field and afterward a crater at 2850 m, confirming a vertical propagation. During the first hours of October 27, a 6 – 7 km long dike propagated from the Northeast crater along the NE rift, intersecting the surface for about 4 km from 2500 m to 1600 m. Twenty small explosive cones formed along its path over a period of about 25 hours demonstrating a clear lateral propagation. About 10  106 m3 of lava erupted in the following 7 days from the NE fracture. Meanwhile, the intrusion on the southern flank continued to feed the explosive cone with spectacularly steady activity (alternating from lava fountains and intense strombolian phases) for nearly two months and with lava effusion that lasted for nearly three months. [4] In this paper we present our observations of ground deformation during the eruption onset period 26 –27 October, recorded through tilt and GPS permanent stations and also we provide source models constrained by these data. We show that this eruption was eccentric in the sense that the vertical intrusion feeding the southern flank eruption was not related to the central conduit feeding system, but rather to vertical dike propagation through the volcano edifice in the same area as the 2001 dike. 2. Networks [5] Continuous ground deformation monitoring at Mt. Etna is performed with permanent tilt and GPS networks (Figure 1). At present the permanent tilt network includes nine bi-axial instruments installed in shallow boreholes at 3– 9 m depth and one fluid long-base instrument [Bonaccorso et al., 2003]. Each tiltmeter is oriented so that one of the perpendicular axes is directed toward the crater (radial component) and a positive signal variation means crater up. The second component is named tangential and a positive signal variation means uplift in the anticlockwise direction. Data are recorded each half-hour for borehole stations and every 10 minutes for the long-base and transmitted to Catania via radio-link. [6] In November 2000 the ground deformation continuous monitoring was upgraded with the installation of the permanent GPS network. The configuration of 13 stations 8 - 1 SDE 8-2 ALOISI ET AL.: ETNA 2002 ERUPTION IMAGED Figure 1. Map of Mt. Etna with the permanent tilt and GPS networks and the 2002 lava flow fields. The zones of the S and NE volcano rifts are indicated. The arrow indicates the sliding direction of the eastern sector [Bonforte and Puglisi, 2003]. mation as measured by EDM, GPS and tiltmeters (Figures 2 and 3). All the tilt stations showed variations (Figure 3); changes of tens of microradians were detected on northern and western flanks while minor variations (a few microradians) characterized the southern stations responses. The PDN long-base station recorded very large changes of about 150 mrad. [8] Different stations recorded tilt changes starting at different times and also showing different durations. From 21:00 to 24:00 GMT on 26 October (grey area in Figure 3) tilt changed on stations close to the south fracture area (PDN, MDZ, CDV and MSC) which was simultaneously affected by seismicity. At 00:10 GMT on October 27 large tilt variations were observed primarily on the summit area and on the northern flank of the volcano as the fracture propagated along the NE Rift. Signals of the northern-east stations (MNR and DAM) revealed that the intrusive process, which propagated for about 4 km and formed about 20 small explosive cones along its path, stopped at about 03:00 GMT on 28 October. [9] Finally, some signals (MSC, MMT, MEG, MDZ) were affected by co-seismic noises related to more energetic earthquakes (the seismic network recorded five events with M  3.5, Mmax = 4.2) accompanying the dike propagation. DAM, MNR, CDV and the PDN long-base station did not show co-seismic interference and therefore demonstrate a clear time evolution of the process. was achieved in the late spring-summer of 2001 [Bonaccorso et al., 2002]. 3. Double Intrusion and Recorded Data [7] The 2002 Mt. Etna eruption started late on the night of 26 October. The eruption caused marked ground defor- Figure 2. Example of selected baseline length variations on the GPS permanent network. This figure clearly shows the sudden variations related to the fast emplacements of the dikes between 26 and 27 October 2003. Figure 3. Tilt signals recorded between 26 and 29 October 2002. The grey area shows the 21:00 –24:00 GMT (of the 26th) time interval in which only tilt stations close to the southern fracture (CDV, MSC, PDN) showed changes. The time of cessation of variations used to model ground deformation sources is also reported. EC10 and EMDZ recorded small variations (1 – 5 mrad) that are not visible at this scale. SPC was not functioning. ALOISI ET AL.: ETNA 2002 ERUPTION IMAGED Figure 4. Modeled and recorded deformation pattern from the ground deformation permanent networks. The recorded vectors are obtained from the interval comparison 26– 27 October. Tilt vectors show the direction of ground uplift. [10] Variations in slope distances between benchmarks of the GPS networks were measured in coincidence with the fast uprising of magma, clearly indicating the occurrence of a wide ground deformation field at the scale of the entire volcano edifice (Figure 2). GPS displacement vectors are shown in Figure 4 and display a clear pattern almost symmetrical to a NE-SW direction, suggesting an important role for a dislocation source along this direction. The vertical component showed a large positive value in the EPDN summit station and a large negative value in the ETDF summit station, while at the remaining stations the height variations are very small or negligible. 4. Modeling [11] The data recorded between 26– 27 October reflect the cumulated effects of a complex mechanism of intrusion on both the southern and north-eastern flank of the volcano. In this manuscript we model only the static offsets; we inverted the tilt components collected by all the permanent stations except PDN and MEG. The PDN tilt signals went off-scale during the 27th, so that we suspect that they do present a lower limit of the cumulative effect of the intrusion. We did not use MEG whose marked tilt variations are currently not clear. Furthermore, we used the GPS horizontal and vertical changes recorded by the Etna permanent network in the time interval 26– 27 October astride the eruption onset. We did not take into account the displacements collected by ETDF station since this benchmark was strongly affected by the near field of the surface fracture system and cannot therefore assume elastic deformation. We did not use the dislocation recorded by EPDN station because the GPS data are not available in the 26– 29 October interval, due to technical problems. [12] In order to image the observed deformation pattern we performed an analytical inversion of a model composed by two tabular dislocation models [Okada, 1985]. We estimated 10 parameters for each dislocation structure: coordinates of the top, dimensions (length and width), SDE 8-3 orientation (azimuth and dip) and displacements (strike slip, dip slip, opening). In this final modeling, the dislocation shear components on the southern structure are kept fixed at zero because the inversions always produced negligible values for these parameters. The inversion is performed using a classic least squares approach. In all the trials, even using different starting models, we always obtained similar solutions. In order to verify the stability of our solution and to solve relative minimum problems we also repeated the analytical process for thousands of models in the space of the parameters with a grid-search technique, testing these results by the c2 test. [13] The modeling results indicated the response to a composite mechanism that included a vertical uprising dike in the upper southern flank and a radial intrusion in the NE sector (Figure 5). The first near vertical tensile dislocation, which triggered the eruption, crossed the volcano edifice in only a few hours and was located in nearly the same area as the 2001 eruption. A shallow tensile dislocation with an opening component of 1.5 m can explain the deformation pattern recorded during the first hours (Table 1). The second lateral dike, crossing the NE flank, was a shear-tensile dislocation with opening and shear component of 1.0 m and 0.5 m, respectively (Table 1). The fit is good and we have a reduced chi-squared c2 equal to 1.0 considering the a-posteriori standard deviations of 0.025 m and 0.030 m for the horizontal and vertical displacement and 8.0 mrad for the tilt. Our model cannot explain all the details in the observations because we use a simple tabular dislocation model [Okada, 1985] in a simple elastic half-space without heterogeneity. Therefore the uncertainty (standard deviation) is more than the instrumental error and the remaining misfit is because a simple rectangular crack cannot completely explain the details in the observations. It’s noteworthy that if we image the recorded deformation pattern using the second lateral dike only, we obtain the near same c2; therefore, the north-eastern intrusion was the primary cause of the recorded deformation pattern, while the southern dike gave only a minimal contribution. [14] We believe that a width of 4.6 km is large for the intrusion propagated along the NE radial direction, since the sliding observed on the eastern flank [Bonforte and Puglisi, Figure 5. 3-D map showing the modeled sources obtained from the inversion of the recorded data. The main calculated values related to the features of the dikes obtained are shown. The NE structure is separated in two parts to indicate that the large width obtained is probably amplified due to the E flank sliding (see text). SDE 8-4 ALOISI ET AL.: ETNA 2002 ERUPTION IMAGED Table 1. Estimated Models Parameters and Uncertainties North Eastern intrusion Xcentre = 500880.5 ± 300 m Ycentre = 4182109.9 ± 300 m Zcentre = 0.0 ± 0.1 km Azimuth = N 24.4 ± 1.4 Dip = 79.0 ± 3 Southern intrusion Xcentre = 500897.4 ± 500 m Ycentre = 4176077.3 ± 500 m Zcentre = 1.0 ± 0.4 km Azimuth = N 17.6 ± 4.7 Dip = 80.5 ± 3.8 Length = 6.6 ± 0.4 km Width = 4.6 ± 0.6 km Opening = 1.0 ± 0.1 m Strike slip = 0.5 ± 0.1 m Dip slip = 0.0 ± 0.1 m Length = 1.6 ± 0.4 km Width = 1.8 ± 0.6 km Opening = 1.5 ± 0.4 m Strike slip = 0.0 m (fixed) Dip slip = 0.0 m (fixed) The reference surface (z = 0) has been assumed at 1500 m, namely the mean elevation of points belonging to the permanent networks. The projection is UTM-WGS84. 2003; see Figure 1] probably amplified the deformative effects in this sector. So we suspect that we obtained a larger width because the model has to take two effects into account: (1) the radial intrusion and (2) the E flank sliding. The observed and calculated horizontal displacement vectors are reported in Figure 4 together with the position and geometry of the modeled dikes. The earthquakes recorded in the time interval 26– 28 October (Figure 4) fit well with the position of the dislocation sources which provoked the main static deformation at the surface. 5. Discussion [15] The previous 2001 eruption was different from other preceding lateral eruptions modelled in the last twenty years [Bonaccorso, 2001]. It was a fast vertical dike emplacement that crossed the entire volcanic edifice below the summit craters area. It occurred after eight years of intensive re-charging but it emitted a small lava output. Moreover, the volume of the modelled dike (ca. 16  106 m3) resulted smaller than the estimated volume of lava output (ca. 48  106 m3). This discrepancy between expected injection volume and lava output was interpreted as a possible evidence of a supply volume from a deeper region [Bonaccorso et al., 2002]. The main outstanding problem is the meaning of this rapid intrusion and what kind of link it has with a possible deeper source. [16] The 2002 vertical emplacement dike occurred again in the same area of the southern flank as the 2001 intrusion and the final uprising was even faster; in fact the accompanying seismicity lasted only a few hours (see INGV-CT web page at http://www.ct.ingv.it). Therefore, the 2002 eruption on the southern flank can be interpreted as the continuation of the 2001 eruption. This is confirmed both by the same area of the eruptive fractures of the two eruptions and by the similar lava composition of the 2002 [M. Pompilio, pers. comm.] compared with the 2001 eruption [Taddeucci et al., 2002]. [17] The main difference between the 2002 and 2001 eruption is the eruptive fracture in the north-eastern flank due to the magma intrusion which propagated laterally from summit craters along the NE rift, pouring out a small lava volume (ca. 10  106 m3) during the following week, although it produced a major part of the ground deformation. [18] Given this small lava volume and since preliminary petrologic data [M. Pompilio, pers. comm.] suggest that the magma feeding the NE fracture was different from the southern, it is reasonable to suppose that the NE fracture system drained the conduits and the very shallow plumbing system. The large ground deformations observed on the NE flank of the volcano are due to the capacity to discharge the stress field induced by the intrusion along the existing structural system, through the entire volcano. [19] However, the longer effusive and explosive phenomena that took place on the southern flank with nearly two months of permanent explosive activity from the cone at 2750 m and nearly three months of lava output from the base of this cone, with a final emitted lava and pyroclastic volume estimated to be 35  106 m3 [S. Calvari, pers. comm.] confirm that the southern dike intrusion is probably the most significant volcanic event of this eruption. [20] For the 2002 eruption the modelled dike volume was much smaller (ca. 4  106 m3) than 2001 and the discrepancy between the modelled dike volume and lava output was more marked. These aspects suggest two important considerations. First, the southern area, already fractured and weakened by the 2001 strong intrusion, favoured an easier and faster final dike emplacement with associated smaller deformation. Second, the 2002 eruption provides further evidence of a magma supply from a deeper storage, which may be consistent with the results of some recent tomography investigations at Mt. Etna [e.g., Chiarabba et al., 2000]. These investigations do not prove the existence of a magma chamber of great size beneath the volcano, instead they show the presence of a central high Vp body, which is interpreted as a solidified intrusive body. Patanè et al. [2003] suggested that the deeper part of the high Vp body represents the plumbing system still used by the magma during its ascent from the mantle source. Instead, the upper part (3 – 5 km depth) of the high Vp body is a region of magma storage. The vertical dike mechanism, like the 2001 eruption, is consistent with over-pressure in the deeper source, which fractures the upper storage limit at about 3 km b.s.l. and causes magma to intrude upwards. We suggest that the 2002 eruption also followed this mechanism and that it represented a continuation of the 2001 eruption. In the last two years we have witnessed a cycle involving pressurization of the plumbing system indicated by the tomography studies and dikes departing from the upper part of it. Therefore, from this plumbing system a cycle opened which involved eccentric vertical dikes emplacing in the southern flank. In the near future the challenge of ground deformation monitoring will be to determine if new magma is still accumulating beneath the volcano in this storage zone. References Bonaccorso, A., Mt. Etna volcano: Modelling of ground deformation patterns of recent eruptions and considerations on the associated precursors, J. Volcanol. Geotherm. Res., 109, 99 – 108, 2001. Bonaccorso, A., M. Aloisi, and M. Mattia, Dike emplacement forerunning the Etna July 2001 eruption modeled through continuous tilt and GPS data, Geophys. Res. Lett., 29(13), 1624, doi:10.1029/2001Gl014397, 2002. Bonaccorso, A., O. Campisi, G. Falzone, and S. Gambino, Continuous Tilt Monitoring: Lesson Learned From 20 Years Experience at Mt. Etna, in The Mt. Etna Volcano, (Geophys. Monogr. Ser.), edited by S. Calvari, A. Bonaccorso, M. Coltelli, C. Del negro, and S. Falsaperla, in press, 2003. Bonforte, A., and G. Puglisi, Magma uprising and flank dynamics on Mt. Etna volcano, studied by GPS data (1994 – 1995), J. Geophys. Res., 108(B3), doi:10.1029/2002JB001845, 2153, 2003. Chiarabba, C., A. Amato, and E. Boschi, Recent seismicity and tomographic modeling of the Mount Etna plumbing system, J. Geophys. Res., 105(B5), 10,923 – 10,938, 2000. ALOISI ET AL.: ETNA 2002 ERUPTION IMAGED INGV-CT, Istituto Nazionale di Geofisica e Vulcanologia-Sezione di Catania, Multidisciplinary approach yields insight Mt. Etna eruption, Eos, Trans., AGU, 82(52), 653 – 656, 2001. Okada, Y., Surface deformation due to shear and tensile faults in half-space, Bull. Seismol. Soc. Am., 75, 1135 – 1154, 1985. Patanè, D., P. De Gori, C. Chiarabba, and A. Bonaccorso, Magma Ascent and the Pressurization of Mount Etna’s Volcanic System, Science, 299, 2061 – 2063, 2003. Puglisi, G., P. Briole, and A. Bonforte, Twelve years of ground deformation studies on Mt. Etna volcano based on GPS surveys, in The Mt. Etna SDE 8-5 Volcano (Geophys. Monogr. Ser.), edited by S. Calvari, A. Bonaccorso, M. Coltelli, C. Del negro, and S. Falsaperla, in press, 2003. Taddeucci, J., M. Pompilio, and P. Scarlato, Monitoring the explosive activity of the July – August 2001 eruption of Mt. Etna (Italy) by ash characterization, Geophys. Res. Lett., 29(8), 1230, doi:10.1029/2001GL014372, 2002. M. Aloisi, A. Bonaccorso, S. Gambino, M. Mattia, and G. Puglisi, Istituto Nazionale di Geofisica e Vulcanologia Sezione di Catania, Piazza Roma, 2, 95123, Catania, Italy. (aloisi@ct.ingv.it)