Seismotectonics of the Nubia plate compressive margin in the south Tyrrhenian region, Italy: Clues for subduction inception
Abstract
[1] Seismological and structural data are used to constrain the active tectonics of the Nubia plate compressive margin in southern Italy. In this region, compressional displacements have resumed at the rear of the Maghrebian orogenic wedge since about 700–500 ka and are presently accommodated along a seismic belt in the south Tyrrhenian area, where a passive margin has developed during late Neogene-Quaternary times. Earthquake data from the south Tyrrhenian belt are analyzed with Bayloc, a probabilistic nonlinear location method. Results show that the seismic belt is segmented and involves a series of NE-SW and NW-SE elongated clusters. The NE-SW clusters are interpreted as high-angle reverse fault zones verging toward the southeast, whereas the perpendicular clusters are interpreted as possible strike-slip fault zones transferring the contractional displacements to advanced segments of the belt in the southeast. Brittle deformations observed in the Quaternary volcanic island of Ustica are consistent with the geometry and kinematics of the seismic belt deduced from the seismological data. The south Tyrrhenian active belt may constitute an early stage of subduction of the Tyrrhenian oceanic crust beneath Sicily. A similar scenario has been hypothesized also for the westward prolongation of the south Tyrrhenian belt off the Algerian coast. By integrating instrumental and historical seismic data with tectonic evidences, we infer that the south Tyrrhenian belt is probably capable of producing earthquakes with a maximum size close to magnitude 7. The analyzed data suggest a multifault system behavior for the studied compressional belt. This inference is important to estimate and mitigate the seismic and tsunamic hazards in such a densely populated region.
1. Introduction
[2] The tectonic mechanisms leading to the conversion of a passive margin into an active one with subsequent subduction of oceanic lithosphere have been widely studied [Dickinson and Seely, 1979; Cloething et al., 1982, 1989; Muller and Phillips, 1991; Erickson, 1993; Toth and Gurnis, 1998; Faccenna et al., 1999; House et al., 2002]; however, their knowledge is still substantially limited because of the absence of obvious, present-day examples of trench initiation. At convergent margins, in fact, the accretion of allochthonous terranes usually masks the early compressional structures including evidences of subduction inception. On the other hand, examples of compressive deformations at passive margins are documented only in a few sites. These include the region off Iberia in the Atlantic Ocean [Cloething et al., 1990; Masson et al., 1994, Alvarez-Marron et al., 1997], the British Isles [Ziegler et al., 1995], the Scotian basin in eastern Canada [Erickson and Arkani-Hamed, 1993], and the Fiordland Region in New Zealand [House et al., 2002].
[3] In this paper, we document the geological and seismological features of an incipient compressional belt developing along a late Neogene-Quaternary passive margin in the central Mediterranean region. In this region (Figure 1), since the Oligocene time at least, the collision between Nubia and Eurasia has produced the north-northwestward subduction of the Calabrian slab for about 400 km and the associated growth of the south verging Maghrebian orogen [Anderson and Jackson, 1987; Faccenna et al., 2004; Rosenbaum and Lister, 2004]. Since about late Miocene time, this process was accompanied by back-arc extension in the modern Tyrrhenian Sea area [Malinverno and Ryan, 1986]. Geological evidence has shown that thrusting at the front of the Maghrebian fold-thrust belt (i.e., in the southern Sicily and in the Sicily Channel) mostly ceased by mid-Pleistocene time [Lickorish et al., 1999]. Seismological and geodetic (GPS) evidences have shown, however, that Nubia and Eurasia are still converging and that the relative contractional displacements are being mainly accommodated at the rear of the Maghrebian belt (Figure 1), namely along an E-W trending compressive belt in the southern sector of the Tyrrhenian Sea [Goes et al., 2004; Montone et al., 2004; Pondrelli et al., 2004; Neri et al., 2005], where the passive margin of Sicily has developed during late Neogene-Quaternary times. Despite several studies on the active tectonics of the south Tyrrhenian compressive belt, the architecture and kinematics of this belt is still mostly unconstrained. In most studies, for instance, the south Tyrrhenian belt is considered as dominated by active, hinterland-verging (i.e., northward) thrusts [e.g., Goes et al., 2004], this tectonic setting implying underthrusting and, possibly, future subduction of the Tyrrhenian oceanic crust toward the south. In contrast, on the basis of seismic reflection data, Pepe et al. [2005] suggested a southward vergence for the south Tyrrhenian active compressive belt.
[4] The aim of this paper is to constrain the active tectonics of the Nubia plate compressive margin in the south Tyrrhenian area. To do so, we analyze a set of earthquake data with Bayloc [Presti et al., 2004], a probabilistic nonlinear location method. A set of structural data from the Quaternary volcanic island of Ustica (Figure 1) are also presented and compared with the seismological data. The analyzed data bear important insights into the past evolution and future scenarios of the Nubia plate margin in the central Mediterranean area. The presented evidences may be interpreted as clues for a very incipient subduction of the Tyrrhenian oceanic lithosphere beneath the Maghrebian orogen. A comparison with tectonic studies on the same compressive margin in a more western location (i.e., off the Algerian coast [Déverchère et al., 2005; Domzig et al., 2006; Stich et al., 2006]) sheds light on the lateral changes of style and activity of the Mediterranean subduction complex. Moreover, Cinti et al. [2004] indicated the compressive margin in the southern Tyrrhenian area as a probable (i.e., probability = 5–10%) site for a M ≥ 5.5 earthquake by about 2015 [see also Jenny et al., 2006]. In this framework, detecting the architecture of the seismogenic apparatus is necessary to assess and mitigate the seismic and tsunamic hazards in this densely populated region.
2. Geological Setting
[5] The study area is located in the southern sector of the Tyrrhenian Sea, at the transition between the Maghrebian fold-thrust belt, in the south, and the Tyrrhenian back-arc oceanic basin, in the north (Figure 1). The Moho is 25–30 km deep in the northern coast of Sicily and shallows toward the north at the continent-ocean transition, where it is approximately 10–15 km deep [Neri et al., 2002]. Similarly, the base of the lithosphere shallows from a depth of about 70 km in Sicily to a depth of less than 40 km in the Tyrrhenian area [Ansorge et al., 1992].
[6] In Sicily, the fold-thrust belt is traditionally divided into two main belts, namely the inner (northern) Kabilian-Peloritan-Calabrian and the outer (southern) Sicilian-Maghrebian belts [Catalano et al., 1996]. The Sicilian-Maghrebian belt involves imbricate sheets of Mesozoic–early Tertiary carbonate rocks, whereas the Kabilian-Peloritan-Calabrian belt includes imbricate sheets of Paleozoic metamorphic and igneous rocks, and Mesozoic sedimentary covers. The Kabilian-Peloritan-Calabrian and the Sicilian-Maghrebian belts form a south vergent orogenic wedge hereafter named Maghrebian fold-thrust belt or Maghrebides. The Maghrebian fold-thrust belt grew upon the Meso-Cenozoic sequences of the African margin during Oligocene–middle Pleistocene times [Lickorish et al., 1999]. Continental collision occurred during Oligocene–early Miocene times and caused the tectonic superposition of the Kabilian-Peloritan-Calabrian belt onto the rock succession of the Sicilian-Maghrebian belt, which mostly formed during Miocene time. The tectonic contact between the Kabilian-Peloritan-Calabrian belt and the underlying Sicilian-Maghrebian belt is a N dipping, S vergent, regional thrust (i.e., also known as the Drepano Thrust [Pepe et al., 2005]), which cuts across the northeastern Sicily and the southern Tyrrhenian area (see inset in Figure 1). Off the northern Sicilian coast, the location of the Drepano Thrust (Figure 1) is uncertain because drawn by interpolating a set of seismic profiles.
[7] Back-arc extension started during late Tortonian time and lasted until early Messinian time [Pepe et al., 2000]. Contraction resumed at the rear of the orogenic wedge during Messinian–early Pliocene times and caused both thrust reactivation and transverse strike-slip faulting [Ghisetti, 1979]. In middle-late Pliocene time, back-arc extension renewed and lithospheric breakup occurred in late Pliocene time. The extensional tectonics led to the formation of the oceanic Tyrrhenian basin during Quaternary time [Faccenna et al., 2004, 2005; Rosenbaum and Lister, 2004; Nicolosi et al., 2006]. Contractional displacements at the front of the Maghrebian fold-thrust belt ceased by mid-Pleistocene time [Lickorish et al., 1999]. Afterward, the southward propagation of the thrust front has been possibly hindered by the thick and buoyant Hyblean-Pelagian continental lithosphere, which has resisted to subduction (Figure 1). Consequently, contractional displacements have shifted to the rear of the belt in the south Tyrrhenian region [Hollenstein et al., 2003; Goes et al., 2004; Pondrelli et al., 2004]. At present, the compressive belt active in this region is laterally (i.e., eastward) bounded by the regional, NNW striking Tindari Fault (Figure 1), along which right-lateral, strike-slip and transtensional displacements are presently accommodated [Billi et al., 2006]. In the east of the Tindari Fault, active NW-SE oriented extension occurs possibly in connection with the rollback of the Calabrian slab [Neri et al., 2003; Montuori et al., 2007]. Toward the north-northwest, the Tindari Fault intersects the Sisifo Fault (Figure 2), a seismic WNW striking structure, which presently accommodates right-lateral strike-slip displacements [Finetti and Del Ben, 1986; Billi et al., 2006].
[8] The Quaternary Aeolian Islands in the southeastern Tyrrhenian Sea (Figure 1) form an island arc whose volcanic activity and products are connected with the subduction of the Calabrian slab [Barberi et al., 1973]. Volcanic activity of the Aeolian volcanoes started about 1.3 Myr ago and is still active in at least three of the seven major islands, namely Lipari (last eruption in 729), Vulcano (last eruptions in 1888–1892), and Stromboli (persistently active). Toward the west, the Quaternary island of Ustica (Figure 1) is the summit of a submarine volcano as high as more than 2 km from the seafloor [Calanchi et al., 1984]. Sodic alkaline basalts are exposed at Ustica, whose magmatism is akin to oceanic island basalts [Trua et al., 2003]. The magmatism of Ustica is possibly connected with mantle return flows around the narrow Calabrian slab [Marani and Trua, 2002; Faccenna et al., 2005; Montuori et al., 2007]. From the age of the exposed rocks, it is inferred that the volcanic activity of Ustica occurred between about 750 and 130 ka [de Vita et al., 1998]. The volcanic activity of Ustica ceased during middle–late Pleistocene time, possibly as a consequence of the onset of the compressional tectonics in the south Tyrrhenian area.
3. Seismological Data and Analysis
[9] The seismic data set used in the present study has been extracted from the earthquake catalog of Istituto Nazionale di Geofisica e Vulcanologia (available at http://www.ingv.it/banchedati/banche.html) and includes P and S wave arrival times of seismic events shallower than 30 km occurred in Sicily during January 1994 to April 2005. We selected the earthquakes with magnitude Md ≥ 3.0 and with a minimum of eight arrival time readings (about 500 events). For a preliminary earthquake location, we used the SIMUL linear algorithm by Evans et al. [1994] and the three-dimensional crustal velocity model proposed for the study region by Barberi et al. [2004]. The epicenter map of Figure 2 shows that, in the studied interval of time, most seismicity of Sicily and surrounding regions occurred in the Etnean active volcanic area and along the south Tyrrhenian compressive belt. The location of this belt is shown with a rectangle in Figure 2.
[10] We focused our attention on the seismic events occurring in the belt and, for this purpose, we used a magnitude threshold of 1.5. We located Md ≥ 1.5 earthquakes in the belt by the Bayloc nonlinear probabilistic method by Presti et al. [2004]. Bayloc uses the arrival times of P and S waves and their velocity models as input data, performs a grid search in the location parameter space, considers the nonlinearity of the location problem, and gives the hypocenter location as a location probability cloud. When drawing cumulative maps or hypocenter cross sections, Bayloc weighs the individual earthquakes inversely to spreading of the respective probability clouds. With respect to linearized methods based on hypocenter point representations, Bayloc was shown to be more effective in the detection of hypocenter trends and seismogenic structures [Presti et al., 2004]. Bayloc uses a Bayesian algorithm to compute the probability density function p(x, y, z) for the spatial coordinates of a seismic event. The spatial distribution of probability density relative to a set of earthquakes is computed as the sum of probability densities of the individual events. Epicentral density maps and hypocentral density cross sections are obtained by integration of cumulative probability densities [Presti et al., 2004; Presti, 2006].
[11] Figure 3 (top) shows the Bayloc epicentral map obtained for the earthquakes of magnitude Md ≥ 1.5 occurring at depths shallower than 30 km in the south Tyrrhenian belt between January 1994 and April 2005 (about 1000 events). Earthquake locations have been computed on a regular three-dimensional grid spaced 2.5 km. In Figure 3, contour levels of probability distribution correspond to 68%, 50% and 25% of earthquake density. Figure 3 (top) shows two main epicentral trends oriented NW-SE and NE-SW. During the studied interval of time, the most active sectors in the belt were: (a) the sector located in the southwest of Ustica and corresponding to the area of the June–July 1998 seismic sequence, which included 80 earthquakes with a maximum magnitude of 5.2; (b) the sector located in the south-southeast of Ustica affected by relatively small phases of seismic activity during August–September 1994 (about 20 earthquakes with Mmax = 3.8); (c) the sector located off the Palermo city and corresponding to the area of the September–December 2002 seismic sequence involving earthquakes with a maximum magnitude of 5.9; (d) the sector including the western Aeolian Islands and corresponding to the Sisifo Fault zone. In the last decade, this fault zone has been characterized by frequent low-magnitude seismic swarms. During 1980 and during the years immediately preceding and following 1980, the Sisifo Fault zone has been moderately active with seismic sequences characterized by earthquakes with a maximum magnitude of about 6 [Neri et al., 1996, 2003].
[12] Figure 3 (bottom) displays vertical cross sections perpendicular to the long axes of individual clusters detected in Figure 3 (top). For sectors a to d in Figure 3, the percentage of events falling in the upper 10 km is as follows: a = 81%, b = 78%, c = 93% and d = 83%. High-angle northward dip of hypocenter distribution is a clear common feature of vertical sections a to d.
[13] Because permanent ocean bottom seismometers (OBS) are absent in the study area, the geometry of the seismic network available for earthquake location is not optimal and, in some cases, almost critical (Figure 2). A first level of check of locations used in drawing maps and cross sections was guaranteed by the Bayloc feature that weighs earthquakes according to their resolution [Presti et al., 2004; Presti, 2006]. An additional, even more conservative effort to reduce the bias introduced by poorly located events has been made in the present study by excluding from maps and cross sections the earthquakes for which the epicentral and hypocentral errors are found greater than predetermined values. We made several attempts by varying the error upper bounds and found that the hypocenter trends are fairly stable in a relatively wide range of the error upper bound. In Figure 3, the map and cross sections obtained by excluding all earthquakes (18% of the total earthquakes) for which the epicentral and hypocentral errors were found greater than 6 and 10 km, respectively, are displayed. A further check of hypocenter trends detected in Figure 3 has been done by synthetic earthquake relocations. We henceforth report the test performed in sector a (Figure 3), which is the most critical one having a network azimuthal gap of almost 180°. The synthetic events were positioned on two-dimensional grids with spacing of 1 km. For each synthetic event, the arrival times of P and S waves were estimated at the stations that recorded a randomly selected event of the real sequence occurring in the same sector. A Gaussian noise with standard deviation of 0.1 and 0.4 s was adopted for perturbation of P and S wave readings, respectively. In Figure 4, we show the results obtained by relocating the earthquakes positioned on two grids significantly rotated with respect to the hypocenter pattern of the real sequence, i.e., a vertical grid oriented N-S (Figure 4, left) and a NW striking grid with a dip of 60° toward the southwest (Figure 4, right). Bayloc relocations of the synthetic events provided a faithful reproduction of the two grids, both in strike and in dip (Figure 4). By making several attempts (i.e., changing dip and strike of the synthetic quake grid), we obtained compelling evidence of the reliability of the real hypocenter trend found in sector a (Figure 3); that is, the trend detected in sector a should not derive from the activity of faults having strike and dip different, respectively, from N45° ± 25°W and 70° ± 20°NE. Similar evidences of reliability of hypocenter trends have been obtained also for sectors b to d of Figure 3. These evidences are not reported for conciseness.
[14] Figure 5 shows the focal mechanisms of the earthquakes with magnitude ≥ 4.0 available for the south Tyrrhenian belt since 1978. The chosen magnitude threshold allows us to better focus our attention on regional-scale geodynamic processes. Nineteen out of 21 focal mechanisms were taken from the literature [Anderson and Jackson, 1987; Pondrelli et al., 2002, 2004; regional centroid moment tensors, 2006, available at http://www.ingv.it/seismoglo/RCMT); Harvard CMT catalog, 2006, available at http://www.seismology.harvard.edu]. Focal mechanisms 5 and 6 have been estimated in the present study by inverting the P-onset polarity data (i.e., 17 and 14 data, respectively) with the FPFIT standard procedure [Reasenberg and Oppenheimer, 1985] and by using the crustal velocity model provided by Barberi et al. [2004].
[15] Figure 5 shows that reverse faulting on E to NE striking planes is dominant within the south Tyrrhenian belt, whereas a combination of dextral strike-slip and normal faulting characterizes the eastern boundary of this belt, i.e., the Tindari Fault system. The polar plots of P and T axes are displayed in the inset of Figure 5. These plots show that seismic activity in the central western sector of the south Tyrrhenian belt occurred in response to a NNW-SSE oriented compressive stress, whereas the seismicity in the eastern sector is connected with a more complex and heterogeneous stress field probably related to a transitional domain between compressional and extensional environments [Billi et al., 2006].
4. Structural Data
[16] The south Tyrrhenian active contractional margin has been thus far considered as dominated by a S dipping N vergent reverse fault system [e.g., Goes et al., 2004; Serpelloni et al., 2005; Billi et al., 2006]. This interpretation was apparently supported by the occurrence of a S dipping, NE striking thrust fault signaled on the island of Ustica [Bousquet and Lanzafame, 1995], which is the only emergent area along the central western sectors of the studied compressive margin. In this section of the paper and in the following ones, the structural evidences from Ustica are critically reconsidered and properly framed into the active tectonic context of the south Tyrrhenian margin.
[17] We analyzed the rock deformations exposed on the volcanic island of Ustica by field surveys and structure mapping at the 1:5,000 scale. The exposed deformations consist of open extensional fractures, Neptunian dikes, and faults (Figure 6a). Extensional fractures and Neptunian dikes were observed in several exposures (e.g., Figure 6b), particularly along the coast. Neptunian dikes have a maximum dilational displacement of about 1.5 m and are usually filled by poorly consolidated marine calcarenites. Fractures and dikes consist of high-angle (dip ≥ 70°) surfaces striking between N30° and N80°. The average strike of the analyzed population of dikes and fractures is N53° (see diagram A in Figure 6a). The relative axis of maximum extension is inferred as trending about NW-SE. In site 3 (Figure 6a), a lava flow emplaced over some Neptunian dikes, this evidence showing that fractures and dikes are, at least in part, synvolcanic.
[18] A set of faults (Figures 6a, 6c, and 6d) was observed in the western sector of the island as already signaled by Bousquet and Lanzafame [1995]. These faults consist of middle- to low-angle surfaces striking preferentially N60°–70° (Figure 6a). Over most fault surfaces, a set of lineations was observed. These structures were dubiously interpreted as kinematic indicators showing a reverse-to-transpressional (left lateral) fault kinematics. In particular, the main sense of tectonic transport for the hanging wall blocks is toward the north-northwest (Figure 6). The relative axis of maximum shortening is inferred as trending about NNW-SSE. In site 2 (Figure 6a), some Neptunian dikes are cut and displaced by the low-angle faults, this evidence showing that the reverse faults postdate the dikes and the extensional fractures.
5. Discussion
5.1. Tectonic Architecture
[19] The architecture of the south Tyrrhenian seismic belt is characterized by an overall E-W trend and a curved shape with a northward convexity. In detail, the belt is split into seismic segments (i.e., elongated clusters or clouds) trending preferentially NW-SE and NE-SW (Figure 3). We interpret these elongated clusters as fault zones activated in different periods between 1994 and 2005. The fault attitude and orientation inferred from the earthquake focal mechanisms (Figure 5) support this interpretation.
[20] We interpret the NW-SE and NE-SW fault zone trends as mostly inherited from previous tectonic processes, namely the compressive tectonics (Oligocene-Pliocene time) that led to the development of the Maghrebian fold-thrust belt [Pepe et al., 2005] and the extensional tectonics (Miocene-Pleistocene time) that led to the development of the Tyrrhenian back-arc basin [Pepe et al., 2000]. In particular, the extensional tectonics led to the partial collapse of the inner (northern) portion of the Maghrebian fold-thrust belt and to the development of several NE and ENE trending basins in the coastal area of northern Sicily and in the southern Tyrrhenian region (e.g., the Cefalù basin in Figure 7, Pepe et al. [2000]). In the oceanic Tyrrhenian basin, Nicolosi et al. [2006] have recently shown a set of NE trending magnetic anomaly stripes, from which a major NE-SW tectonic trend (i.e., strike of faults) is inferred. The complex tectonic setting inherited from the Paleogene-Quaternary contractional and extensional tectonic phases is probably the cause for the observed segmentation and trends of the south Tyrrhenian seismic belt (Figure 7). This hypothesis is also consistent with recent seafloor bathymetric maps of the Tyrrhenian Sea (Figure 7) [Marani et al. [2004]), where a morphologically very heterogeneous passive margin occurs between the continental domain of Sicily and the oceanic one of the Tyrrhenian region. Along this margin, alternated basins and ranges are mostly NE-SW oriented.
[21] In cross-sectional views (Figure 3), the south Tyrrhenian active belt is a zone of seismic deformation inclined toward the north with dip angles greater than 60°. Such high angles may be explained by two models. In a first model, the high-angle seismic zones are interpreted as inherited thrust zones, which originally developed as low-angle and then became high-angle zones with the emplacement, at their footwall, of forelandward piggyback thrust sheets. Such a thrust imbrication typically produces the progressive steepening of the inner early thrusts [e.g., Storti et al., 2000]. In an alternative model, the high-angle seismic zones may be interpreted as connected with the inversion of basin-bounding high-angle normal faults.
[22] The fault zones imaged through the Bayloc seismological analysis (Figures 3 and 7) apparently do not coincide with the Drepano Thrust (i.e., as drawn by Pepe et al. [2005] by interpolating data from seismic profiles), which marks the tectonic boundary between the inner crystalline thrust sheets, in the north, and the outer sedimentary ones, in the south. This suggests that the present seismic activity occurs along faults younger than the Drepano Thrust (i.e., note in Figure 7 that most seismic clusters cut across the Drepano Thrust), such as basin-bounding normal faults generated during the late Neogene-Quaternary back-arc extension or such as newly generated faults. Against this hypothesis, Pepe et al. [2005] suggested that an inherited thrust (i.e., Oligocene–early Miocene in age) was the hypocenter for the compressional seismic sequence of September 2002. A proper campaign of seismic reflection profiling in the southern Tyrrhenian Sea may clarify the origin of the observed high-angle fault zones and their tectonic relationships with the Drepano Thrust.
5.2. Present Kinematics
[23] Focal mechanisms show that the south Tyrrhenian seismic belt is mostly accommodating reverse displacements in response to a maximum compression oriented approximately NNW-SSE (Figure 5). This evidence, combined with the cross-sectional attitude of the seismic segments (i.e., dipping toward the north, Figure 3), shows that the general vergence of the belt is presently toward the south.
[24] In previous papers, it was demonstrated that on the NW striking faults at the eastern edge of the south Tyrrhenian seismic belt (i.e., the Sisifo and Tindari faults [Finetti and De Ben, 1986; Billi et al., 2006]) right-lateral, strike-slip displacements are presently accommodated. These faults presumably transfer the reverse displacements, which are presently active on the south Tyrrhenian belt, toward the southeast in the Calabrian arc [Goes et al., 2004; Pondrelli et al., 2004; Gutscher et al., 2006]. By analogy with the kinematics of the Sisifo and Tindari faults, the NW striking seismic zones included in the south Tyrrhenian belt may act as transfer structures shifting the reverse displacements accommodated on the NE striking compressional zones. The focal mechanisms available for the study area (i.e., M ≥ 4, Figure 5) provide, however, evidence only for the reverse displacements accommodated on the NE striking faults and not for possible strike-slip displacements accommodated on the NW striking faults. This can be explained by assuming that minor strain release occurs on the transfer faults (i.e., M < 4). Compressional earthquakes on the master reverse structures should be, in fact, greater in magnitude than the strike-slip earthquakes on the transfer structures because the amount of slip accommodated on the transfer structures is only a fraction of that occurring on the master reverse structures. According to this model, strike-slip earthquakes along NW striking faults should generally be less than 4 in magnitude and are therefore absent in Figure 5, where only focal mechanisms of M ≥ 4 earthquakes are displayed. In the south Tyrrhenian area, the impossibility to estimate fault plane solutions of M < 4 earthquakes is connected with the nonhomogeneous geometry of the seismic network (i.e., absence of offshore stations, Figure 2).
[25] The right-stepping pattern of the NE striking reverse structures is very similar to that observed along the same plate margin in Algeria and off the Algerian coast [Meghraoui et al., 1986], where focal mechanisms from recent earthquakes are very similar to those presented in this paper [Stich et al., 2003].
[26] Indirect evidence for strike-slip faulting along the south Tyrrhenian belt is provided by the occurrence of ENE striking reverse and thrust faults at Ustica. These faults imply considerable horizontal oblique-compressional displacements. Reverse and thrust faults, in fact, develop as low-angle surfaces for the effect of contractional strain, which is horizontally transmitted through a continuum medium (i.e., rocks). Because the faults observed at Ustica are at the summit of a conic volcano as high as 2000 m at least [Marani et al., 2004], the only way to transmit some horizontal tectonic contraction in such a location is via a high-angle transpressional fault cutting across the entire volcano. In such a model, the observed surficial faults would represent the expression of reverse-to-transpressional fault segments occurring within a larger WNW striking strike-slip-to-transpressive fault system (i.e., positive flower structure) across the Ustica volcano [Bousquet and Lanzafame, 1995]. The occurrence of differently oriented fault segments within a larger strike-slip fault system is common across orogenic wedges [e.g., Sylvester, 1988; Holdsworth et al., 1998]. This model may also explain some of the observed high-angle zones of seismic deformation in the south Tyrrhenian belt (Figure 3).
[27] Recently published GPS data for the Sicilian-Tyrrhenian region [D'Agostino and Selvaggi, 2004] are used to estimate the present relative motion across the south Tyrrhenian belt (Figure 7). By combining the velocity vectors from the Ustica station with those from the Trapani and Fossa stations along the northern coast of Sicily (note that Fossa is right on the western block of the Tindari Fault, Figure 7), we obtained that the two blocks that face along the south Tyrrhenian seismic belt are horizontally approaching by about 7 mm/yr along the N174° direction, in the eastern sector (i.e., result obtained by combining the velocity vectors from Ustica and Fossa stations), and by about 3 mm/yr along the N124° direction, in the western sector (i.e., result obtained by combining the velocity vectors from Ustica and Trapani stations). These estimated motions across the south Tyrrhenian belt are consistent with the data presented in this paper (see diagrams A and B in the inset of Figure 4 and diagram B in Figure 6a), according to which the south Tyrrhenian seismic belt mostly involves ENE and NE striking reverse faults and NW striking right-lateral strike- or oblique-slip faults. A greater slip rate in the eastern sector of the south Tyrrhenian belt is consistent with the occurrence of a retreating slab beneath the Calabrian arc [Faccenna et al., 2004; Montuori et al., 2007].
[28] By considering about 500 kyr, at least, as the period of activity for the south Tyrrhenian compressive belt [Goes et al., 2004] and a reverse heave rate between 3 mm/yr (western sector) and 7 mm/yr (eastern sector), we obtained an estimate of the total reverse heave produced since 500 ka between about 1500 and 3500 m for the western and eastern sectors, respectively. Because of errors in the original data [D'Agostino and Selvaggi, 2004], however, the above estimates based on GPS data should be considered as moderately accurate.
5.3. Seismic Potential
[29] The south Tyrrhenian compressive belt has been so far depicted as a continuous structure along the entire northern coast of Sicily (i.e., an about 250 km long thrust [e.g., Serpelloni et al., 2005; Billi et al., 2006]). A maximum seismic potential of magnitude between 7.1 and 7.6 or even M ≥ 8 has been recently proposed for this structure [Jenny et al., 2006]. The segmented geometry of the belt as shown in Figure 3 gives new insight about its seismic potential. Previously published tectonic [Giunta et al., 2004; Pepe et al., 2005] and morphological [Marani et al., 2004] data in the same area support our results concerning the segmented geometry of the belt.
[30] By using the standard relationships by Wells and Coppersmith [1994], a rupture area of about 1800 km2 can be estimated for an earthquake with reverse mechanism and magnitude in the 7.1–7.6 range. Such an earthquake implies a rupture length on the order of 100 km, at least, if crustal thickness and fault dip values of 15 km and 60°, respectively, are assumed. The epicentral map obtained through the Bayloc method (Figure 3) suggests that no fault segments with such a length should reasonably occur in the belt.
[31] The maximum magnitude recorded in the belt during the last decades has been 5.9 (i.e., the 6 September 2002 earthquake; earthquake 14 in Table 1). Also, recent reviews of historical data [e.g., Guidoboni et al., 2003; Azzaro et al., 2004; Working Group CPTI, Catalogo parametrico dei terremoti Italiani, versione 2004 (CPTI04), Istituto Nazionale di Geofisica e Vulcanologia, Bologna, Italy, 2004, available at http://emidius.mi.ingv.it/CPTI04/] have led to estimate offshore locations (i.e., within the south Tyrrhenian belt) and magnitude values of about 6 for the three earthquakes (1726, 1823 and 1940), which, during the last five centuries, have produced the most severe effects along the northern coast of Sicily.
| N | Dateb | OT | Lon, °E | Lat, °N | Depth, km | Mag | P_az | P_plg | T_az | T_plg | Sourcesc |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 780415 | 23.33 | 15.07 | 38.39 | 21.0 | 5.7 | 18 | 8 | 116 | 42 | A and J |
| 2 | 800528 | 5.57 | 14.25 | 38.48 | 12.0 | 5.5 | 159 | 5 | 249 | 5 | A and J |
| 3 | 800601 | 5.26 | 14.33 | 38.39 | 10.0 | 4.8 | 334 | 6 | 148 | 84 | POND04 |
| 4 | 810622 | 9.36 | 14.09 | 38.49 | 13.0 | 4.8 | 323 | 1 | 56 | 71 | POND04 |
| 5 | 900328 | 5.47 | 14.91 | 38.16 | 24.1 | 4.2 | 195 | 20 | 280 | 0 | this study |
| 6 | 950827 | 19.42 | 15.17 | 38.28 | 8.9 | 4.0 | 185 | 15 | 90 | 5 | this study |
| 7 | 980117 | 12.32 | 12.75 | 38.45 | 10.0 | 4.8 | 342 | 17 | 193 | 71 | POND02 |
| 8 | 980620 | 2.25 | 13.08 | 38.46 | 10.0 | 5.2 | 349 | 23 | 184 | 66 | RCMT |
| 9 | 980621 | 8.59 | 12.67 | 38.43 | 10.0 | 4.6 | 348 | 10 | 208 | 78 | RCMT |
| 10 | 980621 | 12.59 | 13.10 | 38.50 | 10.0 | 4.6 | 349 | 7 | 123 | 79 | RCMT |
| 11 | 980914 | 5.24 | 13.60 | 38.46 | 10.0 | 5.0 | 349 | 15 | 187 | 74 | RCMT |
| 12 | 990214 | 11.45 | 15.06 | 38.17 | 17.7 | 4.7 | 179 | 77 | 300 | 7 | RCMT |
| 13 | 020405 | 4.52 | 14.74 | 38.48 | 15.0 | 4.4 | 347 | 6 | 102 | 77 | RCMT |
| 14 | 020906 | 1.21 | 13.57 | 38.42 | 15.0 | 5.9 | 329 | 6 | 231 | 53 | CMT |
| 15 | 020906 | 1.45 | 13.73 | 38.44 | 15.0 | 4.7 | 137 | 2 | 232 | 64 | RCMT |
| 16 | 020910 | 2.32 | 13.70 | 38.47 | 15.0 | 4.4 | 315 | 20 | 91 | 64 | RCMT |
| 17 | 020920 | 23.06 | 13.74 | 38.46 | 16.1 | 4.7 | 325 | 12 | 177 | 76 | RCMT |
| 18 | 020927 | 6.10 | 13.66 | 38.41 | 15.0 | 5.2 | 324 | 22 | 170 | 65 | CMT |
| 19 | 020928 | 2.46 | 13.71 | 38.47 | 15.0 | 4.6 | 340 | 7 | 109 | 80 | RCMT |
| 20 | 021002 | 22.57 | 13.72 | 38.46 | 15.0 | 4.9 | 325 | 7 | 215 | 69 | RCMT |
| 21 | 060227 | 4.34 | 15.18 | 38.15 | 15.0 | 4.5 | 251 | 67 | 108 | 18 | QRCMT |
- a Abbreviations are N, order number; OT, origin time; and Mag, magnitude. Hypocentral parameters are longitude (Lon), latitude (Lat), and depth.
- b Date is given in year, month, and day format.
- c Bibliographic sources are abbreviated as follows: A and J, Anderson and Jackson [1987]; POND04, Pondrelli et al. [2004]; POND02, Pondrelli et al. [2002]; RCMT, regional centroid moment tensor; and QRMCT, quick regional centroid moment tensor the Istituto Nazionale di Geofisica e Vulcanologia catalog (available at http://www.ingv.it); and CMT, the Harvard CMT catalog [Dziewonski et al., 1981] (available at http://www.seismology.harvard.edu).
[32] From the above considerations, we conclude that all available data, including the geologic and tectonic information, suggest a maximum seismic potential for the south Tyrrhenian belt lower than 7.1–7.6 or even 8. It is evident that intrinsic approximations of the available relationships between earthquake magnitude and rupture size [e.g., Field et al., 1999; Dowrick and Rhoades, 2004], as well as inaccuracy of historical earthquake catalogs do not allow us to definitely rule out the existence of a structure in the study area that may generate an earthquake with magnitude over 7; however, the presently available information lead us to consider a maximum magnitude close to 7 as more realistic for earthquakes within the south Tyrrhenian belt.
5.4. Implications for the Regional Plate Tectonics and for Future Scenarios
[33] Since about 700–500 ka, contractional deformations across the Maghrebides in Sicily have resumed at the rear of the orogenic wedge and are presently active across the south Tyrrhenian belt [Goes et al., 2004]. The causes for this process are presumably connected with the continental collision and the associated substantial cessation of subduction across the Maghrebides during Quaternary times [Lickorish et al., 1999; Faccenna et al., 2004]. Tomographic images show a large slab window beneath Sicily and the southern Tyrrhenian Sea, thus supporting the hypothesis of a substantial cessation of subduction in Sicily [Faccenna et al., 2005; Montuori et al., 2007]; however, geodetic and seismic data show that the Eurasia-Nubia convergence in this region is still ongoing and progresses with high convergence rates [Hollenstein et al., 2003; D'Agostino and Selvaggi, 2004; Pondrelli et al., 2004]. This explains why the Maghrebian fold-thrust belt is still active in its south Tyrrhenian portion, where the back-arc passive margin is being compressionally inverted (Figures 5 and 7). A similar tectonic history has been inferred for the northern passive margin of the Tyrrhenian ocean, (i.e., in the Ligurian Sea, Figure 1), where compressional inversion started about 5–3 Myr ago [Bigot-Cormier et al., 2004; Chardon et al., 2005], and for the westward prolongation of the south Tyrrhenian margin in the Alboran Sea and off the Algerian coast (Figure 1) [Comas et al. [1992]; Stich et al. [2003]; Déverchère et al. [2005]; Domzig et al. [2006]. In the Alboran-Algerian segments of the Nubia compressive margin, the compressional inversion of the back-arc passive margin started earlier (i.e., about 7–5 Myr ago) than in the south Tyrrhenian region. The eastward migration of the inversion tectonics along this margin (i.e., from the Alboran-Algerian to the Tyrrhenian regions) is probably connected with a parallel migration of continental collision and associated cessation of subduction across the west Mediterranean subduction zone [Faccenna et al., 2004]. The temporal gap between onsets of inversion tectonics in the Alboran-Algerian margin and in the south Tyrrhenian one may explain the different, present kinematics of thrusts. Whereas the south Tyrrhenian active thrusts mostly verge toward the foreland (i.e., southward), the Alboran-Algerian active thrusts mostly verge toward the hinterland (i.e., northward). This evidence suggests different evolutionary stages for the Alboran-Algerian and south Tyrrhenian compressional margins. The older Alboran-Algerian margin may truly represent a case of incipient southward subduction of the back-arc basin as proposed by Déverchère et al. [2005], whereas the younger south Tyrrhenian margin may still be in a very early stage of contraction involving the compressional reactivation or inversion of inherited, north dipping faults [Pepe et al., 2005]. However, because further northward subduction of the Hyblean-Pelagian foreland beneath the Maghrebian fold-thrust belt is highly unlikely for the continental nature of this foreland lithosphere, with the progression of the Eurasia-Nubia convergence, a likely future scenario may involve the subduction of the Tyrrhenian oceanic lithosphere toward the south beneath the Maghrebian fold-thrust belt. The nucleation of subduction may occur earlier off the Algerian coast and then propagate toward the east involving the south Tyrrhenian region.
[34] Models of subduction inception predict a dynamic uplift for the overriding plate during the early stages of compression [Cloething et al., 1990; Shemenda, 1992; Toth and Gurnis, 1998; Faccenna et al., 1999]. The Holocene tectonic uplift of northern Sicily (highest uplift rate probably exceeding 2 mm/yr [Rust and Kershaw, 2000]) may thus be explained, at least in part, by a process of subduction inception in the south Tyrrhenian region. Further geophysical studies off the Sicilian coast are necessary to constrain both temporally and spatially the uplift process. It should be noted also that both the Ligurian (northern Tyrrhenian Sea) and the Alboran-Algerian margins are presently undergoing uplift [Bigot-Cormier et al., 2004; Domzig et al., 2006].
[35] In the south Tyrrhenian area, the inversion of the passive margin and the subsequent subduction could be favored by the young and thin Tyrrhenian oceanic lithosphere [Cloething et al., 1982, 1989; Muller and Phillips, 1991; Erickson and Arkani-Hamed, 1993]. On the other hand, the young and poorly dense Tyrrhenian lithosphere will be not enough negatively buoyant to initiate a self-sustained subduction [Cloos, 1993]. This suggests that, in the Tyrrhenian region, the Eurasia-Nubia convergence will likely lead to an Alpine-type collisional scenario [e.g., Beaumont et al., 1996], where continental collision will follow a brief shortening and partial subduction of a small and young oceanic basin. A Pacific-type subduction, where an old and negatively buoyant ocean is able to long subduct beneath a continent, is far a less appropriate scenario for the future evolution of the Eurasia-Nubia margin in the Tyrrhenian region.
6. Conclusions
[36] 1. The use of a new technique of earthquake location (Bayloc) in the south Tyrrhenian region has significantly contributed to define the tectonic architecture and kinematics of the seismogenic apparatus of the Nubia plate compressive margin in this region. Main results show that the studied compressive belt is characterized by spatially discontinuous epicentral clusters with major NW-SE and SW-NE trends and by hypocentral clusters dipping toward the north by angles greater than 60°. These seismic clusters are here interpreted as high-angle, S verging, reverse fault zones forming an early configuration of a young compressive belt (i.e., younger than about 700–500 ka). This belt occurs along and inverts a late Neogene–Quaternary passive margin, which developed at the southern border of the Tyrrhenian back-arc basin.
[37] 2. On the basis of instrumental and historical seismic data combined with tectonic and morphological information, earthquakes with a maximum size close to magnitude 7 are proposed for the south Tyrrhenian compressive seismic belt. At present, the segmented geometry of this belt should prevent from the occurrence of stronger earthquakes.
[38] 3. An integrated monitoring system including measurements of deformation and seismicity also in offshore areas [e.g., Beranzoli et al., 1998, 2000] will significantly contribute in knowing the evolution of the seismogenic apparatus and therefore in mitigating the earthquake and tsunami hazards in this densely populated region.
[39] 4. Results from this work combined with previously published geodetic data provide clues to predict possible future scenarios, one of which involves the development of a new subduction zone in the southern Tyrrhenian area where the Tyrrhenian oceanic lithosphere would subduct toward the south beneath the Maghrebian fold-thrust belt in Sicily.
Acknowledgments
[40] R. Arculus, S. Goes, F. Pepe, and an anonymous Associate Editor are thanked for patient editorial handling and for insightful comments. L.A. is thanked for a text proofreading.
