Integrated geological, geomorphological, and differential interferometry synthetic aperture radar (DInSAR) data are used to constrain the timing and modes of activity of Quaternary fault systems in the Hyblean Plateau. This area, which represents a unique natural laboratory for studying surface deformation in relation to deep slab dynamics, has grown since middle Miocene times as a doubly plunging forebulge associated with slab rollback during NW-directed subduction. Bimodal extension has produced two mutually orthogonal normal fault systems. The detailed stratigraphic record provided by synrift sediments and postrift marine terraces allowed us to define the timing of activity of an early Pleistocene, flexure-related fault system, thus constraining the duration of a typical foreland extensional tectonic event to ∼1.5 m.y. Subsequent late Quaternary to present deformation was dominated by strike-slip faulting associated with NW-oriented horizontal compression. During this latest stage, regional uplift progressively increased toward the thrust front to the NW and was accompanied by differential uplift accommodated by dip-slip components of motion along active NNW-trending faults. The general active tectonic setting of the study area, characterized by NW-oriented horizontal compression consistent with major plate convergence, and the regional uplift pattern can both be explained within the framework of intraplate shortening and foreland rebound following complete slab detachment, a major geodynamic event interpreted to have taken place at ca. 0.7 Ma in southern Italy.


For decades, studies on fold-and-thrust belts all over the world have described the foreland plate sector in front of an orogen-foreland basin system as the “undeformed foreland.” Starting from the 1980s, far-field propagation of a stress field over an orogen has been assumed as an explanation of weak strain recorded in large foreland sectors ahead of the thrust front (Geiser and Engelder, 1983). Evidence for significant brittle deformation of foreland areas has been gathered during the years, and in recent times, bending of the foreland lithosphere under the weight of the growing orogen and associated forebulge development have been invoked to explain extensional faulting and fracturing in foreland sectors, a classical example being represented by the Hyblean Plateau of SE Sicily (Pedley and Grasso, 1992). Here, conjugate normal faults and fractures that form two orthogonal sets have been related to lithospheric flexure by Billi and Salvini (2003), Billi (2005), and Billi et al. (2006). The process of foreland flexuring is well known to produce outer-arc extension in the peripheral bulge and in the outermost sector of the foredeep in response to lithospheric bending (Tankard, 1986; Bradley and Kidd, 1991; Doglioni, 1995; Sinclair, 1997; Turcotte and Schubert, 2001; Langhi et al., 2011). “Tangential” (i.e., parallel to basin strike) normal faults in foredeep and forebulge settings are frequently imaged in seismic profiles (e.g., Lorenzo et al., 1998; Matenco and Bertotti, 2000; Mazzoli et al., 2001, 2008; Ranero et al., 2003; Shiner et al., 2004). These structures, which show dominant forelandward, but also hinterlandward, dips (e.g., Maillard et al., 1992; Scisciani et al., 2001; Krzywiec, 2001; Tavarnelli and Peacock, 2002; Bolis et al., 2003; Tinterri and Muzzi Magalhaes, 2011), form half-grabens, grabens, and horsts that substantially control the distribution and accumulation of foredeep deposits (e.g., Casnedi, 1988; Tinterri and Muzzi Magalhaes, 2011).

Transversal extensional deformation structures, oriented perpendicular to “tangential” ones, can also form during flexure (e.g., Destro, 1995; Medwedeff and Krantz, 2002; Quintà and Tavani, 2012). “Radial” (i.e., normal to basin strike) normal faults form due to the arcuate shape of the foredeep, inducing arc-parallel stretching (e.g., Doglioni, 1995). The amount of along-strike foredeep stretching increases with orogen curvature (e.g., Zhao and Jacobi, 1997; Whitaker and Engelder, 2006), as it derives from the sum of cross-sectional and along-strike curvature.

Foredeep extensional structures, particularly “tangential” faults and associated basin depocenters and structural highs, are well known to play a key role in subsequent tectonic processes associated with thrust propagation into the foredeep (e.g., Butler, 1989; Scisciani et al., 2001; Mazzoli et al., 2002; Tavarnelli and Peacock, 2002). Increasing evidence suggests that these features represent very important, widespread background structures in rocks later involved in fold-and-thrust belts (Calamita and Deiana, 1980; Scisciani et al., 2001; Lash and Engelder, 2007; Casini et al., 2011; Beaudoin et al., 2012; Quintà and Tavani, 2012; Tavani et al., 2012). For example, foreland flexure–related inherited fracture networks, being independent of structural position within folds, likely play an important role for fluid flow, e.g., in fractured carbonate reservoirs (Vitale et al., 2012).

The peculiar foreland continental sector represented by the African plate indentor of SE Sicily (e.g., Yellin-Dror et al., 1997), located in front of a recess and partly surrounded by two salients of the orogen (Fig. 1), is characterized by vigorous tectonic activity beyond that expected by the relatively simple foreland flexuring processes described here.

In this paper, we discuss well-exposed fault systems in the Augusta area, in the eastern portion of the Hyblean Plateau. The sedimentary and geomorphological record is well preserved in this area and provides detailed stratigraphic constraints on the onset and end of activity of various Quaternary fault systems. These fault systems show different trends and kinematics, as well as cumulative displacements on the order of hundreds of meters to kilometers. We particularly focus on an early Pleistocene faulting event that controlled the deposition of a thick carbonate-clastic succession. The detailed stratigraphic record provided by synrift sediments and postrift marine terraces allows us to define the timing of activity of related fault systems, thereby constraining the complete duration of a typical extensional tectonic event affecting a foreland area. As this area is still tectonically active, we also used multitemporal interferometry (differential interferometry synthetic aperture radar [DInSAR]) analysis to unravel the behavior of the late Quaternary fault system that controls the seismotectonic setting of SE Sicily. However, since high-rate vertical motions involving subsidence phenomena on short time scales—comparable to those recorded by radar techniques—may be induced by significant water-level decline due to large-volume groundwater pumping, available hydrogeological information has also been taken into account.

The features analyzed in this study provide clear evidence of the complexity and relevance of foreland structures, well beyond the traditional notions of far-field foreland stress or bending-related brittle deformation of forebulge sectors. In particular, the relationships between surface deformation and deep geodynamic processes (i.e., slab rollback vs. slab breakoff) are emphasized. These results have important implications for a better understanding of tectonic processes—including active faulting and seismogenesis—in foreland areas, which provide further insights into the widespread occurrence and important role of inherited prethrusting structures in fold-and-thrust belt evolution (e.g., Butler, 1989).


This study was carried out through a combined approach that integrates topography analysis, DInSAR analysis, and field controls with hydrogeological information.

The hydrogeological information was used to reconstruct the groundwater surface and flow orientations of the main aquifer in the Augusta basin.

The geomorphological analysis was carried out through the analysis of topographic maps from the Italian Istituto Geografico Militare (IGM 1:25,000 scale maps), and Regione Sicilia (Carta Tecnica Regionale 1:10,000 scale maps). The geomorphological analysis is aimed at identifying and further constraining late Quaternary differential vertical motions, taking into account the poor constraints on recent fault activity provided by stratigraphy alone. In particular, we focused on crosscutting relationships between the main faults and the wave-cut/wave-built marine terraces. Keys to the identification of vertical offsets that postdate terrace formation are the spatial distribution, and shape, of the marine terraces, and alignments of straight scarps with straight valleys or valley trunks that affect the terraced surfaces.

The DInSAR technique is able to detect displacements that occur between subsequent radar acquisitions in the sensor-target direction (line of sight [LOS]) with subcentimetric precision. In the surface deformation detection field, the DInSAR technique allows one to obtain spatially and temporally dense measurements over large areas (from 5 km × 5 km to 100 km × 100 km). In addition, the availability of a now several-decades-long (since 1992) image archive allows us to reconstruct the history and evolution of previously unmonitored phenomena. Over the past 20 yr, synthetic aperture radar (SAR) technology has greatly improved, and many satellite constellations have been launched: ERS1/2 and ENVISTAT ASAR (European Space Agency), JERS-1 SAR (Japanese Aerospace Exploration Agency), RADARSAT-1/2 (Canadian Space Agency), TerraSAR-X and TanDEM-X (Infoterra, Germany), COSMO-SKYMed (Italian Space Agency), and SENTINEL (European Space Agency). The development of SAR technology has allowed the implementation of image-processing algorithms that produce increasingly reliable velocity maps and temporal series of deformation. Two main groups of algorithms may be distinguished: algorithms that work at “full resolution” (i.e., 4 × 20 m for medium-resolution SAR images or 3 × 3 m for high-resolution SAR images), which consider only those points characterized by radar backscattering signal (phase) stable in time, called persistent scatterer (PS) algorithms (Ferretti et al., 2000; Werner et al., 2003; Arnaud et al., 2003; Hooper et al., 2004; Duro et al., 2005; Costantini et al., 2008; Iglesias et al., 2014), and those working at “medium resolution” (usually at 60 × 60 m for medium-resolution SAR images or ∼10 × 10 m for high-resolution SAR images), which evaluate the coherence value. The latter, as reported in Seymour and Cumming (1994), represents the maximum likelihood estimator of the phase quality over an estimation window (also called multilook). This technique is named small baselines subset (SBAS), or coherence-based (Berardino et al., 2002; Mora et al., 2003; Lanari et al., 2004; Prati et al., 2010; Sowter et al., 2013). In recent years, algorithms that incorporate both the PS and SBAS approaches have been proposed (Hooper, 2008; Ferretti et al., 2011).

The DInSAR technique is based on the computation of interferometric phase differences (δϕint) for each pixel, constituting the interferogram, between two SAR images acquired in different spatial positions (spatial baseline) with a difference of time (temporal baseline) using the following formula (Hanssen, 2001): 
where δϕflat is the flat Earth component related to range distance in absence of topography; δϕtopo is the topographic phase; δϕdispl is the component due to the displacement of the terrain in the LOS direction (line that goes from the radar to the observed point) between the SAR acquisitions; δϕatm is the phase related to atmospheric artifacts; and δϕnoise accounts for degradation factors related to temporal decorrelation. It is worth pointing out that the δϕflat, δϕtopo, and δϕatm terms can be removed because the orbital and topographical parameters are known. In particular, as far as δϕtopo concerns, if a digital terrain model (DTM) with an adequate resolution is available, the contribution of known topography can be almost completely removed from the interferometric phase, δφint, such as to detect the ground motions in the so-called differential interferogram.

In this study, DInSAR was performed by means of a SUBSOFT processor, based on the coherent pixels technique algorithm (CPT; Mora et al., 2003; Blanco-Sánchez et al., 2008; Iglesias et al., 2014) developed by the researchers of the Remote Sensing Laboratory (RSLab) of the Universitat Politecnica de Catalunya (UPC). We use Environmental Satellite–Advanced Synthetic Aperture Radar (ENVISAT-ASAR) images spanning the time period 2003–2010. In detail, the data sets consist of 45 and 50 single look complex (SLC) images in ascending and descending orbit, respectively, acquired between June 2004 and July 2010, and April 2003 and June 2010.

The interferometric chain implemented in the SUBSOFT processor is reported in the Appendix.


The study area forms part of the Hyblean Plateau of SE Sicily (Fig. 1), which consists of a doubly plunging forebulge characterized by the occurrence of preexisting crustal heterogeneities that developed during a long-lasting tectonic evolution since Mesozoic times. The inherited crustal architecture enhanced differential retreating processes of the foreland during subsequent subduction, with the associated development of orogenic salients and recesses (Billi et al., 2006). Various generations of foreland fault systems characterized by different trends and/or kinematics have developed since middle Miocene times, which have important implications for the seismotectonic behavior of the studied area (e.g., Adam et al., 2000). This area is located within a general context encompassing the whole of Sicily and surrounding onshore and offshore areas, which are characterized by intense active tectonics witnessed by historical and recent destructive earthquakes and tsunamis, late Quaternary faulting, and fast rates of vertical and horizontal crustal motions, and quiescent as well as active volcanoes (including Etna; Billi et al., 2010, and references therein).

The Hyblean Plateau, together with the Sicily Channel and the adjacent continental shelf areas, belongs to a largely submerged portion of the African foreland domain known as the Pelagian block (Burollet et al., 1978). This block represents an indentor of the African plate during its Neogene convergence with the Eurasian plate (Caire, 1970; Bianchi et al., 1987; Butler et al., 1992), and it is characterized by crustal segmentation partly controlled by inherited structures that originated during the Mesozoic evolution of the African passive margin (Reuther et al., 1993; Robertson and Grasso, 1995). The Hyblean Plateau consists of a continental basement of unknown age, overlain by a Triassic to Quaternary sedimentary succession including dominantly carbonate sediments intercalated with Upper Triassic–Lower Jurassic, Upper Cretaceous, and Upper Miocene to Quaternary mafic volcanics (Fig. 2; Barberi et el., 1974; Patacca et al., 1979; Grasso et al., 1983). The whole volcano-sedimentary succession of the plateau ranges in thickness from ∼5 to 6 km at its northern and eastern edges to ∼10 km in its central sector (Agocs, 1959; Zarudski, 1972; Lentini, 1983; Bianchi et al., 1987; Antonelli et al., 1988). The oldest outcropping rocks consist of Cretaceous carbonates. These include both cherty-marly limestones of basin origin to the west, and platform carbonates along the Ionian margin to the east (Lentini et al., 1984). The overlying deposits include Oligocene to Miocene carbonates and marls, which represent the most abundantly outcropping units of the plateau, followed by Quaternary sediments occurring mainly along its margins (Fig. 2).

The plateau is bound to the east by the Malta Escarpment, a steep submarine slope that drops into the Ionian Sea down to a depth in excess of 3000 m (Fig. 1). The escarpment, most probably controlled by an inherited Mesozoic crustal structure (Scandone et al., 1981), may be subdivided into two different portions (Argnani and Bonazzi, 2005): The segment north of Siracusa (location in Fig. 2) is characterized by the occurrence of NNW-SSE east-dipping recent extensional faults and related sedimentary basins, whereas the segment south of Siracusa appears to not be affected by recent faulting and consists of a steep morphological surface that flattens out toward the Ionian basin. A significant change in crustal thickness occurs between the block to the west, consisting of thick continental crust extending via the Malta platform to the Maltese islands, and the thin crust of the Ionian Sea to the east.

The Hyblean Plateau is flexured to the northwest and dips below the frontal part of the Maghrebian-Sicilian fold-and-thrust belt, represented in this area by the Pliocene–Quaternary Gela Nappe. The resulting flexural basin is known as the Gela-Catania foredeep (Di Geronimo et al., 1978; Grasso et al., 1983). Magnetic and gravity data clearly outline the sharp boundaries of the Hyblean Plateau with respect to adjacent units of the African continental margin (AGIP, 1978, 1982). Bouguer anomalies are positive over the plateau, while a pronounced negative anomaly occurs west of it, where a thick pile of sediments and allochthonous thrust sheets sit on top of the flexured and downfaulted foreland carbonate succession (Grasso et al., 1990). Extensional faulting and crustal thinning are marked by significant volcanic activity during the last 8 m.y. along the northern margin of the Hyblean Plateau. This area underwent uplift associated with the formation of a broad carbonate platform on its eastern sector during the Late Tertiary (Grasso and Lentini, 1982). As a result of regional uplift, slow rates of sedimentation—or even erosion associated with local emergence—characterized the northern Hyblean area. This uplift event was coeval with slow tectonic subsidence in the western part of the Hyblean Plateau, which underwent greater rates of flexural downbending than the eastern sectors. As the Messinian sea-level fall was coeval with uplift, evaporites were only locally deposited within NE-SW-trending narrow grabens located along the northern sector of the plateau (Pedley and Grasso, 1991). Subsequent subsidence, coeval with a major early Pliocene rise in Mediterranean sea level (Butler et al., 1995), led to flooding and chalk deposition (Trubi Formation). These new conditions created the accommodation space for Pliocene–Pleistocene sediment accumulation within the Gela-Catania foredeep, a foreland basin filled with several hundred meters of marine clays and sands unconformably overlying older volcano-sedimentary successions. Following a massive late Miocene–early Pleistocene basaltic volcanic flare-up (Schmincke et al., 1997), renewed uplift characterized the Hyblean Plateau during the middle to late Pleistocene, coeval with the end of thrusting along the frontal part of the Sicilian-Maghrebian fold-and-thrust belt. This renewed uplift is inferred from the occurrence of Lower Pleistocene shallow-water sediments at elevations in excess of 600 m in the northwestern part of the plateau (Schmincke et al., 1997), and from uplifted marine terraces developed along the whole coastal area of southeastern Sicily. Based on marine terrace elevations, uplift rate values of ∼2 mm/yr in the north (Taormina area), decreasing southward, have been estimated (Carbone et al., 1982a; Westaway, 1993; Bianca et al., 1999; Bordoni and Valensise, 1998; Monaco et al., 2002; Di Stefano and Branca, 2002; Antonioli et al., 2003, 2006, 2009; Catalano and De Guidi, 2003; Ferranti et al., 2006; Catalano et al., 2008, 2010; Lambeck et al., 2011). However, recent studies (e.g., Scicchitano et al., 2008) suggest that the area is undergoing differential displacements upon a regional, long-term uplift process.

Fault Systems and Seismotectonic Setting of the Hyblean Plateau

The main fault trends of the study area, outlined by Carbone et al. (1982b), include: (1) an older, inactive, NW-SE–trending graben system, which was already delineated during the early Pleistocene, and the related faults of which controlled paleocoastlines; and (2) younger, NNW-SSE– and ENE-WSW–striking fault systems that, besides offsetting Quaternary deposits, also dissect the preexisting NW-SE–trending fault system. For the latter, an early Pleistocene age has been proposed by Adam et al. (2000). These authors analyzed the dominant active stress field over the study area, suggesting that earthquake faulting is mainly of strike-slip type, with a NW-SE–trending maximum compression (σ1) and a NE-SW–oriented minimum compression (σ3). The occurrence of active strike-slip faulting is suggested by Azzaro et al. (2000) for the so-called “Scicli Line” (Fig. 2), a major wrench fault extending for ∼100 km from the Sicily Channel to the northern margin of the Hyblean Plateau. Although there is no direct evidence of activity postdating the middle Pleistocene, earthquake distribution (events in years A.D. 1698, 1818, 1895, 1949, 1980, 1990) is interpreted by the latter authors as indicating the existence of minor seismogenic sources associated with this structure. However, according to Azzaro and Barbano (2000), the main active fault system in SE Sicily is that associated with the Malta Escarpment. This fault system extends for more than 200 km from North Africa to Sicily, and controls the trend of the east coast of the island. It includes NNW-SSE–trending, dominantly extensional fault segments producing a cumulative vertical displacement of ∼3000 m. The northernmost fault segment is exposed onshore in the Mount Etna area (Continisio et al., 1997; Billi et al., 2010). In the Siracusa-Augusta onshore area, clear evidence of activity of this fault system dates back to the middle Pleistocene (Carbone, 1985), while offshore seismic profiles image middle Pleistocene to Holocene faulted deposits (Hirn et al., 1997; Argnani and Bonazzi, 2005; Firetto Carlino et al., 2013). Azzaro and Barbano (2000) suggested that the Malta Escarpment fault system is the source for the large earthquakes (M ≥ 7.0) that destroyed eastern Sicily in A.D. 1169 and 1693, and also of minor events such as those of A.D. 1818 and 1848.

Although subsurface stratigraphic data suggest that the forebulge of SE Sicily grew mainly as an ENE-trending foreland monocline (Grasso and Pedley, 1990), modeling of convergent deformation in Sicily (Ben-Avraham et al., 1995) as well as structural data (Barrier, 1992; Sirovich and Pettenati, 1999; Billi et al., 2006) point to a noncylindrical, complex tectonic evolution for this foreland sector. Indeed, the structural architecture of the Hyblean Plateau is characterized by multiple sets of strike-slip and normal faults (Fig. 2) for which trends and kinematics cannot be reconciled within the framework of a single regional stress field (Adam et al., 2000). In particular, only the older NW-SE–trending normal fault systems appear to be consistent with the horizontal extension expected for the outer arc of a cylindrical forebulge (Turcotte and Schubert, 2001). Moreover, focal mechanisms of upper-crustal earthquakes recorded in the Hyblean area between 1977 and 2002 show both strike-slip and normal fault solutions (Fig. 1; Goes et al., 2004). The strike-slip solutions are consistent with the dominant kinematics of active regional faults in this area (Adam et al., 2000; Grasso et al., 2000), as well as with the NW-trending maximum shortening axis (SHmax) obtained by the analysis of borehole breakout data (Ragg et al., 1999). On the other hand, the normal fault solutions are mostly compatible with a NE-SW–oriented, horizontal maximum extension.


The Augusta Basin (Fig. 3A) is a tectonic depression, ∼20 km long and 12 km wide, located in the eastern part of the Hyblean Plateau (Firetto Carlino et al., 2013). The basin is controlled by two main fault segments; namely, it is separated southward from the Magnisi–St. Panagia ridge by the NW-SE–oriented, NE-dipping Mount Climiti fault (Bianca et al., 1999; Catalano et al., 2010) and northward by the 7-km-long, NNW-SSE–trending Mount Tauro fault, which borders the Mount Tauro horst with a roughly 35-m-high scarp. From a stratigraphic point of view, the Augusta Basin is characterized by Lower Pleistocene calcarenites and sands passing upward and laterally to deep-water clays (Di Grande, 1972; Di Grande and Scamarda, 1973; Di Grande and Raimondo, 1982; Carbone, 1985; Carbone et al., 1986, 1987; I.S.P.R.A., 2011a, 2011b). These sediments rest unconformably upon an Upper Cretaceous–Miocene basement succession made of carbonates with volcanic intercalations, outcropping in the horst blocks bounding the Augusta Basin (Patacca et al., 1979; Carbone et al., 1982b, 1982c, 1986, 1987; Carbone, 1985; Bianchi et al., 1987; Dall’Antonia et al., 2001; I.S.P.R.A., 2011a). The stratigraphic gap indicates a prolonged emergence of the area during the Pliocene before tectonic collapse that led to graben formation during the late Pliocene, according to Carbone (1985), or at the onset of the Pleistocene, according to Adam et al. (2000). Younger deposits consist of widespread fossiliferous calcarenites and sands (“Panchina” Formation; Accordi, 1962) that are related to the middle Pleistocene (e.g., Accordi, 1962, 1963; Ruggieri and Greco, 1965; Carbone et al., 1982a; I.S.P.R.A., 2011b). The Panchina Formation deposits form few-meter-thick blankets on top of the wide marine terraces, which, in the Augusta area, form a flight reaching ∼200 m above sea level (a.s.l.). Conversely, Upper Pleistocene shallow-marine deposits are more localized and associated with marine terraces in the coastal belts of the Mount Tauro horst and Gesira headlands (e.g., Di Grande and Scamarda, 1973; Bordonaro et al., 1984; Di Grande and Neri, 1988). The Upper Pleistocene deposits are characterized by faunal assemblages including specimens of Strombus bubonius, which in the Mediterranean area marks fossils associations correlated with the Last Interglacial (MIS 5) highstand (Gignoux, 1913).

Costa Mendola Area

Within the Augusta Basin, in the Costa Mendola area, an elevated ridge known as Mendola horst occurs (I.S.P.R.A., 2011a, 2011b; Fig. 3A). The Mendola Horst mainly consists of calcarenites and calcirudites of the Miocene Monti Climiti Formation. Younger deposits in contact with the Miocene carbonates include: (1) Lower Pleistocene calcarenites and shales; and (2) Middle Pleistocene veneers of calcarenites and sands of the Panchina Formation, associated with marine terraces (refer to section “Quaternary Marine Terraces and Vertical Motions of the Augusta Area”). In the northern sector of the Mendola horst, the Lower Pleistocene calcarenites rest unconformably on top of the Miocene limestones. The Lower Pleistocene shales, cropping out along the eastern flank of the Mulinello River valley, pinch out to the west and onlap the structural high. The stratigraphically overlying deposits of the Panchina Formation form discontinuous patches unconformably covering both the Lower Pleistocene calcarenites and the Miocene limestones, in both instances with the interposition—at some places—of a paleosol (I.S.P.R.A., 2011b, and references therein). Along the southern edge of the structural high, between Masseria Mendola and Balatelle (Fig. 3A), the contact between the Miocene limestones and the Lower Pleistocene shales occurs along a WNW-ESE–striking, SSW-dipping, high-angle fault. The deposits of the Panchina Formation in this area either sit stratigraphically on top of the Lower Pleistocene shales, or they are in tectonic contact with the Miocene limestones along the same high-angle fault. The highly degraded and eroded fault surface does not preserve indicators of precise fault kinematics, such as shear fibers or abrasion striae. However, it is clear from the stratigraphic separation that the fault has a significant normal dip-slip component of motion. WNW-ESE–striking, minor normal faults form a conjugate extensional set that is well exposed in Miocene limestones in the interior of the Mendola horst (Fig. 3A), roughly parallel to the main fault bounding the elevated ridge to the south.

Carbone (1985) showed how the Lower Pleistocene calcarenites represent a coarse-grained facies that is generally proximal to paleofault scarps. These deposits pass both laterally (toward the basin depocenter) and vertically to the Lower Pleistocene shales, which constitute a more distal facies and reach a thickness of several hundreds of meters in the center of fault-bounded tectonic depressions. Significant thickness variations of the Lower Pleistocene deposits along faults (e.g., the WNW-ESE–trending Costa Mendola fault; Fig. 3B) are confirmed by well log data recently provided by Firetto Carlino et al. (2013). Common synsedimentary deformation of the Lower Pleistocene calcarenites and yellow sands is mentioned by Carbone (1985).

Quaternary Marine Terraces and Vertical Motions of the Augusta Area

A flight of marine terraces and shorelines testifies to uplift coeval to Quaternary sea-level fluctuations of the Augusta area. However, the chronostratigraphic framework of marine terraces, and related uplift evaluation, is debated.

The lowest paleo–sea level indicators consist of notches carved at around 2 and 5 m a.s.l. in the Miocene carbonates of the Mount Tauro horst (I.S.P.R.A., 2011b). Higher marine terraces standing between 5 and 10 m (with inner rim around +15 m; I.S.P.R.A., 2011b) occur in the Mount Tauro and Gesira headlands. These terraces are blanketed with 1–5-m-thick Upper Pleistocene conglomerates and sands bearing Strombus bubonius (Di Grande and Scamarda, 1973; Bordonaro et al., 1984; Di Grande and Neri, 1988). Based on the recovery of further Strombus-bearing deposits, Cosentino and Gliozzi (1988; see also Bordoni and Valensise, 1998) related marine terraces at about +30–35 m in the Mount Tauro horst to the Last Interglacial. These authors estimated an uplift rate around 0.2 mm/yr since the late Pleistocene for the Mount Tauro headland. More recently, Antonioli et al. (2006) correlated the shorelines at +16 m at Mount Tauro (with a resulting 0.07 mm/yr uplift rate), and the +32 m marine terrace in the Climiti Mountains–Belvedere ridge with the 126 ka marine isotope stage (MIS) 5.5 (Lisiecki and Raymo, 2005). Marine terraces around 30 m a.s.l. at the northwestern boundary of Augusta Bay were correlated to the Last Interglacial by I.S.P.R.A. (2011a, 2011b). Higher (≥50 m a.s.l.) shorelines consist of tidal notches carved in the Climiti Mountains ridge, and large terraces generally topped by middle Pleistocene Panchina Formation deposits (I.S.P.R.A., 2011b). Collectively, six marine terraces with up to 150 m of elevation have been identified by Bianca et al. (1999) and Monaco et al. (2002). Catalano et al. (2010) distinguished 10 marine terraces, including higher subplanar, mostly erosional, surfaces resting up to 325 m a.s.l. in the uplands to the west of Augusta. These authors estimated a mean uplift rate of around 0.7 mm/yr for the last 240 k.y. (Bianca et al., 1999), and for the last 520 k.y. (Catalano et al., 2010). However, these estimates rest on a rather questionable marine terrace chronostratigraphic framework, which is based on dating of mammal fauna from continental deposits either overlying or underlying marine deposits in the Climiti Mountains–Belvedere ridge. Within this framework, the terraces at 15 m and 105 m a.s.l. are correlated with the 60 ka MIS 3.3 and 126 ka MIS 5.5, respectively, while the occurrence of the Strombus-bearing deposits and notches at around +2 and +5 m in the Mount Tauro horst are neglected. The resulting mean uplift rate of ∼0.7 mm/yr for the middle Pleistocene to present time span therefore appears to represent a large overestimation.

Although a large-scale southward tilting of the Hyblean Plateau during the Quaternary evolution of the forebulge area is generally recognized (see section “Geological Setting”), the Augusta area is interpreted as less uplifted with respect to the more southerly Siracusa area based on the elevation of the Last Interglacial markers and submerged Holocene markers (Ferranti et al., 2006; Antonioli et al., 2009; Lambeck et al., 2011). Based on submerged archaeological markers, Scicchitano et al. (2008) suggested that in the late Holocene, the central part of the Augusta Bay coastal belt recorded slower uplift (∼0.30 mm/yr for the last 2.6 k.y. in the Megara Hyblaea site; location in Fig. 2) with respect to the morphostructural highs to the south (0.68 mm/yr for the last 3.5 k.y. in the Magnisi peninsula; location in Fig. 2).

Hydrogeological Setting of the Augusta Area

The Augusta area hosts the urban center of Augusta (the second largest town in the province of Siracusa, with ∼36,000 inhabitants) and several industrial installations such as petrochemical manufacturing facilities, the activity of which started in the 1950s and was supported by the pumping of large volumes of groundwater.

The carbonate block of Mount Tauro hosts an unconfined aquifer (Mc, in Fig. 3), while the hydrogeological system of the Augusta Basin includes two carbonate aquifers. An upper, low-thickness (1–10 m) aquifer coincides with the Middle Pleistocene Panchina Formation sands and calcarenites, whereas a deeper aquifer is located in the Lower Pleistocene calcarenites and calcirudites (Qc, in Fig. 3) and the underlying Miocene carbonates (Aureli et al., 1987a, 1987b). The two aquifers are separated by an impermeable layer represented by overconsolidated Lower Pleistocene clays (Canova et al., 2012), the thickness of which ranges from a few meters to over 300 m in the offshore Augusta Bay (Carbone, 1985). Both aquifers of the Augusta Basin are in lateral contact with that of Mount Tauro along the central sector of the Mount Tauro fault. The permeability of the carbonate complexes is variable: The primary permeability ranges between 10−3 and 10−5 m/s, while the secondary permeability (favored by joints and karst caves, especially in the lower aquifer) is slightly higher, ranging from 10−1 to 10−2 m/s. Measured transmissivity ranges from 0.1 to 9 × 10−3 m2/s.

The aquifers are exploited by deep wells (depths > 400 m) for agricultural, industrial, and potable purposes. Since the upper aquifer is characterized by low thickness, the deeper aquifer is the most exploited. The piezometric level has been affected by a progressive decline caused by overexploitation essentially by the industrial installations (I.S.P.R.A., 2011b). The groundwater surface decline started in the 1960s and increased, particularly in the coastal area, in the 1980s (I.S.P.R.A., 2011b).

The groundwater surface and flow orientations of the lower, confined aquifer have been reconstructed based on measurements dating back to 2004–2005 (Fig. 4). A comparison of the groundwater surface shown in Figure 4 with that mapped by Carbone et al. (1986) based on data dating back to 1983 points to a decline of the piezometric surface of up to 70 m in some areas. In the coastal area, such a decline caused saltwater intrusion, as documented by an official report on the 2003–2006 monitoring by the Regional Hydrographic Office (Ufficio Idrografico Regionale, 2007).


In the mainland area to the west of Mount Tauro, seven main marine terraces (Fig. 5) in the elevation range of ∼170 m to 10–15 m a.s.l. have been identified. They are carved on variable bedrock types and generally topped by calcarenites of the Panchina Formation deposits, or younger—Upper Pleistocene—conglomerate and sands. The Upper Pleistocene terraces include the terraced surface standing at about +30 m to the NW of the Augusta peninsula (I.S.P.R.A., 2011a), and labeled t2 in Figure 5. Among such terraces, terrace t3 (standing at elevations ranging from ∼50 to 60 m and rising inland up to ∼65 m; Fig. 5) and the higher terrace t4 (elevation in the 90–105 m range) are carved in the Lower Pleistocene Augusta Basin deposits and in the underlying pre-Quaternary bedrock, which includes carbonates of the Mendola horst. Such relationships indicate that fault activity bounding the Mendola horst predated formation of terraces t3 and t4. Formation of terraces t3 and t4, which stand higher than—and predate—the Upper Pleistocene t2 terrace, may be related to the late part of the middle Pleistocene. The age of terraces t3 and t4 may be better constrained through correlation of the Augusta terrace flight with sea-level highstands. The correlation has been based on the sea-level curve derived by Waelbroeck et al. (2002), and the uplift rate (ranging between 0.2 and 0.3 mm/yr) inferred from elevation of the Last Interglacial shorelines in the Augusta area (e.g., Cosentino and Gliozzi, 1988; Bordoni and Valensise, 1998; Antonioli et al., 2006; Ferranti et al., 2006). Using such information, terrace t3 may be tentatively correlated with the MIS 7 highstands (with formation of its large surface mirroring successive reworking by the sea level in the around 200–240 ka time span), and terrace t4 can be tentatively correlated with MIS 9.3. Based on such correlation, fault activity at the Mendola horst boundaries is at least older than 330 ka.

Vertical motions along two fault systems that postdate the formation of some of the marine terraces are suggested by several pieces of evidence. One system, consisting of two main strands with a roughly N-S trend and length of ∼8 km, has been identified to the west of Costa Mendola by a distinct alignment of rectilinear scarps and straight stream incisions affecting terrace t4 (Fig. 5). The activity of such structures was probably responsible for capture of the Maccaudo River (which shows a distinct change in flow orientation to the west of the fault trace; Fig. 5), and for the outline of the coastal belt perimeter. This is inferred from both the straight—roughly N-S—trend the of rims of terrace t4 compared with the more sinuous boundaries of higher terraces, and the strongly variable width of terrace t3 from the north to the south (Fig. 5).

More recent activity is recorded by the NNW-SSE–oriented Mount Tauro fault system (Fig. 5). This activity is inferred from the lack of both terraces younger than terrace t3 to the north of the Augusta peninsula, and coastal landforms carved in the fault scarp bounding the Mount Tauro fault system to the west. These observations are in contrast with evidence from the remaining Mount Tauro perimeter belt, which is characterized by bays and coastal caves raised up to 20 m a.s.l. (I.S.P.R.A., 2011b, and references therein), and marine terraces. These terraces stand around 15 m (terrace t1), 30–40 m (terrace T2), and 50 m a.s.l. (terrace T3; Fig. 5). Recent vertical motions along the Mount Tauro fault system are also suggested by deformation of terrace t3. Terrace t3 is affected by segmented rectilinear scarps oriented roughly parallel to the main Mount Tauro fault and around E-W, pointing to progressive lowering toward both the west and the southeast of the terrace surface, and by slight tilting shown by the gentle dip of such surface toward the E-SE down to 40 m. All these observations points to activity of the Mount Tauro fault system that postdate formation of both terrace t3 and the lower (Upper Pleistocene) terrace t2 (Fig. 5). More importantly, the mapped fault scarps control the perimeter of the coastal plain to the north of the Augusta peninsula, suggesting that fault activity in the forebulge area has continued until the Holocene.


For the Augusta Basin area, Canova et al. (2012) discussed the results of an analysis of surface deformations obtained by the DInSAR technique using ERS (European Remote Sensing) images from years 1992–2000. Significant land subsidence in a NW-SE–trending area (including the Augusta urban area) was identified and primarily related to groundwater overexploitation.

Our DInSAR analysis is aimed at assessing the deformation trends of the area spanning from the coastline to the Costa Mendola horst to the west in the 2003–2010 time span. The first step of the analysis was the cut of a ROI (Region of Interest) of ∼12 × 8 km (Fig. 6) starting from the SLC images, which cover an area of 100 × 100 km. In order to perform a fine coregistration of SAR images and to remove the topographic contribution to the interferometric phase, we used an external digital terrain model (DTM) with a 10 × 10 m resolution cell (obtained from the Tinitaly Project; Tarquini et al., 2012). In this work, a SBAS approach was used, and from the whole set of interferograms, only those with a perpendicular spatial baseline smaller than 250 m and a temporal baseline shorter than 211 d were selected. Due to the use of a multilook factor of 15 × 3 pixels (azimuth × range), the ground resolution of the outputs was resampled to 60 × 60 m. Finally, in order to both have a sufficient number of points covering the entire study area, and display a phase standard deviation of ∼20° (which corresponds to a displacement standard deviation of ∼1.5 mm), two coherence threshold values (0,50 and 0,40 in more than 50% of the interferograms) have been set. In total, 45 images were available for the ascending track (track 129, frame 729), and 58 interferograms were selected based on the aforementioned baseline thresholds; the mean coherence map was then constructed (Fig. 7). This map allows identification of points characterized by different stable phase quality during the time acquisition; starting from the coherence map, the next step was to identify those points with above-threshold coherence values. Such processing allows the construction of a mean velocities map (Fig. 8). The map in Figure 8 shows that the highest displacement rates (∼25 mm/yr) are focused to the north of the Augusta urban area, while the area of Mount Tauro is identified as stable. Such displacement rate values are in good agreement with the results of Canova et al. (2012), based on processing of SAR data acquired in the 1992–2000 time span. It is worth nothing that both in the map of Canova et al. (2012) and in our map, the stable area (Mount Tauro horst) is separated from the unstable area (Augusta graben) by a NNW-SSE–trending lineament, which coincides with the Mount Tauro fault trace.

Fifty descending images (track 222, frame 2853) were processed. The same maximum baselines as those of the ascending images were used. In this case, and for the reasons discussed already, two coherence threshold values were identified and set at 0.50 and 0.40, respectively. A measured displacement rate map was then constructed (Fig. 9).

The results of both processing analyses show the same sign (positive, i.e., displacements in the sensor-target direction), indicating that the predominant component of the displacement is the vertical over the horizontal one, as schematically shown in Figure 10.

As reported in Cascini et al. (2010), the combination of ascending and descending results allows the evaluation of vertical displacement rate (Fig. 11). The map in Figure 11 shows that the subsidence rate is highest to the north and west of the Augusta peninsula (in particular, around Contrada Saline site). It is worth pointing out that the spatial distribution of the highest displacement rates is independent from that of the wells, suggesting that the subsidence phenomenon is not correlated with water extraction. Figure 11 also shows that such subsidence rate tends to decrease with a radial pattern toward the west and south. On the other hand, a correlation of the deformation trend with anthropogenic factors may be envisaged in the Augusta industrial area. In this area, a slight subsidence increase is observed. Such increase is probably related to the presence of the main industrial installations (petrochemical) of the Augusta area, as suggested by Canova et al. (2012).


Data Interpretation

The onset of activity of the NW-SE–trending fault system controlling the Mendola horst may be inferred using the sedimentary record associated with related tectonic subsidence, as fault activity produced the accommodation space for syntectonic (i.e., synrift) deposits in this sector of the forebulge. Although a late Pliocene initiation of faulting, as originally suggested by Carbone (1985) and recently reproposed by I.S.P.R.A. (2011b) cannot be ruled out, it seems likely that the deposition of the Lower Pleistocene proximal calcarenites marks fault activity. These deposits may be interpreted as the coarser facies of the synrift succession, deposited close to the fault scarps. As demonstrated by Carbone (1985), these sediments pass laterally (toward the basin depocenters) and upward to Lower Pleistocene gray-bluish clays, which appear to represent the distal facies of the synrift succession, at least for their lower portion. In fact, the upper part of the gray-bluish clays, stratigraphically overlying the Lower Pleistocene proximal calcarenites and locally resting unconformably on top of the Miocene substratum, may be associated with a generalized postrift subsidence. It is worth noting that the model proposed by Carbone (1985) involves faulting during the Pliocene—a period characterized by a lack of sedimentation in the study area—and a cessation of fault activity by the early Pleistocene, when, according to Carbone, sedimentation would have been passively controlled by the preexisting paleobathymetry. Such a model appears to be inconsistent with the available stratigraphic data, as it implies a lack of a synrift sequence, which typically marks the activity of normal fault systems producing significant tectonic subsidence. Our interpretation is consistent with that proposed by Adam et al. (2000), who, for the NW-SE–trending fault system controlling the Mendola horst, referred to a tectonic collapse that led to the development of horsts and grabens at the beginning of the Pleistocene. It is worth noting that Carbone (1985), who suggested that the NW-SE–trending fault system was already “delineated” during the Pliocene, does not rule out its “reactivation” during the early Quaternary. On the other hand, cessation of differential vertical motions of the Mendola horst during the middle Pleistocene is constrained by coupled stratigraphic and geomorphological evidence. In fact, the faults bounding the horst are sealed by marine terraces and correlative deposits, which developed roughly parallel to the coast both across the Mendola horst and the adjacent grabens. In particular, the crosscutting relationships between the fault bounding the Costa Mendola ridge to the southwest and the marine terraces indicate that relative uplift of the Mendola horst at least predated the formation of the t4 marine terrace, which we tentatively correlate with the 330 ka MIS 9.3. The end of Mendola horst relative uplift is also inferred from the spatial distribution of the younger t3 marine terrace (that is carved in both the Mendola horst and adjacent graben blocks), and by the subsequent superimposition of the Torrente Molinello—and its left tributary—valley through the Costa Mendola ridge (Fig. 12). All this evidence indicates that normal fault activity is at least older than 400–300 ka. The DInSAR analysis clearly shows the lack of present-day vertical motions along the Costa Mendola fault.

On the other hand, the geomorphological evidence indicates that the NNW-SSE–trending Mount Tauro fault system has been active (or reactivated) in very recent times and has controlled the relative lowering of an area including the Augusta coastal plain. Significant constraints on the vertical motions in the town of Augusta and surrounding area have been obtained by the DInSAR analysis performed in this work. The analysis clearly highlights different vertical velocities affecting the footwall and hanging-wall blocks of the Mount Tauro fault system, with the latter block being subject to subsidence in the 2003–2010 time span. The pattern of vertical motions clearly points to a tectonic origin of the recorded vertical displacements, with the anthropogenic effects suggested by Canova et al. (2012) being minimal or even negligible, as is also inferred from hydrogeological information. This information indicates that the spatial distribution of high subsidence values is not correlated with that of the wells for water extraction (Fig. 11), and the lows in the groundwater surface of Figure 4. This lack of correlation suggests that the strong subsidence affecting the Augusta area in the 2003–2010 time span cannot be interpreted as the direct response to consolidation triggered by groundwater extraction/overexploitation, which dates to previous decades (a strong decline of the piezometric surface of the lower aquifer has been framed in the time span ranging from the 1960s to the 1980s; I.S.P.R.A., 2011b). In addition, since the Lower Pleistocene clays are overconsolidated, consolidation triggered by the decline of the piezometric surface would have affected only a limited part (i.e., the alluvial deposits and Lower Pleistocene calcarenites/sands; Fig. 3B) of the sedimentary succession overlying the Miocene carbonates in the Augusta Basin. All this evidence indicates that water extraction and related phenomena (essentially consolidation, taking into account that salt-wedge intrusion, although causing changes in the groundwater chemistry, is not responsible for subsidence) may only account for a small amount of the subsidence observed in the area to the southwest of the Mount Tauro fault system, although an anthropogenic contribution to subsidence of the Augusta industrial area (see section on “Interferometric Data Set and Results”) may be envisaged. Furthermore, evidence for recent activity of NNW-SSE–trending faults in the Augusta area is consistent with information from offshore Augusta from Firetto Carlino et al. (2013) and Pirrotta et al. (2013), which points to vertical offset affecting seismic reflectors associated with Upper Pleistocene–Holocene deposits along NNW-SSE–trending faults (e.g., the Augusta offshore fault in Fig. 3B).

Tectonic Implications

Forebulge-related tectonic features discussed in this study provide further insights into the magnitude, modes, and timing of development of structures that are commonly incorporated in fold-and-thrust belts. These prethrusting structures may play an important role in subsequent deformation (e.g., Butler, 1989; Scisciani et al., 2001; Mazzoli et al., 2002; Tavarnelli and Peacock, 2002). Increasing evidence suggests that these features represent very important, widespread background structures in rocks later involved in fold-and-thrust belts (Calamita and Deiana, 1980; Scisciani et al., 2001; Lash and Engelder, 2007; Casini et al., 2011; Beaudoin et al., 2012; Quintà and Tavani, 2012; Tavani et al., 2012). Detailed structural surveys carried out by Billi et al. (2006) over the whole Hyblean Plateau documented coeval, multiple structural associations consisting of normal faults and joint systems. According to these authors, the development of two mutually orthogonal sets of major extensional structures occurred since Langhian time in response to bimodal stretching along both NW-SE and NE-SW directions. These results were consistently interpreted as evidence for the growth of a doubly plunging forebulge, controlled by the occurrence of foreland crustal heterogeneities. The latter would have also enhanced differential retreating processes of the foreland along the subduction zone and the subsequent formation of orogenic salients and recesses (Billi et al., 2006). Within this framework, strike-slip and extensional faulting along the Malta Escarpment and the Scicli Line (Grasso and Reuther, 1988; Grasso, 1993; Adam et al., 2000; Argnani and Bonazzi, 2005) is inferred to have accommodated part of the differential retreating kinematics of the Hyblean foreland monocline and part of the extensional strain induced by the doubly plunging flexure (Billi et al., 2006). The Costa Mendola normal fault, which was active for a time span in excess of 1 m.y. during the early to middle Pleistocene, appears to represent one of the most recent extensional structures associated with development of such a doubly plunging forebulge. In fact, earthquake fault plane solutions and borehole breakout data indicate that the present-day tectonic setting of the study area is dominated by NW-SE–oriented shortening and NE-SW extension, within the general framework of an active strike-slip regime (Ragg et al., 1999; Adam et al., 2000; Goes et al., 2004). This is also consistent with the principal strain-rate axes obtained from interpolation of the global positioning system (GPS)–derived velocity field over the study area (Devoti et al., 2011). Active shortening normal to the strike of the forebulge axis is clearly incompatible with the tectonic setting expected by bending around a roughly NE-trending horizontal axis as a result of flexure of the foreland plate. The deactivation of NW-SE–striking extensional structures such as the Costa Mendola fault discussed in this study further suggests that orthogonal bending around a NW-trending horizontal axis, characterizing the development of the doubly plunging forebulge (Billi et al., 2006), also ceased during the middle Pleistocene. All together, this evidence indicates that the rollback process controlling the tectonic evolution of the Maghrebian-Sicilian fold-and-thrust belt during Neogene to early Pleistocene times has come to a stop, similar to what has been documented for the Southern Apennine segment of the same orogen. There, a major geodynamic change occurred at the beginning of the middle Pleistocene (at ca. 0.7 Ma; e.g., Cinque et al., 1993), when complete slab detachment is inferred to have taken place, following a southeastward along-strike slab tear migration that initiated during the Pliocene, at ca. 4 Ma (Ascione et al., 2012). A similar process probably occurred in Sicily as a result of an eastward slab tear migration along strike of the Maghrebian-Sicilian belt (Neri et al., 2009), leading to the final complete slab breakoff in the Calabrian arc sector at ca. 0.7 Ma. Following this event, NW-oriented shortening associated with Africa-Eurasia plate convergence (Mazzoli and Helman, 1994) became dominant over both the Adriatic-Apulian and Hyblean forelands. Active deformation of the latter is consistent with that recorded across the whole North African continental margin south of Sicily, which is characterized by major E-W–trending, right-lateral, strike-slip fault zones along the Sicily Channel (e.g., Jongsma et al., 1985). This consistency provides further evidence for the Hyblean Plateau presently behaving as part of the larger Pelagian block, where lithospheric bending–related deformation has ceased in the forebulge sector. Indeed, significantly reduced or even negligible rates of southeastward motion of Calabria characterize the last 1 m.y. (Mattei et al., 2007; Johnston and Mazzoli, 2009), while GPS geodesy indicates that rapid southeastward motion of Calabria has stopped (D’Agostino and Selvaggi, 2004). Cessation of subduction (e.g., Westaway, 1993) is consistent with the observation that slow convergence across the Africa-Eurasia plate boundary (Mazzoli and Helman, 1994) is presently accommodated along a seismic belt dominated by thrust faulting offshore northern Sicily (e.g., Billi et al., 2007), rather than at the thrust front.

The earthquake focal mechanism pattern depicted in Figure 1 and the short-term vertical motions documented in this study in the Augusta–Mount Tauro area indicate that significant dip-slip extensional movements accompanied the dominant strike-slip tectonic regime that is active in SE Sicily. The middle Pleistocene to present uplift of SE Sicily, characterized by maximum values along the front of the Maghrebian-Sicilian belt and by a pattern of decreasing values to the south, is also consistent with the rebound effect following complete slab breakoff (Cinque et al., 1993; Ascione et al., 2012).


The integrated approach used in this study constrains the modes and duration of activity of Quaternary fault systems in a peculiar area of the African continental margin representing a foreland indentor during major plate convergence in the Mediterranean area. Modes of forebulge deformation, being strongly controlled by subduction dynamics, record a dramatic change in the slab configuration beneath the study area (Fig. 13).

Bimodal extension associated with bending of the foreland plate during the development of the doubly plunging Hyblean forebulge produced two mutually orthogonal normal fault systems, one parallel to the NE-trending main forebulge axis, and another parallel to the secondary NW-trending axis. Detailed analysis of the stratigraphic record allowed us to constrain the timing of activity—not older than ca. 1.8 Ma and not younger than ca. 0.3 Ma—of an extensional structure (the Costa Mendola fault) belonging to the NW-trending system. Therefore, this well-documented Pleistocene structure appears to have been active for not longer than ∼1.5 m.y. during the latest stages of rollback associated with NW-directed subduction.

Subsequent late Quaternary to present deformation is dominated by regional uplift, progressively increasing toward the thrust front to the NW, and strike-slip faulting controlled by NW-oriented horizontal compression. Differential uplift within the Hyblean Plateau is accommodated by dip-slip components of motion along active NNW-trending structures such as the Monte Tauro fault. This fault probably forms part of the larger Malta Escarpment fault system, which is known to be active and characterized by large recent vertical offsets at the same latitude (Argnani and Bonazzi, 2005). The overall active tectonic setting—dominated by NW-oriented horizontal compression consistent with major plate convergence—and the regional uplift pattern characterizing the study area may be explained within the framework of intraplate shortening and foreland rebound following complete slab breakoff, a process that has been inferred to have occurred at ca. 0.7 Ma (Cinque et al., 1993; Ascione et al., 2012).

We wish to thank Andrea Billi, an anonymous reviewer, and Editor Arlo Weil for their thoughtful and constructive reviews, which substantially helped us to improve the manuscript.


SUBSOFT processor software, based on the coherent pixels technique (Mora et al., 2003; Blanco-Sánchez et al., 2008; Iglesias et al., 2014), implemented at the Remote Sensing Laboratory (RSLab) of the Universitat Politecnica de Catalunya (UPC), is able to extract the evolution of deformation from a stack of differential interferograms over wide areas during large time spans.

The SUBSOFT processor is composed of two utilities for interferometric processing, PRISAR and SUBSOFT. PRISAR consists of a set of routines implemented for the computation of the differential interferograms and the coherence maps; SUBSOFT is an application that uses PRISAR’s output (differential interferograms and coherence maps) to extract the linear and nonlinear deformation evolution of the study area.

The interferometric process starts with PRISAR routine steps:

  • (1) The first step is selection of the best SLC image pairs among all the available images of the area under investigation. In order to carry out the selection, from all SLC images, the spatial baseline (distance between orbital positions), the temporal baseline (temporal distance between acquisitions), and the Doppler frequency (Df) are considered. The aim of this step is to identify the minimum number of interferograms in the stack that have the maximum quality overall.

  • (2) Starting from SLC images, coregistration between each image is carried out in order to evaluate the interferometric phase differences (interferogram). The coregistration consists in spatial registration and eventually resampling (SLC with different pixel size) in order to adjust for relative translational shift and rotational and scale differences.

  • (3) Precise satellite orbits and DTMs are used to create the differential interferograms and interferometric coherence maps.

From these outcomes, the SUBSOFT routine is implemented:

  • (4) Since not all pixels are characterized by stable phase, a pixel selection is carried out. The SUBSOFT processor is implemented with different criteria of selection: coherence stability (Berardino et al., 2002), amplitude dispersion (Ferretti et al., 2000), and temporal sublook coherence (Iglesias et al., 2014).

  • (5) The last step consists of phase analysis to calculate linear and nonlinear components of deformation. Through the linear model, a measured displacement rate map is obtained, while the nonlinear model allows time series of deformations to be obtained.