Geodetic surface measurements in SE Sicily include GNSS e.g.,, PS-InSAR/DInSAR (Differential Interferometry Synthetic Aperture Radar) e.g.,, and leveling datasets e.g.,.

The instrumental seismicity map of SE Sicily, derived from INGV and the datasets (Fig. ), shows minor to moderate events (M<5) with deep crustal hypocenters (15–30 km). Over the Hyblean Plateau, earthquake hypocenters tend to roughly align along the inferred active, N–S trending, Scicli–Ragusa strike-slip fault e.g., and near the Cavagrande Canyon fault system (Fig. ). Most of these faults are probably inherited from the Plio-Quaternary tectonomagmatic phase of deformation and were partly re-activated in response to the ongoing Africa–Nubia/Eurasia plate convergence e.g.,. In this framework, the identification of the seismogenic source that triggered the 1693 event remains debated e.g.,. The isoseists of the Mw∼7.4 Noto earthquake appear largely open toward the Malta Escarpment and Ionian Sea domains, suggesting the seismogenic fault is located offshore (Fig. ). East of the Hyblean Plateau, earthquakes essentially distribute along the Malta Escarpment where a normal fault system, potentially responsible for the 1693 earthquake, has been identified e.g., (Fig. ).

The focal mechanisms over the Hyblean Plateau have dominant strike-slip characteristics, contrasting with the extensional deformation characterizing the NE corner of Sicily (Fig. ).

To estimate the present-day regional stress field across SE Sicily, we analyzed the available focal mechanisms using Vavryčuk's numerical model that is based on Michael's method . Results show that the regional stress across SE Sicily (Fig. ) is homogeneous (Figs. S7 and S8). The maximum compressive stress (σ1) is horizontal and oriented N154° E ± 7°, compatible with the N160° E Africa/Eurasia plate convergence e.g.,. The minimum stress (σ3) is oriented N64° E ± 7°, compatible with the extension rate derived from GNSS data inversion (Fig. ).

If this regional stress field is compatible with the PS-InSAR surface deformation data (E–W bending generating extensional stress), it does not explain the observed eastward-increasing subsidence rate across the HP.

To constrain the deep structure and rheology of the studied area, we synthesize the available geological and geophysical data into a 200 km long simplified crustal-scale structural cross section following the N30° E AB profile. This section incorporates part of the Hyblean Platform, the Malta Escarpment, the western Ionian domain, and cuts, almost perpendicularly, the offshore normal faults along the Malta Escarpment and the Alfeo and Ionian strike-slip fault systems, extending eastward (Figs. –). The eastern part of the synthetic structural profile is mainly based on seismic refraction profiles from , particularly the DY-P3 profile running sub-parallel to the AB profile and located 20 km further north, as well as seismic reflection profiles from , , , , and (Fig. c). The structure of the western section is constrained by onshore and offshore geology, well log stratigraphy, geophysics, seismic reflection profiles, and geological cross sections from the project, , , , , and .

In the Hyblean domain, geophysical data e.g., indicate that the crust has an average thickness of ∼30–35 km, with a notable difference in the Hyblean Plateau region marked by a huge positive Bouguer anomaly. Based on gravity data modeling, showed that this gravity anomaly can be explained by a 100 km large, high-density, and lower-crustal body, compatible with a local Moho uplift to a depth of about 20–25 km. This last interpretation also seems to be supported by recent tomographic data . We constrain the geometries of the Quaternary to Mesozoic sedimentary units of the Hyblean Platform and Gela basin, and these are constrained using the Monterosso 1, Plinio Sud 1, Troitta 1, Vittoria 3, and Vizzini 1 wells from the project (in pink; Figs. c and S9); the Chiaramonte 1 and Mellili 1 wells from ; and Buccheri 1–2, Comiso 3, Giarratana 1, and Licodia 1 wells from (in purple; Figs. c and S9). We also used the top of the Upper Triassic (Gela formation) isobaths published by .

In the DY-P3 seismic refraction profile , the 6.0 and 6.5 km s-1 velocity contours delimit two main steps deepening eastward at the junction between the Hyblean continental and Ionian oceanic domains (Fig. a and b). Considering their locations along the Malta Escarpment that outline the continent–ocean transition (COT), we interpret these velocity variations as the deepening of the sediment/basement boundary, potentially related to tilted blocks of thinned continental crust formed during the Permo-Triassic–Early Jurassic rifting phase (see Sect. 1) e.g.,. Our interpretation of tilted blocks at the continent–ocean transition is consistent with similar considerations analyzing seismic reflection/refraction profiles e.g.,.

As documented in , , and , the seismic reflection profiles (MESC-O6, MESC-11, CIR-01, MESC-08, and MESC-09) show several normal faults bounding and crossing the Turbiditic Valley, extending along the base of the Malta Escarpment . The Turbiditic Valley fault system is constituted by three parallel normal faults, ∼60 km long, producing a marked morphological offset of the Ionian seafloor from the latitudes of Catania to Siracusa (Fig. a and b). These faults dip 35–50° to the east and most probably merge at depth into a single major fault plane (; see MESC-08 and MESC-09 seismic reflection profiles in ). These offshore normal faults could be linked to the recent re-activation of crustal faults at the ocean–continent transition inherited from the Early Mesozoic rifting phase (Fig. a and b).

On the eastern side of the Hyblean domain, the Moho is constrained by DY-P3 and DY-P1 refraction profiles to a depth of ∼30 km below the Malta Escarpment. To the east, in response to the bending of the Ionian slab, the Moho deepens northward from 20 km (DY-P1) to 32 km (DY-P3). Based on these data and the DY-P4 refraction profile , we estimate the depth of the Moho below the Ionian oceanic crust to be about 25–30 km in the eastern part of the AB synthetic profile. In this region, the domain delimited by the seismic refraction velocities of 3.8–5.1 km s-1 has been interpreted as corresponding to the deformed sediments of the Calabrian accretionary prism (CAP) . Its thickness increases from 5 km (DY-P1) to 15 km (DY-P3), and it is evaluated to be ∼15 km along the AB profile (Fig. a and b). Note that a portion of the southern termination of the Calabrian Arc (i.e., Hercynian basement) is probably present in the AB profile according to the seismic refraction DY-P4 profiles (Fig. a and b). The location of the main subduction décollement along the AB profile has been estimated at a depth of ∼20 km (thick red line in Fig. a), using the velocity of 6.75 km s-1 seismic refraction in the DY-P3 and DY-P4 profiles .

To better constrain key flexural parameters, such as the rigidity of the Hyblean and Ionian crust/lithospheres and the slab pull force, and to investigate the impact of the Ionian slab roll-back, we first model the bending of the subducting Ionian slab along a NNW–SSE profile (CD profile) trending orthogonal to the AB profile (Fig. a). We compare the Ionian slab geometries with the and datasets, with the depth of the top oceanic crust from the seismic refraction data (Fig. S10). In the southern part of the CD profile, the dataset indicates shallower depths (∼5 km) compared to the and data because the main décollement jumps away from the top of the Ionian oceanic crust to a higher level in the sedimentary cover (Fig. S10). Note that in the northern part of the CD profile, the dataset indicates also a shallower depth compared to the dataset.

Finally, we decided to use, as a structural reference, the isobaths of the top of the Ionian slab published by because it correlates with the top of the oceanic crust depths derived from the seismic refraction data (Fig. a).

The lithosphere flexure models (as well as those in section 3.2) are calculated using the gFlex software . We impose a no-displacement condition at the southern profile boundary and a broken plate with no bending moment and no shear at the northern boundary. The Ionian oceanic lithosphere is modeled assuming an effective elastic thickness (Te) ranging from 25 to 37 km (Figs. b and S11) that is compatible with its Triassic to Early Jurassic age e.g., and consistent with other publications e.g.,.

The flexure of the subducting slab depends on its mechanical properties and the loads induced by the sedimentary cover, the accretionary prism, and the slab pull force (Fig. b). According to seismic refraction profiles DY-P1 and DY-P4 , the undeformed ante-Messinian sedimentary cover overlying the Ionian crust has a thickness of about 5 km. Thus, taking into account a depth of the Ionian Sea of 5–6 km, we consider that the top of the Ionian crust was lying at a uniform depth of 10–11 km before the onset of the Calabrian subduction system (Fig. b). This depth corresponds to the isostatic equilibrium for the Ionian crust. It determines the initial geometry of the flexural model from which we calculate the bending induced by the Calabrian accretionary prism (CAP) load.

Based on seismic refraction profiles DY-P4, DY-P1, and DY-P3 , the Calabrian accretionary prism thickness increases northward from 5 to 15 km. By removing the initial 5 km thick Ionian sedimentary cover, the CAP load represents an increase in sediment thickness from 0 km at the southern end of the CD profile to 10 km at the northern end. The Calabrian backstop, made of Hercynian continental crust, is not taken into account (Fig. b).

The CAP load is calculated by 1FCAP=ρgh, with a sediment density (ρ) of 2500–2800 kg m-2 (profile 2D), using ; a gravity acceleration (g) of 9.81 m s-2; and an increase in the CAP thicknesses (h) from 0 to 10 km. We also calculated the CAP load using an end-member density of 2800 kg m-2 (Fig. b), which resulted in a variation in flexure amplitude of a few percent and thus not affecting the results of continental flexural models.

The CAP load (FCAP) is applied on the CD profile divided into 1 km long segments by imposing a northward linear gradient from 0 to 2.45×108 N m-2 (Eq. 1) on the first 250 km of the profile (Fig. b and c). We perform several tests with different maximum CAP load (FCAP) and elastic thicknesses (Te) ranging from 2×108 to 3×108 N m-2 and 25 to 37 km, respectively. Models are tested with a constant mantle density of 3300 kg m-2 and no filling density for a mantle restoration force (Fig. c). The resulting flexure (∼8 km maximum), even if significant, is not sufficient to fit the Ionian slab profile (gray line in Fig. b and c).

The slab pull force is then added to the northern termination of the Ionian lithosphere as a point load (Fig. b). We tested with different slab pull forces ranging from 1×1012 to 4×112 N, consistent with other publications reviewing slab rollback mechanical properties e.g., and the same range of elastic thicknesses from 25 to 37 km (Figs. c and S11). The best fit to the Ionian slab-top profile is obtained for elastic thicknesses (Te) of 30–32 km, a maximum accretionary wedge load (FCAP) of 3×108 N m-2, and a slab pull force (FSP) of 2×1012 N (Figs. c and S11). It is worth noting that including the CAP load significantly reduces the amplitude of the fore bulge associated with slab bending, resulting in a flat-and-ramp geometry similar to that of the Ionian slab.

The impact of the Ionian subduction roll-back on the deformation of the Hyblean Plateau is evaluated along the N30° E trending AB profile (Fig. a), considering the following simplifications: (1) the ongoing roll-back induces incremental changes in the slab profile that can be matched with a southward translation of the slab geometry, inducing a local deepening. (2) This results in a local incremental increase in the accretionary prism thickness. (3) Due to the mechanical coupling of the Ionian slab and Hyblean lithosphere, the slab deepening exerts an incremental downward force on the COT (Fig. ).

The effective elastic thickness of the Hyblean lithosphere is less constrainable than that of the Ionian lithosphere but should remain within standard values for a regular undeformed continental crust with an average geotherm. We test elastic thicknesses (Te) ranging from 25 to 40 km (Fig. ), assuming a uniform thickness, considering that the continent–ocean transition and the oceanic lithosphere have the same elastic rigidity as the Hyblean crust. Finally, we also considered that none of the fault systems offshore SE Sicily are mature enough to significantly affect the mechanical properties of the abovementioned crustal/lithospheric blocks e.g.,.

We first evaluate the flexural response due solely to the local incremental increase in the CAP load induced by its southward migration, using our previous analysis of the bending of the Ionian slab. Based on the velocities of the GNSS stations situated in Calabria, we estimate the southward migration to 3 mm yr-1 compared to a fixed Hyblean Plateau . At the intersection between the AB and CD profiles, at the 170 km length mark in the CD profile, the Ionian slab dips 6±1° toward the north (Fig. d). Taking into account the CAP geometry, its southward motion, and the slab geometry, we calculate a local incremental thickening of the CAP of 3×10-4 m yr-1 (equivalent to 300 m Myr-1) and a resulting load (FCAPa) of about 5–10 N m-2 yr-1 (Fig. d). Applying a linear load gradient starting from zero at the base of the Malta Escarpment (140 km marks of the AB profile) to 5–10 N m-2 yr-1 at the end of the continent–ocean transition (165 km marks of the AB profile) and then applying this load constantly until the end of the AB profile results in a slow onshore subsidence rate of 1.5×10-4±5×10-5 mm yr-1 maximum, 20 000 times smaller than the PS-InSAR subsidence rate measured in the same area (∼3 mm yr-1).

We then investigate the effect of the southward Ionian slab roll-back and associated downward pull on the COT. We first calculate the flexural rigidity of the oceanic lithosphere , 2D=ETe312(1-ν2), with a Young modulus (E) of 1×1011 Pa, a Poisson's ratio (ν) of 0.25, and an effective elastic thicknesses (Te) of 30–32 km (see Sect. 3.1). We obtain a flexural rigidity (D) of the Ionian lithosphere of 2.4–2.9×1023 Pa m-3.

To simulate the Ionian slab retreat, we translate the slab profile southward, assuming a slab retreat velocity of ∼3 mm yr-1 (Fig. d). At the intersection of profiles AB and CD, this induces an incremental deepening of the Ionian slab of about 3×10-4 m yr-1 (equivalent to 300 m Myr-1), which defines the equivalent downward force at the same location along the CD flexure profile , 3FB=ω2Dx2L-x3, with an incremental deflection (ω) of 3×10-4 m yr-1 (Fig. d) and a flexural rigidity (D) of 2.4–2.9×1023 Pa m-3. The total profile length L corresponds to the point of the Hyblean lithosphere where the deflection (ω) is null, with ∼200 km based on the PS-InSAR and structural data (Fig. ). The distance x corresponds to the point at which the deflection (ω) is estimated (intersection with profile CD). Considering L=250±50 km and x=150 km, the equivalent incremental downward force is about 1–6.5×104 N m-1 yr-1.

This equivalent force (FB) is then applied on the AB profile to model, with gFlex, the resulting flexure of the Hyblean crust/lithosphere. Flexural models are calculated with a no-displacement boundary condition at the southwestern end of the profile (20 km west of Gela) and a free displacement of a horizontally clamped boundary condition at its northeastern end (80 km east of the Malta Escarpment). Flexural models are run with a fill density of 2500 kg m-2 (2D profile) solely for the CAP load. The downward force (FB) and CAP load (FCAPa) are applied as constant loads (on 1 km long segments) over the 35 or 60 km long portion of the AB profile corresponding to the only adjacent Ionian crustal domain and from the base of the Malta Escarpment to the end of the COT as a linear load gradient evolving from zero to the maximum calculated load. We test different elastic thicknesses (Te) and a bending force (FB) ranging from 25 to 40 km and 1×104 to 6.5×104 N m-1 yr-1, respectively (Figs. b and S12).

To determine the best Hyblean crustal flexure models, we first filter the PS-InSAR vertical velocities (5 km stacked in the AB profile) using a 5 km width median filter with a step of 1 km. Comparing the resulting long-wavelength trend of the PS-InSAR data with all flexural models shows maximum misfits of about 12 mm yr-1. The comparison between the GNSS data (20 km stacked of the AB profile and 5 km large median filter with a step of 1 km) shows a little bit of a higher maximum misfit of about 13 mm yr-1 due to a variable spatial density and quality of GNSS stations over the Hyblean Plateau (Fig. S12c). The best models (0.5 mm yr-1 RMS PS-InSAR) have a CAP load plus a bending force ranging from 1×104 to 3×104 N m-1 yr-1 distributed on a 35 km long portion of the AB profile and also between 1×104 to 1.5×104 N m-1 yr-1 distributed on a 60 km long portion of the AB profile, with the effective elastic thicknesses ranging from 25 to 40 km (Figs. b and S12b, c). None of the tested continental crustal flexure models reproduce the short-wavelength deformations observed in the Gela region (slow uplift of ∼0.5 mm yr-1) or along the Augusta–Siracusa coastal area (relative uplift of 1–2 mm yr-1).

Along the coast, from Augusta to Siracusa, PS-InSAR vertical velocities vary at a kilometer scale and appear 1–3 mm yr-1 slower than the general trend of subsidence affecting the eastern Hyblean Plateau (Figs. a and b). Interestingly, these short-wavelength signals show triangular patterns similar to those produced by shallow faulting in an elastic domain. To investigate the sources of these surface deformations, we test several scenarios involving interseismic loading and aseismic creep on coastal and offshore faults.

Offshore, several active normal faults, outcropping along the base of the Malta Escarpment, have been identified, imaged, and documented in detail by , , and . Close to the coastline, the offshore Augusta–Siracusa fault (Fig. ) has also been considered a potentially active fault e.g.,. We use the Coulomb 3.4 software to impose different fault slip rates and geometric boundary conditions on these fault systems, assuming standard elastic properties (Poisson's ratio of 0.25; Young modulus of 80 GPa).

The fault plane geometries tested (strike and dip) are based on published field-trip observations and measurements . Fault locations are based on published geological/structural maps and on the presence of sharp gradients in the PS-InSAR velocity pattern. The imposed fault slip velocities result from a trial-and-error empirical approach. The objective, essentially, is to evaluate if aseismic slip on known and unknown faults could generate sufficient surface deformation to explain the measured surface deformation pattern.

The model predictions are compared to the PS-InSAR short-wavelength signals (Fig. b) obtained by removing the mean of the best-fitting flexural models (see Sect. 3.2) from the original geodetic dataset. Two patterns of relative uplifts of about 2.5±0.5 mm yr-1, gently tapering westward, can be identified near and to the SE of Augusta with a zone of relative subsidence of about -2±1 mm yr-1 in between them (Fig. a). We hypothesized that these surface deformations could be induced by fault slip along ENE-dipping normal fault systems (Fig. ).

The first set of models corresponds to interseismic locking of the shallow (0 to 10–15 km depth) sections of the main normal faults identified in the study area (Fig. b) and elastic loading by deep (>15 km depth) creeping sections. Regardless of the deep-fault geometry or slip rates, all of these models generate generalized long-wavelength subsidence rates incompatible with the geodetic data (dotted green line; Fig. S13). Thus, we dismiss interseismic loading as a potential mechanism to explain the short-wavelength surface deformation patterns.

The second set of models corresponds to shallow aseismic slip imposed on three offshore normal faults: the Augusta–Siracusa fault , the Malta Escarpment fault, and the Turbiditic Valley fault (Figs. a and S13). We decided to test the Malta Escarpment fault because it lies between the Turbiditic Valley active fault and the Augusta–Siracusa fault for which evidence of activity has been documented by as-yet-unpublished sparker lines acquired in the Augusta Bay (Giovanni Barreca , Carmelo Monaco, personal communication, 2024). The modeled faults (Fig. a) share a similar listric geometry, with a first fault plane dipping 70° NE and extending from the surface to 12 km depth (inferred brittle/ductile transition zone) and a second one dipping 20° NE and extending from 12 to 50 km depth (to limit boundary effects). We imposed the slip rates of 5 mm yr-1 on the first fault plane, based on the model (Fig. S13), and 1 mm yr-1 on the second plane to dampen the elastic deformation produced by slip on the shallow fault (Fig. a). Aseismic slip on these various faults produces coastal uplift rates, reaching at most ∼1 mm yr-1 for the Augusta–Siracusa fault, consistent with the PS-InSAR measurements east of Augusta (Fig. a). However, all the modeled offshore faults failed to reproduce the ∼2–3 mm yr-1 relative uplift rates measured west of Augusta (Fig. a and b).

The third set of models focuses on the surface deformation generated by aseismic creep on 70–80° ENE-dipping shallow coastal and onshore fault planes. We first simulate slip on the upper portion of the Augusta–Siracusa fault, but if this model succeeds in producing sufficient uplift east of Augusta, it fails to reproduce the relative uplift west of Augusta. Based on PS-InSAR data and structural evidence of regional onshore normal faulting e.g.,, we added to the previous Augusta–Siracusa fault model an 80° dipping onshore normal fault outcropping at the 106 km mark of the AB profile (sharp velocity gradient in the PS-InSAR data), with a slip rate of 3 mm yr-1 down to 10 km depth (light blue lines in Fig. a). The surface deformation generated by this dual creeping fault can explain the observed PS-InSAR relative uplift between the 103 and 106 km profile marks and 110 and 112 km. Note that imposing aseismic slip on the onshore normal fault alone fails to reproduce the subsidence east of Augusta (dark blue line in Fig. a).

The triangular patterns of sharp steps and associated lows in the PS-InSAR data could be also fitted by a three-fault model involving shallower aseismic creep (up to 5 to 8 km depth) and combining the onshore ENE-dipping fault (106 km mark), creeping at 3–4 mm yr-1, with an antithetic onshore WSW-dipping fault (110 km mark), creeping at 1 mm yr-1, and the Augusta–Siracusa coastal fault (112 km mark), creeping at 3–4 mm yr-1 (brown lines in Fig. a). We test the same configuration (two onshore faults and the Augusta–Siracusa coastal fault) with a fault plane propagating to the surface up to 500 m depth (Fig. a). This model, equivalent to a blind fault, induces vertical surface deformation (between the 106 and 110 km marks) about 0.2 mm yr-1 slower than the model starting to creep from the surface but remains consistent with the PS-InSAR data.

At present, however, there is no evidence of the existence of faults matching the ones used in the third set of models. All these ad hoc models illustrate that the short-wavelength geodetic signal along the eastern Hyblean Plateau coast could be explained by ongoing extension tectonics and creep on coastal normal faults.

To explore if other natural processes could explain part of the observed geodetic velocity patterns, we briefly investigate three alternative models.

We explain the eastward tilt and subsidence rates of the Hyblean Plateau as the flexure of the Hyblean continental crust/lithosphere induced by the southward migration of the Calabrian accretionary prism (CAP) and retreat of the Ionian subducting slab (Sect. 3.1 and 3.2). This model is based on the assumption that the geodetic data (GNSS and PS-InSAR), measured over a short period (5–15 years), are representative of the kinematic evolution of the studied region at the scale of a few hundred to 1000 years. In the absence of significant seismic events during the period of geodetic data acquisition, and considering that major earthquakes (M>7) in SE Sicily probably have a return period of more than 500 years, geodetic data are mainly recording interseismic elastic deformation and, possibly, a minor permanent one (fault creep, folding, and human-related surface deformation). Flexural modeling indicated that the increasing loading of the COT, induced by the southward propagation of the CAP, is not sufficient (Fig. b). The increase in bending force, imposed by a ∼3 mm yr-1 southward retreat of the Ionian slab, gives interesting positive results. This process could be strong enough to pull down the eastern termination of the Hyblean crust at velocities compatible with PS-InSAR measurements. However, we obtained this result considering that the Hyblean crust/lithosphere, the continent–ocean transition (COT), and the Ionian crust/lithosphere have similar mechanical properties. The role of the nearby Alfeo–Etna Fault system (AEF) is still under debate; even though it cuts the entire lithosphere starting in recent times (probably Middle–Late Pleistocene), it has probably not reached yet the stage of a slab tear (STEP) . We, therefore, considered the AEF to not be mature enough offshore SE Sicily to significantly alter the mechanical properties of the abovementioned crustal/lithospheric blocks. This assumption implies that the COT has a significantly rigid and potentially too strong rheology (Fig. ), as discussed hereafter (Sect. 4.2).

We used simple 2D elastic models based on parameters determined through analytical modeling of the Ionian oceanic lithosphere flexure using, as a reference, the Ionian slab geometry determined by and data (depth of the top of the Ionian crust) extracted from the refraction profiles published in . The use of more advanced numerical models (finite element model, FEM), including 3D modeling methods, would likely improve our first-order estimates. Similarly, the lateral variations in the Hyblean continental crust thickness and elastic properties are not accurately known. We used the available geophysical data , but it was not possible to constrain the Hyblean crust/lithosphere rheology with better confidence (Fig. ). Should such parameters become available in the future, they could be used to refine our Hyblean crust/lithosphere flexure calculations.

One of the other assumptions we made concerns the rate of increase in the slab-bending force due to the southward propagation of the Ionian slab roll-back. The calculated increase in the slab-bending force east of the HP is based on the estimated rate of southward retreat of the Ionian slab defined by the mean of the GNSS N–S horizontal velocities in southwest Calabria (using Malta island as a reference). However, this estimate may be understated if the Calabrian Arc migrates southward more slowly than the Ionian slab retreat, due to lateral mechanical interactions with the Apulian and African margins.

The short-wavelength relative uplift signal, observed in the geodetic data along the southeastern Sicily coast, must be driven by more shallow deformation mechanisms than those responsible for the long-wavelength eastward flexure of the HP (Fig. b). Kilometer-long surface deformations are typically related to upper crustal deformation processes e.g.,, so we test interseismic loading models on the inferred and identified onshore and offshore fault systems.

Slip on the Malta Escarpment and Turbiditic Valley normal fault cannot explain the observed deformation of the eastern coast of the Hyblean Plateau. Only creep on the Augusta–Siracusa coastal fault and antithetic structure induces onshore vertical deformation compatible with the geodetic data near Augusta. Interseismic slip (creep) on two onshore ENE and WSW 80° dipping faults and the Augusta–Siracusa coastal fault fit with the PS-InSAR data to the east of the AB profile. These faults could re-activate inherited Permo-Triassic to Early Jurassic NW–SE extensional structures, leading to the formation of the Augusta Graben and extending up to Siracusa e.g.,. Even if some seismic activity affects this region e.g.,, field evidence of recent (Holocene) tectonic activity has yet to be demonstrated.

Our results suggest that these faults should creep up to the surface or the near-surface (blind fault) to produce sufficient interseismic surface deformation in the footwall. In that later case, their surface expressions could correspond to gentle surface folding or to fold scarp morphologies e.g., rather than localized cumulated fault scarps.

High-precision leveling data acquired between 1970–1991 and analyzed by , reveals a remarkable ∼4 mm yr-1 velocity offset between benchmarks 107 and 113, both situated near the coast 5 km west of Augusta (Fig. c). This sharp vertical velocity gradient is correlated with a marked topographic step trending N–S and descending toward the sea. Northwest of Augusta, the leveling dataset also shows a ∼2 mm yr-1 offset between benchmarks 119 and 120 associated with a topographic step oriented E–W and facing north (Fig. b and c).

A morphostructural analysis of this region, using a 2 m resolution DEM, outlines sharp drainage incision anomalies oriented perpendicular to the identified topographic steps and potentially related to tectonic surface uplift (Fig. b). The topographic step between benchmarks 119 and 120 (Fig. a and d) could correspond to the Scordia–Lentini graben border e.g.,. The topographic anomaly between benchmarks 113 and 107, extending to the north up to the Ionian Sea and to the south toward Siracusa, was not previously identified as a tectonic feature. It could correspond to the implemented creeping fault used to match the PS-InSAR data. Uplifted late Quaternary marine terraces have been evidenced in this region , but the authors did not mention a tectonic origin for the measured coastal uplift. Finally, the measured fast surface uplift (1–2 mm yr-1) could be considered inconsistent with the low amplitude of the topographic scarp measurable in the field (a few tens of meters). This point is discussed hereafter (Sect. 4.2).

The subsidence and tilt patterns observed in the geodetic data can be explained by the combination of (1) the flexure of the Hyblean continental crust induced by the bending forces generated by the Ionian subduction roll-back (slab pull) and the CAP overload, explaining the long-wavelength deformation affecting the HP, and (2) the aseismic activity of the Augusta–Siracusa fault system, potentially extending onshore an inferred tectonic structure and explaining the short-wavelength deformation signal affecting the Augusta/Siracusa region (Fig. ). In this section, we discuss how this short-term (geodetic) model could be combined with long-term geological and tectonic observations.

Interestingly, along the N30° E trending AB synthetic profile, a ∼1° generalized eastward tilting of the HP topography can be evidenced (Fig. a). The origin of this tilt, in apparent agreement with the geodetic data, could be linked to the Plio-Quaternary formation of the HP . Indeed, geological analyses suggest that the eastern coast of SE Sicily has been relatively stable over the last million years, with maximal subsidence and uplift amplitudes of ±0.2 mm yr-1 . More recently, dating of late Quaternary marine terraces along the Siracusa–Augusta coastal domain indicates that the eastern coast of the Hyblean Plateau has experienced a slow constant uplift during the last 500 kyr, increasing northward from 0.1 to 0.4 mm yr-1 . On a shorter historical timescale based on Roman archeological site studies, propose that the Siracusa coast has been slowly uplifting during the last 4 kyr, albeit with significant uncertainties. These long-term observations, extending from the Quaternary to historic time, point to slow regional uplift that is apparently at odds with geodetic data. However, it should be remembered that we have considered that PS-InSAR measurements primarily document the interseismic phase. At this stage, the part of the seismic cycle that generates uplift has not yet been taken into account. Previous calculations show that a Mw=7 earthquake on the active fault system at the base of the Malta Escarpment generates little coastal uplift, but early and late post-seismic deformation were not considered. In addition, a 500-year seismic cycle contains other earthquakes contributing to surface deformation than just a single M=7 event. To reconcile long- and short-timescale surface motions, we propose an original seismic cycle model driven by the southward roll-back of the Ionian oceanic slab (Fig. ).

During the interseismic phase, the active offshore normal faults affecting the eastern HP and Malta Escarpment are locked. The Hyblean and Ionian crusts are coupled and can be compared to an elastic beam, bending eastward in response to an increasing downward vertical force, namely the slab pull induced by the Ionian slab roll-back (Fig. a). Considering a minimum 500-year return period for major earthquakes such as the 1693 Val di Noto event , and extrapolating the PS-InSAR measurements over this period, coastal subsidence along the Siracusa–Augusta region could reach 1–2 m. This subsidence could be dampened to 0.5–1 m if, at the same time, the onshore faults, potentially related to extrados deformation, creep aseismically. During the coseismic phase, the offshore fault unlocks, and the seismic slip induces (for a Mw>7 earthquake) a multi-metric subsidence of the hanging wall and an associated decimetric-to-metric uplift of the footwall e.g., (Fig. b).

The cumulated succession of interseismic coastal subsidence and coseismic uplift could result in three scenarios (Fig. c). If the coseismic coastal uplift equals the cumulated interseismic subsidence, the coastal domain remains stable in the long term. If the former is lower than the latter, as predicted by elastic modeling (Fig. a), the coast subsides. Conversely, long-term coastal uplift occurs if coseismic uplift surpasses interseismic subsidence. Considering that geological data suggest a slow coastal uplift, this last scenario should be preferred, but additional sources of footwall uplift should be identified . At this stage, we can only evoke the following raw hypothesis:

The buoyancy of the flexed Hyblean crust could significantly increase post-seismic slip after major earthquakes and thus increase footwall uplift in the coastal region.