Buried Alive: Imaging the 9 November 2022, Mw 5.5 Earthquake Source on the Offshore Adriatic Blind Thrust Front of the Northern Apennines (Italy)

The prompt identification of faults responsible for moderate‐to‐large earthquakes is fundamental for understanding the likelihood of further, potentially damaging events. This is increasingly challenging when the activated fault is an offshore buried thrust, where neither coseismic surface ruptures nor GPS/InSAR deformation data are available after an earthquake. We show that on 9 November 2022, an Mw 5.5 earthquake offshore Pesaro ruptured a portion of the buried Northern Apennines thrust front (the Cornelia thrust system [CTS]). By post‐processing and interpreting the seismic reflection profiles crossing this thrust system, we determined that the activated fault (CTS) is an arcuate 30‐km‐long, NW‐SE striking, SW dipping thrust and that older structures at its footwall possibly influenced its position and geometry. The activation of adjacent segments of the thrust system is a plausible scenario that deserves to be further investigated to understand the full earthquake potential of this offshore seismogenic source.

The buried geometry of the NA outer thrusts led to a long-lasting debate concerning its recent and present-day activity. Such questioning rested on the weak geologic-geomorphic deformation signature onshore, the lack of bathymetric signature offshore, and focal mechanisms of small-magnitude earthquakes (Coward et al., 1999;Di Bucci & Mazzoli, 2002;Mazzoli et al., 2015;Troiani & Della Seta, 2011). The recent offshore activity was also attributed to NW-SE transpressional tectonics (Costa et al., 2021;Pierantoni et al., 2019). These conclusions have been highly disputed based on the integration of different observations, including geomorphology, extensive geophysical and subsurface geological datasets, historical seismicity, and focal mechanisms of major earthquakes (Basili & Barba, 2007;Burrato et al., 2003;DISS Working Group, 2021;Maesano et al., 2013;Panara et al., 2021;Vannoli et al., 2004Vannoli et al., , 2015. The seismogenic character of the NA buried thrust system became dramatically evident in May 2012, when two earthquakes (Mw 5.9 and Mw 6.1, respectively) occurred within 10 days from one another near Mirandola town, causing 27 casualties and significant damage in a densely populated and heavily industrialized region of the Po Plain (Bonini et al., 2014;Burrato et al., 2012;Nespoli et al., 2018;Pezzo et al., 2013;Scognamiglio et al., 2012;Tertulliani et al., 2012;Tizzani et al., 2013), about 200 km to the NW of Pesaro.
In many tectonic contexts, multichannel seismic reflection profiles are one of the most powerful tools to unravel the fault geometries, their associated structures at depth, and their activity rates (e.g., Bergen & Shaw, 2010;Buttinelli et al., 2021;Morley, 2009), and are essential in the case of offshore blind faults (e.g., Rivero & Shaw, 2011).
In this contribution, we show how the analyses and interpretation of publicly available seismic reflection profiles and well-logs allowed us to infer the causative fault of the 9 November 2022 mainshock, since its offshore location makes other direct and indirect methods of investigation inapplicable.
The seismic profiles from the ViDEPI database were converted from raster to SEG-Y format to enhance their integration and interpretation. We adopted the Wiggle2segy code (Sopher, 2018) recently tested and calibrated on this same dataset (Buttinelli et al., 2022). We applied an Automatic Gain Control scaling of the amplitude to improve the signal-to-noise ratio of the SEG-Ys.
We used a multilayer seismic velocity model to depth-convert the seismic reflection profiles. For the Plio-Pleistocene stratigraphy, we relied on an extensive analysis of sonic logs for the central Adriatic (Mancinelli & Scisciani, 2020). This model considers that progressive sediment compaction with depth affects the sandy-marly foredeep sequence, gradually increasing the velocity. For the pre-Pliocene units, we used fixed velocities for each seismo-stratigraphic interval (Table 1) based on the analysis of wells data in adjacent areas (Montone & Mariucci, 2020;Trippetta et al., 2021) and on the sonic log of the Elga-1 well within the study area ( Figure 2a).

Results
The interpretation of three seismic reflection profiles crossing the Pesaro Offshore earthquake epicentral area, from north to south CROP M16, B-402, and B-403 (Figures 2a and 3; see Figures S1 and S2 in Supporting Information S1 for the uninterpreted version), provided an accurate picture of the geological structures involved in the earthquake sequence.
The eastern part of the CROP M16 profile shows a gentle west-dipping monocline of the Triassic-Miocene succession (Adria monocline), associated with some normal faults within the Mesozoic succession, cutting up to the Miocene with moderate offset (Figure 3a). The Plio-Pleistocene foredeep deposits unconformably cover the Adria monocline with an onlap transgressive contact on the upper-Messinian regional unconformity (Base PL1, Figure 2b).
The Tamara-1 well-log projected onto the western portion of the CROP M16 seismic profile shows the repetition of the Miocene-Pliocene sequences due to the offset of various thrust splays connected at depth on a main ramp (Tamara Thrust System [TTS]). The CROP M16 profile also shows a deeper and less developed thrust below the Tamara-1 well, likely connected to the main ramp (Figure 3  The Pesaro Mare-4 well (projected onto the CROP M16, Figure 3a) intercepts the Pleistocene deposits unconformably overlying the Miocene succession. This unconformity is generated by the inner thrust ramp, which controls the uplift of the Mesozoic and Miocene successions. We named this set of faults "Pesaro thrust system" (PTS).
Section B-402 shows two well-developed thrust-related anticlines involving the Mesozoic succession at their cores ( Figure 3b). The anticline drilled by the Pesaro Mare-4 well correlates with the western one in the CROP M16 profile, showing a pronounced thickness reduction of the Pliocene deposits in the axial zone, accompanied by a marked unconformity at the base of lower-Pleistocene deposits. The PTS displaces the lower Pleistocene unconformity; its Pleistocene activity also influenced the geometry of the prograding complex bottom and top surfaces (Lower-Middle Pleistocene, Figure 3 inset B1). The Cornelia-1 well penetrates the eastern anticline, highlighting a reverse fault that brings Jurassic dolostones at the hanging wall into contact with lower Cretaceous calcarenites in the footwall. Two thrusts control the anticline (Figure 3, inset B2), the upper acting as the main ramp while the lower splay shows only limited displacement. The Plio-Pleistocene unconformities record the anticline deformation, with an abrupt truncation of the Pliocene reflectors against the Base PS2 unconformity. The latter is tilted eastward, whereas the base and top of the prograding complex recorded only subtle deformation in the anticline forelimb. We interpreted the fault controlling this anticline as correlated with the deeper thrust observed in the CROP M16 profile below the Tamara-1 well and named it the "Cornelia thrust system" (CTS).
Section B-403 is located ∼20 km southeast of the mainshock; similarly to B-402, it shows that CTS controls an anticline intercepted by the Cornelia-1 well with an imbricated frontal thrust and a minor backthrust (Figure 3c). In this section, we did not identify the PTS that controls the western anticline in section B-402. Conversely, a minor fault, detached within the Messinian succession in the western part of the profile, may represent the northern closure of the Elga thrust system (ETS) controlling the anticline located to the southeast and drilled by the Elga-1 well. This profile also shows Mesozoic normal faults that displace the Adria monocline and control the transition to the shelf facies of the Adriatic Carbonate Platform (e.g., Casero et al., 1990;Vlahović et al., 2005). The Adriatic Carbonate Platform is recognized in the structural high at the NE end of the section (Figure 3c from Note. The fourth column indicates the velocity values used in the time-to-depth conversion derived from the analyzed well logs (Montone & Mariucci, 2020;Trippetta et al., 2021). V 0 and k are the initial velocity at the top of the PL-PLS unit and the gradient describing the velocity variation with depth, respectively (Mancinelli & Scisciani, 2020); V i , interval velocity. 23 to 28 km), where Pliocene deposits unconformably rest upon Eocene successions (in correspondence with Malachite-1 well). The Eocene-Miocene succession is fully preserved in the western-dipping normal fault's hanging wall interpreted in the section's eastern part. In all analyzed seismic profiles it is also possible to tentatively recognize a horst structure in the downgoing Adria monocline at the footwall of the CTS (Figure 3, insets A1 and C1).
Integrating the seismic interpretation with the newly re-interpolated top FUC horizon and fault meshes (Figure 4a) highlights that the outermost Apennines front is characterized by various segments (TTS, CTS, and ETS from north to south). CTS strikes N135 and is at least 28 km long and 10-15 km wide; PTS strikes N150 and is ∼25 km  Figure S1 in Supporting Information S1; location in Figure 2a). The interpreted top horizons refer to the seismo-stratigraphic scheme of Figure 2b. Insets A1, B1, B2, and C1 show details of the interpretation.
long and 10 km wide. The two thrust systems diverge from north to south (Figures 4b and 4c), giving room to a depression bounded by the two anticlines and filled by a Pleistocene prograding complex, deformed by the growing anticlines (B-403 in Figure 3c from 0 to 5 km). The CTS ramp position may be related to a Mesozoic structural high in the footwall of the thrust system (the horst imaged below the Cornelia-1 well, Figure 4d). This structural high bounds a Miocene basin at the CTS footwall (Figure 3c, from 11 to 22 km), confined to the east by west-dipping normal faults, separating it from the Adriatic Carbonate Platform.
To compare the geological reconstruction with the ongoing seismic sequence, we plotted the latest reviewed earthquake locations available from INGV (ISIDe Working Group, 2007) onto our map and sections. The geological cross-sections derive from the time-to-depth conversion of the seismic profiles ( Figure 4d) and were validated through restoration and balancing ( Figure S3 in Supporting Information S1). Within the intrinsic uncertainties associated with the network geometry and resolution, the Mw 5.5 mainshock is located ∼4 km north of the CROP M16 profile, and its hypocenter lies just 1 km from the CTS, near its upper tip. The Ml 5.2 event lies close to the B-402 section and projects at the base of the Cornelia ramp. At the time of writing, the aftershocks of the Pesaro Offshore sequence still affect the area between the CROP M16 and B-402 seismic profiles (Figure 4b). The seismicity projected on these geological cross-sections suggests that most aftershocks concentrate on the CTS (Figure 4d), mimicking its geometry.

Discussion and Conclusions
The integration of the geological cross-sections, the distribution of aftershocks, and the re-located mainshock suggests that the Pesaro Offshore sequence occurred on the outermost front of the Northern Apennines (marked as NAF-10 in Panara et al., 2021) and specifically on the buried CTS. This thrust is part of a well-known fault system (Maesano et al., 2013;Panara et al., 2021) already deemed seismogenic (DISS Working Group, 2015).
The mainshock causative fault is characterized by a steep ramp in its central part and becomes more gently dipping toward its northern termination. The structural position of the ramp is possibly controlled by pre-existing  Figure S3 in Supporting Information S1). Mw 5.5 mainshock projected along the strike of the TDMT (128°, Figure 1); aftershocks projected normal-to-section in a buffer of 5 and 10 km for CROP M16 and B-402 profiles, respectively. normal faults and associated structural highs in the downgoing Adria monocline (Figure 4c), in agreement with observations in other segments of the NA front (Amadori et al., 2019;Livani et al., 2018).
The mainshock focal mechanisms obtained by various agencies (Table S2 in Supporting Information S1) show an average SW-dipping plane with strike 140°, dip 26°, rake 95°, and a variability of over 20° on each angular estimate. Such values agree very well with the fault geometry and kinematics observed in the geological cross-sections. Due to its offshore setting, the earthquake sequence location may be affected by significant uncertainties arising from the large asymmetry of the recording network and the azimuthal gap. Uncertainties in the horizontal coordinates may vary between 1 and 2 km and can be greater for earthquake depth, which can explain the aftershocks scatter around faults in Figure 4d. Future relocations may reduce uncertainties and allow for further comparison.
The Pesaro Offshore earthquake occurred in a tectonic setting where it is very challenging to identify its causative fault. The moment magnitude implied an earthquake rupture of relatively small size (25-40 km 2 ), according to common scaling relations (Leonard, 2014;Thingbaijam et al., 2017), and remained blind. Also, the hosting fault is part of a larger and articulated offshore fault system. The CTS size (∼300 km 2 ) suggests it could host ruptures of up to Mw 6.5 or even larger considering the possible activation with adjacent thrusts. If this were the case, the buried faults of the Adriatic offshore could also generate sizable tsunamis, as observed in the past (Table S1 in Supporting Information S1) and predicted by tsunami scenarios Tiberti et al., 2008). Therefore, this earthquake demonstrated that the CTS is an active and seismogenic fault belonging to a larger fault system extending north and south and toward the Italian coast. Similar faults were previously identified in the area (Argnani, 1998;Bigi et al., 1992;Fantoni & Franciosi, 2010;Maesano et al., 2013;Panara et al., 2021) and should be considered capable of releasing this type of seismicity as already demonstrated by the 1930 Senigallia (Vannoli et al., 2015) and 1987 Porto San Giorgio (Battimelli et al., 2019) earthquakes.
The NA outermost thrusts have generally low activity rates (<1 mm/year during the Pleistocene), and the sedimentation rates in the area are one order of magnitude larger (Maesano et al., 2013(Maesano et al., , 2015Maesano & D'Ambrogi, 2016;Panara et al., 2021). Hence, the faults are deeply buried below the Plio-Pleistocene sedimentary succession and, for this reason, were considered inactive (Coward et al., 1999;Di Bucci & Mazzoli, 2002;Mazzoli et al., 2015). Using a constant moment rate approach (Youngs & Coppersmith, 1985), the low slip rates of the NA outermost thrusts imply that larger events have an average recurrence interval of thousands of years. This explains the lack of documented historical earthquakes referable to the CTS. The Pesaro Offshore earthquake demonstrated that the absence of historical and recent seismicity could not exclude the seismogenic potential of active faults, even if they are blind and buried but are geologically documented.
Furthermore, although the Pesaro Offshore earthquake occurred ∼30 km offshore, the shaking level onshore (i.e., in the near field) was significant and caused some damage to the built environment (Tertulliani et al., 2022). All these considerations play an important role in devising input data for long-term earthquake hazard models since geological information can provide the accurate location and rupture mechanism of future potential earthquakes. Among the most recent hazard models, such geologic fault information was used to devise earthquake rate forecast input for the new Italian seismic hazard model, MPS19 Visini et al., 2021), the NEAM Tsunami Hazard Model 2018 , and the European Seismic Hazard Model 2020 (Danciu et al., 2022).
From the short-term perspective, the 3D geometry and structural position of the Pesaro Offshore earthquake causative fault are key elements for studies on the seismic source and the unfolding of the ongoing seismic sequence. The structural interlinkages with adjacent thrusts suggest that activating multiple faults is a plausible scenario that deserves further consideration.

Conflict of Interest
The authors declare no conflicts of interest relevant to this study.

Data Availability Statement
• All data used in this work are publicly accessible. All databases are cited in the relative figure caption and references list.