Central Italy’s current tectonic setting is the result of two main tectonic phases: (ii) the contractional phase (middle Oligocene to lower Pliocene) during which the formation of the Apennine fold and thrust belt occurred. This belt is a classic example of an ensialic post-collisional chain, characterized by an eastward-migrating piggyback sequence of the principal cover thrust-sheets (Cosentino et al., 2010; Cosentino et al., 2017). (ii) The extensional phase (Messinian-Quaternary). This phase, characterized by the opening of the Tyrrhenian Sea basin, led to the development of intermontane basins within the central Apennine chain (Carminati et al., 2010; Cavinato and De Celles, 1999).

In central Italy, the dominant extensional tectonic setting, relevant to the second phase, is characterized by SW- and S-dipping normal faults. These faults exhibit dip- and oblique-slip kinematics, resulting in complex graben or half-graben features corresponding to the intermontane basins, including the L’Aquila basin (Barchi et al., 2000). PSCSA is located within L’Aquila basin.

These basins were filled with lacustrine, slope, and alluvial deposits, beginning in the Upper Pliocene and continuing throughout the Quaternary to the present day (Cosentino et al., 2017). Many of the extensional faults, responsible for the formation of these basins, remain active and seismogenic. They are accountable for both present-day and historical seismic activity, as evidenced by seismological, GPS, and paleoseismological data (e.g., Boncio et al., 2004; Devoti et al., 2010; Pondrelli et al., 2006). These faults have the potential to generate earthquakes with maximum expected magnitudes of up to 6.5–7 (e.g., Galadini and Galli, 2000) (Figure 1).

The active normal faults of central Italy are split into a western set and an eastern one (Barchi et al., 2000; Falcucci et al., 2011). Several main historical earthquakes have likely been caused by the active extensional faults within the western set (Figure 1), such as the 1,349 (Mw 6.6), 1703 (Jan. 14; Mw 6.7), 1703 (Feb. 2; Mw 6.7) and 1915 (Mw 7.0) earthquakes (Rovida et al., 2022). This is supported by evidence, from the comparisons between the active faults pattern, the macroseismic damage distribution associated with these earthquakes and paleoseismological data (Galli et al., 2008 and references therein).

Conversely, the active faults of the eastern set, such as those of Mt. Vettore-Mt. Bove fault system (MVBF), Campotosto fault (CPF), Assergi-Campo Imperatore (ACIF), Upper Aterno Valley-Paganica fault system (UAVPF), Middle Aterno Valley-Subequana fault system (MAVSF) and Mt. Morrone fault (MF), defined each as a possible seismic gap since they did not activate during historical times (Galadini and Galli, 2000) (Figure 1). The 2016–2017 earthquake sequence (events with Mw up to 6.5), caused by two of the mentioned faults (MVBF and CPF; Figure 1), confirmed the preceding geological inferences.

Concerning the local seismotectonic setting of PSCSA and surroundings (Figure 1), the Mt. Pettino fault (PEF) represents the main active fault within ASB, bordering the NE slope of the basin. PEF dips to the south and southwest, extends for 14 km, and exhibits variable orientation. It pertains to the Upper Aterno Valley active fault system (UAVPF), whose movement throughout the Quaternary conditioned the tectonic and sedimentary evolution of the western portion of the L’Aquila basin (Figures 1–3). The UAVPF completely ruptured during the 2 February 1703 Mw 6.7 earthquake (Rovida et al., 2022; Figure 1), resulting in significant surface faulting, as observed through paleoseismological investigations (Galadini and Galli, 2000; Moro et al., 2002; 2013; Galli et al., 2011). PEF is in right en-échelon relationship with the near northernmost NW-SE trending 15 km-long extensional Mt. Marine fault (MAF). This latter is characterized by dip- and left oblique-slip (Moro et al., 2016). Significant surface faulting induced by the 2 February 1703 (Mw 6.7) earthquake has been recognized along the MAF, which caused considerable damage to L'Aquila town (Blumetti, 1995; Moro et al., 2002; Chiaradonna et al., 2021). To the east and southeast of L’Aquila basin, the seismogenic faults are the Paganica fault (PAF) and the Middle Aterno Valley-Subequana Valley fault system (MAVSF), which exhibit a NW-SE average direction and dip-slip kinematics (Galli et al., 2010; Figure 1). Geological, seismological and geodetic data show that the 6 April 2009 (Mw 6.3) L’Aquila earthquake (e.g., Hermann et al., 2011; Chiaraluce, 2012; Valoroso et al., 2013) was caused by the activation of the PAF, with an overall surface coseismic length of approximately 10 km (e.g., Falcucci et al., 2009; Galli et al., 2010; EMERGEO Working Group, 2010; Boncio et al., 2010; Gori et al., 2012; Moro et al., 2013).

PSCSA is placed within L’Aquila-Scoppito basin (ASB) sensu Nocentini et al. (2017), an E-W oriented asymmetrical graben. ASB, which is placed in turn within the broad L’Aquila basin, is filled up by chiefly Quaternary detrital deposits pertaining to alluvial and slope environments. It is bordered to the north mainly by the Mt. Pettino active fault (PEF in Figure 1; Galli et al., 2011) and, to the south, by its antithetic PSF, which is crossed by the Aterno River and its tributaries (Cosentino et al., 2017; Nocentini et al., 2017; Spadi et al., 2022) (Figures 1, 2).

Starting from the upper Piacenzian, ASB was filled up by a thick sequence of continental deposits laying via an unconformity surface onto the Upper Miocene terrigenous units and Meso-Cenozoic carbonate units (Figure 2).

ASB oldest post-orogenic deposits belong to the Colle Cantaro-Cave Formation (CCF in Figure 2) (upper Piacenzian-Gelasian) and include slope breccias, debris-flow deposits and matrix-supported alluvial conglomerates (Nocentini et al., 2017).

The Madonna della Strada Synthem (MDS), which is separated from the lower CCF by an unconformity boundary, is made up of clayey-sandy silts and sands. In the lower and middle part of MDS there are gravel levels with variable granulometry, from coarse to medium, and with a good rounding degree. Furthermore, levels of lignite and peaty pelite were found inside MDS (Mancini et al., 2012). The environment of MDS deposit can be referred to an alluvial system meandering within a wide and swampy floodplain. MDS is dated back to the Calabrian age due to the reverse magnetic polarity of its deposits and the finding of specific mammalian fauna (Cosentino et al., 2017).

Above MDS, separated by unconformity boundaries, several deposits consisting of gravelly-sandy from Middle to Upper Pleistocene sediments of plain and alluvial fan environment (e.g., FGS) and late Middle Pleistocene calcareous breccia complete the filling of ASB (Antonielli et al., 2020; Nocentini et al., 2017; Tallini et al., 2019).

The most recent ASB deposits (Upper Pleistocene? – Holocene), outcropping in the study area, are represented by slope and colluvial (COL) deposits bordering the base of the surrounding reliefs and by the recent alluvial deposits of the Aterno River and its tributaries (ATF) (COL and ATF in Figure 2 are named together as ALCO).

PSF is a N-dipping extensional fault and borders the southern boundary of ASB graben, representing an antithetical fault of PEF (Moro et al., 2016; Galli et al., 2011). Along the PSF, evidence of Quaternary activity was detected. This activity lasted at least until the Middle Pleistocene and possibly into the Late Pleistocene (Nocentini et al., 2017).

PSF was considered in the seismic microzonation studies as a “Potentially Active and Capable fault” with an uncertain/presumed trace (Tallini et al., 2014) (Figure 2). As noted above, according to the Italian guidelines of seismic microzonation (Commissione tecnica per la microzonazione sismica 2015), an “Active and Capable” fault exhibits evidence of faulting within the last 40 ka, whereas a “Potentially Active and Capable fault” shows evidence of activity in the Middle-Late Pleistocene, but its activity over the last 40 kyr remains unknown and needs to be demonstrated.

In the section of the seismic microzonation map (Figure 2), PSF affects the MDS unit, but it is sealed by recent colluvial deposits (Late Pleistocene? - Holocene). In this context, Durante et al. (2017) demonstrate that FGS is displaced by PSF by approximately100 m east of PSCSA. Furthermore, an anthropic excavation reveals shear planes, associated with the PSF deformation zone, that affect the MDS. These shear planes are overlain by gravels of the CPF unit, dated to approximately 40 kyr (Cosentino et al., 2017) (Supplementary Figure 1). Therefore, the paleoseismological study performed at PSCSA was carried out considering: (i) that the 40 kyr constitutes the lower chronological boundary for the Italian definition of active and capable fault; (ii) that the aforementioned recent faulting is very close to this time boundary (Supplementary Figure 1); and (iii) the presence of two escarpments at the base of the slope, bordering the alluvial plain, which may represent fault-derived scarps (Figure 3).

Finally, it should be noted that the fault is not mapped in the geological sheet no 358 Pescorocchiano of CARG Project (ISPRA, 2010) (scale 1:50.000), in the ITHACA Catalog (ITHACA Working Group, 2019) and in the DISS Database (DISS Working Group, 2018).

Prior to digging the paleoseismological trenches, two electrical resistivity tomographies (hereafter ERT 1 and ERT2) were performed.

The ERTs in general highlight medium-to-low resistivity values, ranging from 5 to 200 Ωm, which are indicative of the clayey-sandy and arenaceous lithologies encountered in the two trenches and in the boreholes S1 and S2, as described below.

In ERT1, two resistivity layers were identified: the superficial layer (“a” in Figure 4) exhibits a resistivity of 55–75 Ωm, while the underlying layer (“b” in Figure 4) has a resistivity of 15–40 Ωm. Furthermore, in the sector where S1 borehole was drilled at approximately 30 m depth, a change in resistivity from approximately 80–200 Ωm is observed (“c” of Figure 4).

ERT2 detects two continuous layers: the shallow one (“d” in Figure 4), a few meters thick, has a resistivity around 60–80 Ωm, the underlying one (“e” in Figure 4), about 5 m thick, with a resistivity equal to 20–35 Ωm.

The general trend of the recorded resistivities does not show lateral contacts or abrupt changes extending from the surface to the depth. Furthermore, the lateral continuity of the shallow resistivity layers (“a”, “b”, “d” and “e” in Figure 4) is evident. These observations do not suggest the presence of discontinuities in the primary attitude of the layer, possibly related to displacements caused by faults.

To verify the presence and recent activity of PSF, trenches T1 and T2 were excavated across the scarp (Figure 3). The trenches overlapped by a few meters to address logistical constraints and ensure the investigation of the entire site.

The two paleoseismological trenches revealed a stratigraphic sequence of alluvial and colluvial units. The deposition of these units is related to the surface water dynamics in the local alluvial plain and to erosion and depositional processes on the slopes around the study area (Figures 5–7).

Chronological information about the deposition of units was acquired by (i) archaeological determination of pottery fragments (Supplementary Table 1; Supplementary Figure 2); (ii) radiocarbon dating of charcoals and organic-rich samples (Supplementary Table 2).

Based on the stratigraphic relationships and correlations between the continental units identified in both trenches, seven main sedimentary units have been distinguished. They are described here from the lowest (oldest unit exposed by Trench 1) to the uppermost unit (Figure 8).

Unit 7 (FGS unit, Middle Pleistocene): it consists of an alternation of decimetric sub-horizontal levels of silt (Munsell colour: 7.5 YR 6/1 – gray) and sands (colour Munsell: 7.5 YR 6/8 – reddish yellow). Horizontal and inclined laminations, ripple and cut-and-fill structures indicate a fluvial braided plain and alluvial fan environment. In Trench 1, a level of sub-angular centimetric carbonate clasts marks a NE-dipping erosion surface that separates unit 7 from unit 6 (PG: 34–37 m) (Figure 5). The sedimentological and depositional characteristics, along with the local morpho-stratigraphic setting, allow this deposit to be correlated with the FGS unit (Middle Pleistocene) (Centamore and Dramis, 2010; Nocentini et al., 2017) (Figure 2).

Unit 6: silty-clayey and sandy massive deposit (Munsell colour: 2.5 Y 6/6 – olive yellow; 2.5 Y 7/8 – yellow) often characterized by mottles, patinas and nodules of Fe and Mn oxides. Small red slags of volcanic origin were found, and some organic sediments were collected and dated with 14C method at 27,378–26779 cal BC (Supplementary Table 2). The lithofacies and grain size of the sediment allow us to refer unit 6 to a deposition in a low-energy alluvial plain environment. In Trench 1, starting from about 95 m, unit 6 shows a carbonate caliche level, most probably related to vertical fluctuations of an ancient water level (Figure 5B). In Trench 1, the lateral continuity of unit 6 is interrupted between the PG 124 m and 126 m by a well dug by hand which was later buried and filled in by anthropic deposit in the 1950s as suggested also by local witnesses (Figures 5C, D).

Units 5, 4, 3 and 2 are found in Trench 2, especially between PG 20 and 38 m. These units consist of silty and sandy deposits of low-energy alluvial plains or fans, exhibiting significant soft sediment deformation-type structures (Figures 6, 7).

Unit 5: it comprises clayey silt (Munsell colour: 7.5 YR 4/6 – strong brown) with variable thickness, overlying unit 6 (Figures 6, 7). This unit contains numerous pottery shards and charcoal shreds, one of which was sampled and dated with 14C method at 1,507–1,407 cal BC (Supplementary Table 2). Given the sedimentological characteristics and the abundant organic material, this unit can be attributed to pedogenic processes affecting unit 6. Evident eroded mushroom-like structures are observed within it.

Unit 4: it outcrops between PG 32 m and 34 m (Trench 2) (Figures 6, 7). It is a body embedded within units 5 and 6 and overlain by unit 3. The deposit exhibits a sandy-silty grain size (Munsell colour: 10 YR 6/6 – brownish yellow) with soft-sediment deformations and sandy dikes. Fragments of ceramic material were also found in its lowermost part (Supplementary Table 1; Supplementary Figure 2).

Unit 3: it consists of a silty-clayey massive deposit (Munsell colour: 7.5 YR 6/4 – light brown) (Figures 6, 7). Significant soft-sediment deformations, such as pseudonodules, load structures, and disturbed laminations were found. A fragment of an olla hem from the Roman period was discovered in unit 3, which can be ascribed to a chronological range between the 1st–2nd century AD (Supplementary Table 1; Supplementary Figure 2). A charcoal sample was collected very near the olla hem and dated with 14C method at 236–385 cal AD, strengthening the archaeological dating (Supplementary Table 2).

Unit 2: the deposit is an E-W trending channeled sandy-silty body underlying units 3 and 5 and covered by unit 1 (Figures 6, 7). It has been divided into two sub-units which show different sedimentary structures. Unit 2b is a sandy deposit (Munsell colour 2.5Y 7/4 – pale yellow) with laminations, cut-and-fill structures and significant soft-sediment deformations such as disturbed laminations, mixed and disturbed layers, mushroom structures, pseudonodules and two whitish sandy dikes. The overlying unit 2a is made up of light brown sands (Munsell colour 2.5Y 8/4 – pale yellow) with flat-parallel laminae and scattered centimeter-long calcareous clasts. It does not exhibit soft-sediment deformations. The two dikes of unit 2b seem to be sealed by the overlying unit 2a.

Unit 1: sandy colluvial deposit (Munsell colour 2.5Y 8/6 – yellow) containing angular and rounded carbonate clasts and abundant fragments of pottery and bricks (Figures 5–7; Supplementary Table 1; Supplementary Figure 2). Unit 1 was locally affected by anthropogenic excavations.

The two trenches did not reveal any shear planes or discontinuities that could be associated with tectonic dislocations after the deposition of the stratigraphic units. In fact, the units exhibited evident stratigraphic lateral continuity, consistent with primary sedimentary bedding, and they are only locally interrupted by structures related to anthropogenic activity and low-slope erosional surfaces.

Considering the age of the charcoal samples obtained by radiocarbon dating and the historical age of the pottery shards, it was possible to rule out surface faulting events at least in the last 25 kyr. However, since the time span (25 kyr) is more recent than the chronological interval (40 kyr) recommended by the Italian Guidelines for Microzonation studies (Commissione tecnica per la Microzonazione Sismica, 2015) to exclude the presence and recent activity of PSF, two continuous core boreholes were drilled near Trench 1 and Trench 2 (boreholes S1 and S2, respectively), each reaching depth of 30 m below ground level (Figures 5, 6). Boreholes S1 and S2 were drilled in the PSF footwall and hanging wall, respectively, to investigate possible tectonic dislocations of deposits older than 25 kyr (Figure 9). To improve the chronostratigraphic setting, a sample of plant remains, obtained at 19 m depth in borehole S1 yielded an age of >43,500 years using the 14C method (Supplementary Table 2). In two samples taken in the borehole S2 at 1.4–1.6 m and 11.70 m depth, the grain size analysis was conducted to determine the sedimentological characteristics of the shallow alluvial deposit (units 2a and 2b) (Figure 9).

The stratigraphy outlined from the S1 and S2 borehole logs, compared with the local geological setting (Centamore and Dramis, 2010; Nocentini et al., 2017), allowed the recognition of the Quaternary FGS and MDS Synthems. At the borehole bottom, greyish well-cemented sandstone was encountered, which outcrop in neighboring areas, particularly just upstream of PSCSA. It is attributable to UAP, the Upper Miocene synorogenic terrigenous unit (Centamore and Dramis, 2010) and corresponds to the pre-Quaternary substratum of PSCSA. Regarding the interpretation in terms of presence/absence of shear planes in the investigated area, the stratigraphic settings derived from the two boreholes will be discussed in the next section.

No very shallow water table was encountered during the drilling of S1 and S2 boreholes. This observation is of significant importance regarding the potential for liquefaction phenomena (see Section 5.2 ahead).

Geophysical investigations (ERT1 and ERT2) primarily evidenced several shallow continuous resistivity layers. Paleoseismological trenches (T1 and T2) enabled the identification of these layers as primarily related to alluvial/slope-environment deposits. In this regard, it is possible to exclude the tectonic origin of the two escarpments at the base of the slope, which could be interpreted as a morphological indicator of PSF activity. These escarpments are, therefore attributed to other phenomena of non-tectonic origin, considering the local geomorphological setting, such as fluvial erosion, slope dynamics and anthropogenic features. Consequently, these data suggest that the presence of escarpments cannot indicate the presence of faulting at or close to the ground surface.

From a chronological perspective, the analyses of the paleoseismological trenches have ruled out faulting events within the last 25 kyr. This is because unit 6, the oldest Late Pleistocene-Holocene unit found in the trenches, dated to 27378–26779 cal BC, remains continuous and is not disrupted by shear planes (Figure 5). Furthermore, considering the mean recurrence interval of major seismogenic fault in the central Apennine, typically ranging from 1,500 to 2000 years (Galadini and Galli, 2000; Galli et al., 2008), a 25 kyr timeframe should encompass a substantial number of possible faulting events (see Galadini et al., 2012 on this aspect). However, according to the aforementioned Italian guidelines for seismic microzonation, this time interval (25 kyr) is considered more recent than the required 40 kyr to definitively exclude surface faulting events. This is especially important given that the PSF displaces the MDS deposits (Early Pleistocene) not so far from PSCSA (Supplementary Figure 1). Therefore, to ensure safety, the implementation of boreholes was deemed necessary to assess vertical displacement related to the PSF activity within older stratigraphic units.

The shallow stratigraphy of the two boreholes is comparable to that recognized in the two trenches, up to a depth of 10–15 m. The boreholes encountered at greater depths the boundary between FGS and MDS, at 7 m in S1 and 11 m depth in S2 and the deeper boundary between MDS and UAP at 25 m in S1 and 32 m depth in S2.

The differences in depth of the boundary between FGS onto MDS and MDS onto the UAP in the two boreholes are about 12 m and 16 m, respectively. Considering the two boreholes are about 100 m apart, the dip towards the northeast of the FGS-MDS and MDS-UAP boundaries is almost the same, at approximately 7.5° and 9°, respectively. These very low values can be simply explained by the morpho-stratigraphic setting of the area, in which the erosional and depositional features of the continental units are consistent with the dip towards the alluvial plain, as would be expected at the basin boundary (Figure 3).

In this regard, the stratigraphic correlation of FGS-MDS and MDS-UAP boundaries, combined with the results obtained from the paleoseismological trenches, allowed us to rule out dislocations in the entire stratigraphic sequence starting from the Miocene bedrock (UAP). This indicates the absence of surface faulting since a period much longer than the last 40 kyr and, therefore, the recent activity of PSF in PSCSA. Thus, even though the important seismogenic faults of Mt. Marine, Mt. Pettino, and Paganica, located close to the study area, have been activated several times in the last 25 kyr, generating surface faulting (Galli et al., 2011; Moro et al., 2013), traces of this activity for the PSF, considered to be antithetical to the Mt. Pettino fault, have not been detected.

Paleoseismological trenches revealed significant soft-sediment deformations affecting the units exposed in the trenches. More precisely, these structures were recognised in units 5, 4, 3, 2b of Trench 2 (PG: 22–34 m) (Figures 5–7). The identified post-depositional deformation structures are similar to those described in Moretti and Sabato (2007) and in Rodriguez-Pascua et al. (2000). They include loop bedding, boudinage-like structure, disturbed varved laminations, mixed layers without fluidization, mushroom-like silts protruding into laminites, large-scale load structures, mixed and disturbed layers with fluidization, pseudonodules, pillow structures, sand dikes and water escape structures (Figure 10). For the sandy dikes, the geometry has been reconstructed by analyzing both the trench walls and the eastern terrace of Trench 2 (Figure 11).

With the achieved 3D view, we estimated two main orientations of the sandy dikes: NNE-SSW and NW-SE. Moreover, both sandy dikes were likely formed simultaneously as they are filled with the same whitish, fine sandy sediment, and at their intersection, the deposit is homogeneous.

In literature, soft-sediment deformations such as those described are related to liquefaction phenomena in water-saturated silty-sandy deposits (Obermeier, 1996; Rodriguez Pascua et al., 2000; Moretti and Sabato, 2007). According to Moretti and Sabato (2007), the mechanisms that trigger liquefaction, and thus the formation of soft-sediment deformations, may be due to lithostatic loading or earthquakes.

To support the hypothesis that the observed soft-sediment deformations mentioned above were generated by earthquakes, the following arguments can be presented:

(1) Conditions potentially predisposing to liquefaction were found in PSCSA: (i) shallow sandy levels with high probability to liquefaction (Figure 12); (ii) the caliche level exposed in the Trench 1 testifies to very shallow fluctuations of the Holocene water table in sandy deposits (Figure 5). This setting highlights the liquefaction-prone conditions in the past (Figure 5).

Therefore, the soft-sediment deformations observed within the predominantly sandy-silt sediments in PSCSA can be interpreted as semites, likely related to the occurrence of seismic events that affected the region in historical times, considering the age of the stratigraphic units that underwent liquefaction.

Conversely, the current liquefaction hazard in PSCSA is considered absent. Although unit 2b exhibits a grain size favorable for liquefaction and Chiaradonna et al. (2024) indicate potential liquefaction, it would not occur due to the absence of very shallow groundwater in boreholes S1 and S2. More specifically, in the Autumn 2020, the water table in boreholes S1 and S2 was encountered at a depth of just over 20 m below ground level. This depth likely represents the shallowest water table level, as the boreholes were drilled in the Autumn which corresponds to the rainy season in Italy. According to the Italian seismic microzonation guidelines (SM Working Group MS (2008), a depth of 20 m is considered the threshold beyond which liquefaction does not occur in sandy deposits. Consequently, in the updated seismic microzonation maps (Tallini et al., 2024) for PSCSA, the liquefaction-prone zone has been removed using the recently acquired water table depth data from boreholes S1 and S2. This pivotal information was unavailable when the map in Figure 2 was created. The absence of liquefaction in PSCSA during the near source 2009 Mw 6.3 L’Aquila earthquake further supports the hypothesis of a negligible liquefaction hazard for PSCSA (Martelli et al., 2012).

The analysis of the western wall of Trench 2 (Figures 6, 7) allowed us to identify at least two earthquake-induced paleoliquefaction events. The methodological approach used to date the paleoliquefaction events was supported by the fact that the pottery shards and charcoal samples were sedimented in alluvial bodies, therefore their age is later (although the exact time difference is unknown) than that of the pottery shards and charcoal samples.

The remarkable mushroom structures, located within unit 5, may be cut by the bottom boundary of unit 2b (Figure 10B), suggesting that unit 5 was involved by a paleoliquefaction event. Its terminus post quem is after the date of charcoal sample found in unit 5 (1,507–1,407 cal BC, SASSA_CARB_5). However, the geometry of the unit 5 and 2b boundary is not very sharp. Therefore, since the relationship between the seismites of unit 5 (Figure 10B) and those of unit 2b is not clear, it cannot be ruled out that the former were generated with the most recent event (PALI2) described below.

The paleoliquefaction event PALI2 occurred after the deposition of unit 4 and prior to the deposition of unit 3 by observing the seismites (deformed varved laminations) in the sands of unit 4 which are cross cut by the erosional surface that separates unit 4 from unit 3 (Figure 10E). These seismites indicate the occurrence of an earthquake immediately following the sedimentation of the sands, which were still well saturated (Moretti and Sabato, 2007). Radiocarbon dating on charcoal collected within Unit 3 (SASSA_CARB_4: 236–385 cal AD in Supplementary Table 2) and the age of pottery fragments in unit 4 (B in Supplementary Table 1) constrain the age of PALI2 within an interval between 3rd-2nd century BC and the deposition of unit 3 occurring after 236–385 cal AD.

The younger paleoliquefaction event PALI1 occurred after the deposition of unit 3 and before unit 2a. In fact, unit 2b, which is intervening in between unit 3 and 2a, is characterized by soft-sediment deformations such as deformed layers and water escape structures, which suggest the occurrence of a seismic event immediately after the deposition of unit 2b. Further, pseudonodules of unit 3 sediment are mixed within unit 2b sediment which testifies to the involvement of both units in the same paleoliquefaction event (PALI1) (Figure 10D). The water escape structures are represented by sandy dikes which cut vertically through the previous seismites up to the erosion surface. This erosion surface separates unit 2a from 2b, truncating the uppermost portion of the sandy dikes and not allowing the preservation of the sandy volcanos (Figures 7, 10A).

It is worth noting that the relationship between the sandy dikes and the deformed layers does not suggest the occurrence of another seismically induced paleoliquefaction event. In fact, the formation of sandy dikes is a direct consequence of the liquefaction and fluidization of water-saturated sands confined within low-permeability deposits (Audemard and De Santis, 1991; Rodriguez-Pascua et al., 2000; Moretti and Sabato, 2007).

In chronological terms, the age of PALI1 can be inferred from the constraints obtained for unit 3 deposition, which provides a terminus post quem for this event. Specifically, the radiocarbon dating of the charcoal sample (236–385 cal AD) and the age of ceramic material (D in Supplementary Table 1; Supplementary Figure 2; 2nd century AD) suggest PALI1 occurred after the 236–385 cal AD. However, the absence of dating from unit 2a and for the more recent units precludes the establishment of a terminus ad quem or ante quem for PALI1.

The paleoseismological investigations allowed us to identify the traces of at least two earthquake-induced paleoliquefaction events. These traces are represented by seismites that affected the stratigraphic sequence, especially in Trench 2, up to unit 2a. These seismites testify to the occurrence of at least two seismic events that impacted the L'Aquila basin during historical times. To clarify which paleoseismic events may have triggered the seismites in PSCSA, several considerations can be made.

According to Rodriguez-Pascua et al. (2000), the seismites of the type here recognized can be attributed to earthquakes of magnitude greater than 6–6.5. The magnitude/epicentral distance relationship proposed by Galli (2000) shows that the liquefaction phenomena in Italy for earthquakes of magnitude greater than 6–6.5 can be triggered at epicentral distances of 45–65 km.

Calculating an average epicentral distance of about 60 km from PSCSA, it is observed that some active faults of the central Apennines (e.g., Galadini and Galli, 2000; Boncio et al., 2004; Falcucci et al., 2016; Galadini et al., 2018), fall within this distance (Figure 1). These active faults have been investigated from a paleoseismological point of view by several Authors, highlighting the occurrence of several surface faulting events during Holocene and in historical times (Galli et al., 2008). In this regard, the analysis of paleoseismological data was carried out to outline the possible paleoearthquakes that may have triggered the seismically induced paleoliquefactions events recognized in PSCSA. Particular attention was paid to those faults whose paleoseismological evidence shows events of activation during a time span consistent with the age of seismites, such as those that occurred in the Roman period (Figure 1). In Table 1 the historical earthquakes associated with the active faults of Central Apennines considered in this study are reported.

Paleoseismological evidence on the Upper Aterno Valley–Paganica fault system (UAVPF) indicates a surface faulting event compatible with the 2 February 1703 historical earthquake on the Mt. Marine fault, the Paganica fault and tentatively also attributed to the Mt. Pettino fault (Moro et al., 2002; 2013; Galli et al., 2011; Cinti et al., 2011). An earlier event was recognised on the Paganica fault and was tentatively attributed by Galli et al. (2011) and Cinti et al. (2011) to the 801 AD seismic event, which struck the central Apennines and Rome (Rovida et al., 2022). However, Moro et al. (2013) provided more precise chronological constraints for this earthquake, suggesting a possible occurrence between the 4th and 6th centuries AD. This conclusion is also supported by archaeoseismological evidence indicating the destruction of the ancient Roman town of Amiternum (northwest of L'Aquila).

Regarding the Middle Aterno Valley–Subequana fault system (MAVSVF), paleoseismological investigations and archaeoseismological data indicate that this fault system ruptured completely during the 2nd-1st century BC (Falcucci et al., 2011; Falcucci et al., 2015).

Paleoseismological and archaeoseismological indications for the Mt. Morrone fault (MF) highlight an activation event in the 2nd century AD (Ceccaroni et al., 2009; Galli et al., 2014).

Regarding the Norcia fault system (NF), paleoseismological investigations performed by Galli et al. (2005) and Galli et al. (2018) indicate that this fault system was responsible for the 14 January 1703 (Mw 7.0) historical earthquake (Rovida et al., 2022) and likely also for an earlier event that occurred in 99 BC.

Paleoseismological indications for the Mt. Vettore-Mt.Bove fault system (MVBF) suggest the activation of the whole structure probably in the 443 AD, an event that caused damages even to Colosseum in Rome (Galli et al., 2019).

Recent paleoseismological investigations suggested that the Assergi-Campo Imperatore fault system (ACIF) caused one of the strongest earthquakes of the 1349 AD seismic sequence (Galli et al., 2022; Gori et al., 2015).

Finally, the Fucino–Mt. Magnola fault system (FMF) is known to be the seismogenic source of the 13 January 1915 historical earthquake. Paleoseismological and archaeoseismological data acquired along this fault indicate that it was responsible for an earlier event tentatively attributed to 508 AD (Galadini and Galli, 1999; Galli et al., 2012; Galadini et al., 2022).

Several hypotheses can be made from data reported in Table 1. Regarding the paleoliquefaction event PALI2, which occurred between the 3rd-2nd century BC and 236-385 cal AD (SASSA_CARB_4), it could have been triggered by: (i) the 2nd-1st century BC event of the Middle Aterno Valley-Subequana Valley fault system (MAVSVF), (ii) the 99 BC event of the Norcia fault system (NF) (99 BC) and (iii) the 2nd century AD earthquake originating from the Mt. Morrone fault (MF). Clearly, as the age of the pottery shards represents a post quem for the liquefaction occurrence, the age of deposition of unit 4 is unknown. Consequently, the seismic event cannot be determined with confidence. Anyway, considering the distance of the mentioned faults from PSCSA, it is more likely that the paleoliquefaction event PALI2 may have been triggered by the 2nd-1st century BC event of the Middle Aterno Valley-Subequana Valley fault system (MAVSVF) (Table 2).

Regarding the younger paleoliquefaction event (PALI1), the chronological constraints derived from the charcoal and pottery fragments in unit 3 suggest that this event occurred after 236–385 cal AD (SASSA_CARB_4). Based on the collected references, the younger paleoliquefaction event could have been triggered by: (i) the events of the Upper Aterno Valley-Paganica fault system (UAVPF) (4th-6th century AD and 2 February 1703), (ii) the 14 January 1703 event of the Norcia fault system (NF), (iii) the 443 AD event of the Mt.Vettore-Mt.Bove fault system (MVBF), (iv) the 1349 AD event of the Assergi-Campo Imperatore fault system (ACIF) or (iv) the events of the Fucino-Mt. Magnola fault system (FMF) (508 AD and 13 January 1915). In this case, considering (i) the proximity to PSCSA, (ii) chronological compatibility and (iii) the fact that the liquefied units are currently located several meters above the present alluvial plain, ruling out a very recent age (the last few centuries), it is possible to hypothesize that PALI1 may have been likely triggered by the 4th-6th century AD event of the Upper Aterno Valley- Paganica fault system (UAVPF). Table 2 illustrates the two recognised paleoliquefaction events (PALI1, PALI2), their dating, the units in which the associated seismites were found in the trenches, the causative fault and the related seismic events.

In Figure 13, the timing of the two recognised paleoliquefaction events (PALI1, PALI2) compared to the timing of the probable causative earthquake and fault are reported.

Lastly, another result concerning the paeloliquefactions studied in this work is that they are not reported in the official Italian databases of earthquake-induced ground effects (CEDIT (https://gdb.ceri.uniroma1.it/index.php/view/map/?repository=cedit&project=Cedit) and CFTI-landslides (https://cfti.ingv.it/landslides/).