Introduction

On a strategic spot of northeastern Sicily, close to the main streams connecting the Tyrrhenian and Ionian coastlines of the island, the Greek colony of Abakainon was probably founded in 1100 BC, although there is a lack of precise information (Oliva 2012). Its location (Fig. 1a) made this settlement very desired, especially for military and trading purposes, leading to a succession of conquests by different people.

Fig. 1
figure 1

a Geolocation of the study area. b 3D surface map of the studied sector. c Location of the main discussed spots (image from Google Earth). TL, Tindari Line; CPO, Calabria-Peloritani Orogen; ME, Malta Escarpment; AMC, Apennine-Maghrebian Chain

Full size image

Nowadays, archaeological excavation campaigns, carried out since 1952, allowed discovering a real buried city, holding precious and well-preserved finds including local coins. However, one of the most relevant archaeological discoveries is a wide necropolis area, with about 150 well-preserved tombs, dated between the end of the fourth century BC and the beginning of the second century BC. This site shows evidence of damage and collapses, and the current setting of unearthed ruins leads to think that the necropolis was destroyed by a catastrophic event. Bottari et al. (2013) carried out an archaeoseimological study, highlighting that the sudden destruction cannot be ascribed either to human action or to building decay but rather to the occurrence of an earthquake. From the analysis of aerial photographs and field investigations, they also hypothesized that the necropolis could have been built on an old landslide body, which could have been reactivated by the seismic shaking, contributing to the destruction of the site. Nevertheless, the dense vegetation, along with the bad accessibility of the study area and a complex geological setting, makes hard to recognize peculiar features that could validate such hypothesis by a simple field survey or by the analysis of aerial photos. Therefore, the need of further investigation is the starting point of this study.

In mountainous region, there are several examples of ancient human settlements constructed on the surface of paleo-landslides, offering a flatter morphology. The Machu Pichu citadel, symbol of the Inca Empire, was probably built on a wide landslide thanks also to the availability of weathered granitic rocks employed for its construction. The presence of such material could be the result of ancient landslides affecting the area, proving the link between geomorphologic processes and development of human settlements (e.g., Margottini et al. 2008; Margottini 2015). A similar case is referred to the Italian village Cortina d’Ampezzo, which was edified on a landslide body dated between fifth and sixth century AD (Panizza 1990).

In this kind of peculiar setting, with specific reference to the study area, two main interacting features must be taken into analysis: landslides and tectonics. It is well known that there is a link between these two elements, since neotectonics can be often related to landslide occurrence, although it is not always the triggering factor (e.g., Willenberg et al. 2008; Tokiwa et al. 2011; Pappalardo et al. 2014; Pappalardo and Mineo 2015; Mineo et al. 2015, 2017). In this specific case, post-orogenic tectonics is likely to be one of the main responsible of the widespread landslide attitude of slopes, because it has strongly affected the study area, with fault activity, uplift phenomena, and earthquakes.

In this perspective, this paper provides the results of new geomorphological and geophysical investigations aimed at looking for evidence of landslides and tectonic structures close to the necropolis area and at finding out if the necropolis was really built on an old landslide body. In detail, field surveys and data retrieved by aerial images were integrated with the analysis of a high resolution 2-m Digital Terrain Model (DTM) (Fig. 1b), which allowed locating the main geomorphological elements to be compared with data reported by official literature. Such procedure highlighted crucial aspects on the interaction between tectonics and main geomorphological features (e.g., river path, crest lines, landslides). Then, geophysical campaigns were aimed at investigating the subsoil of the necropolis area, looking for evidence relatable to a landslide body. In particular, the study of the environmental noise (microtremors), through passive seismic recordings of single-station H/V (or horizontal to vertical spectral ratio (HVSR)), allowed highlighting the frequencies at which the ground motion is amplified by stratigraphic resonance, both inside and in the close proximity of the necropolis. Furthermore, since this method can also give information on the stratigraphy (e.g., Amorosi et al. 2008; Pappalardo et al. 2016), the registered impedance contrasts have been related to stratigraphic changes or to peculiar structures affecting the subsoil. Finally, in order to achieve a detailed knowledge of the shallow subsoil, an Electrical Resistivity Tomography (ERT) was performed to investigate low depths and specific matches with the results of passive seismic recordings were found.

Geological and seismotectonic framework

The necropolis of Abakainon is located in northeastern Sicily, on the border between the northeastern sector of Nebrodi Mountains and the northwestern sector of the Peloritani mountain belt (Fig. 1). The latter is the southernmost portion of the Calabria-Peloritani Orogen (CPO), an arched belt connecting the Southern Apennine and Maghrebian chains (AMC), and made of a set of south-verging basement nappes with metamorphic grade increasing upwards and remnants of a Mesozoic–Cenozoic cover sequence (Atzori and Vezzani 1974). To the south, Mount Etna and the Malta Escarpment (ME) (Fig. 1a) represent the most relevant elements from a regional geodynamic point of view (e.g., Barbano et al. 2014a).

The metamorphic rocks cropping out in the study area mainly belong to the Aspromonte Peloritani Unit (APU), which is represented by middle crustal biotite paragneisses and augen gneisses with minor amphibolites, mica schists, and marbles (Appel et al. 2011). The top of the local stratigraphic succession is represented by sedimentary deposit, known as Capo D’Orlando Flysch (CDOF), showing alternation of yellow-grayish sandstones and clay-silt layers (Lentini et al. 2000) (Fig. 2).

Fig. 2
figure 2

Schematic geological map of the study area (modified after Carbone et al. 1993)

Full size image

From a structural point of view, a NNW-SSE striking extensional fault belt (Ghisetti 1979), the Tindari Line (TL), marks the transition between two compressional and extensional crustal sectors in Northeastern Sicily produced by the Neogene-Quaternary convergence between Africa and Eurasia (e.g., Lentini et al. 2006). Onshore neotectonic data provide evidence that the Tindari Fault System is still active in the central and northern sectors and mostly accommodates extensional or right-lateral and left-lateral transtensional displacements on a diffuse array of faults (Billi et al. 2010). The vertical dislocation of Tyrrhenian terraces and the right-lateral movements of Plio-Quaternary terrains morphologically mark such structure, whose activity lasts from middle-upper Pleistocene to the present (e.g., Ghisetti 1979), as also shown by geodetic data (e.g., Palano et al. 2012). In the studied sector, fault segments related to this system affect the formations around the necropolis area (Fig. 2).

Furthermore, along the northern segment of TL, low energy seismicity is frequently recorded (Bottari et al. 2013 and reference therein). On the other hand, the Italian seismic catalog CPTI15 (Rovida et al. 2016) reports a scant moderate seismicity (4.5 < Mw < 5.5) all over the region. The main historical earthquakes affecting the coastal area are the 10 March 1786 (Mw 6.14), the 5 March 1823 (Mw 5.81), and the 15 April 1978 (Mw 6.03). These events caused extensive damage (Intensity I = VIII–XI MCS) in a broad sector of northern segment of TL, which is considered the causative fault (e.g., Basili et al. 2008). The 1978 earthquake heavily damaged Tripi village (I = VII–VIII MCS), placed 2 km far from the necropolis of Abakainon (Fig. 1c) (Guidoboni et al. 2007 and reference therein). The seismic history of this area starts from the 1490, but most of the earthquakes are poorly defined and catalogue parameters are based on 2–10 observation points (Barbano et al. 2017). Although the Italian earthquake catalogues cover a long time span (461 BC–2009 AD) (Guidoboni et al. 2007; Rovida et al. 2016), descriptive data on earthquake effects are generally scarce until the 13th–14th centuries. Therefore, the epicenters and magnitudes of the first millennium events, that affected northeastern Sicily, such as the 1st and 4th centuries, 650 AD and 853 AD earthquakes (Guidoboni et al. 2007; Bottari et al. 2013; Barbano et al. 2014b) are badly constrained or unknown. Furthermore, large earthquakes occurring in nearby seismogenic zones (Messina Straits, Southern Calabria, and Southern Sicily) locally produced heavy damage.

Historical background on the necropolis of Abakainon

Abakainon necropolis, located in Contrada Cardussa (Fig. 1c), is part of a wider archaeological complex, which includes also some caves and rocky necropolis in Contrada Pertusa (south of Pizzo Cisterna) (Fig. 3) and the ancient town in the district today named Piano. The town of Abakainon was built in an already inhabited area in Middle Neolithic (the tomb caves testify this period); it was then Grecized (sixth century BC–third century BC) and finally Romanized (in 262 BC with the name of Abacaenum) (Oliva 2012). In the fourth century BC, the city of Tyndaris was founded north of Abakainon territory. The increasing prosperity of this settlement is probably linked to the decline of Abakainon itself, which used to be a place of power and importance until then. In fact, from the time of Hieron, it disappears from history and no mention on it is found in the Verrine orations of Cicero (e.g., Diodorus Siculus 1st century BC; Crisà et al. 2015).

Fig. 3
figure 3

a Sketch map of part of Abakainon necropolis (modified after Bacci and Coppolino, 2009). b Photo of a sector of some caves in the rocky necropolis. c Ruins at Abakainon necropolis. d View of the site from upstream. e One of the best-preserved tombs

Full size image

Extensive excavation surveys have never been carried out and we still do not know the real planimetric or topographic development of the town. Fazello (1558) provided a vague description of the ruins of the ancient town: “below the city walls there are remains which appear completely destroyed: collapsed stone columns and ruined fortresses.” Villard (1954), who documented some remains of a Roman Villa (La Torre 2009), provided further elements on the old town.

Since 1994, the Superintendence of the archaeological heritage of Messina has performed systematic excavations, uncovering most of the necropolis area with numerous tombs (Fig. 3c–e). These were found well-preserved and show different constructive features (Bacci and Coppolino 2009); moreover, some of them appear partially collapsed or broken.

The necropolis was used since the fourth century BC to the beginning of II century BC and subsequently abandoned for unknown causes, probably due to the arrival of the Romans. The stratigraphy proposed by Bottari et al. (2013) highlights that the necropolis had already been abandoned at the time of collapse. Indeed, the ruins of the tombs lie on a deposit made up of coarse sand and medium fine gravel up to 20 cm thick and were covered by a further similar loose deposit. Such a colluvium deposit was originated by the erosion of the CDOF formation and completely covered the site before the recent archaeological excavations (carried out between 1994 and 2004).

Geological-geomorphological data, site stratigraphy, and actual state of tombs led Bottari et al. (2013) to suggest that the Greek necropolis of Abakainon was suddenly destroyed, sometime after the second century BC by a seismic event. Optically Stimulated Luminescence dating constrained the age of the collapse to the first century AD, probably during the time span 14–37 AD, when other nearby sites (e.g., Tyndaris) were also damaged by an earthquake, as reported by historical sources (Pliny the Elder 1st century; Phlegon Trallensis 2nd century; Barbano et al. 2014b and further references therein).

Geomorphological analysis

Evidence of neotectonics in the study area

Abakainon necropolis lies on the flank of a NNE-SSW trending steep slope (on average > 50°), on a portion characterized by a slightly flatter morphology, which was thought as an old landslide body by previous studies (Bottari et al. 2013). Along the valley bottom, the perennial Tallarita stream flows towards NNE following a quite straight path. Nevertheless, downstream of the necropolis a marked meandering aspect characterizes such torrent, with relevant loop deviation of the water flow (Fig. 4). This is a key point to address, because literature data report the presence of several normal fault segments surrounding but not crossing the necropolis area (Carbone et al. 1993; ISPRA 2013), which have a strong influence both on the path of the stream and on the neighboring crest lines. In fact, the analysis of a 2-m × 2-m DTM, provided by Regione Siciliana-Assessorato Regionale Territorio e Ambiente, allowed noticing that the meandering path of Tallarita stream seems to have a tectonic origin, because it is affected by some faults. Among these, F1 segment is a NNW-SSE trending fault belonging to TL system, which gives rise to a subvertical scarp along the left bank of the stream. Although morphological evidence suggests that this structure seems to cross the stream, running towards SW and bordering the necropolis area, where a marked scar in the southeastern flank of the slope is likely to testify its presence, official literature reports its termination at the stream (Fig. 4b).

Fig. 4
figure 4

a DTM showing the main morphostructural elements. b DTM with traces of fault reported in official literature (Carbone et al., 1993). c DTM showing the main landslide bodies. d Slope-terrain map

Full size image

Moving towards southwest, another fault (F2) stops at the stream according to Carbone et al. (1993), whereas F3 and F4 are two of the most evident structures of the area (Fig. 5). The first one borders the necropolis along its southwestern margin and brings into contact CDOF formation with APU. This segment is visible by looking at the outcrops along the Mazzarà Stream (Fig. 5a). F4 segment has an almost perpendicular trend, with respect to F3, and gives rise to a tectonic contact between CDOF and APU in the southwestern sector of the study area (Fig. 5b).

Fig. 5
figure 5

a F3 fault segment surveyed along the Mazzarà Stream. b F4 segment surveyed southwest of the necropolis area. c, d Foot of landslides along the Mazzarà Stream. e CDOF boulder surveyed at the foot of M. Sceti. f CDOF boulder surveyed along the trail. g Panoramic view of the studied spot

Full size image

A further element, testifying the relevant tectonic activity of this spot, is the orientation of crest lines. These, usually oriented NNE-SSW, are locally interrupted at the intersection with faults (Fig. 4a). In detail, the ridge of Monte Sceti, standing upstream of the necropolis, shows several plano-altimetric discontinuities suggesting a certain relation with faults (F1 and F2), which are likely to border the necropolis spot. In fact, it is well known that a strong relationship between past/recent tectonics and morphological elements exists, since tectonic structures interact with the morphological features of an area modifying its landscape (Panizza 2014).

Paleosurfaces at different altitudes represent further morphological elements proving the uplift of the area. These, crossed by TL, have been also documented in several international papers (e.g., Lanzafame and Bousquet 1997; De Guidi et al. 2013).

Landslides in the study area

The study area shows signs of slope instability, mainly represented by landslide bodies lying along the slopes around the necropolis site. In particular, landslides mainly occur within CDOF formation, more rarely at the crystalline APU outcrops. Because of active tectonics and environmental conditions, flysch deposits are very prone to fast weathering and weakening, resulting in different types of mass movements, such as rotational slides, mud flows, as well as rockfalls and debris fall. The common characteristic of landslides in flysch deposits is the complexity of the phenomenon itself, which is connected with the softening of clay-bearing rock layers and which may be activated or reactivated by extreme weather events (e.g., Petkovšek et al. 2011; Peternel et al. 2015). Several examples of landslides in flysch deposits are available in the international literature testifying the wide occurrence of such type of mass movements worldwide (e.g., Tarchi et al. 2003; Morales et al. 2004; Sabatakakis et al. 2005; Arbanas et al. 2014; Berti et al. 2017). In the study area, the most common movement is represented by roto-translational slides, with associated earth slides and flows when the resulting failed material is weathered and involved in sliding or flowing (Fig. 4c). Such events modified the original morphology of the slopes, locally giving rise to gentle dips as highlighted by Fig. 4d. The main evidence of landslides is found along the watercourses, where foots of previous events can be surveyed (Fig. 5c, d).

On the other hand, rock outcrops, which are intensely jointed and show local signs of deep weathering, are affected by rockfalls involving boulders with variable size (up to 250 cm of edge). In fact, several sandstone blocks have been surveyed in the study area (Fig. 5e, f), especially along a SW-NE trail running northwest of the necropolis (from now on referred to as “the trail”), testifying the intense predisposition of slopes to this type of movement.

With specific reference to the necropolis site, the morphological evidences of a landslide (L1 in Fig. 4c), represented by a flatter morphology in the upper sector of the slope and by counter slope sectors with swamps at its foot, support the theory that Abakainon necropolis was built on an old landslide body, which is likely to have occurred before its construction. In fact, the flatter morphology of this spot would have probably favored this settlement (Fig. 5g). Nevertheless, as already introduced, dense vegetation and bad accessibility of this spot make hard recognizing further peculiar features relatable to a landslide on site. Moreover, if the landslide occurred before the settlement (i.e., it is older than 2000 years), most of the in situ evidences are likely to have been obliterated by time. The only elements deserving a mention in this section are the anomalous presence of a thick talus deposit, filling part of the incision southwest to the archaeological site, and the occurrence of two swamp spots downstream of the necropolis, where counter slope sectors favor the stagnation of water. This particular morphological trend, with an anomalous drainage pattern, is typical of rotational or complex landslides.

Geophysical data: methods, acquisition, and processing

Geophysical surveys were carried out with the aim of investigating the subsoil at the necropolis area and at its surroundings, looking for evidences of landslide body and/or tectonic structures. Due to the archaeological relevance of the site, active seismic profiles with appropriate energization techniques were forbidden. Therefore, we carried out a “passive seismic single-station” survey, based on the acquisition of environmental seismic noise, and an ERT survey. The main advantage of these methodologies is related to their noninvasiveness and quick execution, since they do not require any artificial energization. Moreover, they highlight the interface between rocks with different physical-mechanical characteristics; therefore, they can be considered suitable methodologies for the aims of this research.

Methods

Passive seismic single-station technique

The “passive seismic single-station” technique is based on the acquisition of environmental noise, constantly present on Earth’s surface. The environmental noise is made up of seismic waves characterized by low energy and amplitudes of the order of 10−4–10−2 mm (Okada 2003).

Seismic noise samplings acquired in this work have been elaborated using the HVSR technique, which allows calculating the spectral ratio between the average of the horizontal on the vertical components of the ground motion (Nakamura 1989). The peaks observed in the spectra are dependent on subsoil seismostratigraphy and are a direct function of the fundamental period of deposits and of site resonance frequency (f0). Therefore, the HVSR technique allows identifying the frequency at which ground motion is amplified due to resonance effects, related to the presence of tectonic and/or stratigraphic discontinuities such as a landslide.

The resonance spectral peaks are indicative of an impedance contrast in the subsoil linked to the transition between layers with different seismic impedance, which is the product of density (ρ) and seismic velocity (V). It varies among layers of different lithology or between homogeneous layers with different mechanical characteristics. The size of the amplification depends on the magnitude of the impedance contrast present in the subsoil.

From a known resonance frequency (f) and shear wave velocity (Vs), it is possible to obtain information on the thickness of the resonating layer (H). In fact, if we consider a simple two-layer system characterized by different densities (ρ1 and ρ2) and velocities (V1 and V2), there is an equation linking the resonance frequency f to the thickness H of the resonating layer:

$$ f=\frac{V_{\mathrm{s}}}{4H} $$
(1)

Several authors, in recent years, used HVSR technique to obtain information on the seismostratigraphy characteristics of the subsoil (Lermo and Chavez-Garcia 1993; Lachet and Bard 1994; Ibs-von Seht and Wohlenberg 1999; Pappalardo et al. 2016; Imposa et al. 2016, 2017).

Assuming that the shear wave velocity increases with the depth due to increased lithostatic load, it is possible to fit a known 1D Vs profile, determined experimentally, using the following function (Ibs-von Seht and Wohlenberg 1999):

$$ {V}_s(Z)={V}_{0\kern0.5em }{\left(1+Z\right)}^{\alpha } $$
(2)

where:

Vs (Z):

Shear wave velocity

V 0 :

Shallow shear wave velocity

Z :

Depth

α :

Proportionality coefficient between depth and velocity

This allows obtaining V0 and α parameters affected by the minimum misfit between the experimental data of Vs–depth profile and the function fitting these data.

Then, by using such two coefficients, through the Ibs-von Seht and Wohlenberg (1999) formula, it is possible to convert the H/V spectra frequency values, in depth:

$$ h={\left[\frac{V_0\left(1-\alpha \right)}{4f}+1\right]}^{\frac{1}{\left(1-\alpha \right)}}-1 $$
(3)

Therefore, it is possible to reconstruct the impedance contrast sections related to acquisitions performed with a regular interdistance along different alignments. In these sections, the contouring of the H/V spectral ratio amplitude values highlights the depth of the impedance contrast producing the resonance peaks.

Electrical resistivity tomography

In recent years, geoelectrical methods have been largely employed in near-surface surveys achieving significant contributions in the subsoil investigation (Pellerin 2002; Auken et al. 2006; Imposa et al. 2007; Reynolds 2011; Loke et al. 2013). The ERT is a methodology aimed at calculating the subsurface distribution of electrical resistivity from a large number of resistance measurements made by electrodes placed in an arbitrary geometric pattern (Daily et al. 2004; Imposa et al. 2004a). The analysis and interpretation of geoelectrical outcomes allow the identification of resistivity contrasts that can be mainly due to the lithological nature of the terrains and the water content variation. The operative procedure is carried out by the use of a multi-electrode cable, placed on the ground with an array of connected steel electrodes placed according to a specific configuration. Such electrodes are used both for the injection of the current (I) in the subsoil and the measurement of the voltage (V). Knowing I and V and the geometrical coefficient depending on the employed electrode configuration, the apparent resistivity values characterizing the subsoil investigated can be calculated (Perrone et al. 2014). These values are positioned at pseudodepths according to a geometrical reconstruction (Edwards 1977), which results in a pseudosection representing an approximate picture of the true subsurface resistivity distribution (Hack 2000). To obtain an ERT, the apparent resistivity values must be inverted by using inversion routines (e.g., Loke and Barker 1996; Loke et al. 2003).

Several authors employed this methodology to investigate terrains affected by landslide movements, proving the reliability of achieved results (e.g., McCann and Forster, 1990; Hack 2000; Jongmans and Garambois 2007; Capizzi and Martorana 2014).

Setting of field measurements

H/V measurements and directional resonance analysis

H/V surveys, performed during a single day characterized by stable weather conditions, allowed recording 83 environmental noise data. All the acquisition points were georeferenced with a GPS; 51 records were acquired along a NNE-SSW profile (profile 1), located along the trail, and the other 32 were carried out along three sections within the perimeter of the archaeological site (Fig. 6) according to the setting reported in Table 1 and Fig. 6.

Fig. 6
figure 6

Location of the ambient noise sampling points. Top-left inset shows a detail of the sampling points within the necropolis perimeter

Full size image
Table 1 H/V acquisition deployment
Full size table

Surveys were carried out using four portable three-component digital seismographs and the ambient noise was acquired using a sampling frequency equal to 128 Hz. Each recording lasted 16 min to ensure the stationarity of the recorded signal, even if no strong anthropogenic sources of noise were present near the site.

Each acquired signal was subdivided into time window of 20 s and a triangular window function with a smoothing equal to 10% of central frequency was applied at each window.

The criteria of execution goodness, analysis, and reliability (statistical tests-SESAME guidelines 2005) are fulfilled by all measurements, while the fundamental spectral peak was not always significant and clear for all the six analyzed criteria.

Furthermore, the spectral ratio was calculated starting from north (0°) to south (180°), rotating the NS and EW components with azimuthal intervals of 10°, in order to verify the possible presence of polarization effects. Polar diagrams, showing the amplitude contour of each spectral ratio for the different directions in the frequency range 0.5–20 Hz, were obtained.

Some of the samplings along the trail were further elaborated by the Directional H/V Peak Occurrence Rates (DHVPOR) technique (Del Gaudio et al. 2013). This method allowed highlighting, among the recording time windows, systematic orientation in site response through the identification of a high percentage of significant H/V peaks around a given frequency-azimuth combination. A H/V peak is considered significant if characterized by an average H/V ratio greater than 2 and maximum and minimum H/V ratio observed at the same frequency (in an orthogonal direction with divergence no more than 30°) at least equal to 1.5 (Del Gaudio et al. 2008).

The DHVPOR technique allows also detecting the directional effect affecting the study area, closely related to the geological, geomorphological, and structural characteristics of the investigated site. This method is useful to distinguish the frequencies, in the range 0.5–20 Hz, affected by a directional effect (thus the directional effect can be correlate to a specific depth in the subsoil) and to identify the possible presence of different directional effects affecting subsoil portions at different depths.

Electrical resistivity tomography

ERT was carried out on a transect located along the trail, parallel to the H/V profile 1, by employing an ABEM Terrameter SAS1000 georesistivimeter, equipped with a self-potential instrument for the measurement of natural DC potentials. By using a configuration array of Wenner-type, 64 electrodes, 3 m spaced, were connected to the acquisition device by a multi-conductor cable. About 500 apparent resistivity values were recorded, for a total investigated depth of about 18 m. The purpose of such indirect investigation was to look for a detailed reconstruction of the shallow subsoil, in order to find similarities with the achieved H/V impedance contrast sections.

Results

Passive seismic survey results

All the 83 spectra, elaborated through the H/V spectral ratio technique in the frequency range 0–64 Hz, show a peak at low frequencies between 0.7 and 1.5 Hz (from now on referred to as Pk1), although with low H/V amplitude values in most cases. The stratigraphic origin of such low frequency peak can be identified by observing the graphs of spectral amplitude components related to each H/V spectrum. Indeed, they show the typical “eye-shape” (Castellaro and Mulargia 2009), which is formed between horizontal components and vertical components when an impedance contrast is present in the subsoil.

As an example, in Fig. 7, some H/V spectra related to the different profiles performed in the study area are reported. Almost all the spectra show other spectral peaks at higher frequencies, related to the presence of impedance contrasts located at lower depths in the subsoil. In particular, two main peaks were recognized, from now on referred to as Pk2 and Pk3. Pk2 occurs at medium frequencies, between 2 and 10 Hz, while Pk3 is at high frequencies (f > 15 Hz).

Fig. 7
figure 7

Examples of H/V spectra, spectral amplitudes of components, and polar diagrams related to profile 1: a red band, b blue band, c violet band, and d green band in Fig. 6. Color bands in the graph of spectral amplitude of components highlight the three different identified peaks: yellow band for Pk1, blue band for Pk2, and pink band for Pk3

Full size image

Along profile 1 (Fig. 6), such peaks are not always present and, sometimes, they occur with different amplitudes; therefore, for an easier description of results, we have divided profile 1 into subsegments identified by bands grouping homogeneous records (Fig. 6).

Starting from northeast, the red band (band 1) (Fig. 6) is characterized by spectra showing the occurrence of both Pk2 and Pk3. In particular, Pk2 has H/V spectral ratio amplitude ranging between 3 and 6, while Pk3 is connected to a very shallow impedance contrast with low value of H/V amplitude.

The polar diagrams related to these acquisitions show a considerable directional effect in ENE-WSW direction, linked with Pk2, which represents in this case the most amplified spectral peak (Fig. 7a).

Moving towards southwest (blue band 2 in Fig. 6), Pk2 disappears almost completely. On the contrary, Pk3, ranging between 10 and 20 Hz, is amplified and reaches H/V amplitude values up to 4. Polar diagrams in Fig. 7b show the directional effects of Pk3, oriented about ENE-WSW. Moreover, 3D histograms obtained by the DHVPOR analysis (Fig. 8) highlight that recordings of this profile band are characterized by a significant occurrence rate percentage of H/V directional peaks.

Fig. 8
figure 8

a Histograms showing the variation of maximum directional peak occurrence rate percentage between some of the acquisitions performed along the profile 1. It is evident that some of the acquisitions at blue and green profile areas are characterized by a high percentage of significant directional peaks compared to adjacent ambient noise records. b Example of DHVPOR histograms related to the acquisitions highlighted in yellow in frame (a).

Full size image

Acquisitions carried out in the central part of profile 1 (violet band 3 in Fig. 6) show, in addition to Pk1, a H/V peak at higher frequency relatable to Pk3, which however does not have a significant H/V spectral ratio amplitude. In fact, polar diagrams do not show the directional effect found along the blue profile band. The percentage of the occurrence rate of the H/V directional peaks is considerably reduced if compared to the adjacent measurements (Fig. 8).

In the southwestern sector of profile 1 (green band 4 in Fig. 6), a significant Pk2 was recorded along with a further weak impedance contrast (Pk2_bis), which may indicate the presence of a slight variation of the physical-mechanical characteristics within the same lithology. In polar diagrams (Fig. 7c), a significant directional effect can be observed at ENE-WSW direction for Pk1 and Pk3, while Pk2 shows a directional effect oriented about N-S. Moreover, the highest percentage of the directional peak occurrence rate (Fig. 8) tends to increase, reaching in some cases occurrence rates greater than 50% (Fig. 7d).

H/V spectra acquired within the necropolis perimeter provide results comparable to the aforementioned ones. In particular, profile 1-bis (Fig. 9a) recorded three main different impedance contrasts in the subsoil corresponding to Pk1, Pk2, and Pk3 detected along the trail. At the central portion of the alignment, a significant increase in the H/V amplitude for Pk2 was recorded; the linked polar diagrams show that this peak is also affected by a clear directional effect (Fig. 9a), which typically occur close to a tectonic structure (Imposa et al. 2004b). On the other hand, in this sector, Pk3 is not identified.

Fig. 9
figure 9

Examples of H/V spectra, spectral amplitudes of components, and polar diagrams related to profile 1_bis (a), profile 2 (b), and profile 3 (c). Color bands in the graph of spectral amplitude of components highlight the three different identified peaks: yellow band for Pk1, blue band for Pk2, and pink band for Pk3

Full size image

Moving towards southwest, the H/V amplitude for Pk2 decreases and a spectral peak with high H/V amplitude at a frequency around 15 Hz is detected (Pk3).

Profile 2 (Fig. 6), performed inside the Necropolis excavation orthogonal to profile 1-bis, is characterized by Pk1 and Pk3 with a weak H/V amplitude and by Pk2 with greater H/V amplitude. Polar diagrams show that the spectral peak at medium frequency has a directional effect in NE-SW direction (Fig. 9b).

Profile 3 (Fig. 6), carried out parallel to profile 2 inside the Necropolis, is comparable to profile 2; furthermore, a directional effect, linked with the spectral peak at medium frequency, can be observed in the polar diagrams, in NNE-SSW and NE-SW direction (Fig. 9c).

Impedance contrast sections

Data related to each profile and commented in the previous paragraph were further processed to obtain four impedance contrast sections. Frequencies were converted into depths by using V0 and α parameters obtained by the fit of 1D velocity profiles previously acquired nearby the study area.

Such sections show the contour plot of H/V amplitude value in the subsoil through color variations associated with a chromatic scale; the distance between acquisition points and depth values are plotted along the abscissa and ordinate axis, respectively.

These sections (Fig. 10) are useful to provide a schematic overview on the distribution of the detected impedance contrasts and to propose an interpretation model of the investigated subsoil. The most common feature in all the sections is the presence of the impedance contrast originating Pk1 at approximately 80 m of depth, which is likely to represent the geological contact between CDOF and the underlying APU. Such depth is in accordance with stratigraphic evidence as shown by Fig. 2, where the geological contact between such formation crops out downstream of the necropolis site.

Fig. 10
figure 10

Impedance contrast sections resulting from the contour plot of H/V amplitude recorded along profile 1 (a), profile 1-bis (b), and profiles 2 and 3 (c). Dashed lines highlight the main impedance contrasts

Full size image

More specifically, section arising from profile 1 (Fig. 10a) shows other significant impedance contrasts, such as the one (relatable to Pk2) at its NNE portion at about 15 m of depth. This, representing a contact between materials with different physical-mechanical properties, is likely to belong to the lateral side of a landslide body as inferable by the concave shape of such contrast. According to field data, this landslide body would be L2 (Fig. 4c), surveyed northeast of the necropolis.

Moving towards southwest, between progressive 50 and 100 m along the profile and at a depth of about 3 m, there is a shallow contrast (linked to Pk3) testifying the presence of buried rock boulders at the foot of a rocky outcrops, where rockfalls signs were surveyed (Fig. 5c, d).

Close to the Necropolis site, a further impedance contrast from a shallow depth to about 20 m and with a concave trend is visible. This main contrast is likely to be related to a landslide body lying right underneath the Abakainon necropolis, confirming the hypothesis that this site was settled on an old landslide. Further impedance contrasts occurring in this sector (such as Pk2_bis) may be associated with the landslide, probably due to variations in the physical-mechanical characteristics between different portions of the mobilized material. Moreover, the extremely superficial impedance contrast (1–2 m depth, relatable to Pk3), identified next to the necropolis, may be associated with the presence of still buried graves.

The section arising from profile 1 bis (Fig. 10b) shows the same impedance contrasts observed in the final sector of profile 1, supporting the hypothesis of the presence of a landslides body underneath the necropolis. The contrast related to the transaction between mobilized material and underling lithotypes, thus indicating the surface of rupture of investigated landslide, is located at a slightly lower depth under the necropolis, testifying that the archaeological site is likely to be located on the margin of this landslide.

Moreover, orthogonal sections (Fig. 10c) show the presence of impedance contrasts located at shallow depths, within the first three meters b.g.l., linked with the possible presence of additional buried graves. The contrast related to the landslide body is located at − 10/− 15 m b.g.l. at profile 2 and at about − 8/− 10 m b.g.l. at the profile 3. This is a relevant datum, because it allows constraining the spatial extension of the landslide body, proving that the necropolis stands at the margin of such landslide, as aforesaid.

ERT section

ERT allowed investigating a depth of about 18 m, in order to achieve a good detail of the shallow subsoil (Fig. 11). According to achieved outcomes, four main elements deserve a comment.

  1. 1.

    Shallow material with the highest resistivity value > 210 Ω·m is found in the necropolis area. Such material can be associated with the buried tombs clearly detected by the passive seismic survey southwest of the necropolis. The same high-resistivity material is found northeast of the archaeological site, testifying the presence of other tombs or buried rock boulders fallen by the upstream rock cliff.

  2. 2.

    The landslide body is identified by a resistivity range ≤ 90 Ω·m, compatible with loosened and weathered material. Sliding surface cannot be identified by ERT section, because it lays at greater depths (about 25 m) according to the passive seismic outcomes.

  3. 3.

    Presence of an abrupt step at about 90 Ω m, which is spatially located where H/V spectra recorded a strong directional effect. This may be related to the presence of a fault, supported also by morphological evidences.

  4. 4.

    Very low-resistivity spot in the middle of the landslide body (5–30 Ω m) related to water circulation. This sector corresponds to the morphological incision and the depth of groundwater is correlated to the downstream ground level where swaps were surveyed. (Fig. 3a).

Fig. 11
figure 11

Location of ERT and tomography section

Full size image

Discussion and conclusions

Results of the multidisciplinary study described herein allowed finding out interesting aspects on the site hosting the archaeological remains of the Greek Abakainon necropolis. The initial hypothesis that the site may have been built on the flat spot offered by an old landslide body is confirmed by morphological and geophysical evidence. In fact, the study area lies in a peculiar sector of Peloritani Mountains, characterized by numerous landslides occurred especially in the close proximity of faults. Several landslide bodies were identified in the surroundings of the necropolis by field surveys coupled with the analysis of a 2-m DTM, which allowed drafting a landslide body (L1) at the necropolis site (Fig. 4). Moreover, Abakainon necropolis lies in an area characterized by a high concentration of faults belonging to the NNW-SSE trending system known as Tindari Line (TL). Most of these fault segments, still locally active, have strongly modified the original morphology of the area, giving rise to local deviation of water flows (meanders at Tallarita Stream), plano-altimetric ruptures, and deviation of crest lines. All these elements concur in predisposing the area to gravitational mass movements.

Indirect geophysical surveys, with particular reference to passive ambient noise measurements, allowed recognizing three main characteristic impedance contrasts, which were correlated with peculiar elements of the study area (Fig. 12). Pk1 is the expression of the geological contact between CDOF and APU, which is located at an indicative depth of about 80 m and is constant along the investigated profiles. The surface of rupture of L1 landslide under the necropolis is highlighted by a 25–30-m-depth impedance contrast, separating the loosened material involved in the movement from the rest of CDOF formation. This impedance contrast, herein referred to as Pk2, was found also in the northeastern sector of the area, where another landslide body was surveyed (L2 in Fig. 4b), thus confirming the reliability of this datum. Such result allows ascertaining that the necropolis was built on the margin of a 500-m-long and 170-m-wide landslide (Fig. 13), probably no longer active at that time. Finally, Pk3 is the expression of the impedance contrast highlighting rock boulders and buried tombs within the necropolis and along the trail (Fig. 12).

Fig. 12
figure 12

Conceptual map on the relation between H/V peaks and corresponding features in the study area

Full size image
Fig. 13
figure 13

Summary of achieved outcomes: a main geomorphological elements and main faults of TL system reported on a DTM of the study area, b landslide depicted on a Google 3D model of the study area with trace of the profile shown in c, and c schematic profile of the slope with a sketch of the landslide

Full size image

Such graves were also highlighted by the electrical tomography, which showed the presence of high-resistivity material within the first meters of subsoil. ERT highlighted also water circulation in the southwestern sector of the study area (Fig. 11), where an evident incision driving meteoric waters towards downstream (where also swamps were surveyed) is present.

Moreover, strong directional effects at some recorded H/V spectra are the evidence of the presence of fault segments bordering and even affecting the landslide body. This is a relevant outcome, because it allows defining the prosecution of two direct faults (F1 and F2), which were not mapped in the study area (Fig. 13), probably because of the presence of the landslide and to the rheological behavior of CDOF. Indeed, this thick sequence (ca. 80 m) may have accommodated the deformations caused by the faults, without a purely brittle behavior, thus hiding such structure on the surface.

According to achieved outcomes, F1 borders the necropolis area, the landslide, and the south of M. Sceti, where an evident scar (Fig. 4b) testifies its presence. F2 segment affects the necropolis site itself cutting also the landslide body as testified by the crest line deviation southeast of the necropolis and by the step highlighted by ERT, respectively. This is a crucial evidence, proving that the necropolis landslide is a relict movement, which occurred before the latest reactivation of F2. In this perspective, it is likely that Abakainon necropolis collapsed under the shaking of an old earthquake, whose action was probably enhanced by a site effect due to the relict landslide body on which the necropolis itself was built.

The archaeoseismological study carried out by Bottari et al. (2013) pointed out that the site could have been destroyed by an earthquake and excluded the landslide occurrence as a destruction cause, since the necropolis was used for about three centuries, and today, after ca. 2000 years, it does not show any evident sign of instability. Authors suggested that the earthquake destroying the site probably occurred in the first century. However, they concluded that the causative seismogenic source that might have produced the collapse of the necropolis of Abakainon could be located either in the Messina Straits or in TL seismic zone. Our findings suggest that the earthquake was more likely linked to TL. Therefore, when the main structure ruptured (for an event in the first century or for another unknown earthquake), also the faults observed in the study site could have moved. This movement can explain the counter slope tilting of the tomb basements, as well as the oriented collapse of columns of the funerary monument observed by Bottari et al. (2013). Indeed, as observed during some Italian recent seismic sequences (at Colfiorito village in 1997, L’Aquila city in 2007 and Amatrice village in 2016), although faulting is related to the main structures, seismically induced gravitational displacements effects can occur in the area around the main fault (e.g., Albano et al. 2017; Pucci et al. 2017). Thus, since a similar seismotectonic setting characterizes Abakainon area, such phenomenon is likely to have occurred also at the studied site.

Moreover, a recent activation of such fault system is testified by F4 segment, which is well recognizable in well-structured young slope deposits south of the studied site, proving that faults in the area recently moved.

As a final remark, this study outlines that the multidisciplinary approach presented herein can help to clarify the geology of an area, especially where faults and landslides are not easy to survey as in restricted areas such as an archaeological site, where active prospecting or drillings are not allowed.