Forensic Investigations of Geohazards: The Norcia 2016 Earthquake

: Earthquakes represent one of the world’s most signiﬁcant hazards in terms of damage to human and animal life, and property. Earthquakes also cause many other related fatalities and damage to urban structures. This paper presents the forensic investigation of failures induced by the Norcia 2016 earthquake in Italy. The detailed geophysical ﬁeld investigations were carried out at selected locations in two cities: Rome and Amelia. The places of investigation were 150 km and 90 km, respectively, from the epicenter. A ground penetrating radar (GPR) survey was carried out at the sites to highlight structural failures, and included a partially damaged urban bridge, and the cracked wall of a private house. These failures have been discussed with reference to the ﬁeld measurements carried out. In both cases, the GPR radargram showed clear lesions along with their geometry and location. This forensic geoscientiﬁc analysis highlights the importance of detecting structural damage immediately after a geohazard event to help plan proper interventions, efforts to prevent human losses and help law enforcement to focus their forensic investigations.


Introduction
The importance of non-destructive techniques such as the ground penetrating radar (GPR) is well known across various fields of its application. In particular, this methodological approach is often used in forensics to help investigators locate bodies or stolen objects that are buried ( [1], and the literature therein). Moreover, this tool is frequently used in geological research and risk studies as it allows mapping of the subsoil, its stratigraphy, and its critical points in an accurate and completely non-invasive manner [2][3][4][5]. When geological risk is intertwined with forensic investigation, we refer to it as a forensic geohazard.
In fact, the use of this geophysical method, GPR, has been recently developed in interconnected environments where several major natural catastrophes (such as earthquakes, floods, landslides, etc.) unfortunately lead to deaths and/or structural damage [6][7][8][9][10][11]. In this context, GPR can play a key role not only in the search for missing persons [12], but also in the monitoring of possible damage to structures and buildings.
Adequate prevention of damage caused by natural disasters requires improving the level of knowledge of buildings, and detecting structural and non-structural anomalies and vulnerabilities that need to be corrected urgently [13]. In the first place, a geometric-structural survey is used to represent the overall geometry of the constructive elements, while also considering the state of conservation of the materials and the lesions (in progress or stable) through identification of the fractures and damage. Fractures in an existing building, depending on the type of the fractures and the circumstances, can be warning signals. Therefore, when analyzing fractures, it is essential to understand whether they are understand whether they are due to settling of the structure or due to a more serious phenomenon that could lead to the collapse of the load-bearing elements ( [14,15], and literature therein).
In a given area and within a given time interval, the seismic risk can be measured by the extent of damage that is expected because of earthquakes. The hazard can be assessed in terms of victims, economic loss, and damage to buildings. Hazards and risks have the same relationship as cause and effect; in fact, risk is represented by the earthquake (the cause) that can strike a certain area, while hazard is represented by its possible consequences, that is, by the damage that can be expected (the effect) [16]. The definition of hazard is influenced not only by the risk, but also by the characteristics of the territory. In order to define the hazard level of a territory, it is necessary to know the territorial seismicity, that is, the frequency of earthquakes and their intensity, the method of construction of the structures, the quantity and quality of the exposed goods, and the density of the population [17]. Considering the same frequency and intensity of earthquakes, the hazard is zero where there are no buildings, exposed goods or population, while on the other hand, densely populated areas or areas characterized by structures that are not resistant to the shaking by a seismic wave represent a high level of hazard [18,19].
The forensic aspect of non-destructive geoscientific investigations lies in its ability to help in rapidly identifying, and thereby, preventing further post-shake collapses and possible victims, in particular in Italy where this approach is not common.
This last aspect, together with an analysis of the structures, will be discussed in detail in the following sections. The study not only tested the GPR's ability to monitor and identify damage during the seismic event known as Norcia 2016, in Italy, in urban areas and densely inhabited places several kilometers away from the epicenter, but also evaluated the logistical contribution that this method can make to law enforcement in preventing post-earthquake damage by reducing collapses [20].

Materials and Methods
The areas under investigation were in the cities of Amelia (in the province of Terni) and Rome, 90 km and 150 km away from the epicenter of the earthquake in Norcia (Italy), respectively. The places in which the GPR surveys were conducted included: (i) a private house in the center of Amelia, and (ii) the Marconi Bridge-one of the busiest bridges in Rome, located on the River Tiber ( Figure  1).

The Norcia 2016 Earthquake
The earthquake that struck Italy on 30 October 2016, at 06:40:17 UTC, (07:40:17, Italian time) was the strongest earthquake since the 1980 Irpinia earthquake. The hypocenter lay at Latitude 42.84 North, Longitude 13.11 East, and at a depth of 9 km. The magnitude calculated in the Istituto Nazionale di Geofisica e Vulcanologia (INGV) monitoring room was 6.1 ML and 6.5 MW as compared to the 6.9 MW of the Irpinia earthquake. The earthquake affected the provinces of Perugia, Rieti and Macerata and was strongly felt in Central Italy; the epicenter was located 5 km from Norcia, 7 from Castelsantangelo sul Nera and Preci, and 10 km from Visso. In case of an earthquake of magnitude 6.5, the fault has an area of several hundred square kilometers and, therefore, the entire area above and around the fault is affected by strong shaking. The earthquake of 30 October was the strongest event of a sequence which began with an earthquake on 24 August of 6.0 MW and another earthquake of 26 October of the magnitude MW = 5.9 [21][22][23][24][25][26][27].
The Italian Accelerometric Network (RAN), managed by the Department of Civil Protection (DPC), and the Italian Seismic Network, managed by the Istituto Nazionale di Geofisica e Vulcanologia (INGV) provided the records of about 650 accelerometric stations. In Figure 2, thanks to both the Italian strong-motion database, ITACA (Italian Accelerometric Archive (http://itaca.mi.ingv.it)) and the USGS database (http://earthquake.usgs.gov), it is possible to calculate the peak ground acceleration (PGA) at the epicenter and at the two study areas (Amelia and Rome). These were 536.48, 18.28, and 24.29 cm/s 2 , respectively [23][24][25].

The GPR Measurements
Over the last thirty years, geophysical techniques have been increasingly used to detect the presence of anomalies in the subsurface, as well as to estimate the hidden targets inside the walls of a building. Geophysical methods can be effectively applied to produce an image of a buried target to prevent extensive, destructive, time consuming and expensive excavations [28].
The scope of a geophysical measurement is the detection of the "boundaries" between objects having different values of a specific physical property. The contrast between the searched target and the background should be strong enough to be detected "at a distance". That is, the target should be able to generate a measurable relative spatial variation of the specific physical property at the surface [29].
GPR operates at frequencies between a few MHz and 3 GHz. GPR's depth of penetration depends on the electrical properties of the ground. For example, in the case of ground with a relatively high conductivity (such as caused by saturated clay), the depth of penetration may be less than 0.5 m, yet in the case of quartz sand, which has low conductivity, GPR penetration may be tens of meters [30].
The radar unit produces a pulsed electromagnetic wave that travels through the ground at a velocity controlled by the electrical properties of the subsurface. Differences in relative permittivity (dielectric constant) or electrical conductivity due to changes in soil/material type or water chemistry result in the waves being reflected. By moving the radar antenna over the material surface, a continuous real-time section (radargram) is built up by arranging each radar record next to the other. The horizontal axis of the section represents the distance (m), and the vertical axis represents the two-way travel times (TWT) of reflections in nanoseconds (ns). To transform this information into distance, the velocity must be known. The lines shown on the radargram represent reflectors and are constructed of coalescing wiggle traces from individual radar records [31,32].
For most survey reports, grayscale plots of the radargrams (i.e., GPR sections or B-scans) are the primary presentation format, and several improvements have been introduced to create a more solid appearance. If one has collected multi-profile grids, one can develop maps (i.e., time/depth slices or C-scans) representing various depths by using the average envelope amplitude and by interpolating among all of the radargrams [33].
In this work, the GPR used in both measurements was the bistatic FINDAR (Sensors & Software, Inc. Mississauga, ON, Canada) system, equipped with a 500 MHz antenna.
In the private apartment in Amelia, a visibly damaged wall was analyzed to understand the internal geometry of the fracture that was assumed to pass through the masonry from the inner part of the apartment to the outer part, which was on the balcony. This type of investigation is well documented in the literature [34][35][36][37][38][39][40], but less used in the post-earthquake situation. The whole house was built in the 1970s, before the new anti-seismic rules and before the modern development of the seismic vulnerability map. The building is a hybrid structure, masonry and reinforced concrete, and this was the most evident and critical crack.
Data was acquired vertically inside a 1.1 m × 1.2 m one-way grid, with a distance of 0.1 m among profiles ( Figure 3). No other processing beyond the standard one (AGC, DeWOW and Average Envelope Amplitude) was required. The signal penetration velocity was calculated using the hyperbola calibration and verified using the actual measurements of the exposed masonry: 0.14 m/ns.
On the Marconi Bridge in Rome, the GPR investigation was focused on the pavements of the bridge. This bridge was built between 1937 and 1955; the construction was interrupted due to the WWII and was only resumed in 1953. The total length is about 235 m and it is the longest bridge in Rome, with six arches and a width of about 31 m [41]. It has a "winged" structure where the central part of the bridge, supported by the pylons, is crossed by vehicles, while the external "wings" are the sidewalks (Figures 4 and 5). This type of investigation is also well known in the literature [34,[42][43][44][45]; however, there has been little post-seismic application, especially in bridges having this particular structure. this was the most evident and critical crack.
Data was acquired vertically inside a 1.1 m × 1.2 m one-way grid, with a distance of 0.1 m among profiles (Figure 3). No other processing beyond the standard one (AGC, DeWOW and Average Envelope Amplitude) was required. The signal penetration velocity was calculated using the hyperbola calibration and verified using the actual measurements of the exposed masonry: 0.14 m/ns.  On the Marconi Bridge in Rome, the GPR investigation was focused on the pavements of the bridge. This bridge was built between 1937 and 1955; the construction was interrupted due to the WWII and was only resumed in 1953. The total length is about 235 m and it is the longest bridge in Rome, with six arches and a width of about 31 m [41]. It has a "winged" structure where the central part of the bridge, supported by the pylons, is crossed by vehicles, while the external "wings" are the sidewalks (Figures 4 and 5). This type of investigation is also well known in the literature [34,[42][43][44][45]; however, there has been little post-seismic application, especially in bridges having this particular structure.   We acquired both single profiles along the two sidewalks, above the "wings", where the greatest damage and detachments were visible after the 2016 earthquake (Figure 6), and one-way grids of 5 m × 3 m, with a distance of 0.25 m between the profiles in the most severely damaged areas. No processing beyond the standard processing (AGC, DeWOW and Average Envelope Amplitude) was required. The signal penetration velocity was calculated by means of the calibration of the hyperboles, and subsequently verified by the real measurements of the thickness of the visible wings. It was 0.14 m/ns. We acquired both single profiles along the two sidewalks, above the "wings", where the greatest damage and detachments were visible after the 2016 earthquake (Figure 6), and one-way grids of 5 m × 3 m, with a distance of 0.25 m between the profiles in the most severely damaged areas. No processing beyond the standard processing (AGC, DeWOW and Average Envelope Amplitude) was required. The signal penetration velocity was calculated by means of the calibration of the hyperboles, and subsequently verified by the real measurements of the thickness of the visible wings. It was 0.14 m/ns. m × 3 m, with a distance of 0.25 m between the profiles in the most severely damaged areas. No processing beyond the standard processing (AGC, DeWOW and Average Envelope Amplitude) was required. The signal penetration velocity was calculated by means of the calibration of the hyperboles, and subsequently verified by the real measurements of the thickness of the visible wings. It was 0.14 m/ns.

Results and Discussion
GPR, a technologically advanced tool, proved to be a suitable solution to the problem of monitoring structures with visible damage caused by catastrophic events, such as this earthquake and the resulting geohazards in the post-seismic phase.
As for the private house in Amelia, the damage to the masonry can be easily related to the earthquake of 2016, because the crack was not present before the earthquake. In the case of the Marconi Bridge, the request for monitoring the internal structure came following the detachments (see Figure 5) that were discovered after the 2016 earthquake. The main findings of the two investigated areas have been reported below. Figure 7 shows the single profile acquired inside the above-mentioned grid and the depth-slices that clearly illustrate the presence of an oblique anomaly, which probably corresponds with the fracture visible on the masonry. It is worth noticing that this fracture passes through the masonry from side to side, to affect the outer wall of the room, also on the balcony side. The last slice in Figure 8, acquired from inside the room at about 0.5 m depth coincides in shape and geometry with the fracture visible outside on the balcony. Thanks to the penetration capacity and the resolution of the GPR, it was possible to understand not only the internal geometry of the fracture but also its gravity, and it facilitated the restructuring that prevented the collapse of the wall and the possible damage to property and people. Given its sudden appearance of this crack following the Norcia 2016 earthquake and its peculiar nature it is also closely related to the similar post-seismic damage found in the area in other buildings [46]. fracture visible outside on the balcony. Thanks to the penetration capacity and the resolution of the GPR, it was possible to understand not only the internal geometry of the fracture but also its gravity, and it facilitated the restructuring that prevented the collapse of the wall and the possible damage to property and people. Given its sudden appearance of this crack following the Norcia 2016 earthquake and its peculiar nature it is also closely related to the similar post-seismic damage found in the area in other buildings [46].   Figure 9 shows the orientation of the profiles acquired along the two "wings" of the Marconi Bridge, and the attempt to overlap one of the GPR profiles to show the thickness of these "wings" and the penetration of the radar signal through the pylons up to the surface of the river water below. Figure 10 shows in detail the most obvious anomalies found from the acquired images. If, on the one hand, the thickness of the "wings" at about 0.6 m is clear, on the other hand, there are evident inconsistent hyperbolic events that can only be traced back to the presence of detachments or cracks inside the sidewalks, as partially shown in Figure 6. Moreover, the detected GPR anomalies are near these visible lesions on the bridge.  Figure 9 shows the orientation of the profiles acquired along the two "wings" of the Marconi Bridge, and the attempt to overlap one of the GPR profiles to show the thickness of these "wings" and the penetration of the radar signal through the pylons up to the surface of the river water below. Figure 10 shows in detail the most obvious anomalies found from the acquired images. If, on the one hand, the thickness of the "wings" at about 0.6 m is clear, on the other hand, there are evident inconsistent hyperbolic events that can only be traced back to the presence of detachments or cracks inside the sidewalks, as partially shown in Figure 6. Moreover, the detected GPR anomalies are near these visible lesions on the bridge. and the penetration of the radar signal through the pylons up to the surface of the river water below. Figure 10 shows in detail the most obvious anomalies found from the acquired images. If, on the one hand, the thickness of the "wings" at about 0.6 m is clear, on the other hand, there are evident inconsistent hyperbolic events that can only be traced back to the presence of detachments or cracks inside the sidewalks, as partially shown in Figure 6. Moreover, the detected GPR anomalies are near these visible lesions on the bridge.  In addition to the internal damage to the "winged" structure of the bridge, there are also more superficial damage linked to the failure of the travertine surface that covers the sidewalks. This is evident from Figure 11, where both the radar profiles and the depth-slices show evident anomalies due to detachments and lesions.

Rome
Note that, although the cracks illustrated in Figure 5 are certainly post-2016 earthquake and Rome suffered similar damage after this seismic event [47,48], the other injuries highlighted by the In addition to the internal damage to the "winged" structure of the bridge, there are also more superficial damage linked to the failure of the travertine surface that covers the sidewalks. This is evident from Figure 11, where both the radar profiles and the depth-slices show evident anomalies due to detachments and lesions.

Conclusions
The necessity to control the territory affected by catastrophic events, such as earthquakes is now back in the limelight. Control facilitates objective assessment of the real degree of vulnerability of the territory and the civil engineering works within it to take up the correct mitigation works in case of seismic events. The characteristics of innovative GPRs are particularly suitable in the field of civil protection activities, for example, in the management of post-seismic emergencies and geohazards, because the GPR not only ensures detailed and constant control but can also help law enforcement during forensic investigations.
The potential level of GPR sensitivity is very high. As illustrated here, even millimetric fractures can be detected by this non-destructive method. By increasing the frequency of the pulses, it is possible to have higher resolution and hidden, minute-even sub-millimetric-cracks can be detected in any type of building, whether modern or ancient [34]. The approach presented here should extensively contribute to the development of specific and ad-hoc classification systems of the vulnerability of structures to earthquakes of various seismic power, within a general seismic classification [30,49]. Note that, although the cracks illustrated in Figure 5 are certainly post-2016 earthquake and Rome suffered similar damage after this seismic event [47,48], the other injuries highlighted by the GPR surveys are not due to this shaking alone.
The use of geoscientific solutions such as GPR for structural monitoring of delicate buildings, infrastructure, etc. in a post-earthquake context, is the basis of any forensic preventive action aimed at improving the control of the territory, geohazards, and its resilience to catastrophic events, and preventing collateral damage to property and people. Non-destructive methods, such as the one used in this work, have identified every single element of the damage suffered by or the cracking in the investigated structures. Thanks to a targeted and timely intervention, the critical points can be flagged in real time.

Conclusions
The necessity to control the territory affected by catastrophic events, such as earthquakes is now back in the limelight. Control facilitates objective assessment of the real degree of vulnerability of the territory and the civil engineering works within it to take up the correct mitigation works in case of seismic events. The characteristics of innovative GPRs are particularly suitable in the field of civil protection activities, for example, in the management of post-seismic emergencies and geohazards, because the GPR not only ensures detailed and constant control but can also help law enforcement during forensic investigations.
The potential level of GPR sensitivity is very high. As illustrated here, even millimetric fractures can be detected by this non-destructive method. By increasing the frequency of the pulses, it is possible to have higher resolution and hidden, minute-even sub-millimetric-cracks can be detected in any type of building, whether modern or ancient [34]. The approach presented here should extensively