Evaluation of a Ground Subsidence Zone in an Urban Area Using Geophysical Methods

1. Introduction

Puglia is one of the Italian regions most affected by the subsidence of the soil due to the subsidence and collapse of natural and anthropic cavities [1,2]. Such events sometimes destroy or threaten the integrity of infrastructures, even of a particular importance, such as roads and railway lines [3]. The repetitiveness of the instability phenomena characterises Salento. This high frequency is linked to the area’s geology and especially to the evolution of karst processes and human activities of exploitation of subsoil resources and settlement in the area. In particular, the city of Gallipoli is located on the Ionian coast of Salento (Figure 1). The eastern part of the town of Gallipoli was affected from the end of the 1800s to the beginning of the 1900s by intense underground mining activity due to calcarenite deposits. These deposits, having good mechanical properties, were used to produce tuff ashlars and as ornamental stones for external surfaces [2,3].

The discovery depth of the calcarenite deposit varies between 2 m and 25 m. The mining activity developed through large halls with randomly arranged square pillars with sides equal to 2 or 3 m (Figure 2). The deposit was reached through vertical wells commonly called bell-shaped due to their shape. At the end of the exploitation, the quarries were abandoned and often used as waste deposits, and the relative accesses were blocked and buried until they lost consciousness and memory. At the end of March 2007, there was an important collapse event of the roadway (Figure 3) [2,3].

The collapse event had as precursors the deformation of the road surface, also indicated by the slippage of a maintenance hole cover placed on the roadway, which then collapsed a few days before the collapse; the subsidence of the street level indicated by cracks detected on the buildings involved in the instability. The sinking was preceded by days of more or less intense and prolonged rain. The collapse, which saw the formation of a chasm 12 m in diameter and 8 m deep, involving two buildings which housed 16 families, was accompanied by the formation of fractures in the vaults with the detachment of large blocks and suspended boulders. For this reason, geophysical investigations were necessary for the area affected by the collapse to identify any other danger zones and to identify safe routes for the transit of mechanical means to carry out the work to make the unsafe buildings safe and prevent new collapses. Given the urgency of this first intervention, it was decided to use the ground penetrating radar (GPR) methodology.

Twenty-two years after the instability phenomenon in the same area, in December 2022, there was a slight sinking of the road pavement (Figure 4). For this reason, further geophysical investigations were necessary to verify the extent of the phenomenon. In this case, electrical resistivity tomography (ERT) and GPR were used.

International scientific literature reports many applications of geophysics in areas at risk of subsidence [4,5,6,7,8,9,10,11,12,13,14]. The risk was often linked to natural cavities formed over the years in karst environments. In the case of this paper, the cavities are known because they are cavities dug by man and abandoned. These cavities, over the years, have become a problem for safety.

The geophysical investigation in the case study made it possible to outline the areas not at risk, facilitating the restoration and safety measures. Fifteen years after the first instability event, the geophysical investigations show that the risk is still present and that the consolidation interventions carried out in 2007 must be redesigned.

2. Materials and Methods

For the first geophysical surveys, those of 2007, the georadar of Mala geoscience with a 250 MHz antenna was used. GPR measurements were performed in areas A, B, C, D and E (Figure 5).

GPR data were acquired in a grid with parallel profiles spaced 0.5 m.

The two-way time window was 100 ns (nanoseconds). The sample interval and the trace interval were, respectively, 512 samples per scan and 0.02m. GPR data were processed using GPR-Slice Software (GPR-SLICE Software (gpr-survey.com)). The 2D data have been processed as the following processing steps: (i) bandpass filter to eliminate both low and high-frequency noise components; (ii) manual adjustment of the gain to try to improve the vision of the deepest reflection events; (iii) background removal filter to eliminate the horizontal components of the signal present on the radar sections and due to ringing; (iv) Kirchhoff migration using an average velocity value of the electromagnetic wave equal to 0.108 m/ns. Electromagnetic (EM) wave velocity can be estimated from GPR data in several ways; the conventional method involves common depth-point (CDP) and wide-angle reflection and refraction (WARR) data sets. Both methods require two antennas in separate units and relatively long acquisition times [9,10]. The EM wave velocity can be more quickly and easily determined from the reflection profiles acquired in continuous mode, using the characteristic hyperbolic shape of reflection from a point source (diffraction hyperbola) [9,10,11,12]. This is a very common method for the EM velocity estimation and based on the phenomenon that a small object (the dimensions of the object are smaller than the wavelength of the EM wave introduced in the ground), reflects EM waves in almost every direction. Figure 6 shows an example of application of the above described method applied to the data acquired in the investigated areas.

The data thus processed were subsequently used for constructed 3D volumes. This allows viewing the data in various ways and subsequently simplifying their interpretation.

For the second geophysical survey of 2022, the georadar Ris Hi-mod with a dual-band antenna of 200-600 MHz was used. The location of GPR profiles is shown in Figure 7. Also, in this case, GPR data were processed using GPR-Slice Software (GPR-SLICE Software (gpr-survey.com)) with the same processing steps.

For the second geophysical survey of 2022, ERT data were acquired using the MAE 600 E georesistivimeter with 32 active channels. In this case, the survey was performed using a particular geometry (L shape) to investigate below the building [13,14,15,16]. The L profile is shown in Figure 7. The dipole-dipole array was considered. It is the most used for cavities detection [13,14,15,16]. The distance between the electrodes was chosen equal to 1 m, and 32 electrodes were used. This type of array is well described in the international literature [14,15,16]. Initially, a 2D survey is conducted along each perpendicular line or transect. In the next step, the current electrodes remain at the end of one line, while the potential is moved, along the line. Then, the current electrodes move one electrode position and the potential electrodes move as previously described. The process is repeated until the current and potential electrodes cover the L geometry. This sequence of observations produces a series of apparent resistivity observations towards and beneath the central portion of the array. The colored circles in Figure 8 represent the attribution points, where the apparent resistivities are measured, for the performed ERT array. This process is discussed in detail by [16].

Electrocardiogram electrodes were used to avoid piercing the floor [17,18,19,20]. Here the contact resistance was low (about 2000-3000 Ω m). This was probably due to the high humidity inside the church.

The ERTLab software (http://www.geostudiastier.it) was used to process the data in 3D mode. The software uses the tetrahedral finite element iterative method (data variance iterative reweighting).

3. Results

3.1. The 2007 GPR Data Analysis

3.1.1. Area A

In the area, A (Figure 5) were acquired, 17 parallel profiles 0.5 m spaced. The processed data (Figure 9) presents several reflection events. It presents a subsoil with backfill material for the first meter of depth. An interesting reflection event is denoted “C” in Figure 9a. It is evident in the first four profiles acquired in area A. The size is about 15 m wide and the depth of the top is between 2.6 and 3.2 m (with an average electromagnetic wave velocity of 0.108 m/ns).

Another interesting reflection event “C” (Figure 9b). The size is about 8 m wide, and the depth of the top, is between 2.4 and 2.8 m. Some considerations can be made to try to understand what C may be due to. Observing them on the radar sections, a change in the polarity of the electromagnetic wave (EM) is noted. An inversion of polarity is produced when the reflection coefficient is negative [10,11]. So, for materials where the wave velocity depends only on dielectric permittivity when radar waves are reflected in a material with a higher dielectric permittivity. A typical case is reflections from air to any other material [10,11]. All these considerations lead to interpreting these anomalies as being due to cavities.

A 3D map of the distribution of the anomalies identified within the 2D radar sections was built. One type of visualization is related to the construction of 2D time slices within specific time intervals [10,11]. In this work, time-slices have been built to visualize the amplitude variations within time intervals Δt = 10.4 ns. The overlay analysis was used [19,20,21,22].

Figure 10 shows the most significant time slices. Here the slices at depths between 2.1 m and 4.7 m are considered. In fact, at these depths, it is possible to highlight the reflected event linked to the C anomalies which correspond to the one indicated with C in the 2D radar section of Figure 9.

Another way to present the 3D data volume is the amplitude iso-surfaces of the EM wave (Leucci, Conyers). Here it is possible to isolate the amplitude values; establishing a minimum threshold value in order to make make the 3D structure of the anomalies identified in the 2D radar sections uniquely visible. The threshold is a very delicate step and depends a lot on the experience of the interpreter to obtain valuable results [10,11].

Figure 11 shows the iso-surfaces amplitude with a threshold value equal to 65% of the total amplitude. Anomalies C can be visualized in their 3D development.

3.1.2. Area B

In the area, B (Figure 5) were acquired 18 parallel profiles 0.5 m spaced. At the end of area B is the collapsed part of the road surface. Here the processed data shows an interesting reflection event “C” in Figure 12. It develops along the entire radar section at a depth between 2.0 m and 4.0 m. An inversion of polarity is noted and therefore this anomaly was probably due to cavities.

Figure 13 shows the most significant time slices. The slices at depths between 2.0 m and 3.1 m are considered. In fact, at these depths, it is possible to highlight the reflected event linked to the anomaly C.

Figure 14 shows the iso-surfaces amplitude with a threshold value equal to 70% of the total amplitude. Anomaly C can be visualized in its 3D development.

3.1.3. Area C

In the area, C (Figure 5) were acquired 16 parallel profiles 0.5 m spaced. The GPR data analysis did not reveal any anomalies attributable to the presence of cavities (Figure 15).

3.1.4. Area D

In the area, D (Figure 5) were acquired 8 parallel profiles 1.0 m spaced. The processed data shows an interesting reflection event “C” in Figure 16. It develops a depth between 1.6 m and 3.0 m. An inversion of polarity (Figure 17) is noted and therefore this anomaly was probably due to cavities.

Figure 18 shows the most significant time slices. The slices at depths between 1.1 m and 3.3 m are considered. In fact, at these depths, it is possible to highlight the reflected event linked to the anomaly C.

Figure 19 shows the iso-surfaces amplitude with a threshold value equal to 70% of the total amplitude. Anomaly C can be visualized in its 3D development.

3.1.5. Area E

In the area, E (Figure 5) were acquired 9 parallel profiles 0.85 m spaced. The processed data did not reveal any anomalies attributable to the presence of cavities (Figure 20).

3.2. The 2022 GPR Data Analysis

The second subsidence event which occurred 15 years after the first made a second campaign of measures necessary. In this case some GPR profiles were performed to understand the extent of the phenomenon. Therefore the investigations were carried out only in the area affected by the second instability event. From the analysis of the data (Figure 21) it is possible to notice reflected events "C" which confirm the presence of the cavity. In fact here it is possible observe a polarity inversion of the reflected EM wave [10,11,20]

3.3. The 2022 ERT Data Analysis

The ERT investigation was necessary to understand the possible development of the cavity underneath the buildings. For this reason, the acquisition geometry shown in Figure 7 was adopted. The 2D distribution of electrical resistivity as a function of the depth are shown in Figure 22.

From the 2D resistivity distribution (Figure 22) the presence of a heterogeneous subsoil with resistivity values between 2000 Ω m and 10000 Ω m is evident. In particular, is possible to note the presence of areas (indicated with C) with resistivity values between 2500 Ω m and 10000 Ω m which suggests the probable presence of cavities. These cavities seem to extend to a depth of about 4 m.

It is possible to build the ERT depth slices in order to visualize the distribution of the resistivity in a 3D mode below the investigated area within depth intervals. Here all the anomalies with the same resistivity values are interpolate and a 3D volume of resistivity data is built. Successively is possible to extract the 2D resistivity distribution at several depths. Figure 23 shows the most significant time slices.

From the analysis of the depth slices (Figure 23) is possible to see the distribution of the resistivity anomalies C below the investigated area.

Figure 24 shows the superimposition of the ERT results on the plan of the investigated area showing the position of probable cavity C.

Here it is possible to see how the cavity develops dangerously underneath the buildings.

4. Discussion

Apart from the area affected by the collapse which was made safe, the geophysical investigation in the adjacent areas had the aim of understanding the stability conditions of the unexplorable cavities. The serious problem is found in areas where the roof thickness of the cavity was less than 5 m (areas A, B and D). In fact, as shown in Figure 7, Figure 10 and Figure 14, the thickness of the roof varies from 2.4 m (Figure 7), 2.9 m (Figure 9) to 1.9 m Figure 14. In this case, the buildings located near these areas could be in danger. Areas C and E were the most suitable for heavy vehicles to travel through for the first consolidation works. The area affected by the collapse was made safe with a consolidation intervention consisting of filling the cavities with concrete and strengthening of the pillars with the use of spritz - beton (shot concrete) inside which a filling of loose material (stone) was inserted. In Figure 25 the photos of the cavities after the intervention are shown.

In December 2022 (15 years after the first collapse event) another small collapse phenomenon occurred in the same area as the first collapse. There was a small hole with subsidence of the road pavement and cracks on the building (Figure 26a). The sagging of the pavement was also noticed inside the building (the photos are not shown for privacy reasons). For this reason, a geophysical survey was decided to be limited to the area affected by the phenomenon. The GPR results show the presence of the cavity with a thinning of the roof thickness ranging from 0.9 m to 1.3 m (Figure 19). Very interesting are the ERT results that help us understand what happens underneath the building. Here we note the presence of cavities already at a depth of 0.5 m (Figure 20 and Figure 21). The inspection of the cavity shows a probable deterioration of the consolidation intervention (Figure 26b)

5. Conclusions

Immediately after the event at the end of March 2007, geophysical investigations became necessary to understand the development of the phenomenon and indicate the safest routes for rescuers to take for the first interventions. In this case, the geophysical survey using the GPR method showed the development of the cavity and the safest ways to go for the first interventions. Subsequently, to make the cavity safe, a series of reinforced concrete pillars were built inside which a filling of loose material (stone) was inserted. Over the years this material (due to the abundant rains) has probably disintegrated causing the second instability event in December 2022. In this case, the integrated geophysical surveys ERT and GPR have highlighted the presence of voids where the fill had been made. Such voids are also present under the houses.