Keywords

1 Introduction

Landslides are caused by disturbances in the natural stability of a slope. They can be activated by heavy rains or follow other natural disasters such as earthquakes or volcanic eruptions.

Landslides are generally classified (Cruden 1991) by type of movement (slides, flows, spreads, topples, or falls) and type of material (rock, debris, or earth).

Sometimes, more than one type of movement occurs within a single landslide, and since the temporal and spatial relationships of these movements are often complex, their analysis often requires a detailed interpretation of both landforms and geological sections or cores (Meng 2023).

Landslides pose a recurrent hazard to human life and livelihood in most parts of the world, especially in some regions that have experienced rapid population and economic growth.

Roughly 4500 people are killed worldwide yearly by landslides (Froude and Petley 2018), and related risk is expected to increase due to climate change and urbanization (Ozturk et al. 2022).

Italy is one of the countries most affected by mass movement, often involving urban areas (Salvati et al. 2010; Haque et al. 2016). According to the most recent mosaics, more than 8% of the Italian peninsula lies at high landslide hazard, with more than one million people and more than five hundred thousand buildings at significant risk (Trigila et al. 2021).

Hazards are mitigated mainly by non-structural measures, by restricting or even removing populations from areas with landslides, by restricting certain types of land use, and by installing early warning systems based on the monitoring of ground conditions such as strain in rocks and soils, slope displacement, and groundwater levels.

Risk mitigation strategies also include structural measures, acting directly on the hazard (Dai et al. 2002), such as internal slope reinforcement, anchors, barriers, slope reshaping, plumbing, and drainage (see, e.g., Genevois et al. 2022), but the implementation of such structural solutions is not always feasible mainly due to cost, environmental impact, and long-term maintenance needs (Anderson et al. 2022).

Moreover, as came out from the Kyoto 2020 Commitment for Global Promotion of Understanding and Reducing Landslide Disaster Risk (KLC2020), a comprehensive approach to landslide risk mitigation should prioritize actions aiming at improving the technologies for monitoring slope stability and supporting early warning. Thus, landslide monitoring became an important prerequisite for proactive landslide risk management, contributing to increased safety and resilience in landslide-prone areas (Huntley et al. 2023).

In order to mitigate landslide-related risk, several landslide-monitoring systems have been developed. Such significant growth was possible due to the development of landslide monitoring technologies (Eberhardt 2012; Klimeš et al. 2017) over the last decades, including both remote sensing (Kimura and Yamaguchi 2000; Casagli et al. 2023) and ground-based solutions (Tarchia et al. 2003; Maheshwari and Bhowmik 2021).

They involve different technologies and processing methods according to landslide type and spatial and temporal scale of analysis. In general, nanotechnology has led to the development of smaller, cheaper, more reliable, and more functional borehole sensors that, together with wireless data acquisition and transmission, have significantly increased the temporal resolution of soil deformation.

Due to their complex kinematics and extension over large and low-accessible areas, landslides often call for the integration of different methods and techniques in order to measure surface and deep modification (Nikolakopoulos et al. 2017). Instruments can include total stations (Artese and Perrelli 2018; Stiros et al. 2004; Tsaia et al. 2012), photogrammetry (Scaioni et al. 2015), laser scanners (Kasperski et al. 2010; Mallet and Bretar 2009), Global Navigation Satellite System (GNSS) (Josep et al. 2000).

However, despite significant progress in Earth observation, satellite-based techniques are mainly used during pre-investigation phases, providing an overview of slope stability issues in the area of interest. On the contrary, ground-based and on-site systems usually provide details for local investigations (Pecoraro et al. 2019).

In particular, on-site monitoring systems involve using sensors and instruments directly within or near the area prone to landslides (see, e.g. Wang et al. 2022) to provide an understanding of slope behavior over time and typical responses to external factors, such as rainfall.

Some common components of on-site monitoring systems include geotechnical and hydrogeological devices such as tiltmeters, inclinometers, piezometers, rainfall gauges, and seismic sensors, together with GPS receivers (Auflič et al. 2023).

Key requirements for a well-designed monitoring system are cost effectiveness, robustness, flexibility, scalability, and suitability for long-term and real-time monitoring (Lau et al. 2023). Recent advances in landslide monitoring include emerging technologies such as the Internet of Things (IoT), allowing interoperability and intelligent communication with services and applications in the cloud using Internet standards (Thirugnanam et al. 2022).

By analyzing collected data from these instruments, often involving low-cost sensors (Glabsch et al. 2009; Artese et al. 2015), on-site monitoring can provide real-time information about changing conditions, allowing effective landslide early warning (Xu et al. 2020).

This information is essential for implementing an evacuation plan and making informed decisions to reduce the potential impact of the hazardous event (Strouth and McDougall 2022).

Within the above framework, this paper describes the almost fully automatic monitoring system in the municipality of Gimigliano, in the Calabria region, South of Italy, as an example of a multi-parametric structure for early warning and risk management purposes.

For monitoring and enforcing the understanding of the mechanism of large-scale landslides, we designed a system that can operate autonomously and in near real-time mode, an essential requirement for risk mitigation procedures. The monitoring system is composed of clusters including sensors for controlling deep movements and pore water pressure, as well as specific strain gauge sensors for the control of urban structures—buildings, walls—and a network of local sensors for topographic surface monitoring using a global positioning system.

After a detailed description of the involved area, the sections provide the analysis of the main components of the monitoring system and the devices adopted for early-communication, then some results of the recent activity focusing on movements recorded in 2022 and 2023.

2 Study Area: Geographical and Geological Setting

The Municipality of Gimigliano covers an area of 33 square kilometers and is located in Calabria, Southern Italy. The site is located within the structural geological framework of northern Calabria and, more precisely, of Sila Piccola. This sector is made up of the metamorphic units of the Alpine chain, which overlay the Apennine carbonates. The alpine and Apennine thrust sheets structures of the Sila Piccola result offset by regional high-angle faults starting from the Neogene-Quaternary time. The occurrence of tectonic deformation phases affecting the study area contributed to the alteration of rocks and the geological predisposition to landslide. The entire territory of the town of Gimigliano consists of a complex and articulated geological setting, both stratigraphically and structurally. The entire structure of the Gimigliano area is made up of overlain Paleozoic crystalline rocks overthrust on Mesozoic metamorphic ophiolithiferous rocks (Chidichimo et al. 2023). From bottom to top, the lithologies out-cropping in the Gimigliano area are composed of metamorphic rocks such as phyllites, schists, and quartzites belonging to the deepest tectonic unit. These rocks are overthrusted by a succession composed of serpentinites and greenschists with a cover of metalimestones, marbles, metarenites, and metapelites belonging to the upper Mesozoic ophiolitic nappe. The uppermost thrust sheet is made up of schists, porphyroids, ortho, and paragneiss of the Paleozoic-derived nappes.

The complex geological framework is affected by brittle deformation with the development of open antiformal folding, extensional low-angle faults (Muto and Perri 2002; Mattei et al. 2002; Rossetti et al. 2001), high-angle tectonic structures such as regional and local strike-slip fault systems cross-cutting, and displaying the metamorphic rocks (Tansi et al. 2007; Brutto et al. 2016; Tripodi et al. 2018). The main landslide is 1.100 m in length and about 600 m wide. Along the contact between the lower and the upper ophiolitic rocks, at a depth of 46–60 m, has been recognized the sliding surfaces of gravitational origin of the main active landslide (Ausilio and Zimmaro 2017; Chidichimo et al. 2023) named GIM 2 in this study (Fig. 1). The Gimigliano area is also affected by several landslide phenomena, characterized by medium (GIM 1 in Fig. 1) to shallow depth with different state of activity, areal distributions, and magnitude (Fig. 1).

Fig. 1
A map of Southern Italy with the location and landslide distribution of Gimigliano. The Quiscent landslides are located in the northeastern section of the map. The rest is under active landslides. The cluster zones 3 and 4 are located centrally and northwestwards from the center, respectively.

Location and landslide distribution of the study area

Full size image

Mass movements are mainly triggered by intense precipitation, usually in single daily rainfall events, which strongly impacts the weathering and degradation of the rocks. The presence of large and medium-sized landslides, as well as movements considered superficial, is evidenced by the continuous deformation of all the infrastructures and buildings studied and surveyed.

Among all the instability phenomena that affect the territory of Gimigliano, the landslides that affect the most recently urbanized area deserve the most attention. They can be divided into two main bodies (GIM1 and GIM2 in Fig. 1) and a series of minor bodies characterized by predominantly sliding kinematics and different activity states.

Previous studies and investigations indicate the Gimigliano landslide as a deep-seated landslide phenomenon with a well-identified sliding surface around the depth of 60 m in the central zone of the landslide body.

The movement appears active, with velocities ranging from slow to very slow (1 cm/year to 5 cm/year). Other landslides in the area of the main scarp cover the main body. These are composite and complex landslide types with depths not exceeding 15 m. Frequent superficial and fast phenomena are triggered as debris flows and avalanches from the crowning-scarp zone.

Previous studies described the highly complex and articulated geological-structural framework of the Gimigliano area (see, e.g., Van Dijk et al. 2000; Tansi et al. 2007; Bianchini et al. 2013; Brutto et al. 2016; Gullà et al. 2021).

The area of the main landslide (GIM 2 in Fig. 1) is also characterized by a complex hydrogeological pattern, which has been the subject of studies and numerical modelling (Chidichimo et al. 2023). Groundwater circulation is strongly influenced by the permeability of the different lithological units, phyllites, serpentinites, greenschists, marbles, and their different degrees of alteration and fracturing. Piezometric measurements and the presence of numerous water springs have provided useful information on the interstitial pressure regime (Chidichimo et al. 2023). The combination of deep and shallow phenomena, whose effects add up in time and space, exposes the entire Gimigliano slope to a very high landslide risk.

In selecting the areas to be monitored, priority was given to areas with high population density, presence of strategic buildings, and communication routes, while areas affected by structural measures to reduce the risk and areas with manual monitoring stations (fixed inclinometers and piezometers) were excluded.

However, in order to monitor and understand the ongoing phenomena, we planned a system capable of operating autonomously and in near real-time mode, a prerequisite for risk mitigation.

The proposed monitoring network is composed of the following elements: cluster stations for the monitoring of deep movements and piezometric levels, as well as for the control of urban structures through specific sensors, and a network of sensors for topographic surface monitoring. They will be discussed in detail in the following sections.

3 Description of the Monitoring Systems

3.1 Deep Movements and Piezometric Levels Monitoring

The monitoring activity of the municipality of Gimigliano was planned to assess the effect of slope movements on different parts of the area of interest, focusing on the definition of underground displacements and consequences for buildings and infrastructure.

These requirements made it necessary to design a system that integrates sensors based on different technologies to provide a comprehensive description of the observed phenomenon.

Following this approach, an automatic monitoring system based on MUMS (Modular Underground Monitoring System) technology was designed and installed on-site.

MUMS is an innovative monitoring system developed and patented by ASE S.r.l. (IT), composed of a series of synthetic resin nodes, named Links, connected by an aramid fibre cable and a single quadrupole electrical cable to form an arbitrarily long array of sensors (Segalini et al. 2014).

MUMS devices are customizable for the sensor’s number, distance, and typology, thus obtaining a multi-parametric structure able to measure different physical quantities depending on the specific case study requirements.

The monitoring process is completely automated, using Internet of Things (IoT) technologies to enhance communication and interaction among system components.

Through this approach, it becomes feasible to achieve high sampling frequencies and gather a significant amount of information about the monitored element.

Moreover, integrating algorithms to identify potentially critical events allows for implementing this system for early warning and risk management. The result is a synergic procedure incorporating data acquisition, processing, storage, and representation (Carri et al. 2021).

For the case study discussed here, the monitoring system was designed to include both traditional instrumentation and innovative MUMS-based devices. Specifically, several crack meters and tilt meters were installed on selected structures of the area (i.e., buildings and retaining walls) in order to detect their interaction with potential slope movements. Moreover, two In Place Array were installed on site, with Links placed at specific depths defined according to previously available data.

Array DT0174 is 70 m long, while the instrumented part of the device is located between 45 and 65 m in depth with a node distance of 2 m. These devices will allow the detection of local and cumulative displacements with a near-real-time approach thanks to the automated process for data acquisition and transmission. Additionally, the Array integrates four piezometers installed at 10, 25, 42, and 67 m of depth in order to measure the water level variation over time.

Array DT0176 is 30 m long and presents a variable distance between Links depending on their position: one every 2 m from 2 to 12 m of depth and one every 3.50 m from 12 to 26 m. It also includes two piezometers located 11 and 28 m in depth.

A comprehensive overview of the position of each device installed in the area of interest is given in Fig. 2.

Fig. 2
A plan of the study area with the monitoring system installed. From left to right the buildings are numbered. The buildings are as follows. Old town hall, private building, town hall, Carabinieri station, Primary School, and Unified School. There are retaining walls laid around the structures.

Monitoring system installed in the study area

Full size image

Due to the system complexity and extension of the area, the site of interest was divided into two main Clusters and five zones to provide a clearer identification of the position of each device, allowing more efficient planning for the installation of data loggers.

Cluster 1 contained the western area of interest, including the current and the Old Town Hall, a private building, the local Police station, and two retaining walls, and it is divided into three zones.

Cluster 2 is located on the eastern side of the municipality and encompasses two schools and three retaining walls, and it is divided into two more zones. Tables 1 and 2 provide additional details regarding the instrumentation features, referring to Cluster 1 and 2, respectively.

Table 1 Monitoring instrumentation installed in Cluster 1
Full size table
Table 2 Monitoring instrumentation installed in Cluster 2
Full size table

3.2 Geodetic Monitoring

The geodetic monitoring of the surface is based on the analysis of the variations over time of the angular and distance measurements of points (inside and/or near the landslide) with respect to reference points.

As a result, the graphical and numerical restitution of the geometric characteristics and the morphological variations are obtained, thus contributing to studying the evolutionary trend of the deep gravitational phenomenon. Since the Gimigliano area is already involved in geodetic monitoring, steps were taken to verify the existing network’s functioning and then define its integration. Specifically, a Leica 1201+ precision robotic station was used for the survey activities. Targets were placed throughout the survey area, almost all anchored to buildings in the city center.

The lack of a stable point within the measuring range of the instrumentation and the visibility problems due to the topography of the area did not allow the installation of a single permanent robotic station that would allow the measurement of the entire area affected by the phenomenon under study.

Two stations were then installed, and periodic monitoring was carried out. The station points were selected to guarantee their stability and to maximize the visibility of the area affected by the landslide and the number of points visible by both stations. Station 1 is situated on the terrace of the rectory church, located in the historic center of the town.

The monumentation of the station was achieved by fixing a tripod for surveying instruments directly on the terrace floor. Station 2 is situated on a terrain in front of the historical centre and opposite to station 1, characterized by rocky outcrops. On this last point, an iron pillar was built with the connection base of the total station at the top.

Three stable orientations were chosen for referencing the monitoring network, visible from the two station points. Each orientation was materialized using two long-range corner reflectors close to each other and oriented to the two station points.

The first corresponds to a rock protruding under a metal cross about 8 m high on Mount Gimigliano, which dominates the historic center. The second orientation points are placed on the north wall of an abandoned building in the valley below the town.

The last orientation points are positioned on the north wall of an isolated house in front of Gimigliano, about 200 m from the Soluri hamlet, on the border between Gimigliano and the near Tiriolo town. These points were used for orientation operations and setting the instrument’s reference system (Fig. 3).

Fig. 3
A satellite map of the locations of the stations and orientation points at the study area. Two of the orientation points are located in the south west section with one at the north east section. The stations are located in the north and south, each.

Location of Station (red dots) and Orientation Points (yellow dots)

Full size image

Figure 4 shows the layout of monitored points. Red vectors are acquired from station 1, while blue vectors are obtained from station 2.

Fig. 4
A map of Gimigliano. Multiple points extend from S 1 on the left towards the right at different points. The points extend from S 2 on the right towards the left section. Each point is numbered and the link is marked by lines. There is a vector measured that directly links S 1 and S 2.

Monitored points. Red arrows are the vectors measured from S1; blue arrows are the vectors measured from S2

Full size image

A few double points are measured from both stations: this allows us to perform some verifying and to link further the surveyings carried out. Figure 5 shows the orientation points. Since the total station is periodically placed and the stations are placed near the landslide, these points are used to recalculate the station position and to compensate for possible displacements. Furthermore, the possibility to measure the three points from both stations, along with the reciprocal monitoring of the station points and the GNSS surveying of their position, allows to link the independent measurements and to have a common referencing of the observations.

Fig. 5
A map of Gimigliano with location of station and orientation points. The vectors extend from S 1 to A Croce, S 2, B Valle, and C Casa Critelli. From S 2, the vectors extend towards the same points as S 1 except when it goes straight to S 1 on the same line as S 1.

Location of Station and Orientation Points. Red arrows are the vectors measured from S1; blue arrows are the vectors measured from S2

Full size image

4 Results and Discussion

4.1 Deep Movements

The availability of multiple devices in the same area makes it possible to detect the occurrence of anomalous events and to verify whether they affect different areas of the monitored site.

In the example discussed here, it is possible to observe the data sampled by tilt meters DT0043 and DT0044, installed respectively on a small wall connected to the old town hall and the building itself. Note that due to their relative position, the X-axis of the first sensor is aligned with the Y-axis of the second, and vice versa (Fig. 6a, b). The dataset of interest reports the tilt variation measured along both instrumental axes for the two sensors. It refers to an event that occurred on May 16, 2022. As shown in Fig. 6c, both instruments recorded a simultaneous variation of 0.18° along the X-axis for DT0043 and the Y-axis for DT0044. The similarity in the magnitude of the tilt variation recorded along two aligned axes is a good indication of the event’s actual occurrence. It allows a reliable evaluation of the tilt direction. In addition, it can be noted that only DT0043 measured a tilt variation along the perpendicular axis, indicating that the retaining wall was more affected compared to the monitored building.

Fig. 6
A plan, typology, and a multiline graph. A, plan with location of D T 0043 and D T 0044. They are located at the right side of the Old town hall with D T 0043 at the top and D T 0044 at the bottom. B, sensor typology has 1 x and 2 has y axis vertically. C, graph plots tilt variation over time with D T 0044 y axis the highest.

(a) position of DT0043 and DT0044 (b) Frame of reference and sensor typology (c) tilt variation over time measured by the two tilt meters in May 2022

Full size image

It should also be noted that the displacement data collected by In Place Array DT0174, the closest to the Old Town Hall included in Cluster 1, did not show any significant movement during this time period. This observation, together with the relatively small amount of tilt variation, allowed this event to be categorized as a non-critical event, and no consequences were noted at the monitored site.

DT0046 Analog Array reads two crack meters installed on the retaining wall in Cluster 2 along the main road that crosses the community. Specifically, the first sensor is installed on the right side of the wall, while the second sensor is installed on the left side at a distance of approximately 12 m (Fig. 7a, b). By comparing the measurements collected by the crack meters during the same time, it is possible to prove whether a movement affects the entire structure or a specific section of the wall.

Fig. 7
A plan, typology, and a multiline graph. A, plan with position of crack meters in D T 0046. There are 2 crack meters, 1 and 2, on the retaining wall. B, sensor typology has 1 and 2 placed diagonally at x. C, graph plots deformation versus date. The plotlines are for 1 and 2. The trend is declining with node 2 higher.

(a) position of the crack meters integrated in DT0046 (b) Frame of reference and sensor typology (c) deformation variation over time measured by the two crack meters in June and July 2023

Full size image

The data set considered for this example was collected over 2 months of monitoring activity, June and July 2023, and is shown in Fig. 7c.

From this graph, it is possible to see some events where the sensor behavior diverged, showing a difference in deformation values. One of these events began on June 13 and involved only Node 1, which measured a positive deformation of 0.8 mm, corresponding to a crack opening, while Node 2 showed no trend deviation. After about 10 days, the deformation regressed, and the crack meter returned to its original position.

Another event can be observed in early July, where the deformation measured by Node 1 began to follow a downward trend, eventually reaching a value of −1.8 mm, while Node 2 showed a relatively similar pattern, although much less obvious. On the other hand, the last part of the dataset shows an example of more similar behavior for both sensors, with a slight positive increase in deformation on July 27 before reaching a stable state.

4.2 Surface Movements

A map of the detected horizontal movements of the monitored points is shown in Fig. 8. The speed of surface displacements, based on the surveys carried out, is in the range of 2.5–5 cm/year. For the same site, Fortunato and Ferrucci (2012) estimated movement speeds of up to 24 mm/year from 1993 to 2000 and up to 14 mm/year from 2002 to 2010 from the analysis of Synthetic Aperture Radar (SAR) data.

Fig. 8
A map with horizontal movements of the monitored points. Arrows represent the movements from S 1 and S 2. Each is numbered. More arrows belong to S 1 as compared to S 2.

The horizontal movements of the monitored points. Red arrows are the movements obtained from S1; blue arrows are the movements obtained from S2

Full size image

5 Conclusions

Landslides represent a significant threat to human life, properties, infrastructure, and natural environments for which in-depth analysis and increasingly reliable systems for risk control and mitigation are required. Extensive and in-depth research projects were started mainly on developing different technologies for detecting, predicting, and monitoring landslides.

Many advanced tools provide the opportunity to apply new techniques for the different stages of landslides and susceptibility, hazard, and risk assessment.

This work describes an integrated monitoring system set up for monitoring Gimigliano village, where the risk of landslide is too high and no longer justifiable for the responsible communities.

It integrates a permanent system, including devices for deep movement and piezometric levels, with periodic surface topographic monitoring. In particular, the system includes both traditional and innovative sensors as the MUMS-based devices supported by modern Internet of Things technologies for communication and interaction procedures.

The periodic monitoring over time revealed considerable deformation cycles corresponding to crack openings and regressions. Tilt meters have measured both positive and negative deformations, and surface monitoring activity has provided the estimation of a wide field of speeds of surface displacement.

The presented system offers the advantages of a low-cost, precise, and accurate measurement system for landslide displacements, low power consumption, continuous measurements, and wireless data communication between the measurement grid and a central server. Aggregated data collected from the system makes possible an important interpretation both for defining the structural measures for risk mitigation plan but also for assessing possible highly critical situations outstanding to issue evacuation orders or orders banning traffic on the main access routes into the municipalities.

The ongoing activity of the network contributes to the collection of accurate and punctual information for future research on landslide behavior. The next goals are concerned with improving the system with an impact on the timely alerts of the surrounding communities. In particular, future activities require an enhancement of the existing system with an extension for the areas not currently monitored and a monitoring activity with highly specialized technicians trained not only for managing the monitoring network but also for direct survey control to support the decision-making phases.