Integration of persistent scatterer interferometry and ground data for landslide monitoring: the Pianello landslide (Bovino, Southern Italy)
- 413 Downloads
We present an example of integration of persistent scatterer interferometry (PSI) and in situ measurements over a landslide in the Bovino hilltop town, in Southern Italy. First, a wide-area analysis of PSI data, derived from legacy ERS and ENVISAT SAR image time series, highlighted the presence of ongoing surface displacements over the known limits of the Pianello landslide, located at the outskirts of the Bovino municipality, in the periods 1995–1999 and 2003–2008, respectively. This prompted local authorities to install borehole inclinometers on suitable locations. Ground data collected by these sensors during the following years were then compared and integrated with more recent PSI data from a series of Sentinel-1 images, acquired from March 2014 to October 2016. The integration allows sketching a consistent qualitative model of the landslide spatial and subsurface structure, leading to a coherent interpretation of remotely sensed and ground measurements. The results were possible thanks to the synergistic operation of local authorities and remote sensing specialists, and could represent an example for best practices in environmental management and protection at the regional scale.
KeywordsPersistent scatterer interferometry Integrated landslide monitoring Inclinometric borehole measurements
The Daunian Apennines in Apulia (Southern Italy) are intensely affected by land instability, mainly related to very shallow soil slips and to relatively deeper (> 20 m) landslides (Ciaranfi et al. 2011; Cotecchia et al. 2015; Cotecchia et al. 2016, among others). These latter represent the most important geological risk factor for the area, as they frequently affect the outskirts of towns and infrastructures (national railways, highways, and other strategic communication routes). The abundance of landslides in the Daunian Apennines is strictly related to the prevailing fine-grained material (silts and clays) forming the flyschoid units cropping out in the area, and to the generally low relief energy of the slopes (10°–15°) (Spalluto et al. 2015 and reference therein). Daunian landslides are usually described as having a complex kinematic behavior, where significant internal distortion of the material occurs. Nevertheless, the dominant movement takes place along shear surfaces bounding the moving mass (e.g., Ciaranfi et al. 2011). This kind of landslide typically starts as a roto-translational slide evolving, after the post-failure stage, into a flow having slow velocity (≲ 1.6 m/year; Cruden and Varnes 1996), occasionally reaching moderate velocities (~ 13 m/month) during main re-activations (e.g., Parise et al. 2012; Cotecchia et al. 2015). The rate of displacements follows seasonal trends, being lowest in summer and highest in late winter/early spring, when very high piezometric heads are recorded down to large depths (Cotecchia et al. 2014).
The ground monitoring of land instability in urbanized areas is a complex, costly, and time-consuming activity. It requires the realization of a network of monitoring stations (e.g., boreholes equipped with inclinometers and piezometers) aimed at measuring several critical parameters, and the work in the field of a highly qualified staff of technicians for collecting the data and for checking the correct functioning of the installed instrumentation. A crucial role for the monitoring of the landslide activity is played by slope inclinometers, installed in boreholes drilled in different sectors of the landslide body. They measure the subsurface horizontal displacement of the landslide mass along the shear surface and are used to determine the magnitude, rate, direction, depth, and type of landslide movement. This information, used in synergy with other surface and subsurface monitoring data, can give important clues for understanding the cause and behavior, and thus devise remediation of a landslide (Stark and Choi 2008).
Recently, remote sensing has come forward as an effective additional tool in the monitoring of landslide terrain displacements, thanks to the synoptic view (large area coverage) and relatively low costs of remotely sensed images (Wasowski and Bovenga 2014). While optical sensors remain the most used kind of data to identify landslide areas, synthetic aperture radar (SAR) sensors, and particularly multi-temporal techniques based on the identification of stable scatterers, known as multi-temporal (SAR) interferometry (MTI), has demonstrated to be a valuable aid in monitoring hazard areas undergoing displacements (Tofani et al. 2013b; Di Martire et al. 2016b).
The joint use of remote sensing MTI and in situ data depends on the characteristics of the displacement phenomenon, on the movement components measured by the in situ instruments, and on the projection of the ground movement along the space-borne SAR line of sight (LOS). Examples aimed at the monitoring of unstable slopes date back more than a decade (see, e.g., Farina et al. 2006; Singhroy 2009; Tofani et al. 2013a), and new experiments are constantly being reported (Del Soldato et al. 2018; Frattini et al. 2018). Examples of synergy between environmental authorities and remote sensing analysts to monitor terrain instabilities are increasing at national (Di Martire et al. 2017; Rosi et al. 2018) and European level (Wilde et al. 2018).
Mostly, ground data are compared with remotely sensed data, for validation purposes. This is the typical situation when both can be assumed to sense the same process, e.g., with a known displacement direction, as in the case of leveling measurements applied to subsidence (Ketelaar 2009) or uplift (Refice et al. 2016), or GPS applied to seismic deformations (Casu et al. 2006). For the specific application to landslides, comparison of ground and remote measurements is possible when processes consist, e.g., of earth or debris flows in the maximum slope direction, which occurs on shallow landslides (Calò et al. 2014; Bovenga et al. 2017). In such cases, scalar measurements can be back-projected onto the assumed direction of motion, and both can be quantitatively compared. More generally, instead, no direct comparison is possible, but integration among independent ground- and satellite-derived measurements can be attempted; experiments in this sense have been presented for wide-area surveys (e.g., Casagli et al. 2016; Calvello et al. 2017) and more local analyses of deep-seated landslides (e.g., Bianchini et al. 2013; Di Martire et al. 2016a).
While comparisons between MTI and in situ data were more frequent in early validation studies, nowadays, as MTI techniques are considered as well assessed, a more interesting use of MTI is as a complementing integration to the in situ measurements.
In this work, we present an example of integrated analysis at a local scale. The present work should be regarded as an example of virtuous synergy/feedback between remote monitoring and in situ inspection. Original knowledge of unstable areas in the wide region of Daunia, Apulia (Southern Italy) led local authorities to team with remote sensing specialists to investigate ongoing displacements through MTI analysis of SAR data from the European ERS and ENVISAT satellites. Results of this collaboration highlighted active displacements affecting critical areas (inhabited, with valuable/historic infrastructures, etc.) over which it was important to focus ground investigations. In the case of the small hilltop town of Bovino, these consisted in the installment and continuous monitoring of boreholes equipped with inclinometers and other instrumentation. Such ground data, mostly sensitive to the horizontal components of terrain displacement, were finally coupled to more recent MTI results (more sensitive to subvertical displacement components) from data acquired by the new European Sentinel-1 constellation, to give a consistent, qualitative model of the full 3D displacements of a local landslide affecting a versant of the southern outskirts of the town.
In the following, after a description of the geological setting of the test site, we describe the used remote sensing and ground instrumentation data and methods. Then, we describe the results of the remote sensing and ground measurements, and, in the “Integrated analysis” section, we investigate in more detail the complementarity between inclinometer deep displacement data and the surface displacements of PS objects close to each of the boreholes. This allows to sketch a qualitative but consistent scheme of the landslide structure and process in the “Discussion” section. Finally, we derive some conclusions.
The allochthonous units of the Apennines chain are thrusted over the terrigenous autochthonous succession of the Bradanic trough foredeep basin, which rests in turn on the Meso-Cenozoic carbonate units of the Apulian foreland.
The highly heterogeneous lithology of the clayey member of the Flysch di Faeto and the deep tectonic deformation seem to be the main driving factors of the widespread landslide activity in the area. Actually, active and dormant landslides mapped in the area mostly formed in correspondence of silty deposits and, subordinately, in the calcareous ones (Fig. 2). According to the Italian Landslide Inventory Project (AAVV 2007), active and dormant landslides of the Daunian Apennines can be classified as flows, as roto-translational slides, and as landslides having a complex kinematic behavior. Many of these landslides, in the last decades, underwent reactivation processes (Parise et al. 2012; Spalluto et al. 2015) producing severe damage to structures and infrastructures in the Daunian Apennines.
The Pianello landslide is set up onto the fold-bent fault (i.e., ramp anticline) formed over the thrust surface crossing the southern border of the old town of Bovino (Fig. 2). This landslide affects only the clayey-silty deposits of the clayey member of the Flysch di Faeto fm and forms a roto-translational body, almost 740 m long and 250 m wide, with toe emerging along the left-side valley of the Biletra River (Cotecchia et al. 2016). The southern outskirt of the town of Bovino extends over the depletion area of the Pianello landslide, where several buildings show structural damage mostly linked to the recent activity of the landslide (Palmisano 2016; Palmisano et al. 2016; Cotecchia et al. 2016).
Data and methods
Legacy SAR data MTI analysis
To understand the extent of occurrence of recent displacements, and thus the activation state of several landslides in the Daunia Apennines, a regional-scale MTI monitoring activity was programmed and carried out in 2008, exploiting available SAR databases (Alemanno et al. 2009).
A dataset of 42 ERS-1/2 images, acquired in descending geometry, i.e., with the satellite trajectory headed at an azimuthal angle of about 188°, covering the interval 1995–1999, and 24 ENVISAT images, acquired in ascending geometry, i.e., with the satellite trajectory oriented at a ~ 352° azimuthal angle, spanning the interval May 2003–June 2008, was processed through the Stable Point Interferometry over Un-urbanized Areas (SPINUA) PSI processing chain (Bovenga et al. 2005). Both satellites observed the Earth surface from about 800-km height to the sensor’s right side along a LOS looking down at an incidence angle of about 23° from the vertical, with a nominal ground resolution of about 20 × 5 m2 (range × azimuth). The two SAR data time series were processed to extract displacement information over persistent scatterers (PS), i.e., objects on the Earth surface with stable scattering characteristics (Ferretti et al. 2001).
The regional analysis of legacy SAR time series highlighted several cases of ongoing displacements over parts of known landslide bodies in the Apulia region. This prompted the Apulia River Basin Authority to carry out a ground survey. In particular, a subsurface monitoring network was installed on the active body of the Pianello landslide and in surrounding areas. From each borehole, a detailed lithostratigraphic log was reconstructed and undisturbed samples for geotechnical laboratory tests were collected. The boreholes were equipped with either inclinometer casings or piezometric cells. Here, we focus on the results of the inclinometric survey in order to quantify the subsurface displacements of the landslide body, in an area where the detected surface PS displacements could be associated to several superficial evidences of ground deformation, as well as damage on structures and infrastructures mostly recorded in the Pianello neighborhood of the Bovino town.
The readings of inclinometric data cover a period of 8 years, and were performed in two different, partly overlapping, monitoring surveys, using two different biaxial probes. The first survey, accomplished by technicians of the Foggia Province, started in December 2008 and ended in November 2015. The second one, accomplished by technicians of the Apulia River Basin Authority, started in July 2015 and ended in September 2016, with an average reading frequency of about one per month. The instrumentation used on each borehole in the first campaign (Foggia Province) consisted of a SISGEO probe and an Archimede Datalogger readout unit, while for the second campaign (Apulia River Basin Authority), it consisted of an RST Digital Inclinometer System and an HP iPAQ hx2410 Pocket PC readout unit. Both probes contain two perpendicular accelerometers, so only two passes of the probe are required to measure movements in the two different directions: one accelerometer measures the tilt in the plane of the inclinometer wheels, tracking the longitudinal grove of the casing, while the other one measures the tilt in the plane perpendicular to the wheels. In both surveys, the measurements were taken starting at the bottom of the inclinometer. Subsequent readings were made as the probes were raised incrementally, in 1-m (Foggia Province) and in 0.5-m intervals (Apulia River Basin Authority), respectively, up to the top of the casing. Since different probes were used, two different zero readings were obtained at the beginning of the two surveys. The difference between the zero and the subsequent readings was used to determine the change in the shape and position of the initially vertical casing. Comparison of the verticality of the casing with time and the width of the slide provides an insight to the magnitude, rate, direction, depth, and type of landslide movement (Stark and Choi 2008).
Sentinel-1 MTI analysis
Following the ground survey, more recent SAR data, acquired by the Sentinel-1 sensor, were processed, with the aim of integrating the in situ information. The Sentinel-1 constellation is currently composed of two twin satellites (Sentinel-1/A and Sentinel-1/B, respectively), the first one active from April 2014 and the second from April 2016, and observes the Earth from about 693-km altitude, at a nominal ground resolution of about 5 × 20 m2 (range × azimuth), thus comparable to that of the legacy sensors. Forty-one Sentinel-1/A images, acquired in ascending geometry at an incident angle of 34°, were processed, through the same SPINUA PSI processing chain used for the legacy ERS and ENVISAT data, spanning the period from October 2014 to March 2016.
SAR MTI results
On the landslide area, the strongest PS motions are recorded close to the detachment area at N, and at the southern limit of the urban area, roughly in the middle of the landslide area. Several PS are located close to the locations of the installed inclinometers.
Borehole inclinometer data results
Figure 4 shows the location of the continuous boreholes drilled for the inclinometric survey. Inclinometers are marked by colored triangles, green in case of no relevant displacement, orange in case of relevant shear displacement, and red in case of rupture of the casing. The boundary between “no relevant” and “significant” displacements was set considering the total inclinometer accuracy, which is a combination of both random and systematic errors of the system. We set for the used probes an empirically assessed accuracy of about 7 mm per 30 m of casing, similarly to what proposed in the recent literature for analogous probes (e.g., Stark and Choi 2008). Displacements measured by the inclinometers located outside the landslide body in the Pianello locality were always below the system accuracy, and thus, they were classified as no relevant (green triangles in Fig. 4). Displacements measured by the inclinometers located inside the main landslide body were always far above the system accuracy, and thus classified as significant because they recorded active shear deformation (orange and red triangles in Fig. 4). More in detail, data plotted on incremental displacement and cumulative displacement graphs outline clear shear bands at different depths, which evolved in the rupture of the casings for boreholes S6, S9, and S12 (red triangles in Fig. 4). In the following section, an integrated analysis is proposed for displacements recorded by each of the borehole inclinometers, and the closest Sentinel-1 PS.
A first qualitative assessment can be made by looking at Fig. 4, which shows Sentinel-1 PS points represented by dots with color proportional to the LOS displacements according to the colorbar in the figure and inclinometers indicated by colored triangles. A good spatial correlation exists between the displacements measured by the inclinometers and those measured by remote sensing. Outside the limits of the Pianello landslide, both measurement sets do not record significant displacements, while within the landslide limits, both record movements, except for the inclinometer installed in the borehole S10, which did not record any displacement, probably because the borehole depth did not reach the stable ground below the shear surface, as detailed later on.
Anyway, as can be seen from the general direction of the slope on the landslide area, in most cases, the horizontal component of motion (recorded by the inclinometers) is almost perpendicular to the LOS projection on the ground, so that the horizontal projection factor in Eq. 2 is considerably small.
Moreover, inclinometric measurements generally apply to points below the surface, as usually measurements close to the borehole surface end are affected by errors due to instrument handling procedures. Thus, in case of complex (not shallow) landslides, this would further complicate a hypothetical direct comparison between ground and satellite-based displacement measurements, which, instead, provide useful complementary information.
This is the case of the present work, where the horizontal components of the displacements, coming from the inclinometers, and the subvertical ones coming from the PS, are well suited for an integrated analysis.
In the following subsections, we analyze the signals from the most relevant inclinometers individually, integrating, wherever possible, the information about subsurface displacements with that about displacements of nearby PS on the surface, in the sense explained above.
The original total depth of the inclinometer casing of the borehole S6 was about 20 m, as confirmed by the zero reading performed at the beginning of the first survey on 09 December 2008. However, the first reading, performed after a long time from the base one, on 28 September 2010, was not able to reach the bottom of the casing, and stopped at the measured depth of about 16 m. The zero reading of the second survey, performed on 28 July 2015, confirmed the results of the first survey, since the reading of the probe stopped at the depth of about 16.5 m. Moreover, the reading carried out on 26 August 2015, and a borehole video camera investigation, outlined a clear rupture of the inclinometer casing due to a shear surface located at this depth. The rupture of the casing occurred in the lithology identified as yellow sandy-clayey silt with scattered gravels (Fig. 7).
The original total depth of the inclinometer casing of borehole S7 was about 28 m, as confirmed by the zero reading performed at the beginning of the first survey on 09 December 2008. The first inclinometric survey recorded 10 readings, after the base one, from 28 September 2010 to 12 November 2015. In this time range, a shear band was detected at a depth between 4 and 8 m, as shown in the resultant incremental displacement (Fig. 9a). The peak of the incremental displacement spike, i.e., the position of the inferred failure plane, is located at the depth of 7 m. The maximum resultant incremental displacement at this depth was about 48 mm, as seen in the reading performed on 12 November 2015 (Fig. 9a).
The greatest horizontal resultant cumulative displacement was measured at the depth of 5 m, where it reached about 105 mm (Fig. 9b). The lateral cumulative displacement showed that the first 2 m was neglected because it was related to the drift of the casing head in the uppermost part of the borehole.
The shear band is located within the lithology identified as yellow clayey-sandy silt (Fig. 9a, b). The direction of the displacement is 158°, about SSE.
Despite the fact that the second monitoring survey was carried out with a different probe and with a smaller sampling interval, the results of the two campaigns are similar. As a matter of fact, a shear zone was detected at similar depths, and the largest spike of incremental displacement was located at the depth of 6.5 m (Fig. 9c). The maximum resultant incremental displacement at this depth was recorded on 16 September 2016 and corresponds to about 13.2 mm (Fig. 9c).
The cumulative displacement graph shows that the top of the shear zone was located at the boundary between two lithologies: the upper black clayey-sandy silt and the lower yellow clayey-sandy silt. However, the strength loss phenomenon occurred only on the lower lithology. The resultant cumulative displacement obtained from the combination of the two cumulative displacements measured on the two axes at the depth of 5 m is about 43 mm (Fig. 9d).
The time series plots of the displacements recorded by the inclinometers at about 7-m depth (Fig. 8c, d) outline the evolution of the strength loss phenomenon recorded in the shear band over time. The two time series (indicated by black squares and red diamonds, respectively) are derived from independent measurement sets, with different zero readings, and have been connected, just for sake of representation, by assuming similar displacement values at near dates. Periods of acceleration, deceleration, or constant rate displacement are detectable on the first survey time series, but a seasonal evolution cannot be clearly discerned because the number of readings is not sufficient (black squares).
In the time plot for the second survey (red diamonds), the rate of movement at the depth of 6.5 m shows accelerations in the following two time intervals: 24 September 2015–27 October 2015 and 23 February 2016–24 March 2016 (Fig. 8c). During the other detected periods, the rate of movement was almost constant. PS time series for the targets within the moving mass, closest to the borehole, denote steady downward displacements with average LOS velocities of − 25, − 25, and − 13 mm/year, respectively (Fig. 8b). In correspondence with the highlighted periods (evidenced by the two gray-shaded rectangles in Fig. 8b, c), some of the PS time series seem to show some sign of slightly abrupt variations in the displacement rates, although no clearer evidence is available. Peaks of precipitations (blue lines in all the plots, right vertical axis) seem to be also particularly strong in correspondence of the evidenced displacement discontinuities. Slight nonlinearities in PS displacement time series seem to appear in several other plots reported in subsequent figures. This would suggest some interesting time correlations with triggering factors, which however would require more in-depth investigations about the hydrology of the area, and is thus left for future work.
The original total depth of the inclinometer casing of borehole S9 was about 40 m, as confirmed by the zero reading performed at the beginning of the first survey on 9 December 2008. The following readings show a clear shear zone at a depth interval between 16 m and 19 m (Fig. 11a). Moreover, the resultant incremental displacement shows a spike at the depth of about 17 m. The maximum resultant incremental displacement at this depth was about 61 mm, as seen in the reading of 18 March 2015 (Fig. 11a). The resultant cumulative displacement at this depth was about 92 mm (Fig. 11b). After this reading, the casing broke off at a depth of about 17 m, confirming, as expected, that the detected spike in the incremental displacement graph evolved in a failure plane.
The time plot of the inclinometric displacements recorded at the depth of about 17 m (Fig. 10c) shows a roughly constant rate of about 17 mm/year starting from the first reading on 23 September 2010, with the exception of the time interval between the two readings carried out on 11 April 2013 and 09 October 2013, when the sliding was more limited (Fig. 11b).
The direction of movement measured during the last reading before the interruption of the casing was of 175°, i.e., approximately in SSE direction.
The zero reading of the second survey performed on 27 July 2015 confirmed the results of the last reading of the first survey, as the probe was not able to reach the bottom of the casing since it stopped at a depth of about 17 m (Fig. 11c). Moreover, the first reading carried out on 27 August 2015, and a borehole camera investigation, outlined a clear rupture of the inclinometer casing due to a shear surface located at this depth. The interruption occurred close to the boundary between the upper black clayey silt and the lower green clayey-sandy silt lithologies (Fig. 11c).
The closest PS (located more than 30 m from the borehole) records an average LOS displacement rate of about − 16 mm/year, with lower rates in the first half of the time series (roughly until October 2015), and then a slightly steeper rate from October 2015 to the end of the series (see plot in Fig. 10b).
The shallow deployment depth of the inclinometer supports the hypothesis that the bottom of the borehole stays actually above the shear surface, so that no horizontal movements are recorded. Further support to this hypothesis comes from the measurements of an inclinometer in a deeper borehole, installed a few tens of meter upslope the one here described, and monitored within a different ground campaign, which recorded shearing at a depth of 46.5–48 m in a period from 2008 to 2014 (Cotecchia et al. 2016). The absence of other displacement discontinuities in the profiles reported in Cotecchia et al. (2016) for this inclinometer excludes the possible presence of a second, shallower shear surface.
The original total depth of the inclinometer casing of borehole S12 was about 40 m. The zero reading was performed at the beginning of the first survey on 18 December 2008. The following readings show an evident shear zone at a depth interval between 24 m and 27 m (Fig. 15a). Moreover, the resultant incremental displacement graph shows a spike at a depth of about 25 m. The maximum resultant incremental and cumulative displacement at this depth was about 40 mm and 68 mm, respectively (Fig. 15a, b). After the last reading, the casing broke off at the depth of about 25 m, confirming that the detected spike in the incremental displacement graph evolved in a failure plane.
The time plot of the inclinometric displacements shows that the shear strength loss at the depth of about 25 m occurred at a mostly constant rate, with only a period of relative acceleration between the readings performed on 19 April 2011 and 02 August 2012 (Fig. 14c).
The direction of the displacements was oriented at about 125°, to SE.
The zero reading of the second survey, performed on 27 July 2015, confirmed that the inclinometer casing broke off at a depth of about 25 m. Moreover, the reading carried out on 28 October 2015, and a borehole camera investigation, outlined a clear rupture of the inclinometer casing due to a shear surface located at this depth. The interruption occurred in the gray clayey-sandy silt lithology (Fig. 15c).
The closest PS is located at a distance of almost 100 m from the borehole, and shows an average LOS displacement rate of − 18 mm/year, again with a few nonlinear spikes (Fig. 14b). We note that it is difficult to associate the two locations to a homogeneous block of terrain. Nevertheless, the overall picture is coherent with the presence of a subvertical displacement component at the location of the PS measurements and the presence of a horizontal component at the location of the inclinometric measurement, which is slightly downslope.
The preceding description of inclinometric and PS displacements allows to get a coherent picture of the landslide phenomenon under study. The joint use of Interferometric Synthetic Aperture Radar (InSAR) and inclinometers allows coupling the advantages provided by the two measurement methods, thus improving the monitoring capability and the knowledge about the spatial and temporal evolution of the displacements within the landslide body.
MTI approaches extend the area of the unstable slopes where displacement measurements are available, with respect to the few locations where inclinometers are deployed. In particular, Sentinel-1 data provide PS maps with improved spatial density with respect to those derived from the legacy ERS/ENVISAT datasets (Fig. 4). MTI analysis improves the monitoring capabilities also in time. By using archived historical datasets, it is possible to go back in time to assess the stability of the area of interest, as performed in this study by using ERS and ENVISAT data (Fig. 4). Moreover, by linking results coming from different datasets, nowadays it is possible to provide monitoring periods of more than two decades (Bovenga et al. 2018). On the contrary, the maintenance of in situ measurement networks for such a long time is demanding, and homogeneous records acquired over long time intervals are difficult to find. Another important parameter for the analysis in time is the frequency of measurements. In the case of inclinometers, it depends on the in situ measurement campaigns, while for the InSAR data, it is determined by the satellite mission repeat time, which as mentioned, goes from 35 days of the legacy ERS and ENVISAT missions to 12 days for Sentinel-1/A, reaching 6 days for the Sentinel-1/A-B constellation, after April 2016. The time series reported in Figs. 6, 8, 10, 12, and 14 show clearly the improved time sampling provided by the InSAR analysis with respect to in situ inclinometric data. This is crucial, in particular, for detecting possible nonlinear displacements.
While InSAR techniques improve the monitoring capability on the ground surface both in time and space, the inclinometer measurements extend the displacement monitoring downward in the subsoil. This is important for modeling the sliding phenomenon by measuring the displacement depth and by relating the displacement to the subsoil lithology.
Finally, since InSAR and inclinometers measure different components of the actual displacement, they provide complementary information that can be used for both qualitative and quantitative assessment of the spatial and temporal evolution of the landslide movement.
This interpretation is coherent with the nature of the landslide. The Pianello landslide shows the typical morphological features of a translational slide having a quite large depletion area upslope and a slide body resting upon a concave up shear surface (see also Cotecchia et al. 2016). Following this morphological evidence, it is expected, as shown in Fig. 16, that the depth of the shear zone is shallower along the border of the slide mass and maximum in its central part. As a matter of fact, the two inclinometers deployed in the S6 and S7 boreholes located close to the upslope boundary of the slide mass showed shallower shear zones than the ones deployed in boreholes S9 and S12, since these latter are located farther from the landslide boundary than the previous two (Fig. 16b). Following this reasoning, it is expected that the deepest position of the shear zone should be recorded by the inclinometer deployed in the borehole S10, but, unfortunately, this one was not drilled to the correct depth. Published data confirms this interpretation, since the shear zone in the central part of the Pianello landslide is located deeper at about 46.5–48 m (Cotecchia et al. 2016). As a result, the S10 borehole, drilled to the depth of about 35 m, was too shallow, and the inclinometric casing translated in conjunction with the main body of the landslide, preventing the reading of any movement (Fig. 16a, c).
PSI data also concur to give credit to this model: as described in the “Data and methods” section, surface displacements recorded along ascending and descending LOS directions by legacy sensors (Fig. 3a, b) show opposite signs of motion for the PS on the center of the landslide body, thus indicating the presence of a consistent horizontal component of motion, typical of roto-translational landslides, although it is worth to note again that the Pianello landslide has also a non-negligible vertical component (Cotecchia et al. 2016).
The more recent Sentinel-1 ascending data are consistent with this interpretation, and their increased spatial density allows to draw a rather coherent picture of the overall shape and limits of the moving part of the landslide, allowing to update to some extent these pieces of information: in particular, areas close to the SW limit of the landslide area (see Fig. 4) appear to be presently stable, while strong motions (up and beyond − 20 mm/year) are detected in the highest portion of the moving area, at NW, suggesting a somewhat rapid evolution of the rupture surface in this area. This can also be seen in Fig. 16, where the PS closest to the three profiles has been drawn on the terrain surface, with the same colorbar proportional to LOS displacement velocity as in the previous figures.
Otherwise, most of the surface displacements recorded by the PS are consistent with the subsurface shear displacements recorded by the inclinometers, both spatially and temporally, for the overlapping parts of the monitored time intervals.
Finally, comparison of the close moving PS data with the measurements of the single borehole inclinometer S10 helped interpreting its rather puzzling absence of any recorded movements, in spite of its position in the middle of the translational zone, by recognizing the abovementioned likely translational displacement of the whole borehole structure, caused by the insufficient borehole depth.
Our experiment is aimed at illustrating the integrated use of multi-temporal InSAR techniques and in situ inclinometers in slope instability monitoring, which allows overcoming some limitations of each of these measurement methods when used separately.
A first investigation, through analysis of legacy ERS and ENVISAT-derived PS data performed over large areas, highlighted the local presence of moving PS points, indicating active deformation within the limits of the known Pianello landslide, in the outskirts of the town of Bovino, in northern Apulia region (Southern Italy). These first results prompted local authorities to allocate resources and efforts, and guided the deployment of ground instrumentation, consisting in several boreholes equipped with inclinometers and other instruments, on locations chosen on the basis of the remotely sensed information, besides terrain knowledge.
Ground data collected during the following decade by the installed instruments were finally compared to new remotely sensed data, acquired by the Sentinel-1 satellites after the deployment of the ground instrumentation. Such new data basically confirm the presence of active displacements over the Pianello landslide, and can be merged with the information coming from the ground sensors to get a consistent picture of the shape and distribution of the spatial borders as well as the underground shear surface of the Pianello landslide.
This successful example of integration between remotely sensed and ground measurements, made possible thanks to the virtuous collaboration between remote sensing specialists and regional institutions, in this case the Basin Authority of the Apulia Region, serve, in our opinion, as an example of best practice in local environmental management and protection and one of the first, to our knowledge, to be adopted in a difficult environment such as a small hilltop town in Southern Italy.
The derived operational protocol is being planned to be repeated in the future over other test sites within the AdB jurisdiction, and could be an example of a standard to be followed at national scale.
Copernicus Sentinel-1 data are provided by European Space Agency (ESA). ENVISAT and ERS data provided by ESA Category-1 project nos. 2653 and 5367.
This work was performed in the framework of the agreement “Convenzione operativa per lo studio del dissesto del suolo nel sub-appennino dauno,” CUP H38C16000050008.
We thank two anonymous reviewers for their insightful comments and mostly constructive criticisms who helped in improving the manuscript.
This research is funded by the Italian Ministry of University and Research (MIUR) in the framework of “APULIA SPACE” project (PON03PE_00067_6), PON Ricerca e competitività 2007–2013.
- AAVV (2007) Italian Landslides Inventory Project (IFFI). http://www.progettoiffi.isprambiente.it/-cartanetiffi/default_nosso.asp?lang=EN. Accessed 19 Dec 2017
- Alemanno D, Caggiano T, Castorani A et al (2009) Sistemi integrati di monitoraggio mediante telerilevamento e strumentazioni in foro di area a criticità geomorfologica. Esempi di applicazioni nell’Appennino Dauno. In: Atti 13a Conferenza Nazionale ASITA. In: pp 77–82Google Scholar
- Bovenga F, Nutricato R, Refice A, Guerriero L, Chiaradia MT (2005) SPINUA: A flexible processing chain for ERS / ENVISAT long term interferometry. In: Proceedings of the 2004 Envisat & ERS Symposium (ESA SP-572). 6-10 September 2004, Salzburg, Austria. Edited by H. Lacoste and L. Ouwehand. Published on CD-Rom., pp 473–478Google Scholar
- Calò F, Ardizzone F, Castaldo R, Lollino P, Tizzani P, Guzzetti F, Lanari R, Angeli MG, Pontoni F, Manunta M (2014) Enhanced landslide investigations through advanced DInSAR techniques: the Ivancich case study, Assisi, Italy. Remote Sens Environ 142:69–82. https://doi.org/10.1016/j.rse.2013.11.003 CrossRefGoogle Scholar
- Casagli N, Cigna F, Bianchini S, Hölbling D, Füreder P, Righini G, del Conte S, Friedl B, Schneiderbauer S, Iasio C, Vlcko J, Greif V, Proske H, Granica K, Falco S, Lozzi S, Mora O, Arnaud A, Novali F, Bianchi M (2016) Landslide mapping and monitoring by using radar and optical remote sensing: examples from the EC-FP7 project SAFER. Remote Sens Appl Soc Environ 4:92–108. https://doi.org/10.1016/j.rsase.2016.07.001 Google Scholar
- Casero P, Roure F, Endignoux L et al (1988) Neogene geodynamic evolution of the southern Apennines. Mem Soc Geol It 41:109–120Google Scholar
- Ciaranfi N, Gallicchio S, Loiacono F (2011) Note illustrative della Carta Geologica d’Italia alla scala 1: 50.000, Foglio 421 “Ascoli Satriano.” Istituto Superiore per la Protezione e la Ricerca AmbientaleGoogle Scholar
- Cotecchia F, Pedone G, Bottiglieri O et al (2014) Slope-atmosphere interaction in a tectonized clayey slope: a case study. Riv Ital di Geotec 48:34–61Google Scholar
- Cotecchia F, Vitone C, Petti R et al (2016) Slow landslides in urbanised clayey slopes: an emblematic case from the south of Italy. In: Landslides and engineered slopes. Experience, theory and practice. CRC Press, pp 691–698Google Scholar
- Cruden DM, Varnes DJ (1996) Landslide types and processes. Special Report, Transportation Research Board, National Academy of Sciences 247:36–75Google Scholar
- Dazzaro L, Di Nocera S, Pescatore T et al (1988) Geologia del margine della catena appenninica tra il F. Fortore e il T. Calaggio (Appennino meridionale). Mem Soc Geol Ital 41:411–422Google Scholar
- Dazzaro L, Rapisardi L (1984) Nuovi dati stratigrafici, tettonici e paleogeografici della parte settentrionale dell’ Appennino Dauno. Ital J Geosci 103:51–58Google Scholar
- Dazzaro L, Rapisardi L (1987) Osservazioni geologiche sull’Appennino dauno. Mem Soc Geol Ital 38:241–246Google Scholar
- Dazzaro L, Rapisardi L (1996) Schema geologico del margine appenninico tra il F. Fortore e il F Ofanto. Mem Soc Geol It 51:143–147Google Scholar
- Di Martire D, Novellino A, Ramondini M, Calcaterra D (2016a) A-differential synthetic aperture radar interferometry analysis of a deep seated gravitational slope deformation occurring at Bisaccia (Italy). Sci Total Environ 550:556–573. https://doi.org/10.1016/j.scitotenv.2016.01.102 CrossRefGoogle Scholar
- Jacobacci A, Martelli G (1967) Note illustrative della carta geologica d’italia F°174. Ariano Irpino. Servizio Geologico D’Italia, RomaGoogle Scholar
- Ketelaar VBH (Gini) (2009) Satellite radar interferometry: subsidence monitoring techniques. Springer Netherlands, DordrechtGoogle Scholar
- Mostardini F, Merlini S (1986) Appennino Centro Meridionale. Sezioni Geologiche e Proposta di Modello Strutturale. Mem Soc Geol Ital 35:177–202Google Scholar
- Parise M, Federico A, Palladino G (2012) Historical evolution of multi-source mudslides. In: Eberhardt E, Froese C, Turner A, Lerouil S (eds) Landslides and engineered slopes. Protecting society through improved understanding. Proceedings 11th Int. Symp. Landslides. Banff (Canada), pp 401–407Google Scholar
- Patacca E, Sartori R, Scandone P (1990) Tyrrhenian basin and Apenninic arcs: kinematic relations since late Tortonian times. Mem Soc Geol Ital 45:425–451Google Scholar
- Patacca E, Scandone P (2007) Geology of the southern Apennines. Boll della Soc Geol Ital Vol Spec 7:75–119Google Scholar
- Refice A, Pasquariello G, Bovenga F, Festa V, Acquafredda P, Spilotro G (2016) Investigating uplift in Lesina Marina (southern Italy) with the aid of persistent scatterer SAR interferometry and in situ measurements. Environ Earth Sci 75:243. https://doi.org/10.1007/s12665-015-4979-1 CrossRefGoogle Scholar
- Spalluto L, Fiore A, Miccoli MN (2015) Mappe di attività delle frane: analisi della riattivazione di alcuni dissesti idrogeologici in Appennino dauno. Geol e Territ 12:3–17Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.