Access the full text.
Sign up today, get DeepDyve free for 14 days.
Original Paper Landslides (2018) 15:2113–2128 M. Del Soldato I A. Riquelme I S. Bianchini I R. Tomàs I D. Di Martire I P. De Vita I S. Moretti I DOI 10.1007/s10346-018-1015-z D. Calcaterra Received: 11 December 2017 Accepted: 11 May 2018 Published online: 6 June 2018 Multisource data integration to investigate one century © The Author(s) 2018 of evolution for the Agnone landslide (Molise, southern Italy) Abstract Landslides are one of the most relevant geohazards world- future urban development (e.g. Van Westen and Lulie Getahun wide, causing direct and indirect costs and fatalities. Italy is one of 2003; Korup et al. 2010; Giordan et al. 2013). Studies of landslide the countries most affected by mass movements, and the Molise evolution that improve the knowledge of ground movements are region, southern Italy, is known to be susceptible to erosional pro- essential to understand the mechanism of deformation (Casson et al. cesses and landslides. In January 2003, a landslide in the municipal- 2003) for landslide-prone territories to mitigate the associated risk ity of Agnone, in the Colle Lapponi-Piano Ovetta (CL-PO) territory, and to prevent landslide occurrences or reactivations. occurred causing substantial damage to both structures and civil Remote sensing techniques such as persistent scatterer interfer- infrastructure. To investigate the evolution of the landslide-affected ometry (PSI) data (e.g. Massonnet and Feigl 1998; Ferretti et al. 2001; catchment over approximately one century, different data were taken Farina et al. 2006; Hooper et al. 2012) and change detection meth- into account: (i) literature information at the beginning of the odologies (e.g. White 1991;Ferretti etal. 2001;Lu et al. 2004;Dekker twentieth century; (ii) historical sets of aerial optical photographs 2005; Farina et al. 2006) have been successfully exploited to detect to analyse the geomorphological evolution from 1945 to 2003; (iii) and map slow-moving landslides at a local or regional scale to SAR (Synthetic Aperture Radar) data from the ERS1/2, ENVISAT and perform back-analysis to better define boundaries and ground rates COSMO-SkyMed satellites to monitor the displacement from 1992 to of movements, as well as to identify the most critical landslide- 2015; (iv) traditional measurements carried out through geological affected sites over wide areas (Solari et al., 2018). Although they are and geomorphological surveys, inclinometers and GPS campaigns to powerful instruments to investigate, map and monitor natural pro- characterize the geological setting of the area; and (v) recent optical cesses, satellite optical and radar images have been available for photographs of the catchment area to determine the enlargement of approximately the last 25 years. A breakthrough in the study of the the landslide. Using the structure from motion technique, a 3D evolution of natural processes was to use different techniques to reconstruction of each set of historical aerial photographs was made investigate the longest possible period (e.g. by means of new tools to to investigate the geomorphological evolution and to trace the analyse historical aerial photographs), which enabled an analysis of boundary of the mass movements. As a result, the combination of approximately the last 70 years (e.g. Van Westen and Lulie Getahun multitemporal and multitechnique analysis of the evolution of the 2003). In the late 1970s, Ullman (1979) developed the structure from CL-PO landslide enabled an assessment of the landslide expansion, motion (SfM) technique that was only recently applied to geomor- which resulted in a maximum length of up to approximately 1500 m. phological studies of the evolution of the Earth’s surface, taking A complete investigation of the past and present deformational advantage of similar tenets from stereophotogrammetry (Snavely sequences of the area was performed to potentially plan further et al. 2008; Westoby et al. 2012;Eltner etal. 2016). This approach mitigation and prevention strategies to avoid possible reactivations. enables the reconstruction of 3D models of various surfaces by overlapping several digital photos. Some common applications of Keywords Landslide evolution Structure from SfM in the literature include 3D reconstruction of external façades of . . . motion Geomorphology A-DInSAR Molise buildings and monuments (e.g. Snavely et al. 2008;Furukawaand Ponce 2010), archaeological sites (e.g. Doneus et al. 2011; Verhoeven Introduction 2011; Verhoeven et al. 2012), generation of digital elevation models Landslides are one of the most important and widespread natural (DEM) by means of Unmanned Aerial Vehicles (UAV) (e.g. Turner hazards affecting the Italian territory (Herrera et al. 2018), partially et al. 2012; Lucieer et al. 2013; James and Robson 2014), change due to climate change with serious effects on the environment and detection (e.g. Abellán et al. 2009) and rock mass characterization geomorphology (Crozier 2010). This affirmation is supported by (Sturzenegger et al. 2011;Riquelmeet al. 2014). several data sets showing the events and people involved in landslide The regions in southern Italy involved in the geologically recent occurrences as well as the economic and social consequences and the Apennines mountain chain are highly susceptible to landslides damage to structures and infrastructure (Schuster and Highland (Cotecchia and Melidoro 1974) due to the complex geological 2001; Kjekstad and Highland 2009; Del Soldato et al. 2017). The setting, characterized by several different structurally convoluted Italian Landslides Database includes up to 528.903 landslides in Italy, lithologies (Esu 1977). More than 4000 mass movements (Rosskopf which is the European country with the highest density of landslides and Aucelli 2007) and incipient erosional processes (Marchetti (the landslide area represents 7.3% of the territory) (Herrera et al. 2016) are known to affect the territory of the Molise region despite 2018). Since the beginning of the twentieth century, more than 1400 its limited area. landslides occurred at approximately 1200 different sites in Italy The aim of this work is an analysis of the evolution of the (Salvati et al. 2015). This suggests that several mass movements are landslide that occurred in the Colle Lapponi-Piano Ovetta catch- reactivations in previously affected areas. Landslide evolution in ment (CL-PO), in the municipality of Agnone (Molise, southern different environments plays a key role in investigating possible Italy), by combining different remote sensing techniques, e.g. Landslides 15 & (2018) 2113 Original Paper SfM and InSAR, and field surveys, i.e. geomorphological inves- categorized by several damage classification approaches (Del tigations and in situ data, i.e. GPS and inclinometer measure- Soldato et al. 2016a). ments. This process will be supported by historical literature The CL-PO territory is geologically characterized by the Mount information and geomorphological and topographical data in Pizzi-Agnone and Colle Albero-Tufillo units outcropping with the order to determine the geomorphological evolution over ap- two members of Agnone Flysch and a lower marly formation proximately one century of the CL-PO landslide-prone area. (Fig. 1a). The members of the Agnone Flysch are catalogued as To achieve this goal, different data were collected, e.g. historical BStructurally Complex Formations^, geological materials charac- optical photographs, ancillary and field data, and radar satellite terized by large and scale-dependent heterogeneity in lithological images and products. Several 3D reconstructions of the CL-PO and structural features that suffered complex compressional and landslide were produced and analysed by using sets of historical extensional geological phases (Esu 1977). The geotechnical param- aerial photographs captured in 1945, 1954, 1981, 1986, 1991 and eters of these types of formations are influenced by heterogeneity 2003. Although historical aerial imagery has several limitations, as well as scaly fabric with the alternation of Bhard^ (rock-like e.g. a bi-dimensional vision of the territory in greyscale, they material) and Bweak^ horizons (soil-like material) (Almagià 1910; stillplayafundamental rolein the studies of environmental and Cotecchia and Melidoro 1974; Guida and Iaccarino 1991; Di Maio landscape evolution (Carrara et al. 2003;Van Westen andLulie et al. 2010). Getahun 2003; Giordan et al. 2013). By analysing all the available The lower member of the Agnone Flysch formation presents data, approximately one century of evolution for the CL-PO alternations of marl limestones, marls and calcarenites, in addition landslide-prone area, from the beginning of the twentieth cen- to deposits of silico-clastic turbidites composed of thin intercala- tury to 2015, was investigated. tions of clayey sandstones, sandstones and arenites. The upper member of the Agnone Flysch is constituted by an alternation of Site description and available data marly, semi-coherent clayey and subordinate greyish sandy layers The municipality of Agnone is a Sannitic archaeology site located in with low mechanical resistance, diffuse alteration traces and the northern part of the Molise region (southern Italy), which has an lithoid sandstones or calcareous intercalations with highly variable area of approximately 96 km , and the territory is strongly affected thicknesses. Inside both members, some olistoliths of older con- by both landslides and erosional processes, as the major part of the glomeratic material are recognizable (Vezzani et al. 2004; Filocamo Molise region. The historical town of Agnone is located at approx- et al. 2015). Furthermore, weathering effects caused discoloration, imately 800 m a.s.l., and it is bathed by Verrino creek. The climate of decomposition and weakening, forming a superficial regolith ho- the region is moderate, with alternating cold temperatures with rain rizon featured by clays, silty clays and subordinate sand with and snow thundershowers during autumn-winter and arid periods diffuse alteration traces, abundant organic material and several during spring-summer. The investigated mass movement is a deep- clasts. A cross-section (A-A′) was traced along the CL-PO landslide seated large roto-translational slide resulting in earth flow in the area and is shown in Fig. 1b. lower portion, characterized by a complex movement (Calcaterra The geotechnical characterization of the area affected by the et al. 2008) affecting the western territory of the municipality in the landslide has been made by the re-interpretation of two geological catchment of the S. Nicola valley, tributary of the Verrino creek on and geotechnical campaigns from 2004 to 2006. The latter consists the hydrographic right side. of 39 boreholes with depths between 10.5 and 40.0 m that have Since the beginning of the twentieth century, the CL-PO catch- allowed the identification of four homogeneous layers (Calcaterra ment has been affected by the landslide. In March 1905, the bridge et al. 2008) from bottom to top as follows: of the main access road to Agnone crossing the Verrino creek was damaged by a gravity phenomenon due to an intense rainfall Level D—marly clays, marls and clayey marls with silty and period combined with snow-melting (Almagià 1910; Calcaterra clay fractions. et al. 2008). Successively, the Agnone municipality has been af- fected by several small and large landslide phenomena reported Thickness: 15.30–20.00; in the national AVI Project (Guzzetti et al. 1994), revealing more than 60 landslides in the territory between 1970 and 1998. Be- tween January 23rd and 27th, 2003, an important remobilization Level C—calcareous layers with thickness up to several metres. involved a large area of the historically dormant CL-PO landslide, causing deformations over the whole basin and forcing the local authorities to adopt restrictive measures for 13 edifices occupied Thickness variable from dm to m; by 17 families located within and nearby the landslide. Further- more, two country roads adjacent to the landslide remain closed Level B—grey clays, silty clays, sandy clay and silty sands due to the substantial damage caused by this event. The mass with a medium plasticity. movement subsequently reactivated in 2004, 2005 and between 2006 and 2007, which induced the local administration to allocate Thickness: 4.00–12.00; resources for some urgent interventions to intercept superficial waters and drain a pond formed in the upper portion of the Level A—chaotic and plastic matrix including dispersed frag- affected area, in addition to geomorphological reshaping work. ments of resistant rock and mudstone. Despite these strategies, the ground displacements increased, which caused the cracking and sliding of structures and of a road Thickness: 6.70–11.20. located uphill of the landslide. These damages were mapped and 2114 Landslides 15 & (2018) Fig. 1 a Localization and geological sketch map (Vezzani et al. 2004) of the area of interest in Agnone with the currently known contour of the landslide reported in red and the possible larger outline in purple. b Geological section A-A′ is traced along the area of interest of the CL-PO landslide (Vezzani et al. 2004) The bottom level (i.e. Level D) was directly involved in the 2003 characterized by clayey Flysch units. Significant erosional processes reactivation of the CL-PO landslide. affect regions with slope angle values greater than 15–20°. Furthermore, some inclinometers and piezometers were The oldest available data used in this work are the historical installed in the boreholes, but they recorded only for a short aerial photographs captured by the Italian Istituto Geografico time due to the continued displacement that caused their Militare (IGM) from 1942 to 2003. Starting from 1942, the IGM failure. made recurring flights covering almost the entire Italian territory Ancillary data (e.g. DEM and derived maps), information ac- shooting greyscale nadir pictures with partial overlap. The first set quired by means of remote sensing techniques (i.e. historical of historical aerial photographs shot on the CL-PO landslide area optical aerial photos, radar satellite images and PSI), as well as was captured in 1945. This was followed by five additional flights direct investigations supported by instrumental monitoring, were recording a series of photographs over the area of interest dating used for the long-term investigation of the landslide. A DEM with back to 1954, 1981, 1986, 1991 and 2003 (Table 1). a 5-m cell resolution was provided by the Molise region and useful Since 1992, the back investigation of the CL-PO landslide information (e.g. aspect, which is a derived map showing the evolution is supported by information from SAR Interferome- directional, with respect to the azimuth, exposures of the slopes try (InSAR). ERS1/2 and ENVISAT data from 1992 to 2000 and of the area; and slope, which is a parameter indicating the incli- from 2002 to 2010, respectively, are available through the PST- nation in degrees of the sides of the areas, with respect to the A(Piano Straordinario di Telerilevamento Ambientale)project horizontal surface) was extrapolated by means of geographical and the Web Map Service (WMS) of the Italian Ministry for the tools in a geographical information system (GIS) environment to Environment, Land and Sea. Furthermore, to cover the period characterize the morphological parameters of the area and to of 2012–2015, 88 images acquired by COSMO-SkyMed constel- better interpret the remote sensing data. lations of the Italian Space Agency (ASI - Agenzia Spaziale The landslide includes the territory from the Verrino creek at Itliana) were collected in the framework of a specific project approximately 660 m a.s.l. up to 870 m a.s.l. and the morphology of (Table 2) that allowed us to extend the investigated period. The the surrounding area strongly controlled by the slope parameters ERS1/2 and ENVISAT data deriving from the Portale and the different local lithotypes. Where high topographical gradi- Cartografico Nazionale were elaborated by the PSInSAR tech- ents are present, from 30° to 35°, the calcareous formations outcrop, nique (Persistent Scatterers Interferometry SAR) developed by while lower slope gradients, from 5° to 10°, correspond to areas TRE-ALTAMIRA (Ferretti et al. 2000, 2001). The COSMO- Landslides 15 & (2018) 2115 Original Paper Table 1 Characteristics of the available historical aerial photographs used in this work Acquisition year Number of photos Estimated scale Flying height (m) Focal length (mm) 1945 4 1:55000 7500 137 1954 8 1:33000 6000 153.01 1981 5 1:30000 5200 152.55 1986 4 1:28000 5100 152.55 1991 6 1:36000 6070 153.22 2003 4 1:35000 5300 153.31 SkyMed images were processed by the Coherence Pixel Tech- successively substituted (S5). Additionally, the inclinometer 2D nique (CPT) developed by the Remote Sensing Laboratory measurements identified the depth of the rupture surface of the (RSLab) of the Universitat Politècnica de Catalunya, Spain landslide and were used to control the efficiency of the adopted (Mora et al. 2003; Blanco-Sanchez et al. 2008; Iglesias et al. mitigation measures. 2015). Methodology As the mobilized material and the widespread vegetation cover caused a low backscatter radar signal from the landslide body, in 2010, The assessment of the landslide evolution has been provided by a eight corner reflectors (CRs) were installed in the mass movement to historical investigation of the literature and analyses conducted with traditional instruments, i.e. inclinometers and GPS cam- enhance the signals reflected in the direction of the radar. The collected data were used to increase the knowledge about the displacement paigns, field surveys and remote sensing techniques. Six sets of affecting the internal portion of the landslide (yellow triangles in Fig. 2). historical aerial photographs were interpreted by means of 3D reconstruction of the landslide area obtained via the SfM tech- Although the corner reflectors were correctly positioned (four visible by the ascending orbit and four by the descending orbit), the high velocity nique (Ullman 1979)(Fig. 3). Hence, a qualitative comparison of displacement affecting the area, some man-operated motions carried between historic and recent available data sets was possible. This analysis shows that several factors, such as the geometrical distor- out by the land owners, damage or vegetation recovery made them invisible to the satellite analysing the entire dataset of CSK images. The tion of the photographs, the ortho-rectification stage and the corner reflectors were also used as benchmarks for differential GPS physical condition of the aerial photos are to be considered in the interpretation. measurements during the campaigns developed since 2010. In addition, GPS measurements were collected for some benchmarks (green circles Furthermore, GPS measurement campaigns, inclinometers and in Fig. 2) to investigate the displacements of the ground surface and remote sensing techniques were applied to monitor the evolution of the CL-PO landslide since the main reactivation in January 2003 increase the spatial point density throughout the entire landslide. During the survey campaigns in 2006, five inclinometers were until 2015. installed (orange circles in Fig. 2). Four of them were placed in the existing boreholes used for geotechnical investigations, while an- other one was placed out of the mobilized area to validate the 3D reconstruction of historical aerial photographs The use of historical imagery allowed the recreation of a 3D image displacement recorded by remote sensing techniques and to mon- by means of a stereoscopic sight to detect, map and monitor itor the kinematic evolution of the landslide after the main reac- tivation. Relevant movements recorded along a sliding surface unique geomorphological shapes across several decades (Hapke 2005). caused the failure of one of the inclinometers (S2), which was Table 2 Main features of the interferometric products dataset used to study the CL-PO landslide Features ERS ERS ENVISAT ENVISAT CSK CSK Wavelength C (~ 5.6 cm) C (~ 5.6 cm) C (~ 5.6 cm) C (~ 5.6 cm) X (~ 3.1 cm) X (~ 3.1 cm) Incidence angle, θ ~ 23° ~ 23° ~ 23° ~ 23° 26.6° 26.6° Orbit Ascending Descending Ascending Descending Ascending Descending Ground resolution, m (azimuth × 4×20 4×20 4×20 4×20 3 × 3 3 ×3 range) Revisit time (days) 35 35 35 35 16 16 Temporal span (day/month/year) 25/04/1993 08/06/1992 29/11/2002 07/11/2002 15/10/2012 13/02/2012 13/12/2000 07/12/2000 30/07/2010 03/06/2010 01/05/2015 15/01/2014 Processing method PSInSAR™ PSInSAR™ PSInSAR™ PSInSAR™ CPT CPT No. of images used 54 78 50 45 41 47 2116 Landslides 15 & (2018) Fig. 2 CL-PO landslide with the location of the benchmarks (green circles), the corner reflectors (yellow triangles) and the inclinometers (orange circles) In this work, the SfM technique for digital photographs and ancillary data and the calculated data for the CL-PO landslide. algorithms optimized for the graphic processing unit (GPU) Hence, a better investigation of the interpretations of the area (Lucieer et al. 2013), implemented in the Agisoft Photoscan Pro- affected by the landslide through time could be conducted. fessional edition software (Agisoft 2016), was applied to the The analysis of the six available sets of photographs covering historical aerial image analysis (Del Soldato et al., 2016b)after the CL-PO landslide-prone area can be subdivided into three adapting some parameters. This software works with sets of main stages: (i) data preparation; (ii) data processing and 3D digital photographs captured at the same time and partially reconstruction; and (iii) extraction of the products and results. overlapped (at least 60%, Agisoft 2016) covering the area of The application of the SfM approach on historical aerial pho- interest. By means of a redundant iterative bundle adjustment tographs requires an adaption to the traditional workflow procedure, the SfM approach extracts information to solve the (Gomez et al. 2015;Ishiguroetal. 2016). The precision of the geometry of the scene, the camera parameters and the orienta- 3D reconstruction depends on several factors such as the grade tion for each set of photographs (Snavely et al. 2008), dramati- of conservation of the photos, based on the scratches or mark- cally reducing the number of unknowns during the alignment ings on them, the scan resolution and the distortion introduced procedure. The 3D Points Cloud (3DPC) represents the geometry/ during this process, the overlap between adjacent photos and structure of the scene that is generated by applying the SfM the identification, in addition to the input of GCPs and Tie approach to the overlapped area between at least two images Points. (Hartley and Zisserman 2003; Szeliski 2010;Fisheretal. 2013)in Historical aerial photos used in this work were captured with local coordinates. The input of several Ground Control Points non-digital cameras. Therefore, they were available in printed (GCPs; points with known object-space coordinates in a system format and were scanned for the 3D reconstruction by the soft- chosen by the operator) and Tie Points (points without coordi- ware. The information of the flight and the camera used for the nates; useful to better correlate the photographs and limit the acquisition of the photographs are reported in a black border distortion between them) allows georeferencing the 3DPC to a around the pictures. During the preparation of the photographs specific coordinate system. At the end, the obtained 3D recon- for the software, these borders have to be carefully removed since structions for all the available sets of photographs were extracted they do not contain useful information for the 3D reconstructions. and imported in the environmental GIS with the collected It is important to maintain the centre of the images and the Landslides 15 & (2018) 2117 Original Paper Fig. 3 Schematic flowchart of the evolution investigation technique applied to the CL-PO landslide number of pixels when cropping the images. This allows the SfM if the GCPs were added to historical aerial photographs scanned software to recognize the set of photographs as captured by the at a higher resolution, i.e. 600 dpi, the maximum resolution that same camera to apply the same camera calibration. Once the couldbeattained was5m. Thesmallervaluesofprecision photographs were prepared, the cameras were aligned, and a resulting from the alignment and georeferencing process indicate sparse point cloud was generated. Subsequently, GCPs were that the algorithm optimized the results. inserted around the landslide to georeference the 3DPC and to For each scenario, the number of inserted GCPs depends on the correct non-linear deformations (Fig. 4a, c) on the mesh. georeferencing precision; the aim was to reach a maximum total The input of GCPs is a very delicate task that has to be error of 5 m. In Table 3, the statistics of the errors derived from conducted with high accuracy by identifying some stable points each set of photographs used for the 3D reconstructions are from the oldest photographs to the newest ones, to obtain the reported. greatest possible precision. For each inserted GCP, the errors in The last step enables the creation of the mesh to reproduce a metres are presented (Fig. 4c, d). Since the coordinates of the georeferenced DEM and the orthomosaic images by means of GCPs were extracted from a DEM with a 5-m cell resolution, even dedicated tools in the PhotoScan Pro software. All the realized 2118 Landslides 15 & (2018) Fig. 4 Some GCPs on the oldest Dense Cloud of the scenario in 1945 (a) and in the newly reconstructed mesh of 2003 (b), with the respective accuracies of some sample GCPs (c and d, respectively) products, i.e. 3DPC, the mesh of the georeferenced DEM and the derived PS (Persistent Scatterer) data were classified using a colour orthomosaic can be extracted in several ways in order to be used scale indicated by hot colour displacement moving away from the by other scientific software for the interpretation. sensor and cold colour data moving towards the satellite. The The 3DPCs are mounted in the Cloud Compare software to stability range (± 1.5 mm/year) is coloured in green, based on the better investigate and compare the resulting products by zooming standard deviation of the processed product. and rotating them in a 3D view. In addition, the colour of each Several studies demonstrated that the in-depth analysis of PS point reproduces the value, in RGB, of the source historical pho- time series located at relevant sites can assist with the under- tographs. This allows better identification and localization of standing of the dynamic and temporal evolution of ground mo- scarps, counterslopes and ground tension cracks by means of both tions for a slope (Meisina et al. 2008; Cigna et al. 2011; Confuorto visual investigation and recognizable geomorphological changes. et al. 2017). Notti et al. (2015) proposed a three-step procedure The realized meshes and DEMs for all the sets of historical aerial that does not interfere with InSAR processing to exploit the photographs were added in a GIS environment to combine them information derived from time series: (a) Pre-Processing evalu- with both sets of ancillary data. Hence, a better interpretation of ating the SAR Dataset Quality Index (SDQI); (b) Post-Processing, the ground displacements of the catchment can be conducted in improving the quality of the previously processed time series data order to define a boundary around the area involved in the applying an empirical/stochastic method to remove single data landslide. anomalies as well as noise and regional trends; and c) detection and correction of possible phase unwrapping errors. The time DInSAR technique series from the area of interest were subjected to a time series The analysed data were acquired by ERS1/2, ENVISAT and improvement in the Post-Processing phase (Notti et al. 2015)to COSMO-SkyMed constellations from 1992 to 2015. ERS1/2 and remove anomalies from the regional trend. This type of effect or ENVISAT data were processed by the PSP (Costantini et al. 2008) anomaly, not related to natural processes affecting the ground and PSInSAR technique developed by TRE-ALTAMIRA (Ferretti surfaces, can be easily detected since the entire dataset is affected. et al. 2000, 2001). COSMO-SkyMed images were processed by the To identify the regional noise, data with high coherence (e.g. SUBSOFT processor using the CPT algorithm developed by the higher than 0.9) and an average LOS velocity between − 0.5 mm/ Remote Sensing Laboratory (RSLab) of the Universitat Politècnica year and + 0.5 mm/year were selected and averaged. The average de Catalunya (Mora et al. 2003; Blanco-Sanchez et al. 2008). The regional trend has to be subtracted from the original trend, Landslides 15 & (2018) 2119 Original Paper Table 3 RMSE control points for each reconstructed set of historical aerial images Set of images Number of Number of tie X error Y error Z error Total error Image (year) GCPs points (m) (m) (m) (m) (pixel) 1945 10 6 3.17 1.39 2.58 4.32 0.657 1954 9 7 2.72 1.59 2.01 3.74 0.873 1981 8 9 0.42 0.48 0.62 1.01 0.539 1986 9 5 1.37 1.37 1.73 2.60 0.521 1991 9 7 0.91 1.11 1.33 1.96 0.362 2003 10 5 0.99 1.41 1.13 2.06 0.392 providing a time series corrected for diffuse noise or trends. The Starting in 2010, several differential GPS measurement cam- data recorded on the landslide-prone area were analysed investi- paigns were conducted to monitor the remedial maintenance to gating the velocity of displacement along the LOS and reprojected intercept and drain surface waters by means of benchmarks and along the slope (Notti et al. 2014). The combination of the ac- corner reflectors, which are currently partially lost and useless due quired velocities by both ascending and descending orbits can to the continued displacement, vegetation and human activities. decompose the detected motion along the LOS into horizontal All campaigns were based on three stable points located outside and vertical components (Manzo et al. 2006; Notti et al. 2014). The the landslide. component conversion is made by formulas that take into con- Furthermore, two recent campaigns were conducted to high- sideration the LOS directional cosines for the ascending and light the geomorphological features to identify possible further descending passes. changes. In 2014 and 2015, several scarps and counterslopes were detected and mapped. Field surveys After the main event in January 2003, several field surveys were Results performed by means of direct investigations and instrumental monitoring. The geology of the landslide-prone area was investi- 3D reconstruction of historical aerial photographs gated by a field campaign and took advantage of several boreholes The application of the SfM technique on sets of historical aerial equipped by inclinometers. photographs was useful to partially fill the temporal gap between Fig. 5 3D reconstruction of the Colle Lapponi-Piano Ovetta landslide-prone area made by a set of historical aerial photographs that date back to 1945 (a), 1954 (b), 1981 (c), 1986 (d), 1991 (e) and 2003 (f). The landslide contour is plotted in blue 2120 Landslides 15 & (2018) the beginning of the twentieth century and the main reactivation recognizable in the widespread vegetation close to Verrino creek in 2003, with no information on the geomorphological evolution (Fig. 6b). of the landslide. The procedure, with dedicated settings for histor- The 3D reconstruction made from the 2003 imagery shows the ical photos, was applied to each scenario to extract the respective involvement of the left flank of the basin (Fig. 5f). Between 3DPC and models. The results were individually analysed zooming January 23rd and 27th, an unusual increase in pore pressures and rotating the 3D model to better detect and map the contour of caused by an intense rainfall of more than 200 mm in 72 h (Fig. 7a) the landslide with respect to the classical stereoscopy by means of occurred, triggering the complex deep-seated mass movement the Cloud Compare software. In addition, the resulting 3D recon- (Calcaterra et al. 2008) consisting of the reactivation of an old struction was integrated in the environmental GIS in order to dormant landslide. The upper member of the Agnone Flysch was obtain a better interpretation combined with the available ancil- involved in a large roto-translational slide resulting in earth flow lary data, e.g., Aspect and Slope. The first set of photographs refer (Cruden and Varnes 1996), including an area of approximately to 1945 (Fig. 5a), in which an area affected by the landslide was 1.21 km that was up to 1 km in length and had an average width of visually recognized at the right flank of the basin damaging the 200 m, which completely altered the local hydrographic network access road to Piano Ovetta. It is difficult to confirm if the phe- (Fig. 7b). nomenon was an activation or reactivation of an existing old mass movement because previous maps are not available. The contour DInSAR technique results map based on the analysis of the 3D reconstruction contains a The ERS1/2 PS data are very sparse for the period of 1992–2000 due to the spread of vegetation and the low presence of outcrops and visible area of approximately 0.86 km that features displaced material with an average length and width of approximately 900 structures (Fig. 8a). They are mainly localized on buildings and and 100 m, respectively. structures in the surrounding area of the landslide. The main movement is recognizable on the left flank (Fig. 8b) and on the The subsequent analysis included the set of photos from 1954 (Fig. 5b), in which no additional differences were recognized with structure of the current crown. respect to the previous set from 1945. In fact, the dimensions of the In March 2004 and between December 2004 and January 2005, several rainfall episodes affected the area again, causing two minor landslide are comparable to those of the previous reconstruction, and only small changes in the length, up to 1 km, are recognizable. reactivations. The mobilized total volume of the whole landslide 6 3 The flight conducted by the IGM over this territory was repeat- area was estimated at 3.5 × 10 m (Calcaterra et al. 2008), and it caused new damage to buildings and infrastructure. An additional ed in 1981 acquiring five additional photographs. After 25 years, the results are comparable to the previous analysis, with some small advancement in the movement of the landslide toe was detected in differences recognizable only on one side of the main body June 2006 and between April 2006 and April 2007. Consequently, new mitigation measures were tracked into the body of the land- (Fig. 5c). Some lateral sliding, probably due to water erosion operating on the toe, is distinguishable. In fact, the length of the slide, i.e. a new reshaping of the slope and 10 trench drains. These landslide is similar to that from the previous years. However, the actions allowed the stabilization of the middle-lower region of the landslide body. width of the landslides increased due to some lateral slides that enlarged the area involved. An analogous boundary, with less The ENVISAT PS data (2003–2010) are concentrated on the evidence of lateral sliding, can be mapped on the 3D reconstruc- structure located on the crown. The ascending PS data show average velocities between − 3.0 and − 4.9 mm/year with a peak tion of the CL-PO landslide-prone area in the 1986 reconstruction (Fig. 5d). of − 7.0 mm/year, while the descending data are close to the In the reconstruction of 1991 (Fig. 5e), remedial work con- stability range (Fig. 9a). The time series of the ENVISAT data, mainly in descending geometry, are noisy, but the trends of ducted between 1986 and 1991 is recognizable (Fig. 6a). These man-made activities are visible as perpendicular white strips both orbits confirm continuous displacement (green line in across the river along the width of the landslide (into Verrino Fig. 9b). Furthermore, from 2012 to 2015, the COSMO-SkyMed data show creek). They consisted of concrete weirs built to avoid the con- tinuous enlargement of the landslide due to span erosion and to displacement, although slow, demonstrating that the landslide control the amount of mobilized material run off. Currently, they movement is still active. The ascending PS data show a relatively constant velocity on buildings located in the current crown of the arenot visibleinsidethe landslide due to the remobilization of material during the 2003 reactivation, but some of them were landslide, exhibiting important displacements with a peak of − Fig. 6 Location of the remedial work visible on the historical aerial reconstruction (a); details of the concrete weirs built along the tributary of the Verrino Creek (b) Landslides 15 & (2018) 2121 Original Paper Fig. 7 Daily (blue) and cumulative (green) rainfall (a); optical image of the study area (the landslide is delineated by a red line) (b) 11.6 mm/year. In descending geometry, despite very noisy data, in addition to the Bzero reading^, were conducted showing im- the movement is confirmed with recorded velocities of + 7 mm/ portant displacements at different depths, after which almost all year moving toward the sensors (Fig. 10a). As shown in this figure, the inclinometer casing tubes were broken (Fig. 11a). In the upper the location of several scarps corresponds to the buildings, thus part of the landslide body, the inclinometer BS4^ exhibits an confirming the retrogressive trend of the landslide. Moreover, important displacement with a sliding surface at approximately these results have also been observed by the damage recorded 25 m under the ground surface (Fig. 11b). The graph of the on the buildings. Figure 10b shows a time series of the ascending inclinometer BS5^ reveals that the instrumentation was not PS data forconstruction onthe crownofthe landslide. Itis locked on a stable bedrock, indicating only that the sliding sur- interesting to note that the building located on the left flank of face is deeper than 25 m from the ground level (Fig. 11c). Using the theboundaryshows adisplacementmovingawayfromthe crown in situ measurements, it was possible to trace the sliding surface (yellow circle in Fig. 10a). This is probably due to the fact that this (Fig. 11d) considering that the CL-PO landslide involves the upper sector is affected by a different mass movement, which induces members of the Agnone Flysch, dated as Lower Messinian displacements in the North-West direction. (Vezzani et al. 2004). The enlargement of the territory involved in the landslide by Field campaigns and in situ results continued displacement is visually recognizable through optical After the main reactivation, several in situ measurements and photographs shot from the same point of view (Fig. 12a–c). field campaigns were performed to monitor the evolution of the The enlargement of the landside area for the period of the GPS in landslide. From 2006 to 2007, eight inclinometer measurements, situ measurement (2010–2016), combined with optical photographs Fig. 8 Distribution of ERS ascending data for the period of 1992–2000 (a); regional trend, original and corrected time series of the points in the white circle (b) 2122 Landslides 15 & (2018) Fig. 9 a PS distribution in the area of interest of the CL-PO landslide. b Regional trend and original and corrected time series of a PS located on the crown of the landslide. The black ellipsis highlights where the main reactivations occurred and field surveys, was estimated to have approximately 350 m of beginning of the twentieth century. Then, focus returned in advancement at the toe and approximately 270 m of retrogression at 2003 when an important reactivation occurred. To investigate the head sector, reaching a total length up to 1500 m. The GPS field the evolution of landslide and erosional processes in the area, survey results in order to record the displacement measurements historical aerial photographs and InSAR techniques were from 2010 to 2016 are visible in Fig. 13a. During the campaign utilized. conducted in November 2015 and July 2016, not all benchmarks were Historical photographs have already been applied to the DEM found because some of them were destroyed for building the drains reconstruction using the SfM technique (Dewitte et al. 2008; or lost because of the continuous displacement affecting the slope. Ishiguro et al. 2016) along with stereophotogrammetry. This data By means of the field surveys, it was possible to design a geomor- along with SfM-MVS has been applied for the diachronic recon- phological map of the area influenced by the landslide and its struction of geomorphological landscape evolution (Gomez surroundings by conducting several geomorphological field cam- et al. 2015), active volcanic areas (Gomez 2014;Ishiguroetal. paigns (Fig. 13a) where several old and recent scarp and counterslope 2016), glaciological monitoring (Kjeldsen et al. 2015;Midgley shapes were noted. and Tonkin 2017) and erosion in river changes (Tonkin et al. In addition to the geomorphological shales, the area influenced 2016). Recently, different multitemporal landslide mapping and by the landslide is currently covered by vegetation (Fig. 13b, c), monitoring by means of the SfM approach was made, using the suggesting a reduction in the movement over the last few years due multi-sensor drone to capture the scene with RGB aerial images to the remedial work. in high-resolution (e.g. Marek et al. 2015; Peternel et al. 2017; Rossi et al. 2018). For analysing the evolution of landslides, Discussion historical aerial images were already used, but without adopting The multitechnique integration for multitemporal investigations the SfM approach (e.g. Casson et al. 2003; Guerriero et al. 2013). enabled the analysis of one century of complex CL-PO mass In this work, several previous applied approaches were merged movements. First, data were found in the literature from the to investigate as longest period as possible of the CL-PO Fig. 10 COSMO-SkyMed PS data in the area of interest of the CL-PO landslide (a). The yellow circle indicates a construction that shows an inversion of the velocity of displacement, while the white circle shows the location of the sample time series (b) corrected by the Notti et al. (2014) approach Landslides 15 & (2018) 2123 Original Paper Fig. 11 Summary of the investigation conducted with the inclinometers (a), with examples of S5 (b) and S4 (c). Section of the CL-PO landslide update with the assumed rupture surface traced with the info derived by the in situ measurements (d) landslide evolution, taking advantage from historical aerial im- Although difficulties arose during the processing due to the age ages not widely exploited by SfM approach despite its great and the state of preservation of the printed and scanned photos, potential. along with their resolution, 3D reconstructions were developed for This paper presents the multitemporal analysis applied to six each set of photographs. Furthermore, stable points recognizable sets of historical photographs from 1945 to 2003, which enabled the in every scenario for good georeferencing were difficult to find due recognition of the geomorphological evolution of the CL-PO land- to the resolution of the photographs. The main differences be- slide over approximately 60 years. The SfM technique, adapting tween the 3D model from 1954 to 2003 were identified in the the standard digital method to non-digital historical aerial photo- medium and upper portion of the mass movement where the graphs, enabled the reconstruction of precise digital models to morphology was strongly affected by the complex landslide that analyse the environmental evolution of the area of interest. even now results in a continuously slow evolution. Less relevant Fig. 12 Optical photographs in 2004 (a), 2005 (b) and 2007 (c) 2124 Landslides 15 & (2018) Fig. 13 Geomorphological maps of the landslide-prone CL-PO area and GPS displacements recorded from 2010 to 2016 (a). Optical photographs of the landslide on November 2015 (b) and July 2016 (c) modifications with enlargements were also recognized on the allowing us to see important changes, including the vegetated lateral sides and the toe of the landslide, probably due to the areas. Three matters have to be addressed: (a) the PS uphill with lateral sliding and material transported by the tributary of the respect to the crown of the landslide show, even if with small Verrino Creek. values, velocities indicating movement in this portion of the Since 1992, ERS, ENVISAT and COSMO-SkyMed PS data basin; (b) COSMO-SkyMed data show continued slow displace- record displacements with high precision where reflectors, i.e. ment, suggesting that the adopted measurements after the main outcrops or infrastructure, were combined with optical photos reactivation were not sufficient to completely stop the Landslides 15 & (2018) 2125 Original Paper advancement of the mass movement; (c) the COSMO-SkyMed in order to extract the geomorphological evolution of the PS data recorded on two buildings that were strongly damaged landslide-prone area by developing a 3D model of each scenario. located on the left flank of the landslide show a displacement The combination and analysis of the 3D reconstructions enable moving through the sensor, while in the ERS 1/2 and ENVISAT the recognition of the oldest landslide affecting the right side of monitoring period, they exhibited movements away from the the catchment, the evolution over time and the subsequent event satellites. The change in movement direction can be ascribed to that occurred in 2003 involving the left side. In 1945, a landslide the involvement of this portion of the landslide by a mass on the right flank of the basin had already occurred. From 1945 to movement affecting the closest basin, which is increasing in 1991, no important landslide evolution was identified, except dimension. Some opening fractures recognizable between two small lateral sliding with material tumbling at the foot of the portions of the building support this hypothesis. It confirms the landslide body. The benefits of the adopted measurement for the relevant impacts of landslide and erosional processes affecting right flank, recognizable by the 3D reconstructions between 1986 the whole territory of the Agnone municipality and the Molise and 1991, e.g. gabions and weirs in the river, were confirmed in region. 2003 when a reactivation landslide involved an extended area on To validate the remote sensing data interpretation, combined the left flank of the basin, which damaged several buildings and with the inclinometers and GPS measurement campaigns, several infrastructure. Then, from 1992 to 2015, PS data of ERS and field and geomorphological surveys were recorded in the landslide ENVISAT, C-band satellite, and COSMO-SkyMed, X-band con- area. Field campaigns were conducted to investigate the geology of stellation, were investigated to monitor the recent evolution. ERS the landslide, to validate the velocity of displacement, and to data exhibit a few stable PS close to the right boundary of the identify the depth of the sliding surface. Recent optical photo- landslide, showing velocities of approximately − 3.5 mm/year graphs show gibbous shapes, as well as direct and indirect field along the LOS of the satellite on structures at the left flank. The measurements conducted specifically to identify the geomorpho- ENVISAT data show important velocity measurements, even if logical evidence, confirming the continued displacement and to- they are noisy, on the buildings at the upstream portion of the pographical evolution of the upper part of the landslide and the basin, confirmed by COSMO-SkyMed data showing continued area uphill from the crown. Continued displacements and slow displacement. All the information recorded by the remote reactivations of a portion or of the total landslide suggested that sensing technique were integrated with field investigations and the designed countermeasures, i.e. drainages and reshaping, are ancillary data derived from geotechnical and geomorphological inadequate to stop the displacement. The countermeasures helped surveys, GPS campaigns, and inclinometer and pluviometer mea- to reduce the velocity of displacement and to stabilize and increase surements. The analysis of the last few decades identified the drainage in the middle-lower region of the landslide but not to portion of the landslide mainly affected by displacement and stabilize the entire region involved in the 2003 reactivation. enabled the monitoring of effects from the remedial work after The continue displacement of the landslide is confirmed also by themainreactivationin2003. Therecentdatademonstratedthat the couple of CR and benchmarks placed out of the boundary of the movements were drastically reduced, if not completely the mass movement. They exhibit a steady displacement due to the halted. continuous enlargement of the influenced area and suggest, ac- The study area has drawn the attention of the scientific com- cording to the analysis of the direct and indirect measurements munity due to the 2003 reactivation that damaged buildings and that the geomorphological evolution of the area could continue in infrastructure, with possible social and economic consequences. the future. The past development of the landslide, the continuous The extent of the damage forced the local administrator to recorded displacement by PS data and the identified geomorpho- propose restrictive measurements for some edifices. In this work, logical shapes recognized in the uphill area with respect to the a complete investigation of the past and present deformational body of the landslide are relevant indicators. Furthermore, the scenarios allowed the planning of potential additional mitigation recognized cracks in the asphalt and weak damage on infrastruc- and prevention strategies to avoid further possible reactivations. ture made to stabilize the upper region of the right flank, e.g. some In the future, the continued analysis by InSAR techniques, e.g. gabions, support the hypothesis that the area could be susceptible high-resolution COSMO-SkyMed images and field measure- to another reactivation. ments, could be useful to monitor the evolution of the landslide The application of the SfM on historical aerial images, in addi- that could involve other structures and infrastructure located tion to the combination of multisource data, can strongly improve upstream. the knowledge on the landslide-prone areas and of the possible Acknowledgements consequences in case of landslide occurrence, being very useful for This work was partially funded by the Spanish Ministry of Economy, avoiding further recurrences. Industry and Competitiveness (MINECO); the State Agency of Re- search (AEI); and the European Funds for Regional Development Conclusions (FEDER) under projects TEC2017-85244-C2-1-P and TIN2014-55413- One century of the complex CL-PO mass movement was inves- C2-2-P and the Spanish Ministry of Education, Culture and Sport tigated by integrating highly heterogeneous multisource data, under project PRX17/00439. Furthermore, the COSMO-SkyMed im- e.g. literature data, historical aerial photographs and InSAR ages were collected by an ad hoc project entitled BGround deforma- products. tion monitoring of slow-moving landslides in Agnone (Molise A literature analysis was conducted to collect information region, Italy) for building damage assessment^,carried outusing about the landslide occurring at the beginning of the twentieth CSK ® Products, © ASI (Italian Space Agency), delivered under an century. Subsequently, six sets of historical aerial photographs ASI license for use. ERS-1/2 PSInSAR data and ENVISAT PSP- from 1945 to 2003 were analysed by means of the SfM approach 2126 Landslides 15 & (2018) Di Maio C, Vassallo R, Vallario M, Pascale S, Sdao F (2010) Structure and kinematics of a DIFSAR data from the Extraordinary Plan of Environmental Remote landslide in a complex clayey formation of the Italian southern Apennines. Eng Geol Sensing (EPRS-E) were made available through the Portale 116:311–322 Cartografico Nazionale of the Italian Ministry of Environment, Ter- Doneus M, Verhoeven G, Fera M, Briese C, Kucera M, Neubauer W (2011) From deposit to ritory and Sea (MATTM) due to a specific agreement. The authors point cloud–a study of low-cost computer vision approaches for the straightforward would like to thank the editor and the three anonymous reviewers documentation of archaeological excavations. Geoinformat FCE CTU 6:81–88 Eltner A, Kaiser A, Castillo C, Rock G, Neugirg F, Abellán A (2016) Image-based surface for their valuable comments and suggestions to improve the quality reconstruction in geomorphometry–merits, limits and developments. Earth Surf Dyn of the paper. 4:359–389 Esu F (1977) Behaviour of slopes in structurally complex formations. In: Proceedings of Open Access This article is distributed under the terms of the the international symposium on the geotechnics of structurally complex formations, Creative Commons Attribution 4.0 International License (http:// pp 292–304 creativecommons.org/licenses/by/4.0/), which permits unrestrict- Farina P, Colombo D, Fumagalli A, Marks F, Moretti S (2006) Permanent scatterers for ed use, distribution, and reproduction in any medium, provided landslide investigations: outcomes from the ESA-SLAM project. Eng Geol 88:200– you give appropriate credit to the original author(s) and the Ferretti A, Prati C, Rocca F (2000) Nonlinear subsidence rate estimation using permanent source, provide a link to the Creative Commons license, and scatterers in differential SAR interferometry. IEEE Trans Geosci Remote Sens 38:2202– indicate if changes were made. Ferretti A, Prati C, Rocca F (2001) Permanent scatterers in SAR interferometry. IEEE Trans Geosci Remote Sens 39:8–20 Filocamo F, Rosskopf CM, Amato V, Cesarano M, Di Paola G (2015) The integrated References exploitation of the geological heritage: a proposal of geotourist itineraries in the Alto Molise area (Italy). Rend Online Soc Geol Ital 33:44–47. https://doi.org/10.3301/ Abellán A, Jaboyedoff M, Oppikofer T, Vilaplana J (2009) Detection of millimetric deformation using a terrestrial laser scanner: experiment and application to a rockfall ROL.2015.11 event. Nat Hazards Earth Syst Sci 9:365–372 Fisher RB, Breckon TP, Dawson-Howe K, Fitzgibbon A, Robertson C, Trucco E, Williams CK Agisoft (2016) Agisoft PhotoScan user manual professional edition, version 1.2. Agisoft (2013) Dictionary of computer vision and image processing. John Wiley & Sons LLC, St Petersburg Furukawa Y, Ponce J (2010) Accurate, dense, and robust multiview stereopsis. IEEE Trans Pattern Anal Mach Intell 32:1362–1376 Almagià R (1910) Studi geografici sulle frane in Italia. Vol. II. L'Apennino centrale e meridionale. Conclusioni generali. Società Geografica Italiana (IS) Giordan D, Allasia P, Manconi A, Baldo M, Santangelo M, Cardinali M, Corazza A, Albanese V, Lollino G, Guzzetti F (2013) Morphological and kinematic evolution of a large earthflow: Blanco-Sanchez P, Mallorquí JJ, Duque S, Monells D (2008) The coherent pixels technique (CPT): an advanced DInSAR technique for nonlinear deformation monitoring. Pure the Montaguto landslide, southern Italy. Geomorphology 187:61–79 Gomez C (2014) Digital photogrammetry and GIS-based analysis of the bio-geomorphological Appl Geophys 165:1167–1193 Calcaterra D, Di Martire D, Ramondini M, Calò F, Parise M (2008) Geotechnical analysis of evolution of Sakurajima volcano, diachronic analysis from 1947 to 2006. J Volcanol Geotherm Res 280:1–13. https://doi.org/10.1016/j.jvolgeores.2014.04.015 a complex slope movement in sedimentary successions of the southern Apennines (Molise, Italy). Landslides Eng Slopes:299–305 Gomez C, Hayakawa Y, Obanawa H (2015) A study of Japanese landscapes using structure from motion derived DSMs and DEMs based on historical aerial photo- Carrara A, Crosta G, Frattini P (2003) Geomorphological and historical data in assessing landslide hazard. Earth Surf Process Landf 28:1125–1142 graphs: new opportunities for vegetation monitoring and diachronic geomorphology. Geomorphology 242:11–20 Casson B, Delacourt C, Baratoux D, Allemand P (2003) Seventeen years of the BLa Clapière^ landslide evolution analysed from ortho-rectified aerial photographs. Eng Guerriero L, Revellino P, Coe JA, Focareta M, Grelle G, Albanese V, Corazza A, Guadagno Geol 68:123–139. https://doi.org/10.1016/S0013-7952(02)00201-6 FM (2013) Multi-temporal maps of the Montaguto earth flow in southern Italy from Cigna F, Del Ventisette C, Liguori V, Casagli N (2011) Advanced radar-interpretation of 1954 to 2010. J Maps 9(1):135–145 InSAR time series for mapping and characterization of geological processes. Nat Guida D, Iaccarino G (1991) Fasi evolutive delle frane di tipo colata nell'alta valle del F. Basento (Potenza). Studi trentini di scienze naturali. Acta geologica 68:127–152 Hazards Earth Syst Sci 11:865–881 Confuorto P, Di Martire D, Centolanza G, Iglesias R, Mallorqui JJ, Novellino A, Plank S, Guzzetti F, Cardinali M, Reichenbach P (1994) The AVI project: a bibliographical and archive inventory of landslides and floods in Italy. Environ Manag 18:623–633 Ramondini M, Thuro K, Calcaterra D (2017) Post-failure evolution analysis of a rainfall- triggered landslide by multi-temporal interferometry SAR approaches integrated with Hapke C (2005) Estimation of regional material yield from coastal landslides based on historical digital terrain modelling. Earth Surf Process Landf 30:679–697 geotechnical analysis. Remote Sens Environ 188:51–72 Costantini M, Falco S, Malvarosa F, Minati F (2008) A new method for identification and Hartley R, Zisserman A (2003) Multiple view geometry in computer vision. Cambridge University Press analysis of persistent scatterers in series of SAR images. In Geoscience and Remote Sensing Symposium, 2008. IGARSS 2008. IEEE international (Vol. 2, pp II-449). IEEE Herrera G, Mateos RM, García-Davalillo JC, Grandjean G, Poyiadji E, Maftei R, Filipciuc TC, Auflič MJ, Jež J, Podolszki L, Trigila A, Iadanza C, Raetzo H, Kociu A, Przyłucka M, Cotecchia V, Melidoro G (1974) Some principal geological aspects of the landslides of southern Italy. Bull Int Assoc Eng Geol-Bulletin de l'Association Internationale de Kułak M, Sheehy M, Pellicer XM, McKeown C, Ryan G, Kopačková V, Frei M, Kuhn D, Géologie de l'Ingénieur 9:23–32 Hermanns RL, Koulermou N, Smith CA, Engdahl M, Buxó P, Gonzalez M, Dashwood C, Crozier MJ (2010) Deciphering the effect of climate change on landslide activity: a Reeves H, Cigna F, Liščák P, Pauditš P, Mikulėnas V, Demir V, Raha M, Quental L, review. Geomorphology 124:260–267 Sandić C, Fusi B, Jensen OA (2018) Landslide databases in the geological surveys of Europe. Landslides:1–21 Cruden DM, Varnes DJ (1996) Landslides: investigation and mitigation. Chapter 3- Landslide types and processes. Transportation research board special report Hooper A, Bekaert D, Spaans K, Arıkan M (2012) Recent advances in SAR interferometry time series analysis for measuring crustal deformation. Tectonophysics 514:1–13 Dekker R (2005) SAR change detection techniques and applications. EARSeL, Munster Del Soldato M, Di Martire D, Tomas R (2016a) Comparison of different approaches for Iglesias R, Monells D, López-Martínez C, Mallorqui JJ, Fabregas X, Aguasca A (2015) Polarimetric optimization of temporal sublook coherence for DInSAR applications. IEEE landslide-induced damage assessment: the case study of Agnone (southern Italy). Rendiconti Online Societa Geologica Italiana 41:139–142 Geosci Remote Sens Lett 12:87–91 Ishiguro S, Yamano H, Oguma H (2016) Evaluation of DSMs generated from multi- Del Soldato M, Riquelme A, Tomàs R, De Vita P, Moretti S (2016b) Geomorphological investigation based on 3D historical reconstruction. In Geosciences on a changing temporal aerial photographs using emerging structure from motion–multi-view stereo technology. Geomorphology 268:64–71 planet: learning from the past, exploring the future. Società Geologica Italiana, Roma Del Soldato M, Bianchini S, Calcaterra D, De Vita P, Martire DD, Tomás R, Casagli N (2017) A James MR, Robson S (2014) Mitigating systematic error in topographic models derived from UAV and ground-based image networks. Earth Surf Process Landf 39:1413–1420 new approach for landslide-induced damage assessment. Geomat Nat Haz Risk:1–14 Dewitte O, Jasselette J-C, Cornet Y, Van Den Eeckhaut M, Collignon A, Poesen J, Kjekstad O, Highland L (2009) Economic and social impacts of landslides Demoulin A (2008) Tracking landslide displacements by multi-temporal DTMs: a Kjeldsen KK, Korsgaard NJ, Bjørk AA, Khan SA, Box JE, Funder S, Larsen NK, Bamber JL, combined aerial stereophotogrammetric and LIDAR approach in western Belgium. Colgan W, van den Broeke M, Siggaard-Andersen ML, Nuth C, Schomacker A, Eng Geol 99(1):11–22. https://doi.org/10.1016/j.enggeo.2008.02.006 Andresen CS, Willerslev E, Kjær KH (2015) Spatial and temporal distribution of mass Landslides 15 & (2018) 2127 Original Paper loss from the Greenland ice sheet since AD 1900. Nature 528(7582):396–400. https:// Solari L, Raspini F, Del Soldato M, Bianchini S, Ciampalini A, Ferrigno F, Tucci S, Casagli N doi.org/10.1038/nature16183 (2018) Satellite radar data for back-analyzing a landslide event: the Ponzano (Central Korup O, Densmore AL, Schlunegger F (2010) The role of landslides in mountain range Italy) case study. Landslides 15:1–10. https://doi.org/10.1007/s10346-018-0952-x evolution. Geomorphology 120:77–90 Sturzenegger M, Stead D, Elmo D (2011) Terrestrial remote sensing-based estimation of Lu D, Mausel P, Brondizio E, Moran E (2004) Change detection techniques. Int J Remote mean trace length, trace intensity and block size/shape. Eng Geol 119:96–111 Sens 25:2365–2401 Szeliski R (2010) Computer vision: algorithms and applications. Springer Science & Lucieer A, de Jong S, Turner D (2013) Mapping landslide displacements using structure Business Media from motion (SfM) and image correlation of multi-temporal UAV photography. Prog Tonkin TN, Midgley NG, Cook SJ, Graham DJ (2016) Ice-cored moraine degradation Phys Geogr 0309133313515293 mapped and quantified using an unmanned aerial vehicle: a case study from a Manzo M, Ricciardi GM, Casu F, Ventura G, Zeni G, Borgströmb S, Berardino P, Del Gaudio polythermal glacier in Svalbard. Geomorphology 258:1–10. https://doi.org/10.1016/ C, Lanaria R (2006) Surface deformation analysis in the Ischia Island (Italy) based on j.geomorph.2015.12.019 spaceborne radar interferometry. J Volcanol Geotherm Res 151:399–416 Turner D, Lucieer A, Watson C (2012) An automated technique for generating Marchetti M (2016) Consumo di suolo, dinamiche territoriali e servizi ecosistemici. georectified mosaics from ultra-high resolution unmanned aerial vehicle (UAV) Edizione imagery, based on structure from motion (SfM) point clouds. Remote Sens 4:1392– Marek L, Miřijovský J, Tuček P (2015) Monitoring of the shallow landslide using UAV photogrammetry and geodetic measurements. In: Lollino G. et al. (eds) Engineering Ullman S (1979) The interpretation of visual motion. Massachusetts Inst of Technology Pr geology for society and territory - volume 2. Springer Van Westen CJ, Lulie Getahun F (2003) Analyzing the evolution of the Tessina landslide Massonnet D, Feigl KL (1998) Radar interferometry and its application to changes in the using aerial photographs and digital elevation models. Geomorphology 54:77–89. Earth’s surface. Rev Geophys 36:441–500 https://doi.org/10.1016/S0169-555X(03)00057-6 Meisina C, Zucca F, Notti D, Colombo A, Cucchi A, Savio G, Giannico C, Bianchi M (2008) Verhoeven G (2011) Taking computer vision aloft–archaeological three-dimensional Geological interpretation of PSInSAR data at regional scale. Sensors 8:7469–7492 reconstructions from aerial photographs with photoscan. Archaeol Prospect 18:67–73 Midgley NG, Tonkin TN (2017) Reconstruction of former glacier surface topography from Verhoeven G, Taelman D, Vermeulen F (2012) Computer vision-based orthophoto archive oblique aerial images. Geomorphology 282:18–26. https://doi.org/10.1016/ mapping of complex archaelogical sites: the ancient quarry of Pitaranha (Portugal- j.geomorph.2017.01.008 Spain). Archaeometry 54:1114–1129 Mora O, Mallorqui JJ, Broquetas A (2003) Linear and nonlinear terrain deformation maps Vezzani L, Ghisetti F, Festa A, Follador U (2004) Carta geologica del Molise. SELCA from a reduced set of interferometric SAR images. IEEE Trans Geosci Remote Sens Westoby M, Brasington J, Glasser N, Hambrey M, Reynolds J (2012) ‘Structure-from- 41:2243–2253 Motion’ photogrammetry: a low-cost, effective tool for geoscience applications. Notti D, Herrera G, Bianchini S, Meisina C, García-Davalillo JC, Zucca F (2014) A Geomorphology 179:300–314 methodology for improving landslide PSI data analysis. Int J Remote Sens 35:2186– White RG (1991) Change detection in SAR imagery. Int J Remote Sens 12:339–360 2214. https://doi.org/10.1080/01431161.2014.889864 Notti D, Fabiano C, Cigna F, Manunta M, Herrera G, Berti M, Meisina C, Tapete D, Zucca F (2015) A user-oriented methodology for DInSAR time series analysis and inter- : : M. Del Soldato ()) S. Bianchini S. Moretti pretation: landslides and subsidence case studies. Pure Appl Geophys 172:3081– Department of Earth Sciences, University of Firenze, Peternel T, Kumelj S, Ostir K, Komac M (2017) Monitoring the Potoška planina landslide Via La Pira 4, 50121, Florence, Italy (NW Slovenia) using UAV photogrammetry and tachymetric measurements. Land- Email: email@example.com slides 14:395–406 Riquelme AJ, Abellán A, Tomás R, Jaboyedoff M (2014) A new approach for semi-automatic rock : : : M. Del Soldato D. Di Martire P. De Vita D. Calcaterra mass joints recognition from 3D point clouds. Comput Geosci 68:38–52 Department of Earth Sciences, Environment and Resources, Rossi G, Tanteri L, Tofani V, Vannocci P, Moretti S, Casagli N (2018) Multitemporal UAV Federico II University of Napoli, surveys for landslide mapping and characterization. Landslides:1–8 Complesso Universitario di Monte Sant’Angelo, Napoli, Italy Rosskopf C, Aucelli P (2007) Analisi del dissesto da frana in Molise Salvati P, Rossi M, Bianchi C, Guzzetti F (2015) Landslide risk to the population of Italy A. Riquelme R. Tomàs and its geographical and temporal variations. Extreme Events Observ Model Econ Departamento de Ingeniería Civil, Escuela Politécnica Superior, 214:177 Universidad de Alicante, Schuster RL, Highland L (2001) Socioeconomic and environmental impacts of landslides P.O. Box 99E-03080, Alicante, Spain in the western hemisphere. US Department of the Interior, US Geological Survey Snavely N, Seitz SM, Szeliski R (2008) Modeling the world from internet photo collections. Int J Comput Vis 80:189–210 2128 Landslides 15 & (2018)
Landslides – Springer Journals
Published: Jun 6, 2018
Access the full text.
Sign up today, get DeepDyve free for 14 days.