Introduction

Deep-seated rock slides affect mountain valleys worldwide by reaching thicknesses from tens to hundreds of metres (e.g. Agliardi et al. 2001; Ganerød et al. 2008; Zangerl et al. 2010, 2015; Crosta et al. 2014; Rechberger et al. 2021). They mostly occur as compound type in foliated, metamorphic rocks (i.e. gneisses, schists or phyllites) with localised deformation within a basal and internal shear zones (Agliardi et al. 2001; Hungr et al. 2014). Shear zone formation of a fully persistent sliding surface is assumed by the time-dependent fracture growth, propagation, and coalescence of existing and new fractures (Terzaghi 1962; Atkinson 1987; Eberhardt et al. 2004; Riva et al. 2018). Characteristic structural features at the surface include the development of a main scarp area with large shear offset of several tens of metres, horst- and graben-structures near the crown, minor uphill- and downhill-facing scarps, trenches, and internal shear zones separating the rock slide mass into individual slabs (Agliardi et al. 2001; Ambrosi and Crosta 2006; Crosta et al. 2014; Hungr et al. 2014; Zangerl et al. 2019; Rechberger et al. 2021).

Deep-seated rock slides can occur individually or in clusters where two or more rock slides are adjacent to each other and share sections with the same boundary or overlap with independently moving slabs (Cruden and Varnes 1996; Agliardi et al. 2012; Strauhal 2015; Rechberger et al. 2021). Clustered rock slides can be of different formation age with truncated, reshaped or new surface features and of different state of activity (Cruden and Varnes 1996; Agliardi et al. 2012; Rechberger et al. 2021). In contrast to inactive rock slides, active ones show measurable slope displacements either characterised by first-time (initial) or reactivated movements along pre-existing shear zones (Cruden and Varnes 1996). From a hazard assessment perspective, challenges arise from the difficulty in predicting the reactivation of an inactive, suspended or dormant rock slide (Ambrosi et al. 2018) and the impact on the formation or deformation behaviour of adjacent slope instabilities. As observed for the Ruinon rock slide in Lombardy (Italy), highly active deep-seated rock slides are often present in the lower parts of extremely slow large-scale mountain slope deformations (Agliardi et al. 2001, 2012; Hungr et al. 2014).

Generally, the evolution of deep-seated rock slides is temporally and spatially constrained by various factors acting on different scales, such as topography, tectonic setting, lithology and structural geology, glacier retreat, permafrost degradation or changing hydrogeological conditions due to snow melt or heavy rainfall events (Ambrosi and Crosta 2006; Crosta et al. 2013). Often, these slope instabilities are bound to active or recently active tectonic zones (Agliardi et al. 2009) or frequently occur in recently deglaciated regions (Ballantyne 2002; McColl and Davies 2013; Glueer 2019; Rechberger and Zangerl 2022). Another major contribution to the location, shape, and deformation of rock slides has been attributed to the pre-existing structural inventory by numerous case studies over the last decades (Agliardi et al. 2001, 2019; Martino et al. 2004; Brideau et al. 2009; Ambrosi and Crosta 2011; Donati et al. 2020; Jaboyedoff et al. 2013; Stead and Eberhardt 2013; Stead and Wolter 2015; Vick et al. 2020).

Usually slow to extremely slow moving over a long period of time (i.e. few millimetres to centimetres per year), deep-seated rock slides can accelerate to velocities of several meters per day (Cruden and Varnes 1996; Crosta and Agliardi 2003; Helmstetter et al. 2004; Hungr et al. 2014; Bonzanigo et al. 2007; Zangerl et al. 2010). Although most rock slides do not collapse during these acceleration phases, but instead slow down and re-stabilise again, some do collapse which can lead to devastating rock avalanches (Heim 1932; Kilburn and Petley 2003; Brideau, et al. 2005; Strouth and Eberhardt 2009; Kalenchuk 2010). In addition to a total and sudden collapse, secondary processes such as rock falls or rock avalanches can develop during these acceleration phases, significantly endangering human life and infrastructure (Willenberg et al. 2008; Zangerl et al. 2019; Rechberger et al. 2021). Therefore, rock slide characterisation by detailed field mapping and deformation monitoring is inevitable to perform comprehensive hazard assessment.

Since the beginning of the twenty-first century, surface deformation monitoring has advanced greatly due to the rapid development of sensors (e.g. laser scanning, photogrammetry, radar interferometry), the increase in computing capabilities, and software innovations that favour the acquisition of high-resolution spatial data. Compared to on-site and punctiform deformation monitoring through classical geodetic methods such as tachymetry or GNSS, advances in remote sensing technologies enable detailed, area-wide, and three-dimensional surveying of terrain surfaces (Jaboyedoff et al. 2012; Scaioni et al. 2014; Lissak et al. 2020). Imagery and laser scan data available in national archives allow to reconstruct rock slide evolution in the past (e.g. Kääb 2002; Delacourt et al. 2007; Bickel et al. 2018; Deane et al. 2020). By combining airborne laser scan (ALS) data from archives with new terrestrial laser scan (TLS) data, the time gap towards the installation of permanent monitoring systems can be closed (Fey 2018).

This study presents the analyses of various types of remote sensing campaigns and geological investigations focusing on the evolution of an active, deep-seated rock slide on top of the Tellakopf/Cima di Tella mountain slope deformation in South Tyrol, Italy and was prepared as part of a doctoral thesis (Rechberger 2022). The rock slide was first noticed in January 2014 after increased rock fall activity, which caused a temporal closure and relocation of the commuter-rich SS41 national road leading to the village of Val Müstair in Switzerland.

We show results of a more than 5-year terrestrial laser scanning (TLS) campaign (2017–2021) combined with over 75 years of remote sensing data from orthoimages and ALS campaigns, in order to better understand the underlying failure and deformation mechanisms as a basis for assessing potential rock slide hazard scenarios. In this context, our research objectives aim (i) to analyse the constraints of the temporal-spatial evolution of the rock slide, (ii) to determine the temporal-variable deformation behaviour and characteristics, i.e. acceleration and deceleration phases, slab formation, and secondary processes, (iii) to define rock slide geometry, and thus, potential failure volume, and whether a basal and internal shear zones have developed, (iv) to identify formation-relevant structural features, i.e. pre-existing discontinuities, and (v) to develop a geological-kinematical rock slide model for subsequent numerical modelling studies regarding rock slide hazard analysis. Field mapping was performed to record undeformed bedrock regarding lithology, structural features, and degree of weathering. A topographic analysis was carried out to document and identify slope changes over time, i.e. the formation of rock slide structures such as scarps, tensile fractures, erosion, and deposition of rock scree. In order to quantitatively determine and analyse the temporal-variable deformation behaviour, slab formation and surface changes, multi-temporal remote sensing data consisting of orthoimages and airborne laser scans (ALS) were analysed together with terrestrial laser scan data (TLS) from an extensive measurement campaign starting in 2017. Through the development of a reliable rock slide model representing the geology, geometry, and kinematics, this research improves the process understanding of deep-seated rock slides in foliated, metamorphic rocks.

Study site

The active Tellakopf (ger.)/Cima di Tella (ita.) rock slide is located on the northern valley flank near the entrance to the Münsteral/Monastero valley, a tributary valley of the Vinschgau/Venosta valley in western South Tyrol, Italy (Laatsch: 46°40′33.85"N, 10°29′50.11"E). The rock slide covers an area of 0.16 km2 of the SE-facing slope of the Tellakopf summit, which is affected by large-scale gravitational slope deformations (i.e. mountain slope deformation, Hungr et al. 2014) and hosts several rock slides (Fig. 1). The active Tellakopf rock slide is located directly above national road SS 41, an important transport connection between the Vinschgau valley (Italy) and canton Graubünden (Switzerland). It extends 400 m in the NE-SW direction and spans 650 m from 1450 m a.s.l. to the main scarp at 2015 m a.s.l., showing a mean slope inclination of approximately 35° (min. 20°, max. 80°).

Fig. 1
figure 1

Topographic map of the Tellakopf/Cima di Tella rock slides and its immediate surroundings in the Münsteral valley of South Tyrol, Italy (Orthoimage from Autonomous Province of Bozen/Bolzano; EPSG: 25832)

Full size image

Geologically, the study site is located within the poly-metamorphic units of the Sesvenna (S-charl) Crystalline of the Upper Austroalpine units between the Engadine and Giudicarie Fault (Schmid et al. 2004; Mair et al. 2007). It borders the Ortler-Campo Crystalline in the E and the Ötztal-Stubai Crystalline in the N, separated by Early Createous cataclasites and mylonites of the Vinschgau shear zone (VSZ) which in this area is delineated by the Schlinig Fault in the western and northern part, the Gallo Line and the Trupchun-Braulio Thrust in the southern and southeastern part, respectively (Fig. 2) (Schmid and Haas 1989; Froitzheim et al. 1997; Mair et al. 2007). At the study site, poly-metamorphic rocks such as orthogneisses are predominant, with intercalations of amphibolites, phyllites, paragneisses, and marbles. In addition, tectonically formed phyllonites can be found at the entrance to the SE-facing slope of the Münstertal valley which can be attributed to the Vinschgau shear zone (Hammer 1912; Schmid 1973; Conti et al. 1994; Froitzheim et al. 1994; Conti 1995; Sennhauser 2007). These prevailing metamorphic rocks are characterised by a penetrative schistosity (Schmid 1973), and a brittle-ductile tectonic deformation forming conjugate sets of dextral NNW- and sinistral NNE-striking strike-slip faults, and N-striking normal dip-slip faults (Conti et al. 1994; Froitzheim et al. 1994; Conti 1995; Agliardi et al. 2009). The Vinschgau and Münstertal area is characterised by moderate uplift rates (approximately 1–1.2 mm/year) and shallow micro-seismicity (< 2 km) with numerous small magnitude earthquakes (Agliardi et al. 2009).

Fig. 2
figure 2

Regional tectonic map of the study site (modified after Froitzheim et al. 1994)

Full size image

During the Last Glacial Maximum (LGM), glacial ice covered the study area up to 2500 to 2600 m a.s.l., reaching ice thicknesses of at least 1200 m (Kelly et al. 2004; Agliardi et al. 2009). During this time, moraine material was deposited, which is largely overlain with quaternary and recent sediments (Bach 1976; van Husen 2004). In the latter, it is mainly fluviatile sediments from the Rambach river and alluvial fans from brooklets or talus cones from lateral ditches that developed after the last glacial cover (Sennhauser 2007).

At the Tellakopf rock slide area from 2011 onwards, increased slope deformations causing rock fall events with volumes of 5–15 m3 were observed, mainly detaching from rock faces between 1620 and 1820 m a.s.l. (Alpin Geologie 2011; Autonomous Province of Bozen 2011). Although most of the rock boulders stopped at the slope foot, some reached the national road SS 41. In 2014, a major rock fall event of 2000 m3 originated from rock faces at approximately 1700 m a.s.l. In addition, the development of a minor scarp reaching approximately 10–20 m shear offset was observed between 1550 and 1950 m a.s.l., indicating the onset of a deep-seated rock slide instability (Alpin Geologie 2014; Autonomous Province of Bozen 2014). As a consequence, the national road SS41 was closed due to safety issues and subsequently relocated to the orographic right valley side (Fig. 1). Additionally, hiking trails and roads in areas endangered by rock fall processes were closed.

Materials and methods

Geomorphological-geological field mapping

A geomorphological-geological field mapping campaign was carried out at the SE-facing slope between 2017 and 2020 based on orthoimages, and shaded relief maps derived from the 2006 and 2014 Digital Elevation Models (DEMs) on a scale of 1:1000 (Table 1).

Table 1 Available remote sensing data used to reconstruct the Tellakopf rock slide
Full size table

In order to develop an integrated geological-kinematical model of the rock slide, the mapping focused on identifying (i) rock slide boundaries, i.e. main scarps and lateral flanks, (ii) internal rock slide structures, i.e. minor scarps, uphill- and downhill-facing scarps, trenches, and tensile fractures (cf. Rechberger et al. 2021), and (iii) the lithological and structural-geological inventory, i.e. different lithologies and discontinuities. Therefore, the undeformed rock and soil cover was mapped area-wide according to the lithology, structural features, and degree of rock mass weathering (ISRM 1978). By mapping the shear offset on scarps, i.e. measured parallel to the scarp surface (Davis et al. 2011) and tensile fractures, deformation features, such as slab formation, area of extension and compression of the rock slide can be reconstructed. The structural-geological mapping focused on different scales, i.e. foliation planes, meso-scale fractures, and fault planes and -zones. Subsequent stereographical analyses using DIPS software (Rocscience Inc. 2020) were performed to identify structural patterns and potential kinematic failure modes according to Hudson and Harrison (2000). The in-situ petrographical and mineralogical composition was determined by X-ray and thin-section analyses using a PANanalytical X’pert PRO X-ray diffractometer and a Leica DM4500P polarization microscope, respectively. Hydrogeological investigations by collecting data on the location, type, and discharge of springs and streams were conducted at the entire SE-facing slope.

Deformation monitoring

Orthoimages and ALS data from archives provided by the Autonomous Province of Bozen and Cartorender GmbH (Table 1) were used to reconstruct the historic slope deformations since 1945. Deformation monitoring was performed when the slope activity increased in 2014 and comprises (i) a helicopter-based photogrammetric survey in 2014 and (ii) eight terrestrial laser scan (TLS) campaigns between 2017 and 2021. Analysis of the remote sensing data allows to (i) reconstruct the temporal evolution of the rock slide, (ii) identify temporal-variable deformation characteristics, and (iii) monitor the current rock slide activity relevant for risk mitigation methods.

Data acquisition and data pre-processing

The orthoimages and ALS were available in federal archives and not acquired with the aim to monitor the rock slide. Thus, some pre-processing is necessary before data analyses. In a first step, all data was transformed into the global coordinate system (UTM 32N ETRS89, EPSG: 25832) and resampled into a common reference frame. The orthoimages were checked for bias and corrected by a translational shift using an image correlation algorithm (Fey et al. 2015). The ALS point cloud from 2006 was cleared of vegetation by applying a bare earth filter implemented in the SAGA GIS add-on LIS Tools (Laserdata GmbH 2018). The filtered point cloud was then generated to a digital elevation model (DEM) with a raster resolution of 0.5 m.

In July 2014, a helicopter-based SfM-MVS (structure-from-motion multi-view-stereophotogrammetry) campaign was performed by Cartorender GmbH with the aim of monitoring the rock slide using a Canon EF camera and the Sierrasoft Geomatics ProSt software. 17 ground control points (GCPs), measured by a total station, were used for geo-referencing. The data was available as digital surface model (DSM). As vegetation causes high uncertainties in photogrammetric data, vegetated areas were excluded from analyses. The point cloud of the data set was not available for this study.

The TLS campaigns were performed with a Riegl VZ4000. In 2017, the first TLS survey was performed from three scan positions on the valley floor. Since 2018, the survey was enhanced by two more scan positions on the counter slope and two scan positions located east of the rock slide (Fig. 1). The distances between scanner positions and scanned slopes range from 400 to 1800 m. The TLS point clouds were registered and geo-referenced using a relative registration approach (Fey 2018). To achieve a coarse alignment into the global coordinate system (UTM 32N ETRS89, EPSG: 25832), the stable slope areas (i.e. not modified by surface processes and vegetation change) of the TLS point cloud from 2017 were matched onto the ALS point cloud from 2006 by applying an iterative closest point (ICP) algorithm implemented in RiScan Pro (Riegl 2013). Successive TLS points from 2018 to 2021 were then matched onto the geo-referenced point clouds from the previous year. The use of ICP-based registration has the advantage that no target objects are required, and thus, (i) the highly active, hazardous areas do not have to be entered for the target installation, and (ii) the data acquisition is simplified and accelerated by the omission of target scanning and measuring. The registered TLS point clouds were then tilted, the normal vectors were calculated, and the vegetation was filtered out. Based on the bare earth points excluding vegetation, a digital elevation model (DEM) with a raster resolution of 0.5 m was generated.

Change detection analyses

The reconstruction of the scarp development on historic orthoimages was carried out using change detection analyses. This was done by visual identification of the structural features, i.e. main and minor scarps, lateral flanks, trenches, and tensile fractures. These linear features were digitalised and the offset between two data sets were measured.

Algorithm-based displacement analyses

Image correlation-based displacement analyses were used to identify the length, direction and dip of the slope displacement. The pre-processed laser scan DEMs were analysed using the image correlation algorithm implemented in the open-source software SAGA GIS (Scambos et al. 1992; Bremer 2012; Fey et al. 2015). The correlation is based on corresponding features detected by an algorithm searching for similar surface textures on two images from different epochs (Fey et al. 2015, 2017a, b). It can be applied on (i) orthoimages or (ii) laser scan data using a shaded relief raster derived from a DEM.

For the application of image correlation on laser scan data, the DEMs were processed to shaded relief images using the non-direct lighting method ambient occlusion (Tarini et al. 2006). This procedure ensures that all objects are displayed independently from the lighting direction and are not obscured by shadows (Fey et al. 2015). Length and direction of displaced features are calculated based on a search window with a user defined size of 128 × 128 pixels (64 m on a raster with a resolution of 0.5 m). A vector is calculated between the centre of the search window (grid point) of the source image and the peak correlation in the target image. Then, a horizontal displacement vector is calculated for each grid point, whose resulting horizontal displacement lengths are aggregated to a displacement raster with 2 m resolution (Fey and Krainer 2020). Additionally, for image correlation based on laser scan data, the total displacement can be determined by considering the DEM elevation of the start and end point of the vector.

A prerequisite for the correlation is the preservation of the same surface texture in both images. No or wrong correlations occur when textures change significantly from one image to the next or are even destroyed, e.g. due to large slope deformations, rock fall events or erosional processes. Additionally, strong shadows in orthoimages and a high surface roughness due to vegetation cause miscorrelations. However, such miscorrelations can be detected by analysing displacement vector patterns within a defined search window. These are characterised by vectors pointing in all directions with strongly varying lengths (Fey et al. 2015, 2017a, b; Fey and Krainer 2020). Further, miscorrelations can also be identified by using DEMs of difference (DoDs), as in areas where the surface texture is destroyed, the elevation change shows high negative and positive values.

In order to distinguish between real terrain changes and measurement uncertainties resulting from data errors, a level of detection (LoD) was derived. Theoretically, the image correlation algorithm can achieve subpixel accuracy (Chang et al. 2006; Rosu et al. 2014). Dealing with data of non-laboratory conditions, the final accuracy is influenced by alignment quality of the input data pair and the surface roughness. When using orthoimages as input data, the geometric errors further influence the accuracy. The alignment quality of a data pair was evaluated by assessing the relative accuracy, i.e. the deviation of two data sets within unchanged areas (Fey 2018). The LoD was determined for each pair of data using the standard deviation of the horizontal displacements measured in areas without terrain changes (Table 3). Displacement values below the LoD and displacement values affected by miscorrelations were not used for further analyses.

Elevation changes and volume quantification

A DEM of differences (DoD) was calculated to analyse the elevation change between the ALS DEM, SfM-MSV DSM, and the TLS DEM.

To analyse the high-resolution TLS point cloud data, a point cloud-based 3D distance measurement approach (Fey and Wichmann 2017) was used. In contrast to the elevation difference from raster DEMs (DoDs), the 3D distance approach measures distances in 3D in the direction of the normal of a tangent plane fitted to the surface, and thus, not necessarily in the vertical direction (Fey and Wichmann 2017). The approach has the advantage that even overhanging areas and very steep terrain can be processed. Volume calculations of rock fall at steep rock walls were carried out by rotating the point cloud with the 3D distance information to a horizontal reference plane and the 3D distance values were triangulated to a raster. The volume was aggregated by summing up the raster values of the generated 3D distance raster.

Results

Geomorphological-geological and structural characterisation

The rock mass at the SE-facing slope of the Tellakopf ridge is composed of mica- and albite-rich, zoned mylonitic and quarzitic gneisses of igneous origin (Fig. 3). A common feature is the absence of biotite and unlike some compact orthogneisses, most show penetrative schistosity caused by the sub-parallel alignment of mica grains. Clasts and porphyroblasts (quartz, feldspar) of up to several centimetres in diameter are intercalated. Tectonically formed phyllonite zones are mapped at approximately 2050 m a.s.l. These moderately to highly weathered mylonitic rocks are largely composed of fine-grained mica, forming a penetrative cleavage (Passchier and Trouw 2005). Deformed quarzitic and feldspar grains indicate significant shearing. Talus and rock scree material can be observed beneath steep rock surfaces. Blocky to very blocky scree material from secondary rock fall processes has accumulated beneath the rock slide down to approximately 1500 m a.s.l. Glacial till deposits cover large areas across the slope and indicate a glacial overprint of the Tellakopf ridge. Fluviatile sediments from the Rambach river and alluvial fans or talus cones mark the valley floor area.

Fig. 3
figure 3

Geological, geomorphological and hydrogeological setting illustrating rock type, structural geological and geomorphological features of the Tellakopf / Cima di Tella rock slides (DEM 2006 from Autonomous Province of Bozen; EPSG: 25832)

Full size image

Springs or wetness zones are occasionally observed at the SE-facing slope of the Tellakopf ridge, specifically in the eastern area of the dormant rock slides (Fig. 3). There, small variable spring discharges temporarily supply some of the channels with water, causing soil waterlogging in the close range. These spring discharges were particularly noticed after rainfall events, indicating strong influence of surface water. Accordingly, the absence of springs in the upper slope area suggests a rather low groundwater level in the study area, probably due to the increased fracturing and loosening of the rock mass caused by tectonic imprinting and gravitational unloading.

Structurally, the rock units show a main NW-SE striking, flat into the slope dipping foliation (s1) (mean dip direction/dip angle: 53°/14°; Table 2; Figs. 3, 4a, 5f). The meso-scale fracture network is characterised by two nearly orthogonally orientated sets which are steeply dipping. Set f1 (203°/73°) is striking ESE-WNW and set f2 (295°/85°) NNE-SSW, respectively (Figs. 4b; 5f). Brittle fault zones were rarely mapped in the field, but lineaments striking either NNE-SSW or NNW-SSE are identified on shaded relief maps, corresponding to the local tectonic setting according to Agliardi et al. (2009). Based on the measured structural data, kinematic analysis was carried out considering a mean slope orientation of 135°/30°, 135°/50°, and 135°/70°, representing the active rock slide area from the head to the slope foot at approximately 1500 m a.s.l. (Figs. 4c, 5f). A friction angle of 30° was applied based on the assumption of a basal friction angle of 27° and a dilatancy for failure along mica-rich gneissic surfaces (Barton and Choubey 1976; Norrish and Wyllie 1996). Analysis show that foliation planes (s1) may act as basal surfaces for directing toppling of block columns formed by the fracture sets #1 (f1) and #2 (f2). Considering the pole density within the critical toppling envelope, flexural toppling is favoured by fracture set #2. Planar and wedge sliding was not determined.

Table 2 Mean orientation of foliation planes and of the meso-scale fractures within the rock slide system (see Fig. 4a, b)
Full size table
Fig. 4
figure 4

Hemispherical projection (lower hemisphere, equal angle) of the mapped structural field data within the Tellakopf rock slide (Table 2). Contoured pole density plots of a foliation planes, and b meso-scale fractures. c Kinematic analyses for planar sliding and flexural toppling considering foliation and fracture sets. Critical pole vector zones and daylight envelopes are shown with slope orientations of 135°/30°,135°/50° and 135°/70° and a friction angle of 30°

Full size image
Fig. 5
figure 5

Geomorphological features of the active Tellakopf rock slide: a Initial stage of rock slide formation in January 2014; b Rock slide situation in late 2018 characterised by large offsets and slab separation; c Minor scarp in the initial formation stage in 2014; d Main scarp area in 2020 with large shear displacements of up to 70 m; e Lateral flank and displaced rock mass at the NE boundary; f Minor scarp with 15 m in offset separating the rock slide mass. The rock mass is transected by two meso-scale fractures (f1, f2) and a low-dipping foliation (s1) into the slope; g Minor scarp and trench formation beneath; and h highly fractured, and loosened rock mass at the slope foot prone to rock fall events

Full size image

The Tellakopf rock slide developed adjacent to a most-likely dormant rock slide on top of a large-scale gravitational slope deformation affecting the entire mountain ridge (Fig. 3). Typical structural features such as scarps, counter scarps, and trenches, are mapped behind a prominent main scarp extending along the ridge top from the NW at 2100 m a.s.l. to the SE at 1500 m a.s.l., indicating the formation of a mountain slope deformation (Hungr et al. 2014). Beneath, rock slides with main scarps altered by weathering and a displaced mass overgrown with trees characterise the eastern slope area. As no movements are detected with our measurement methods, a dormant activity state is assumed.

To the west, the active Tellakopf rock slide developed with a crescent-shaped main scarp area between 1550 and 2015 m a.s.l. (Figs. 3, 5a, b). Displacements along the main scarp range up to approximately 74 m (Fig. 5d). Well-defined left and right lateral flanks separate the rock slide mass from the stable slope. Here, offsets between 10 and 50 m were measured (Figs. 3, 5e). Within the sliding mass, NE-SW striking minor scarps with 10–20 m high exposed rock surfaces developed (Figs. 3, 5f–h), which separate the rock slide mass into individual slabs. However, with respect to the distinct minor scarp at 1925 ma a.s.l. which was first recognised in January 2014 reaching shear offsets in the range between 10 and 20 m (Fig. 5c), displacements increased up to 77 m in offset since 2014. In total, the formation of seven different rock slide slabs (slabs I–VII) can be observed (Fig. 6f), separated by main and minor scarps along which shear displacement accumulated. Minor scarps preferably evolved along fractures of set#2 (f2) acting as predefined weakness zones. At the foot of these minor scarps, up to one meter deep trenches were observed (Fig. 5g). Each of the individual slabs is composed of a highly fractured, disturbed, loosened rock mass, separated by tensile fractures or trenches (Figs. 3, 5h).

Fig. 6
figure 6

Temporal and spatial rock slide evolution at the SE-facing slope at the Tellakopf site: a 1945, b 1985, c 2000, d 2011, e 2014, and f 2020, showing surface changes, and the development of scarps and individual slabs. (Orthoimages from Autonomous Province of Bozen, EPSG: 25832)

Full size image

Initial rock slide formation process

Surface deformation features (i.e. main and minor scarps) identified by orthoimagery illustrate the spatio-temporal evolution of the Tellakopf rock slide between 1945 and 2020 (Fig. 6).

The 1945 orthoimage shows no newly formed visible structural features across the entire slope, suggesting stable slope conditions at the time of acquisition (Fig. 6a). However, an inactive or dormant rock slide is suspected in the eastern slope area, characterised by a tree-covered, depression-like terrain that bulges slightly towards the valley floor between 1500 and 1000 m a.s.l.

In 1985, first traces of the formation of a left lateral flank between 1800 and 2050 m a.s.l. and the formation of two minor scarps at 1800 m a.s.l. in the upper part of the western slope area (Fig. 6b) indicate beginning slope deformations. In addition, rock fall activity originating from these minor scarps can be identified due to the accumulation of fresh rock scree deposits directly beneath and down to approximately 1600 m a.s.l.

The orthoimage taken in the year 2000 indicates that the left lateral flank and the minor scarps have slightly extended, as well as that a single minor scarp has formed at approximately 1925 m a.s.l. (Fig. 6c). The successive expansion of rock scree deposit areas suggests increasing rock fall activity.

The 2011 orthoimage shows the development of new minor scarps between 2000 and 2050 m a.s.l. of which some merge into the left lateral flank (Fig. 6d). Increased slope deformations are suspected as numerous minor scarps have formed between 1900 and 1550 m a.s.l. with accumulation of rock scree deposits below. This agrees with the increased rock fall activity evident from field observations (Alpin Geologie 2011; Autonomous Province of Bozen 2011), where rock falls of 5–15 m3 in volume were noted, with block fragments reaching the national road SS41.

The orthoimage of the year 2014 depicts accumulated slope deformations and terrain changes until July 2014. The development of a crescent-shaped main scarp with distinct left and right lateral flanks and large shear displacements is observed between 1620 and 2015 m a.s.l., indicating clearly that the rock slide has formed (Fig. 6e). A minor back scarp is mapped at 2050 m a.s.l., while two distinct minor scarps separate the displaced rock slide mass at 1925 m a.s.l. (Fig. 6e) and 1620 m a.s.l., respectively. Traces of two smaller minor scarps are observed within the separated mass. At the foot of the newly formed rock slide at an elevation of about 1620 m a.s.l., immense rock scree accumulation down to 1300 m a.s.l. occurs due to increased rock fall activity.

The initial formation of the rock slide associated with increased slope displacements can be dated to January 2014, as first traces of the prominent minor scarp were observed on 13th of January 2014 (Alpin Geologie 2014; Autonomous Province of Bozen 2014). On this day at 11:40 a.m., a 2000 m3 large rock fall event detached from the rock face at the foot of the slide at about 1700 m a.s.l., nearly simultaneously with the formation of the minor scarp at 1925 m a.s.l., reaching an offset of 10–20 m (Fig. 5a, c). Between January and July 2014, ongoing slope displacements led to the aforementioned structural deformation features in Fig. 6e.

The 2020 orthoimage indicates further shear displacement at the main scarp and particularly at the minor scarp at 1925 m a.s.l. (Fig. 6f). Additional minor scarps have formed between 1700 and 1925 m a.s.l. separating the rock slide into slabs. Presently, the evolution of seven individual slabs (slabs I-VII) could be proven based on topographical analysis. Tensile fractures at some of these slabs indicate zones of extensional strain of the rock slide mass. New and slightly enlarged rock scree deposits beneath the rock faces at approximately 1620 m a.s.l. indicate ongoing rock fall activity during the period 2014 and 2020.

The temporal-variable deformation characteristics

The onset of rock slide formation until early 2014 could be detected by the visual identification of the structural features on orthoimages (Fig. 6). To quantify changes and to analyse the temporal-variable deformation characteristics, laser scan and photogrammetric data between 2006 and 2021 were analysed by (i) DEMs of difference (DoD), and (ii) image correlation-based slope displacement analyses.

The DoD of the ALS 2006 and TLS 2021 pair illustrates surface elevation changes between the pre- and post-failure terrain of the active Tellakopf rock slide (Fig. 7). Between 2006 and 2021, areas of terrain subsidence (i.e. negative elevation change) are primarily below the main scarp, while areas of terrain uplift (i.e. positive elevation change) mark the lower part of the slope. Slabs I-IV are entirely affected by subsidence. While the upper slabs I and II only show minor subsidence of -1 to -2.5 m, large values of up to -20 m are obtained for the area of slab III, which coincides with the large shear displacements along the internal scarp between slab III to slabs I and II (Figs. 5d, e, 7). Maximum subsidence of < -30 m is reached below the minor scarp of slab IV at 1900 m a.s.l., reflecting large shear offset and the separation into rock slide slabs of different thickness (Figs. 5f, 7). Within slab V, rock slide deformation changes from subsidence in the upper part to uplift and indicates the area where the rock slide mass starts to move out of the slope and rock scree accumulates. The area below 1800 m a.s.l. is mainly characterised by an uplift reaching maximum values of up to +50 m in the central slope area (Fig. 7). While these positive elevation changes within slab VI are mainly attributed to displacement out of slope, the uplift of the area below the steep rock face is related to large rock scree accumulations (Figs. 6e, f, 7). In terms of the small slab VII, the area is affected by both, sliding of debris material and multiple rock fall events leading to negative elevation changes in the source area and positive ones below due to rock scree accumulation.

Fig. 7
figure 7

DEM of difference obtained between ALS 2006 and TLS 2021, illustrating rock slide mass balance and kinematics. Rock slide slabs I-VII are shown based on structural features (minor scarps). (DEM 2006 from Autonomous Province of Bozen, EPSG: 25832)

Full size image

The results from image correlation-based displacement analyses provides detailed information on the temporal-variable rock slide deformation behaviour from 2014 to 2021 (Fig. 8). The results show the same trend of slope deformation towards SE, slipping over the adjacent lower situated dormant rock slide (see Fig. 3).

Fig. 8
figure 8

Image correlation-based displacement analyses showing the mean annual velocity [m/year] for the period a ALS 06/2006–SfM-MSV 07/2014, b SfM-MSV 07/2014–TLS 10/2017, c TLS 10/2017–TLS 10/2018, d TLS 10/2018–TLS 10/2019, e TLS 10/2019–TLS 10/2020, and f TLS 10/2020–TLS 10/2021. Control points are illustrated in (a). (2020 Orthoimage from Autonomous Province of Bozen, EPSG: 25832)

Full size image

In contrast to the displacement direction, the displacement length shows a temporal and spatial variability. To compare the rock slide displacements of ALS, SfM-MSV, and TLS data from the different periods (Table 3, Fig. 8), the mean annual velocity [m/year] was calculated by dividing the absolute displacement length by the number of days of the period. Given that continuous displacement measurements are not available, we calculated the average for the respective time periods. Figure 8 shows the displacement excluding displacements smaller than the LoD (Table 3) and miscorrelations. The displacement values in Table 4 are the image correlation results aggregated on selected control points.

Table 3 Level of detection (LoD) of image correlation-based displacement analysis (Fey 2018)
Full size table
Table 4 Accumulated total displacements [m] and annual velocities [m/year] of control points of slabs I, and III-VI within the measuring period 06/2006 and 10/2021
Full size table

Between 2006 and 2014 (Fig. 8a), slab III is affected by miscorrelation due to the large displacements and strong change of the surface structure. Here, the image correlation algorithm is not able to find surface patterns with a similar surface structure in both raster images. However, between 2006 and 2014, mean annual velocities of up to 6 m/year can be observed for the lower slope area (slabs IV–VI), while the rock slide top (slabs I–II) shows velocities of < 1 m/year (Fig. 8a; Table 4). Due to scan shadows in the TLS data set of 2017, no information for the period 2014–2017 is available from the upper slope parts (Fig. 8a). Therefore, an additional image correlation based on the SfM-MSV DEM 2014 and TLS DEM 2018 data pair was performed to gain information on the upper slope which is shown in Table 4. During the period from 2014 to 2017/2018, slope deformation slightly decreased for slabs IV, V, and VI, characterised by annual mean velocities of 1 m/year near the lateral flanks and up to 5 m/year in the central rock slide area (Fig. 8b). Although a general slowdown to velocities of 2–3 m/year is also observed between 2017 and 2018, some areas of slabs III and VII show still high annual velocities of up to 6 m/year and 5 m/year, respectively (Fig. 8c). This trend is also recognised for the periods 2018–2019 and 2019–2020 (Fig. 8d, e, Table 4). Mean annual velocities of 1–2 m/year characterise slope deformation with areas of higher velocities of up to 4 m/year (slab III) and 3 m/year (slab VII). Between 2020 and 2021, rock slide deformation further decreases, characterised by mean annual velocities of 0.5–1 m/year at the main scarp area, near the lateral flanks and for slab V (Fig. 8f). The central areas of slabs II, III, VI, and VII show slightly higher velocities of 1–2 m/year. As of the TLS 2018–2019 pair, velocities of < 0.5 m/year are obtained below slabs VI and VII from 1600 down to 1100 m a.s.l., indicating increased rockfall activity.

The accumulated displacements between 2006 and 2021 are shown in Fig. 9 and Table 4. While slab I shows significantly smaller displacements of up to 16 m between 2006 and 2021, slabs III to VI reach total max. displacements in the range between 71 and 77 m. Within these relatively uniformly moving slabs III to VI, it can be observed that deformation increases slightly towards the central part of the slope (i.e. slab IV) and then decreases towards the rock slide foot (cf. Figs. 7, 8, Table 4). After the initial rock slide formation phase in 2014, in which total displacements between 41 and 54 m were reached, the rock slide mass experiences a continuous velocity reduction until 2021. Between the ALS 2006 and the SfM-MSV 2014, a total displacement of 54 m was measured. Assuming that large slope displacements started in January 2014 and lasted until the acquisition date of the SfM-MSV DSM July 2014 (Fig. 6a, e), a mean velocity of up to 9 m/month can be derived. Most probable the displacement velocity was not continuous within the onset phase, but since we have no monitoring data in between, this cannot be proven. The subsequent deceleration phase is reflected in a successive decrease of the mean annual velocities of slabs III to VI from 4.4–6.0 m/year to 1.4–6.0 m/year for the period 2014–2018, to 0.5–3.3 m/year for the period 2018–2020, and to 0.7–2.5 m/year for the period 2020–2021. Remarkably, a slight increase in velocity is noted for slab I since 2018.

Fig. 9
figure 9

Total displacements [m] of selected control points within slabs I, III, IV, V and VI from 06/2006 (reference survey) to 10/2021 derived by image correlation. The location of control points is shown in Fig. 8a

Full size image

Secondary processes triggered by rock slide deformation

In order to identify secondary landslide processes such as rock falls, 3D distance change analyses were carried out for the TLS point clouds between 2017 and 2021. Increased rock fall activity was determined for steep, overhanging rock faces in the lower slope area.

Figure 10 shows the 3D distance change map (i.e. mass change) derived from the TLS measurements between 2017 and 2021, identifying four zones with minor scarps (niches) at an elevation between 1900 and 1500 m a.s.l. Areas with negative mass changes are identified as rock fall source areas, while areas with positive mass changes show rock scree accumulation. Further, the sliding processes cause negative values in areas with subsidence and positive values in areas where the rock slide moves out of the slope. Areas with vegetation lead to errors at distance measurement and need to be excluded from analyses. Between 2017 and 2018 (Fig. 10a), no larger rockfall events can be identified. The positive distance change of 1.5–3 m is consistent with the results of the image correlation-based displacement analysis and shows the rock slide movement out of the slope (cf. Fig. 8c). The rock fall activity increased from 2018 onwards, reaching a total failure volume of 24,000 m3 until 2021 (Table 5). For the period 2018–2019 and 2019–2020, high rock fall activity occurs at the central slope foot (i.e. domain 1; Fig. 10b) with volumes of up to 9000 m3 and 6000 m3, respectively. Between 2019 and 2020, increased rock fall is also observed in the eastern (i.e. domain 2) and western slope foot area (i.e. domain 3) with 2500 m3 and 3200 m3 in volume, respectively (Fig. 10c). From 2020 to 2021, reduced rock fall activity was recorded from rock faces in the central (i.e. domain 1) and western slope foot area (i.e. domain 2), characterised by a volume of 400 m3 and 1400 m3, respectively (Fig. 10d). Increased activity was recorded in the lower central area (i.e. domain 4) of 1500 m3 in volume. No rock fall was observed in the east (i.e. domain 3).

Fig. 10
figure 10

3D distance change (mass change) derived from TLS measurements for the period a 10/2017–10/2018, b 10/2018–10/2019, c 10/2019–10/2020, and d 10/2020–10/2021, showing four areas with minor scarps (niches) of high rock fall activity at an elevation between 1900 and 1500 m a.s.l. (2020 Orthoimage from Autonomous Province of Bozen, EPSG: 25832)

Full size image
Table 5 Volume estimation [in m3] of rock fall events at the lower rock slide area (slab IV-VII) between 2018 and 2021. The domains with the niches of the rock fall events are shown in Fig. 10
Full size table

Geological-kinematical model

The reconstruction of a geological-kinematical rock slide model provides detailed information on type of movement, geometry of the rock slide, i.e. slab thickness and volume, and whether internal and basal shear zones have developed. In order to present the rock slide evolution with its changing geometry, two rock slide models were derived along the NE-SW directed longitudinal section A-B-C orientated in the direction of movement (Figs. 3, 11), which represent (i) the slope situation in 2006 before initial rock slide formation (A–C), and (ii) the slope situation in 2021 after the rock slide evolved and large-scale displacement accumulated (A-B).

Fig. 11
figure 11

NW-SE directed longitudinal sections showing rock slide geometry, geological setting, kinematics, and the dip of the total displacement vectors of survey points: a along the profile A-C (see Fig. 3) displaying the situation in 2006 before the formation, b along the profile A-B (see Fig. 3) displaying the situation in 2021. The kinematic behaviour of the active rock slide is mainly determined by discrete shear offsets along discrete discontinuities, i.e. main scarp, and minor scarps dipping either downhill or uphill

Full size image

In Fig. 11a, the 2D model extends from the mountain crest down to the valley floor of the Münstertal valley and depicts the location of the active Tellakopf rock slide before its initial formation. At this stage, only the much larger, most likely dormant rock slide at the lower section of the flank was formed, characterised by a bulging slope topography. Gneisses of different composition dipping flat into the slope mark the area down to the valley floor. There, valley infillings of approximately up to 200 m in thickness mainly composed of quaternary deposits overlain by fluvial sediments are expected. Considering the large shear displacements along the altered main scarps of the dormant rock slide (Figs. 3, 6), we infer the existence of a fully persistent basal shear zone. We conclude that the basal shear zone outcrops as left lateral flank at approximately 1620 m a.s.l. From there, the basal shear zone may flatten to form a curvilinear shape towards the slope toe where it impinges with the valley infillings at approximately 880 m a.s.l. Accordingly, a maximum rock slide thickness of approximately 150 m is assumed. A smaller rock slide is suspected on top of this large-scale slope instability, whose basal shear zone extends from its lateral flank at 1380 m a.s.l. down to the valley floor which outcrops just below the alluvial deposits at 950 m a.s.l. A maximum slab thickness of approximately 50 m is expected for this slide (Fig. 11a). Increased slope deformations between 2050 and 1600 m a.s.l. are assumed due to the formation of multiple minor scarps and the high rock fall activity originating there (Fig. 6), preparing the formation of the basal shear zone of the currently active Tellakopf rock slide.

In Fig. 11b, a detailed view of the active Tellakopf rock slide in 2021 shows the post-failure geometry, slab development, and kinematic indicators with the dip of the total displacement vectors of the control points. Considering that the rock slide mass has detached and separated several tens of metres along the main and minor scarps, fully continuous internal and basal shear zones are expected. Based on detailed field mapping and remote sensing data analysis, the basal shear zone is assumed to outcrop as a main scarp at approximately 2015 m a.s.l. with an offset of > 70 m, separating the largely displaced slabs III-VI from the back-scarp area (slabs I and II, Figs. 7, 11b). From there (dip angle between 40 and 44°), the inferred basal shear zone extends in a slightly curvilinear shape to the lower slope (dip angle between 33 and 42°) where it outcrops just above the left lateral flank of the dormant rock slide at approximately 1620 m a.s.l. Based on detailed analyses of multi-temporal TLS data, we observed no displacements below 1500 m a.s.l. Thus, we conclude that the rock slide mass continuously slides in a translational manner over the dormant rock slide down to an elevation of 1500 m a.s.l. where exposure of the basal shear zone is covered by large rock scree accumulations. Moderate to steeply dipping minor scarps (dip angle between 60 and 85°) with discrete shear displacements of several tens of metres separate the rock slide mass into slabs III, IV, and V. These scarp surfaces face downslope, with the exception of the 4-m high uphill-facing scarp between slabs IV and V. Several tensile fractures transect the lower slabs V and VI, indicating extensional stress regimes. These extensional processes at the lower slope area are consistent with the increased rock fall activity there and large amounts of rock scree deposits below 1620 m a.s.l.

The active Tellakopf rock slide comprises a maximum slab thickness of approximately of 60 m. A total rock slide volume of 4.08 and 3.77 Mm3 was determined with respect to the pre- and post-failure topography from 2006 and 2021, respectively (i.e. DEM from 2006 and 2021) and the reconstructed topography of the inferred basal shear zone (Table 6). A rock mass volume loss of approximately 0.31 Mm3 was determined, which is related to rock fall events that deposited debris further down to the valley floor (Fig. 8e). Pre-failure volumes of the individual slabs I-VI range between 0.03 and 1.33 Mm3. Considering rock mass volume balance analysis of the rock slide toe until 1500 m a.s.l. (Fig. 6f), an increase from 0.27 Mm3 in 2006 to 0.95 Mm3 in 2021 is noted, probably attributed to (i) the out-of-slope movement of the rock mass (i.e. bulging), and (ii) the accumulation of large amounts of rock scree below the bottom rock face at 1620 m a.s.l. (Fig. 11b; Table 5).

Table 6 Volume estimation [in Mm3 = million m3] of slabs I–VII, and the rock scree accumulation with respect to the pre-failure (2006) and the post-failure topography (2021)
Full size table

Discussion and conclusion

Deep-seated rock slides in mountainous regions are often difficult to investigate due to inaccessible and hazardous terrain. Challenges arise in monitoring and collecting field data to characterise their spatially and temporally complex failure and deformation processes. However, the application of different remote sensing techniques and analytical methods has emerged as a very important tool for advancing the characterisation of such slope instabilities (Jaboyedoff et al. 2009; Fey 2018; Zangerl et al. 2019; Hormes et al. 2020; Crippa and Agliardi 2021; Rechberger et al. 2021). Our research presents a multi-disciplinary methodological approach of detailed geological field mapping and remote sensing analysis that successfully captures the temporal-variable deformation behaviour and formation processes of the active, deep-seated Tellakopf rock slide. The application of image correlation-based displacement analyses and 3D distance change on the basis of different laser scan and imagery sources (Fey 2018) was proven helpful to comprehensively characterise slope surface changes as a basis for developing a geometrical-kinematical rock slide model. However, the applicability of these methods depends greatly on the available data set and its relation between data accuracy and magnitude in slope changes. In the case of the Tellakopf rock slide, the large displacements of several tens of metres were occurring during the last years. Considering that most deep-seated rock slides rather show very small displacements over long periods (i.e. few millimetres to centimetres per year) (Cruden and Varnes 1996; Agliardi et al. 2001; Hungr et al. 2014), measurement uncertainties in the centimetres to decimetres range (Table 3; Fey 2018) limit the application of such methods.

In terms of the Tellakopf rock slide with current displacement rates of 1–6 m/year from 2014 onwards, these measurement uncertainties were below the real displacement values, and thus, negligible. The high geometric and geo-referencing errors of the orthoimages from 1945–2011 does not allow displacement analyses based on image correlation. Here, the visual detection of structural topographic changes such as scarps and rock scree deposits enabled a reconstruction of the rock slide behaviour before 2014. The combination of data from different sensors and platforms (ALS, SfM-MSV and TLS) made it possible to quantify displacement and volume changes from 2014 onwards. The LoDs for the ALS and SfM-MVS datasets are much higher. The accuracy of the ALS 2006 dataset is limited by the low point density especially in areas with dense vegetation. The SfM-MVS data set was only available as DSM, hence a point cloud-based improvement of the relative accuracy by ICP was not possible. Further, vegetation causes high uncertainties and no data areas in photogrammetric data. In contrast, with laser scan data, the multi-pulse detection allows to record ground points underneath vegetation and vegetation can be filtered by a bare-earth filter. However, since the displacements are in the range of meters, the higher data uncertainties of these datasets, have no influence on the interpretation of the rock slide process mechanisms between 2006 and 2017. At the Tellakopf rock slide, TLS is well suited for period monitoring. Compared to SfM-MVS from UAVs (uncrewed aerial vehicle) or helicopters, measurement of GCPs within the rock slide area is not required and the vegetation can be filtered out. The change of the scan settings after the first TLS campaign in 2017 with scan positions at the upper area of the counter slope and the east side of the rock slide enabled to cover almost the entirety of the rock slide area. Only the areas behind the counter scarps are not covered by TLS.

The image correlation-based displacement analysis works well on non-vegetated areas with en-block displacements. If the displacements are associated with a change of surface structure (e.g. large rock fall areas or strong structural displacements), the algorithm fails and causes miscorrelations. A decimetre precise delineation between stable terrain and the rock slide is not possible due to the averaging effect of the image correlation. In the case, the search window covers terrain, which displaces at different displacement rates, an unambiguous assignment is not possible. The resulting displacement vector of this search window is averaged, and the real displacement of the corresponding grid point is over- or underestimated. The boarder can be located within an area of about 32 m (half of a search window). Hence, to delineate between slabs and stable terrain, further information from the DoDs, the shaded relief images and geomorphological maps are considered. Consequently, the interpretation of the image correlation results requires a knowledge of the geological processes.

In the case of rock slides with large displacement rates, as at the Tellakopf, a close-meshed data acquisition with annual TLS measurements is essential and provides, both an intermediate and an overall impression of the deformation behaviour. In addition, the understanding of promoting and triggering factors of deep-seated rock slides is crucial for predicting potential failure scenarios in the world’s increasingly populated mountainous regions. A major preparatory factor influencing the spatio-temporal evolution of deep-seated rock slides is attributed to the pre-existing structural inventory (Agliardi et al. 2001, 2019; Brideau et al. 2009; Ambrosi and Crosta 2011; Stead and Wolter 2015; Zangerl et al. 2019; Donati et al. 2020; Vick et al. 2020; Rechberger et al. 2021; Rechberger and Zangerl 2022). However, at the SE-facing slope of the Tellakopf, slope-parallel structures favouring a sliding mechanism are missing. In contrast, the rock slide is situated in foliated metamorphic gneisses, whose schistosity dips flat into the slope, i.e. orientated favourably for slope stability. However, the proximity to the Vinschgau shear zone and the Schlinig fault affects the rock mass by strong fracturing, fragmentation, and strength degradation (Fig. 5h). This tectonic imprint is also reflected in (i) the appearance of phyllonites on the ridge top, and (ii) the development of a meso-scale fracture network of NE-SW to NW-SE striking fracture sets orientated sub-parallel to local tectonic structures (Agliardi et al. 2009). Field observations identify that these steeply dipping fracture sets act as kinematic release surfaces that (i) influence rock slide geometry in form of head scarps and lateral flanks, and (ii) separate the rock slide mass into individual slabs in form of minor scarps (Fig. 5f–h). Additional kinematic analyses show that the pre-existing discontinuity network is preferably orientated towards direct and flexural toppling mechanisms (Fig. 4). While flexural toppling of block columns can be excluded, due to the extensively widely-spaced fracture network (Wyllie and Mah 2004), deep-seated direct toppling where the flat foliation planes serve as basal rupture surfaces for the widely-spaced rock blocks is assumed likely. However, toppling failure mechanisms are often associated by (i) decreasing slope movement from the rock slide’s crest to the slope toe, and (ii) significantly increasing dip angles over time (Amann 2006; Glueer et al. 2019), neither of which is observed at the Tellakopf rock slide. In addition, no indications for deep-seated toppling was observed outside the rock slide of the SE-facing Tellakopf slope. Field observations suggest only toppling mechanism as a secondary process promoting rock fall at steep rock faces (Fig. 5h). We infer that even though no brittle prominent structures dipping parallel to the slope are determined, a slip mechanism along a fully persistent basal shear zone is still expected. Due to the close proximity to major tectonic fault zones which may have weakened the strength of the rock mass it is assumed that progressive failure processes based on time-dependent growth, propagation, and linkage of pre-existing and new fractures must have taken place, which finally lead to the formation of a fully persistent basal shear zone. Our assumption of a sliding mechanism along a continuous basal shear zone is based on (i) the large discrete shear offset at the main scarp area (Figs. 5d, 6e, f), (ii) a spatially uniform deformation pattern of movement rates and constant dip angles of 30–45° (Figs. 8, 11b), and (iii) an out-of-slope movement of the lower slope sliding continuously over the dormant rock slide below (Figs. 8, 11b). Apart from the structural geological impact, a further possible cause favouring rock slide initiation is attributed to retrogressive slope processes, due to long-term stress release caused by the downwards movement of the lower slide accompanied by exposure of the rock slide foot.

In addition, increased precipitation of 244 mm when compared to the mean value of 163 mm (1924–2021) was recorded for the last quarter of 2013 at the nearby weather station Marienberg at 1311 m a.s.l. (46°42′20.52" N, 10°31′16.68" E) (Autonomous Province of Bozen 2022a). Notable is a precipitation event of snow on 5th January 2014, which corresponds to the monthly mean value (1924–2021) of 32 mm, and was followed by warm temperatures immediately prior to the onset of the rock slide evolution (Autonomous Province of Bozen 2022b). We therefore assume that slightly increased water pressures in the slope, due to a rainy autumn in 2013 and the preceding snow melt, may have exerted an additional destabilising effect on the slope stability.

Deep-seated rock slides, especially those in foliated metamorphic rocks, usually show slow to extremely slow movement rates over a long period of time, but may episodically accelerate to moderate rates, then decelerate again and tend to re-stabilise (Crosta and Agliardi 2003; Helmstetter et al. 2004; Bonzanigo et al. 2007; Zangerl et al. 2010). At the Tellakopf, the temporally variable deformation behaviour is characterised by a continuous slowdown in velocity after the initial formation phase. From the onset of rock slide formation in January 2014, displacements of up to approximately 54 m accumulated until July 2014, suggesting a lower bound of a mean velocity of 9 m/month. Interestingly, no sudden collapse of the rock mass at these alarming slope displacements was observed. Instead, the rock slide mass separated into individual slabs and successively reduces slope velocity (Figs. 6, 8, 9). Such accumulation of large displacements within a short period of time without the entire mass suddenly collapsing has also been observed for example at Marzellkamm, Aletsch glacier, and Bliggspitze (Glueer et al. 2019; Zangerl et al. 2019; Rechberger and Zangerl 2022). Nevertheless, important to consider are smaller rock fall events originating from the highly fractured and loosened lower slope area, which can adversely affect infrastructure and settlements. Ongoing deposition of large amount of rock scree at the rock slide toe has a stabilising effect and could contribute to the observed velocity reduction since the initial formation phase. Further investigations based on geomechanical modelling are needed to prove the impact of the lower situated rock slide and deposition of debris at the foot on slope stability.

In terms of risk mitigation measurements in densely populated mountainous regions, reliable hazard assessment of potential failure scenarios with regard to infrastructure and human lives is a prerequisite (Fell et al. 2000; UNISDR 2017). However, difficulties arise in identifying hazard scenarios, due to the complex history of evolution, temporal changes, and deformation behaviour of deep-seated rock slides (Crosta and Agliardi 2003; Strouth and Eberhardt 2009; Rechberger et al. 2021). In this study, an interdisciplinary methodological approach has therefore proven successful in capturing relevant key findings, such as (i) rock slide geometry and slab formation, (ii) whether internal or basal shear zones may have developed, (iii) type and rate of slope displacement, and (iv) whether secondary slope processes are occurring and which one (e.g. rock falls). Different remote sensing techniques combined with field mapping campaigns enabled the development of a geometrical-kinematical rock slide model representing the pre- and post-failure state. Results indicate a sliding mechanism along a curvilinear basal shear zone with an estimated total volume of 4.1 Mm3. Separation of the rock slide mass into single slabs of varying volume between 0.03 and 1.33 Mm3 but of relatively uniform displacement rates was observed. Secondary slope affecting processes such as single rock fall events from the lower slope area (Fig. 10) pose a hazard scenario when reaching long travel distances down to the valley floor. A substantial rock fall failure of up to 0.24 Mm3 has been recorded between 2018 and 2021. In general, the large amount of debris determined at the rock slide toe below 1620 m a.s.l. (Figs. 5b, 6f, 11) can serve as a source area for debris flows when mobilised by heavy precipitation and channelized into gullies (Hungr et al. 2014; Zangerl et al. 2019). In summary, landslide hazard scenarios are not exclusively linked to one specific landslide type, but rather transitions in between and cascading effects should be considered.

Building upon these results, subsequent research is proposed on numerical modelling based on the derived geometrical model (i) to identify potential hazard scenarios with respect to different sliding masses and whether these can lead to severe runouts, (ii) to study progressive failure processes with regard to the development and possible structural geological constraints of shear zone formation, (iii) to analyse the impact of in-situ stress changes due to rock slide displacement in the lower section of the slope and (iv) to determine whether increased water pressure due to snow melt can trigger the onset of rock slide formation.