A new–old approach for shallow landslide analysis and susceptibility zoning in fine-grained weathered soils of southern Italy

Geomorphology 241 (2015) 371–381

Contents lists available at ScienceDirect

Geomorphology journal homepage: www.elsevier.com/locate/geomorph

A new–old approach for shallow landslide analysis and susceptibility zoning in fine-grained weathered soils of southern Italy Leonardo Cascini a, Mariantonietta Ciurleo a,⁎, Silvio Di Nocera b, Giovanni Gullà c a b c

University of Salerno, Italy University of Naples, Federico II, Italy National Research Council, Research Institute for Geo-Hydrologic Protection, Cosenza, Italy

a r t i c l e

i n f o

Article history: Received 18 June 2014 Received in revised form 9 April 2015 Accepted 16 April 2015 Available online 24 April 2015 Keywords: Shallow landslides Weathered clays Triggering stage Susceptibility Multi-scale approach

a b s t r a c t Rainfall-induced shallow landslides involve several geo-environmental contexts and different types of soils. In clayey soils, they affect the most superficial layer, which is generally constituted by physically weathered soils characterised by a diffuse pattern of cracks. This type of landslide most commonly occurs in the form of multiple-occurrence landslide phenomena simultaneously involving large areas and thus has several consequences in terms of environmental and economic damage. Indeed, landslide susceptibility zoning is a relevant issue for land use planning and/or design purposes. This study proposes a multi-scale approach to reach this goal. The proposed approach is tested and validated over an area in southern Italy affected by widespread shallow landslides that can be classified as earth slides and earth slide-flows. Specifically, by moving from a small (1:100,000) to a medium scale (1:25,000), with the aid of heuristic and statistical methods, the approach identifies the main factors leading to landslide occurrence and effectively detects the areas potentially affected by these phenomena. Finally, at a larger scale (1:5000), deterministic methods, i.e., physically based models (TRIGRS and TRIGRS-unsaturated), allow quantitative landslide susceptibility assessment, starting from sample areas representative of those that can be affected by shallow landslides. Considering the reliability of the obtained results, the proposed approach seems useful for analysing other case studies in similar geological contexts. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Shallow landslides are one of the most common categories of landslides. They frequently involve large areas and different soils in various climatic zones (e.g., Kirkby, 1987; Benda and Cundy, 1990; Selby, 1993 Antronico et al., 2004; Borrelli et al., in press), and they often cause environmental and economic damage (Crozier, 2005; Glade et al., 2005) (Fig. 1). In clayey soils, shallow landslides affect the most superficial layers of soil, which are generally composed of physically weathered soils of variable thickness along the slope and are characterised by a spatially diffused and time dependent pattern of cracks (Fig. 2), usually attributed to an alternating process of wetting and drying, insolation and frost (Blight, 1997; Gullà et al., 2006; Fredlund et al., 2010). Shallow landslides exhibit different morphometric features depending on their localisation along the slope, where they have widths ranging from 3 to 15 m and lengths ranging from 10 to 100 m; the sliding surface can reach depths varying from a few decimetres to 3 m (Rogers and Selby, 1980; Gullà et al., 2004; Crozier, 2005). Some scientific publications on shallow landslides address hydrological, geological and geomorphological aspects (Antronico and Gullà, ⁎ Corresponding author. Tel.: +39 089964329. E-mail address: [email protected] (M. Ciurleo).

http://dx.doi.org/10.1016/j.geomorph.2015.04.017 0169-555X/© 2015 Elsevier B.V. All rights reserved.

2000; Sorriso-Valvo et al., 2004; Crozier, 2005; Guzzetti, 2008), and others focus on geotechnical characterisation and numerical modelling (Eden and Mitchell, 1969; Lim et al., 1996; Eigenbrod and Kaluza, 1999; Claessens et al., 2007). However, in spite of many contributions, a rational framework able to link the predisposing factors of shallow landslides at small, medium and large scales has not yet been provided. This paper is aimed at overcoming this lack of methods by proposing a so-called “new–old” approach, which is based on classic geological, geomorphological and geotechnical analyses, as well as on a new framework allowing for rational and quantitative correlations between landslide occurrence and regional and local factors in a test area in southern Italy. The procedure is based on the following simple ideas: i) the wide diffusion of shallow landslides in the test area is not casual, being strictly related to regional factors affecting the morphological evolution of the slopes, and ii) the limited thickness of these landslides is related to local factors that cause an increase in soil weathering grade until the instability conditions are reached. The procedure first distinguishes and quantifies the landslide predisposing factors and regional shallow landslide susceptibility at small and medium scales. Once the most prone areas have been clearly identified, attention is devoted to finding quantitative relationships between weathered soil and shallow landslides for the local level. The proposed procedure applies a general framework

372

L. Cascini et al. / Geomorphology 241 (2015) 371–381

Fig. 1. Multiple rainfall-triggered shallow landslides. (a) Air photo of July 1977, Wairarapa, North Island, New Zealand (Crozier, 2005); (b) Air photo of events in New Zealand in February 2004 (Hancox and Wright, 2005); (c) 3D view from Google Earth of March 2010, Catanzaro, Calabria, Italy; (d) Typical phenomena in a morphological hollow, Catanzaro Graben, Calabria, Italy.

provided by Fell et al. (2008a) who link the size of the study areas, the scales of analysis (small, medium and large scale) and the methods of zoning (heuristic, statistical and deterministic methods), even though they do not use the proposed framework to solve practical problems. The usefulness of the framework is addressed by Cascini (2008), who refers to different geological contexts at different topographic scales as well as to landslides and soils that are different from those studied in the present work.

rates (Fig. 3). Sector II coincides with the horst-graben system of the Coastal Chain–Sila Massif, which is characterised by predominantly N–S faults. The maximum uplift rate in this sector in the last million years, affecting the Coastal Chain, was approximately 1 mm/y, whereas the maximum uplift rate in the Sila Massif was 0.8 mm/y (SorrisoValvo and Tansi, 1996). Sector III corresponds to the Catanzaro graben and is characterised by the Lamezia–Catanzaro fault, with its southern end at the Maida–Girifalco–Squillace fault line. A 0.2 mm/y uplift rate was calculated for the Catanzaro graben (Sorriso-Valvo and Tansi, 1996). Considering that the proposed approach uses a multi-scale analysis, a gradually smaller reference area is analysed when moving from a small to large scale (Fig. 3). In particular, at the small scale (1:100,000), the reference area extends 2000 km2 and falls within the provinces of Catanzaro and Crotone; to the north, the area borders the Sila horst, to the south it borders the Serre horst, and to the east and the west it is delimited by the Ionian and the Tyrrhenian seas, respectively. At the medium scale (1:25,000), only a portion of the territory in the province of Catanzaro, extending for 150 km2, is analysed because shallow landslides are prevalent there. The borders of this portion are the Sila horst to the north and the Ionian Sea to the south, and it is delimited to the east and west by the watersheds of the basins located to the hydrographical left and right of the Corace and La Fiumarella rivers. Finally, at the large scale (1:5000), the attention is focused on a morphological hollow that is chosen, investigated and analysed as being representative of the studied landslides. 2.2. Data 2.2.1. Geological and morphological data At the small scale, geological and structural data were obtained by merging the data available from the structural geological map proposed

2. Materials and methods 2.1. Study areas The study area is located in central Calabria, southern Italy, and falls into two distinct morphostructural sectors named Sectors II and III by Sorriso-Valvo and Tansi (1996) and characterised by different uplift

Fig. 2. Crack pattern at the soil surface. (a) Spatial distribution of cracks in a soil ploughed in spring. The numbers indicate the crack depth in centimetres (Meisina, 2006); (b) Crosssection showing the crack depths (Meisina, 2006); (c) Typical section of weathered soils in the study area.

Fig. 3. Main morphotectonic structures and uplift rates in Calabria during the Quaternary Era (from Sorriso-Valvo and Tansi, 1996 modified), and the test areas at different topographic scales.

L. Cascini et al. / Geomorphology 241 (2015) 371–381

by Van Dijk and Okkes (1991) (at a 1:100,000 scale), the litho-structural map proposed by Antronico et al. (2001) (at a 1:50,000 scale), and the Geological map of Calabria (at a 1:25,000 scale). Through this procedure, a small-scale map of the main lithologies was created, simplifying the different geological complexes (Fig. 4a). The geo-structural map

373

shows, at the foot of the Serre Massif, the presence of a Palaeozoic crystalline basement, while along the pediment linking the Sila Massif and the sedimentary basins of the Catanzaro graben and the Crotone basin, discordant sedimentary Miocene rocks unconformably overlie the Palaeozoic crystalline basement. Upwards, the Miocene succession becomes a Plio-Pleistocene post-orogenic sedimentary succession composed of marine deposits originating from the Catanzaro and Crotone basins (Longhitano et al., 2014; Perri et al., 2014). Specifically, in their lower portion, these lithotypes consist of thick clay deposits of Pliocene age (approximately 800 m thick for the Catanzaro basin and 1200 m thick for the Crotone basin, based on AGIP boreholes), which become primarily sand deposits of Pleistocene age in the upper portion of these lithotypes. Pleistocenic terraced deposits of primarily sand and gravel are found at the end of the succession, are well preserved in the area of Crotone, become sporadic in the transitional zone between the Crotone basin and the Catanzaro graben, and then reappear in the form of preserved terraces in the central part of the Catanzaro graben. Referring to the main macro-structural elements, at small scale, an initial NW–SE fault system characterises the northern and southern boundaries of the study area and delimits the tectonic depression of the Catanzaro graben. These faults have been overlapped by a second fault system, NE–SW, which predominates in the transition area between the Catanzaro and Crotone basins, where a series of terraced faults dip towards the Ionian Sea. In the Crotone area, a local horst has been delimited by two shear zones that isolate, identify and preserve the structural high of Isola Capo Rizzuto (Fig. 4a). At the medium scale, the outcropping lithologies have been identified by integrating the Geological map of Calabria and the lithostructural map proposed by Borrelli et al. (in press) (available at 1:25,000 scale) with aerial photo interpretation and in situ surveys, thus allowing for the identification and mapping of different geological units. These units are briefly described from the lowest to highest elevations as follows: sporadic evidence of Palaeozoic crystalline basement (schists) and Miocene rocks (conglomerates, sandstones and evaporitic limestones); Pliocene light blue-grey silty clays partially affected by intercalations of sand and silt; an alternation of Pliocene sands and sandstone; and overlying Pleistocene sands, gravels, and brown and redbrown conglomerates (Fig. 4b). The aerial photograph interpretation performed by black-and-white images (at 1:33,000 scale), derived from an IGM flight from 1991, allows us i) to confirm the structural complexity of the test area, which was already highlighted by Van Dijk and Okkes (1991) and Tansi et al. (2006), and ii) to distinguish some relevant structural features (Burbank and Anderson, 2001). At the large scale, the main stratigraphic contacts were obtained through photograph interpretation and in situ investigations, which also allowed for the identification of the weathered thickness of soil. Fig. 4c shows the presence of Pliocene light blue-grey silty clays that are partially affected, in the upper part of the stratigraphic sequence, by intercalation of Pliocene sands and sandstones. The rapid evolution of these phenomena, which undergo substantial morphometric changes over time, renders the updating of the inventory through traditional techniques extremely difficult, as testified by the systematic lack of adequate inventory maps in the scientific and technical literature. With reference to the study area, the only available document was created on 2001 by the Calabria River Basin Authority at the 1:10,000 scale for residential areas and small villages with more than 200

Fig. 4. The dataset used. (a) The geo-structural map of the study area at small scale (1:100,000). (b) The geo-structural map of the test area at medium scale (1:25,000); Legend: 1. Holocene alluvial deposits and eolian sands; 2. Pleistocene sands, gravels, brown and red-brown conglomerates; 3. Pliocene sands and sandstones; 4. Pliocene light bluegrey silty clays; 5. Miocenic evaporitic limestones; 6. Miocene sandstones and sands; 7. Miocene conglomerates; 8. Paleozoic schists; 9. normal faults; 10. Strike–slip faults; 11. The test area at large scale. (c) The geological map of the test area at large scale (1:5000). Legend: 1. Pliocene light blue-grey silty clays; 2. Pliocene sands and sandstones.

374

L. Cascini et al. / Geomorphology 241 (2015) 371–381

Fig. 5. In situ investigations at the TS1, TS2 and TS3 test sites. a) Spatial location; b) boreholes log; c) monitoring stations; d) daily values of rain depth, and pore water pressure measured by tensiometers (modified after Gullà et al., 2004).

inhabitants, therefore underestimating the areas affected by landslides. These data were only used for analysis at a small scale (1:100,000) after the transformation of landslides into dots with attributes (Fell et al., 2008b). At a larger scale, the landslides were inventoried using aerial photos and the satellite images from Google Earth as input data. The latter is not considered a scientific method to map shallow landslides, but some authors have recently noted its potential for the creation and upgrading of inventory maps (Sato and Harp, 2009; Guzzetti et al., 2012; Borrelli et al., in press). Considering the limitations of the adopted procedure, the data provided by Google Earth were locally validated by in situ surveys. Currently, few hydro-geological data are available in the test area, and a limited number of sites have been properly monitored for an adequate period of time. In that context, the Explanatory Notes of the Hydrogeological Map of Southern Italy, at 1:250,000 scale (Celico et al., 2005) was used as a reference at the small scale; in contrast, the data provided by Gullà et al. (2004) for one well-monitored study site were considered for the analysis of the pore water pressure regime at the large scale. In this study, Digital Elevation Models (DEMs) with different resolutions (of 95 × 95 m at the small scale, 25 × 25 m at the medium scale, and 5 × 5 m at the large scale) were used in geotechnical modelling and for acquisition of the parameters generally used in GIS procedures, such as slope angle, curvature, and flow direction. The DEMs used refer to the elevations at the time when they were created, and they consequently do not take account of the natural slope evolution. 2.2.2. Geotechnical characterisation Among the data available in the literature, the extensive database published in Gullà et al. (2004, 2008) and the data available in Cascini

and Matano (2010) have been used (Figs. 5 and 6). In particular, these studies distinguish between intact and weathered rock and identify a stratigraphic succession composed of silts with clay or clay with silt with sporadic sand intercalations (Figs. 5b and 6a). The soil is classified as inorganic, inactive clay having high plasticity and a high liquidity limit (Fig. 6b). Gullà et al. (2004, 2008) and Cascini and Matano (2010) noted that the natural unit weight (γn) varies between 15.0 and 20.5 kN/m3, the saturated unit weight (γs) varies from 16 to

Fig. 6. Physical and mechanical properties of weathered clays: (a) grain size distribution envelope; (b) plasticity chart; (c) index properties; (d) shear strength properties (modified after Gullà et al., 2008).

L. Cascini et al. / Geomorphology 241 (2015) 371–381

375

21.0 kN/m3, and the dry unit weight (γd) varies between 13.7 and 19.7 kN/m3; the values assumed for soil porosity (n) range between 0.30 and 0.48, and the index void (e) ranges from 0.5 to 1.0 (Fig. 6c). In this study, based on the shear strength tests reported in Gullà et al. (2004, 2008), a linear shear strength envelope is assumed for low values of vertical tension (Fig. 6d). This figure corresponds to cohesion values in the range of 2–24.3 kPa and friction angles in the range of 22–35°, which demonstrates the heterogeneity of the weathered clays. The data given in Gullà et al. (2004) and in the database of the CNR-IRPI of Cosenza have been integrated by saturated and unsaturated laboratory tests as described in Ciurleo (2012). The overall data set obtained provides saturated permeability values varying in the range Ksat = 3.1 × 10−8 to 7.65 × 10−7 m/s and saturated and residual volumetric water contents varying within the ranges θs = 0.48–0.43 and θr = 0.28–0.0002, respectively. Finally, with regard to the boundary hydraulic conditions, Fig. 5c, d shows the rainfall data and the soil suction values recorded at a monitoring station by a rain gauge and several “Jetfill” tensiometers installed at various depths below the ground surface (0.10, 0.15, 0.30, 0.45 and 1.20 m). These values show that the suction ranges vary between 0 and 80 kPa during the hydrological year, up to a depth of 0.45 m, a strong correlation exists between the pore water pressure regime and the daily cumulative rainfall, and near-saturated conditions were reached at a depth of 1.20 m between October and May, when many shallow landslides were recorded in the study area. Moreover, when examining previous landslide occurrences, a critical cumulative rainfall for the study site of 43 mm in 2 days was determined by Gullà et al. (2004). 2.3. Proposed approach The proposed approach is characterised by three subsequent steps (at small, medium and large scales); each step provides significant elements, and once completed, it allows for both the establishment of a connection between the landslide predisposing factors (at each scale of analysis) and the landslide susceptibility zoning (Fig. 7). This approach differs from others insofar as it uses two distinct substeps for each step: substep I in which predisposing factors are identified, and substep II in which the factors are quantified. From an operational perspective, the proposed approach begins at a small scale, following a logical process that considers the output of substep II for each scale as the input to substep I for the scale that follows. Such conditions enable a strict evaluation of the validity of both the methods used and the results obtained. At a small-scale analysis (step 1), substep I is developed by applying the heuristic methods such as geomorphic analysis to i) link the morphological evolution of the slopes with the small scale predisposing factors and ii) individuate and qualitatively rank the spatial density of landslides (SLD). To pursue the first aim, the potential correlations between the morphological evolution of the area and the slopes generally affected by landslides were analysed, comparing the main lithological, stratigraphic and tectonic factors governing the morphological evolution of the region with the landslides inventory via geomorphic analysis (Kienholz, 1978; Soeters and van Westen, 1996). To pursue the second aim, a spatial landslide density map was created by cross-referencing with the information gathered from the official landslide inventory and using the radial basic method for interpolation (Silverman, 1986) as implemented in the ArcGIS 10 code. The SLD is a key point of the procedure. Indeed, if the spatial density of landslides appears consistent with the distribution of shallow landslides, the user can go on to the following step; on the contrary, at the small scale, a further iteration is necessary to properly select the predisposing factors of the landslides at hand. The objective of substep II is to verify the factors recognised in the previous substep through the “qualitative map combination”—a

Fig. 7. The new–old approach for landslide susceptibility zoning at different topographic scales.

procedure universally adopted for landslide susceptibility and hazards (Barredo et al., 2000; Van Westen et al., 2003; Perotto-Baldiviezo et al., 2004)—thus ranking the areas most prone to landslides. In this procedure, expert judgement is used to assign weighting values to a series of thematic maps; the sum of the weights leads to susceptibility values that can be grouped into susceptibility classes. The problem with the application of this method lies in determining reliable weighting of the parameters, which necessarily depends on the availability of both field knowledge and a good landslide inventory. When insufficient data prevent the proper establishment of factor weights, the use of weighted values assigned by other scientific papers that address similar geological contexts and use an accurate landslide inventory map can be considered. In this paper, we use the weighted values, which were further amended by the information gathered by Sorriso-Valvo and Tansi (1996) regarding differential uplift rates (Table 1), and the data reported by Del Monte et al. (2002) for the Trionto basin (Calabria Region), which is characterised by a stratigraphic succession comparable to those recognised in the Catanzaro and Crotone areas. The analysis at a medium scale (step 2) by substep I which is implemented using more detailed input data, allows us to i) distinguish shallow landslides from erosion phenomena, thus excluding the areas potentially affected by erosion, and ii) identify shallow landslide predisposing factors and the area to be investigated in depth at a larger scale. Moreover, to distinguish any possible relationships between shallow landslides and the morphological evolution of the area, we proposed the use of a Coe's Density Index (Coe et al., 2004).

376

L. Cascini et al. / Geomorphology 241 (2015) 371–381

In the first case, the statistical indicator is the standard deviation, which measures the compactness of the distribution by providing a single value representing the dispersion of features around the average value. In the second case, the analysis can be based on the use of a two-by-two contingency table, in which the success and error indexes at medium scale are calculated using the following expression:

Table 1 Weights used in the analysis (Del Monte et al., 2002 mod.). Predisposing Factors

W

Lithologies Alluvial deposits Clay Sand, gravel and conglomerates Grey limestone, marl and sandstone Phyllite, micashist and gneiss Granite and granodiorite

1 37 8 15 16 3

Slope gradient 0 b S ≤ 10 10 b S ≤ 20 20 b S ≤ 30 30 b S ≤ 40 S N 40

5 5 12 12 14

Drainage density 0bD≤2 2bD≤4 4bD≤6 6bD≤8 DN8

0 26 17 7 7

Differential uplift rates DUA1 DUA2 DUA3

2 6 0

Substep II quantifies both the shallow landslide predisposing factors and the shallow landslide susceptibility zoning (SLS) by any statistical method that is able to identify a number of independent variables based on the information provided by a landslide inventory. The statistical method used in the paper is the “information value” bivariate analysis (Yin and Yan, 1988), which is based on the evaluation of relative landslide density (IRLDi), the value of which can be computed for each terrain-mapping unit (TMU) and then summed to compute the global index of landslide density (ILD): 3 NumLi 6 Denclasi Areai 7 7 ¼ ln 6 IRLDi ¼ ln 6 NumLtot 7 Densmap 6 7 6 Area 7




2

ð1Þ

tot

ILD ¼

X i

IRLDi

ð2Þ

where Densclasi and Densmap are, respectively, the density of landslides in the i-th class and the density of landslides in the entire study area, NumLi is the number of landslides in the i-th class, NumLtot is the total number of landslides, Areai is the area of the i-th class, and Areatot is the total area. In the statistical analysis and in general for landslide susceptibility zoning, the discretisation of the territory into TMUs depends on several factors, including the quality and the resolution of the required thematic information, the scale of the analysis and the type of landslide phenomena under investigation (Calvello et al., 2013). Considering the relevance of this issue, following Calvello et al. (2013), two different TMUs have been identified in this paper: terrain computational units (TCUs) and terrain zoning units (TZUs), whose use and size depend on both the level and the scale of analysis. The variables introduced in the model were defined on the basis of the substep I analysis. They were divided into eight classes following the “quantile” method, except for the geological map and uplift rate map for which the number of classes depends, respectively, on the outcropping of geological complexes and on the results of substep I. The importance of each variable can be determined by defining two statistical parameters, i.e., the individual discrimination capability of a single variable and the contribution of the variables to the success of the analysis (Calvello, 2012).

SI ðMSÞ ¼ ðN nlm =N nli Þ=ðNlm =Nli Þ

ð3Þ

EI ðMSÞ ¼ 1−Nnlm =Nnli

ð4Þ

where Nnlm and Nlm are, respectively, the number of TCUs computed as stable and unstable by the method, whereas Nli and Nnli are the number of TCUs that are affected and not affected, respectively, by the landslide events. Finally, at a large scale (step 3), substep I individuates the typical shallow landslides mechanisms and their spatial distribution along the slopes, after which it examines the available geotechnical data with the aim of identifying the following: i) representative geotechnical properties of the lithotypes involved in landslides, excluding shear strength parameters for which substep II has been implemented; ii) the pore water pressure regime in the slopes; and iii) the spatial distribution of the weathered thickness of soils. In the present paper, the map of the weathered thickness of soil was initially created by starting from the identification and localisation of the different triggering mechanisms, and the thicknesses were verified by in situ investigations carried out with a steel bar and some exploration trenches at a second stage. Based on these data, substep II focuses on: the identification of the most representative shear strength parameters over a large area; the quantification of the triggering mechanisms and the quantitative susceptibility assessment (QSA) through physically based models. These models are characterised by a general grid-based structure supported by geographic information systems, thus allowing for geotechnical analysis over a wide area. This approach generally couples a hydrologic model for the analysis of the pore-water pressure regime with an infinite slope stability model for the computation of the factor of safety. Among the distributed models proposed in the scientific literature (e.g., Montgomery and Dietrich, 1994; Baum et al., 2002; Savage et al., 2004), TRIGRS (Baum et al., 2002) and TRIGRS-unsaturated (Savage et al., 2004) were selected because they use analytical solutions for the pore pressure response to rainfall and provide reliable results for similar problems (Sorbino et al., 2010). Indeed, TRIGRS was used to analyse saturated conditions that presumably occur in association with a densely cracked soil cover. Under such conditions, a mean value of friction angle was assumed, varying the cohesion in a range consistent with laboratory values and vice versa. For operating the model for this type of soil, we assumed the following: i) a friction angle of 27° and cohesion values between 2 and 14 kPa and ii) a cohesion value of 5 kPa and a friction angle between 20° and 40°. On the other hand, TRIGRS-unsaturated analysed the unsaturated conditions of the soil cover—representing the initial landsliding stage—by assuming the representative geotechnical values derived by TRIGRS analyses. To quantify the reliability of the analysis performed, the unstable areas computed by the model were compared with the landslide inventory. The obtained results were quantified following Sorbino et al. (2007), where two percentage indexes named “Success Index SI(LS)” and “Error Index EI(LS)” were introduced and used to refer to the event that occurred in 2008–2010. For each source area, SI(LS) is the proportion (in percent) of the observed source area computed to be unstable by the model, and EI(LS) represents the percentage ratio between the unstable areas located outside of the observed triggering areas (Aout) and the area not affected by instability (Astab).

L. Cascini et al. / Geomorphology 241 (2015) 371–381

377

3. Results and discussion 3.1. Small scale — step 1 Substep I, by geomorphic analysis, highlights the correlation between the evolution of the slopes and the uplift rate that occurred and continues to occur at different velocities. Indeed, the fluvial system is conditioned by this factor as well as by the outcropping lithotypes and their stratigraphic and tectonic characteristics. In particular, such a system is influenced by the local horst of Isola Capo Rizzuto in the Crotone area and by the Sila and Serre horsts in the Catanzaro graben. Consequently, the highest-gradient rivers are located in the transitional zone and flow directly into the Ionian Sea, whereas in other areas, the primary rivers diverge due to the presence of the Serre and Isola Capo Rizzuto horsts. Based on this information, three sub-zones, referred to as A1 (Catanzaro Graben), A2 (Transitional zone) and A3 (Crotone basin), can be distinguished. The relationships between the geological evolution of these three sub-zones and the landsides, represented at a small scale by dots with attributes, are highlighted by the landslide density map (Fig. 8). As illustrated in Fig. 8, the area between sub-sectors A1 and A2 is the most affected by landslides, i.e., the area where the highest differential uplifts, due to the active tectonics, caused the erosion of sand and gravel deposits, exposing the clay slopes to weathering and subsequent instability phenomena. Substep II, by qualitative map combination, validates the role played by the predisposing factors using the weighted values reported by Del Monte et al. (2002) for the Trionto basin (Calabria Region), which is an area characterised by a stratigraphic succession comparable to those recognised in the Catanzaro and the Crotone basins. Del Monte et al. (2002) identified four predisposing factors of morphological evolution affecting the area, i.e., lithology, slope, drainage density and land use. Each factor was split into several classes with different weights depending on the percentage of the area affected by mass movements, which was obtained by a landslide inventory map at a 1:5000 scale. Given the scale of analysis (1:100,000), land use is considered less significant than the other factors mentioned above, whereas the differential uplift rate is paramount, as noted in the substep I analysis. The sum of all of the weights (Fig. 9) indicates that one of the areas most prone to landslides is the one located near the city of Catanzaro, where clay soils predominate. This area is characterised by a slope gradient in the range from 20° to 40° and a drainage density between 2 and 4 km−1, and it is highly affected by the uplift of the Sila Massif. In spite of the use of weights identified by other authors in a different test area, the adopted approach allows for an automatic identification of a small area of 150 km2 characterised by a more rapid

Fig. 8. Landslide density map with the identified sub-zones A1, A2 and A3. In legend, LD = landslides density (number of shallow landslides per km2).

Fig. 9. Geomorphic evolution map. Legend: the sum of the weights.

morphological evolution that is essentially related to the differential uplifts, the prevalent lithotypes, and the characteristics of the fluvial system.

3.2. Medium scale — step 2 Within the 150 km2 area identified by the small scale analysis as having experienced the most rapid morphological evolution, substep I of the medium scale analysis is devoted to identifying the predisposing factors of shallow landslides, beginning with the geological map and the structural data previously reported in Fig. 4b. Geological and structural features of the area reveal the presence of NW–SE normal faults linked to the tectonic system of the Catanzaro graben (Fig. 4b) and identify four morpho-structures denoted by two-digit ID codes (1.3, 1.2, 3.1, 2.4) (Fig. 10a). The first digit refers to the surfacing geological period (1 = Lower Pliocene, 2 = Lower Pleistocene, 3 = Middle Pleistocene), and the second digit refers to the uplift rate qualitatively defined on the basis of the scientific literature (Sorriso-Valvo and Tansi, 1996; Tansi et al., 2006; Ciurleo, 2012). Focusing on the landslide inventory, which was developed using the procedures introduced in Section 2.2.1, it is evident that shallow landslides, at a medium scale, can be identified only by the landslide area, whereas deep-seated landslides (with a slip surface deeper than 3 m) can be recognised by all of their morphological features. Moreover, the southwest border of the study area (morpho-structure 2.4), where sands and conglomerates prevail, is affected by deep-seated landslides; in contrast, on the northern border of the area (morpho-structure 1.3) where clay deposits are predominant, both deep-seated and shallow landslides were found. In morpho-structure 1.3, shallow landslides are localised in the upper layer of the slopes and particularly in welldefined morphological hollows of clay, whereas the northwest portion of the study area, a 14 km2 area where Plio-Pleistocenic sands and conglomerates outcrop, is mainly affected by erosional processes (Fig. 10a). Thus, the test area has been reduced from 150 km2 to 136 km2, and the computed Coe's Index confirmed that shallow instability phenomena are concentrated within subsector 1.3. This is most likely due to the localisation of this sector closest to the Sila Massif; hence, it is characterised by the highest uplift rate. The uplift, together with the NW–SE and NE–SW directional fault systems, produced the highest slope energy, which influenced the stream network and the formation of the morphological hollows in which shallow landslides frequently occur. In substep II of the analysis, the landslide predisposing factors are quantified by applying the “information value” (Yin and Yan, 1988), and the TCUs used for computational analysis comprise a regular square grid with a cell size (or pixel) of 25 × 25 m, whereas the dimension of the TZUs is set to 16 elementary pixels.

378

L. Cascini et al. / Geomorphology 241 (2015) 371–381

3.3. Large scale — step 3 The analysis at the medium scale has identified the morphological hollows located at the toe of the Sila Massif as the areas most susceptible to the occurrence of shallow landslides. Therefore, a better understanding of the genesis and evolution of these landslides was pursued by analysing one of these morphological hollows.

Fig. 10. Medium-scale analysis: (a) Results of Substep I. Legend: 1. Holocene alluvial deposits and eolian sands; 2. Pleistocene sands gravels, brown and red-brown conglomerates; 3. Pliocene sands and sandstones; 4. Pliocene light blue-grey silty clays; 5. Miocenic evaporitic limestones; 6. Miocene sandstones and sands; 7. Miocene conglomerates; 8. Paleozoic schists; 9. Normal faults; 10. Strike–slip faults; 11. Erosion areas; 12. Shallow landslides; 13. Deep-seated landslide triggering areas; 14. Deep-seated landslide bodies; 15. Morpho-structures. Red bars represent the Coe's Index for shallow landslides. (b) Results of Substep II: Bivariate analysis.

All of the considered variables were divided into eight classes, and only the uplift rate was divided into four classes depending on the number of morpho-structures identified in substep I of the medium scale analysis (Fig. 10a). Following Calvello (2012), the importance of each variable can be determined assuming values of the significance threshold of deviation, DEV.ST(Vi)*, and SI(MS) greater than 0.7 and 70%, respectively. The variables that simultaneously satisfy these conditions are elevation zones, slope gradient, lithological distributions, and curvature. The ILD map obtained by considering these four variables is reported in Fig. 10b, in which four susceptibility descriptors are defined: not susceptible for ILD b 0; low for ILD in the range of [0–0.5]; medium for ILD between 0.5 and 1; and high for ILD N 1. A comparison between the obtained susceptibility map and the landslide inventory from 2010 clearly shows the success of the statistical analysis, establishing the role of the main predisposing factors already identified in substep I at the medium scale. This success is further confirmed by the values of the SI(MS) and EI(MS) indexes, respectively equal to 92% and 22%, which were obtained by considering all of the TCUs with ILD greater than 0, i.e., the TCUs classified as low, medium and high susceptibility.

3.3.1. Substep I At the large scale, these landslides are clearly visible and can be mapped; thus, the starting point of the proposed approach is the identification and classification of the main landslide mechanisms (Antronico and Gullà, 2000). For this purpose, and focusing only on the event that occurred in the winters of 2008–2010, an accurate landslide inventory was preliminarily developed through in situ surveys, whereas the available geotechnical data were used to provide a classification of the involved soils. According to Varnes (1978) and Leroueil et al. (1996), the investigated phenomena can be generally classified as translational and/or complex slides involving clay and silt, as confirmed by the stratigraphic successions composed of silts with clay or clays with silt with sporadic sand intercalations (Figs. 5a and 6b). Moreover, the borehole data show the presence of a weathered thickness of clay up to 3 m, a thickness confirmed by a geotechnical dataset that shows substantial differences in index properties and shear strength beyond 3 m (Gullà et al., 2004; Cascini and Matano, 2010). With reference to the stage of activity, these landslides can be considered to be prevalently “first failure” phenomena, although post-failure stage movements have been found in some portions of the morphological hollows. Then, on the basis of the main features observed by in situ surveys, three mechanisms named MORLE-CZ1, MORLE-CZ2, and MORLE-CZ3 have been identified (Fig. 11). The primary characteristics of these mechanisms are strictly related to their different locations along the slope, which also affect their evolution and magnitude. Specifically, MORLE-CZ1 can be classified as an earth slide; this mechanism always occurs at the top of the open slopes, either on old landslide deposits or near a change in slope gradient. The surface and subsurface runoff, strongly influenced by slope morphology, are characterised by straight lines and can be considered responsible for a weathering process that involves the most superficial soil. This phenomenon is usually rectangular along the maximum gradient direction; it has a triggering area that is a few metres wide and a total length of generally less than 10 m. The maximum depth of the sliding surface is located approximately 1 m from the ground surface, and the first failure stage is frequently preceded by the opening of cracks at the top of the slope; just after the first failure stage, the unstable mass dislocates from the landslide source area.

Fig. 11. Shallow landslide mechanisms and their spatial distribution.

L. Cascini et al. / Geomorphology 241 (2015) 371–381

379

MORLE-CZ3 generally occurs within morphological hollows wider than those affected by the mechanisms described above and is characterised by a radially convergent surface and subsurface runoff towards the main drainage lines, which is mainly produced by cracked zones associated with tectonic structures. In this case, the weathering process is very rapid and is influenced by lateral runoff, and the weathered thickness becomes deeper. In general, the involved areas are between 10 and 30 m wide and between 50 and 100 m long, with a maximum depth of the sliding surface of 3 m, and this mechanism occurs over a longer time span than the other mechanisms. This is essentially due to the weathering process, which involves greater depths and volumes than with the other mechanisms; MORLE-CZ3 begins with the opening of a tension crack at the top of the slope, and its evolution is progressively accelerated by a pattern of cracks inside the landslide body that facilitates the infiltration of runoff water. Starting from the localisation of the MORLE-CZ1, MORLE-CZ2, MORLE-CZ3 mechanisms, and after further verification by in situ investigations, the weathered thickness map was created (Fig. 12) (Ciurleo, 2012). Finally, among the available laboratory values, the geotechnical parameters considered representative of weathered clays over a large area include the main value of γn (=18 kN/m3) and the highest values of porosity and saturated conductivity (n = 0.48, Ksat = 5 × 10-07 m/s) to consider the effects of weathering. Additionally, we used TRIGRS to identify representative values of cohesion and friction angle until limiting equilibrium conditions were reached.

Fig. 12. Cover thickness maps. (a) Created by starting from the localisation of the triggering mechanisms; (b) Verified by in situ surveys (black dots).

MORLE-CZ2 can be classified as a complex earth slide-earth flow phenomenon; the mechanism usually occurs in the upper portions of secondary morphological hollows characterised by convergent surface and subsurface runoff towards the main drainage line, where the overall weathering intensity is very high. It typically has an oblong shape, and it is a few metres wide and between 10 and 50 m in length. The sliding surface is at a maximum depth of approximately 1.5 m below the ground surface. This mechanism has a very rapid triggering stage, and the volume involved in the landslide is distributed along the main drainage line, although secondary phenomena can occur along the lateral flanks.

3.3.2. Substep II The prerequisites for the application of physically based models are the availability of reliable geotechnical and topographical input data, which must be properly integrated with i) the initial and boundary conditions (pore water pressure regime and critical rainfall values) and ii) an accurate map of the weathered clay. The geotechnical input data used for the analyses were deduced in substep I and are summarised in Table 2. Among the hydraulic parameters that significantly influence the physically based models, TRIGRS and TRIGRS-unsaturated include Ksat, θs, θr, and the diffusivity value (D). The parametric analyses based on the data summarised in Table 2 show that the highest value of SI(LS) (90%), corresponding to cohesion c′ = 2 kPa and friction angle φ′ = 27° (Fig. 13), is associated with a very high value of EI(LS) (56%), and the computed unstable area is approximately 3 times larger than the observed area. Fig. 13 also shows that in the range of c′ = 4.5–6 kPa and φ′ = 26°–31°, TRIGRS yields more satisfactory results as it systematically provides values of SI(LS) greater than 60% for values of EI(LS) less than 30%. Moreover, the geotechnical parameters that better fit the TRIGRS analysis correspond to the lower value of cohesion and the average value of the friction angle

Table 2 The geotechnical input data used for TRIGRS and TRIGRS unsaturated analyses. TRIGRS Unit weight 3

γ (kN/m ) 18

Effective cohesion

Friction angle

Soil depth

Hydraulic conductivity

Diffusivity

c′ (kPa) 2–14 5

φ′ (°) 27 20–40

h trigrs (m)

K (m/s)

D TRIGRS (m2/s)

Variable

5 × 10-07

3.49E-05

TRIGRS-unsaturated Unit weight

Effective cohesion

Friction angle

Soil depth

Hydraulic conductivity

Diffusivity

θs

θr

γ (kN/m3)

c′ (kPa) 5

φ′ (°) 27

h trigrs (m)

K (m/s)

D TRIGRS_unsaturated (m2/s)

(−)

(−)

Variable

5 × 10-07

3.49E-05

0.48

0.10

18

380

L. Cascini et al. / Geomorphology 241 (2015) 371–381

An overview of the obtained results highlights two interesting aspects. The back analysis confirms that in undisturbed clayey soils, weathering affects cohesion (by the saturation and desaturation cycles) much more than the friction angle, as already discussed by Gullà et al. (2006). TRIGRS and TRIGRS-unsaturated (Fig. 14) seem able to capture the first and the last slope evolution stages that are strongly influenced by the significant change in the fissure patterns from the occurrence of the first failure phenomena to a completely unstable slope (Ciurleo, 2012). 4. Conclusions

Fig. 13. SI(LS) and EI(LS) vs. cohesion c′ and friction angle φ′.

that characterise the linear shear strength envelopes assumed by the laboratory tests. Based on these results, a TRIGRS-unsaturated model was used to determine the role played by the unsaturated conditions in the landslide triggering mechanism. A comparison of the obtained results is provided in Fig. 14 together with the landslide shapes of the 2008–2010 winter events.

Shallow landslides in fine-grained, weathered soils are natural phenomena that generally affect wide areas in several geo-environmental contexts. However, the scientific literature only addresses specific problems and does not provide a methodological approach to properly study all of the aspects related to these phenomena and, particularly, of aspects pertinent to landslide susceptibility assessment. The latter is a complex issue because landslide inventories, when available, only localise the existing landslides and usually do not provide any further information about the area potentially affected by landsliding. To address the topic, this paper proposes a multi-scale approach that firstly allows, by heuristic and statistic methods, the discrimination and zoning of the areas most prone to instability phenomena at small and medium scales. Upon enlarging the scale, and with the aid of detailed in situ surveys and geotechnical analyses, the procedure is able to provide quantitative elements in both the inventory and the susceptibility of the slope to shallow landslides. The validity of the proposed approach is testified by its application to the test area, for which the obtained results underscore a good correlation between the shallow landslides affecting some morphological hollows and the complex geological characteristics of the much wider area in which the hollows are located. Considering that the innovative aspect is represented by the ability to properly link the classical geological and geotechnical methods in a consistent way, it is the authors' opinion that the proposed approach can be adopted by expert users in other geoenvironmental contexts, where landslides and their consequences represent a serious problem. Acknowledgments This work was carried out under the Commessa TA.P05.012 Tipizzazione di eventi naturali e antropici ad elevato impatto sociale ed economico of the CNR Department Scienze del sistema Terra e Tecnologie per l'Ambiente. The authors would like to thank the reviewers for their constructive and useful remarks. We are grateful to Dr. Takashi Oguchi, Editor of the journal, for his valuable comments and suggestions, which helped us to improve the scientific quality of the manuscript. References

Fig. 14. Instability scenarios obtained with TRIGRS and TRIGRS-unsaturated considering c ′ = 5 kPa and φ′ = 27°. Landslide inventory from 2010.

Antronico, L., Gullà, G., 2000. Slopes affected by soil slips: validation of an evolutive model. Proceedings of the VIII International Symposium on Landslides — Landslide in Research, Theory and Practice. Thomas Telford, London, UK, pp. 77–84. Antronico, L., Gullà, G., Borrelli, L., 2004. Shallow instabilities for sliding flow: regional influence and area affects. Proceedings of the IX International Symposium on Landslides, Rio de Janeiro, Brasile. Lacerda. Ehrlich, Fontoura & Sayao, London, UK, pp. 1381–1388. Antronico, L., Gullà, G., Oddo, B., 2001. Considerazioni preliminari sui risultati relativi alla caratterizzazione di soil slip in alcune zone di studio della Stretta di Catanzaro (Calabria). Mem. Soc. Geol. Ital. 56, 1–10. Barredo, J.I., Benavides, A., Hervas, J., Van Westen, C.J., 2000. Comparing heuristic landslide hazard assessment techniques using GIS in the Tirajana basin, Gran Canaria Island, Spain. ITC J. (Issue 1), 9–23. Baum, R.L., Savage, W.Z., Godt, J.W., 2002. TRIGRS-A FORTRAN program for transient rainfall infiltration and grid-based regional slope-stability analysis. US Geological Survey Open-File Report 02-0424 (Available via: http://pubs.usgs.gov/of/2002/ofr-02-424/). Benda, L.E., Cundy, T.W., 1990. Predicting deposition of debris flows in mountain channels. Can. Geotech. J. 27, 409–417. Blight, G.E., 1997. Interactions between the atmosphere and the Earth. Geotechnique 47, 715–767.

L. Cascini et al. / Geomorphology 241 (2015) 371–381 Borrelli, L., Cofone, G., Coscarelli, R., Gullà, G., 2014. Shallow landslides triggered by consecutive rainfall events at Catanzaro Strait (Calabria–Southern Italy). J. Maps http:// dx.doi.org/10.1080/17445647.2014.943814 (in press). Burbank, D.W., Anderson, R.S., 2001. Tectonic Geomorphology. Blackwell Scientific, Oxford. Calvello, M., 2012. Applicazione dei metodi intermedi alla Regione Campania. In: Cascini, L. (Ed.), Criteri di zonazione della suscettibilità e della pericolosità da frane innescate da eventi estremi (piogge e sisma). Composervice S.r.l, Padova, Italy. ISBN: 9788890687334, pp. 187–198. Calvello, M., Cascini, L., Mastroianni, S., 2013. Landslide zoning over large areas from a sample inventory by means of scale-dependent terrain units. Geomorphology 182, 33–48. Cascini, L., 2008. Applicability of landslide susceptibility and hazard zoning at different scales. Eng. Geol. 102, 164–177. Cascini, L., Matano, F., 2010. Scivolamento del muro in conci prefabbricati realizzato a valle della Centrale tecnologica, nell'ambito dei lavori di completamento del nuovo complesso demaniale della legione dei Carabinieri di Catanzaro. Technical report. Celico, P., De Vita, P., Monacelli, G., Scalise, A., Tranfaglia, G., 2005. Note illustrative della Carta Idrogeologica dell'Italia Meridionale. Tav. 2: Arco Ionico Tarantino, Tav. 3: Arco Calabro Peloritano. APAT and University of Naples Federico II. Ciurleo, M., 2012. A Methodological Approach for the Analysis of Shallow Landslides in Non-collapsible Soils. Ph.D. thesis. University of Salerno, Italy. Claessens, L., Schoorl, J.M., Veldkamp, A., 2007. Modelling the location of shallow landslides and their effects on landscape dynamics in large watersheds: an application for Northern New Zealand. Geomorphology 87, 16–27. Coe, J.A., Godt, J.W., Baum, R.L., Bucknam, R.C., Michael, J.A., 2004. Landslide susceptibility from topography in Guatemala. In: Lacerda, W.A., Erlich, M., Fontoura, S.A.B., Sayao, A.S.F. (Eds.), Landslides — Evaluation and Stabilization. Proceedings of the 9th International Symposium on Landslides. vol. 1. A.A. Balkema Publishers, London, pp. 69–78. Crozier, M.J., 2005. Multiple-occurrence regional landslide events in New Zealand: hazard management issues. Landslides 2, 247–256. Del Monte, M., Fredi, P., Lupia Palmieri, E., Marini, R., 2002. Contribution of quantitative geomorphic analysis to the evaluation of geomorphological hazards: case study in Italy. In: Allison, R.J. (Ed.), Applied Geomorphology: Theory and Practice. John Wiley and Sons Ltd., Chichester, UK, pp. 335–358. Eden, W.J., Mitchell, R.J., 1969. The mechanics of landslides in Leda clay. Can. Geotech. J. 7, 286–296. Eigenbrod, K.D., Kaluza, D., 1999. Shallow slope failures in clays as a result of decreased evapotranspiration subsequent to forest clearing. Can. Geotech. J. 36, 111–118. Fell, R., Corominas, J., Bonnard, C., Cascini, L., Leroi, E., Savage, W.Z., On behalf of the JTC-1 Joint Technical Committee on Landslides and Engineered Slopes, 2008a. Guidelines for landslide susceptibility, hazard and risk zoning for land-use planning. Eng. Geol. 102, 85–98. Fell, R., Corominas, J., Bonnard, C., Cascini, L., Leroi, E., Savage, W.Z., On behalf of the JTC-1 Joint Technical Committee on Landslides and Engineered Slopes, 2008b. Commentary of Guidelines for landslide susceptibility, hazard and risk zoning for land-use planning. Eng. Geol. 102, 99–111. Fredlund, D.G., Houston, S.L., Nguyen, Q., Fredlund, M.D., 2010. Moisture movement through cracked clay soil profiles. Geotech. Geol. Eng. 28, 865–888. Glade, T., Anderson, M., Crozier, M.J., 2005. Landslide Hazard and Risk. John Wiley & Sons Ltd., Chichester, UK. Gullà, G., Aceto, L., Niceforo, D., 2004. Geotechnical characterisation of fine-grained soils affected by soil slips. Proc. of the 9th International Symposium on Landslides, Rio de Janeiro, Brazil, pp. 663–668. Gullà, G., Mandaglio, M.C., Moraci, N., 2006. Effect of weathering on the compressibility and shear strength of a natural clay. Can. Geotech. J. 43, 618–625. Gullà, G., Aceto, L., Critelli, S., Perri, F., 2008. Geotechnical and mineralogical characterization of fine grained soils affected by soil slips. Proc. of the 10th International Symposium on Landslides, Xi'an, China, pp. 373–379. Guzzetti, F., 2008. The rainfall intensity–duration control of shallow landslides and debris flows: an update. Landslides 5, 3–17. Guzzetti, F., Mondini, A.C., Cardinali, M., Fiorucci, F., Santangelo, M., Chang, K.T., 2012. Landslide inventory maps: new tools for an old problem. Earth-Sci. Rev. 112, 42–66. Hancox, G.T., Wright, K., 2005. Analysis of terrain and landsliding caused by the 15–17 February 2004 rainstorm in the Wanganui–Manawatu Hill Country, southern North Island, New Zealand. Science Report 2005/11. Institute of Geological & Nuclear Sciences, Lower Hutt, New Zealand.

381

Kienholz, H., 1978. Maps of geomorphology and natural hazards of Grindelwald, Switzerland, scale 1:10,000. Arct. Alp. Res. 10, 169–184. Kirkby, M.J., 1987. General models of long-term slope evolution through mass movement. In: Anderson, M.G., Richards, K.S. (Eds.), Slope Stability. John Wiley and Sons Ltd., Chichester, UK, pp. 359–379. Leroueil, S., Vaunat, J., Picarelli, L., Locat, J., Lee, H.J., Faure, R., 1996. Geotechnical characterization of slope movements. Proc. 7th Int. Symp. on Landslides, Trondheim, Norway, pp. 53–74. Lim, T.T., Rahardjo, H., Chang, M.F., Fredlund, D.G., 1996. Effect of rainfall on matric suctions in a residual soil slope. Can. Geotech. J. 33, 618–628. Longhitano, S.G., Chiarella, D., Muto, F., 2014. Three-dimensional to two-dimensional cross-strata transition in the lower Pleistocene Catanzaro tidal strait transgressive succession (southern Italy). Sedimentology http://dx.doi.org/10.1111/sed.12138 (in press). Meisina, C., 2006. Characterisation of weathered clayey soils responsible for shallow landslides. Nat. Hazards Earth Syst. Sci. 6, 825–838. Montgomery, D.R., Dietrich, W.E., 1994. A physically based model for the topographic control on shallow landsliding. Water Resour. Res. 30, 1153–1171. Perri, F., Borrelli, L., Critelli, S., Gullà, G., 2014. Chemical and minero-petrographic features of Plio-Pleistocene fine-grained sediments in Calabria (southern Italy). Ital. J. Geosci. 133, 101–115. Perotto-Baldiviezo, H.L., Thurow, T.L., Smith, C.T., Fisher, R.F., Wu, X.B., 2004. GIS-based spatial analysis and modeling for landslide hazard assessment in steeplands, southern Honduras. Agric. Ecosyst. Environ. Rogers, N.W., Selby, M.J., 1980. Mechanisms of shallow translational landsliding during summer rainstorms: North Island, New Zealand. Geogr. Ann. 62 (A), 1–2. Sato, H.P., Harp, E.L., 2009. Interpretation of earthquake-induced landslides triggered by the 12 May 2008, M7.9 Wenchuan earthquake in the Beichuan area, Sichuan Province, China, using satellite imagery and Google Earth. Landslides 6, 153–159. http:// dx.doi.org/10.1007/s10346-009-0147-6. Savage, W.Z., Godt, J.W., Baum, R.L., 2004. Modeling time-dependent areal slope stability. In: Lacerda, W.A., Ehrlich, M., Fontoura, S.A.B., Sayao, A.S.F. (Eds.), Landslides: Evaluation and Stabilization. Balkema, Taylor & Francis Group, London, pp. 23–36. Selby, M., 1993. Hillslope Materials and Processes. Oxford University Press, Oxford. Silverman, B.W., 1986. Density Estimation for Statistics and Data Analysis. Chapman and Hall, New York. Soeters, R., van Westen, C.J., 1996. Slope instability recognition, analysis and zonation. In: Turner, A.K., Schuster, R.L. (Eds.), Transportation Research Board Special Report 247. Landslide Investigation and Mitigation, National Research Council, pp. 129–177. Sorbino, G., Sica, C., Cascini, L., Cuomo, S., 2007. On the forecasting of flowslides triggering areas using physically based models. Proceedings of 1st North American Landslides Conference, AEG Special Publication. 23, pp. 305–315. Sorbino, G., Sica, C., Cascini, L., 2010. Susceptibility analysis of shallow landslides source areas using physically based models. Nat. Hazards 53, 313–332. Sorriso-Valvo, M., Tansi, C., 1996. Grandi frane e deformazioni gravitative profonde di versante della Calabria. Note illustrative della Carta 1:250,000. Geogr. Fis. Dinam. Quat. 19, 395–408. Sorriso-Valvo, M., Antronico, L., Gaudio, R., Gullà, G., Iovine, G., Merenda, L., Minervino, I., Nicoletti, P.G., Petrucci, O., Terranova, O., 2004. Carta dei dissesti causati in Calabria meridionale dall'evento meteorologico dell'8–10 settembre 2000. CNR-GNDCI Publication. vol. 2859. Rubbettino Publisher, Soveria Mannelli, Italy. Tansi, C., Muto, F., Critelli, S., Iovine, G., 2006. Neogene–Quaternary strike–slip tectonics in the central Calabrian Arc (southern Italy). J. Geodyn. 43, 393–414. Van Dijk, J., Okkes, M., 1991. Neogene tectonostratigraphy and kinematics of Calabrian basins; implications for the geodynamics of the Central Mediterranean. Tectonophysics 196, 23–60. Van Westen, C.J., Rengers, N., Soeters, R., 2003. Use of Geomorphological Information in Indirect Landslide Susceptibility Assessment. Nat. Hazards 30 (3), 399–419. Varnes, D.J., 1978. Slope movement types and processes. In: Schuster, R.L., Krizek, R.J. (Eds.), Special Report 176: Landslides: Analysis and Control. Transportation and Road Research Board, National Academy of Science, Washington, D.C., pp. 11–33. Yin, K.J., Yan, T.Z., 1988. Statistical prediction model for slope instability of metamorphosed rock. Proc. 5th Int.l Symposium on Landslides. Lausanne, Switzerland, pp. 1269–1272.