A NEW APPROACH TO ASSESS THE SUSCEPTIBILITY TO SHALLOW LANDSLIDES AT REGIONAL SCALE AS INFLUENCED BY BEDROCK GEO-MECHANICAL PROPERTIES

Due to high velocity, high frequency and the lack of warning signs, shallow landslides represent a major hazardous factor in mountain regions. Moreover, increasing urbanisation and climate changes triggering intense rainfall events make shallow landslides a source of widespread risk. The interest of the scientific community in this process has grown in the last three decades with the aim to perform robust shallow landslide hazard assessment at regional scale. Generally, these slope failures involve relatively small volumes of material sliding along with a planar shallow rupture surface. In the literature it is widely accepted that shallow landslides involve only slope deposit (or colluvium) and the sliding surface correspond to the discontinuity between bedrock and the overlying loose material. The fieldwork conducted in this thesis highlighted that often shallow landslides involve also the weathered and fractured portion of bedrock. In this framework, the implementation of shallow landslides susceptibility modelling should take into account the engineering geological properties of slope deposits, as well as of the underlying bedrock. In this thesis a fieldwork-based method is proposed to acquire, process and spatialize engineering geological properties of slope deposits and bedrock. The aims of this thesis were to compile a new multi-temporal shallow landslide inventory, characterize the engineering geological properties of slope deposits and bedrock, implement and compare shallow landslide susceptibility modelling by means a physically-based and a data-driven methods and explore the role of bedrock in shallow slope failures. The study area corresponds to a 242 km2 portion of the Garfagnana basin (Northern Apennines), a mountainous region where the elevation ranges between 150 and 2000 m a.s.l. characterized by an incised and rugged morphology with steep slopes (average 28° degrees) and a mean annual rainfall between 1500 and 2500 mm/year. From a geological point of view, the Garfagnana basin is a narrow intra-mountainous valley, interposed betweeen the Alpi Apuane metamorphic complex to the east and the sedimentary northern Apennine’s ridge to the west. The fieldwork and laboratory tasks carried out to map engineering geology characters of slope deposits consisted on a set of hundreds of field sampling points, with the acquisition of depth to the bedrock, geotechnical horizons, unit weight, as well as soil samples for lab analysis. The distribution of points was chosen by observing that engineering geology properties of slope deposits depend on both bedrock lithology and morphometric conditions. In order to obtain the map distribution of engineering geology parameters, we implemented a spatial analysis by clustering morphometric variables stratified as a function of bedrock lithological units. In order to investigate the engineering geology characteristics of the bedrock, a field survey aimed to classify rock masses was conducted. For each survey site, 200-400 Schmidt hammer rebound measures, bedding and joint data, GSI (Geological Strenght Index) and samples for laboratory analyses (unit weight and slake durability test) were collected. The field data were processed and spatially analyzed by means uni-variate and multi-variate cluster analysis in order to delineate domains with different bedrock geo-mechanical properties. The shallow landslide susceptibility analysis was performed using both data-driven, Information Value, and physically-based, a modified version of SHALSTAB model (PROBSS), methods. The numerical modelling faced three issues: a) the comparison of PROBSS and Information value (IV) in the prediction of shallow landslides involving SD; b) the training and cross-validation of IV models using shallow landslides involving bedrock or not; c) implementation of a physically-based model to predict involving bedrock shallow landslides. First of all, the results highlight that the field-based methods proposed here to evaluate engineering geological properties of slope deposits and bedrock are adequate for the implementation of regionalised physically-based susceptibility models. The comparison between PROBSS and IV highlights that the simplification of shallow landslides adopted by the infinite slope model which do not take into account the occurrence of a sliding surface located below the slope deposits / bedrock discontinuity, may affect the performance of physically-based susceptibility models. The accuracy of IV model is slightly better that PROBSS model. Having implemented two data-driven susceptibility models using two different training datasets highlighted the different characteristics that slope deposits and bedrock involving shallow landslides have, suggesting and demonstrating that the latter occur in conditions that the physically based model cannot predict. By placing the slip surface below the discontinuity between slope deposits and bedrock and providing shear strength parameters compatible with a weathered and fractured rock material, satisfactory accuracy result was obtained with PROBSS model.