Studio degli effetti della vegetazione sulla stabilità dei pendii e sullo sviluppo delle frane superficiali

Rainfall-induced shallow landslides are among the most common gravitational mass movements on natural and artificial slopes; due to high frequency, high propagation velocity and lack of warning signs represent one of the most dangerous and destructive instability phenomena occurring in the world. The interest of the scientific community in this process has grown in the last three decades with the aim to perform shallow landslide susceptibility assessments at regional scale. The assessment of shallow landslide susceptibility by physically based methods usually focus on the parametrization of hydraulic and geotechnical features of soils, while the role of vegetation is generally overlooked. Instead the implementation of shallow landslide susceptibility models should consider both the engineering geological properties and vegetational characters of hillslope deposits. In this PhD thesis a fieldwork-based method is proposed to acquire, process and spatialize engineering geological, above-ground and below-ground vegetation features of hillslope deposits. The main goal of this research consists on characterizing the comprehensive role (at both macro and micro-scale) of vegetation towards the occurrence of shallow landslides. The thesis is organized in the following main topics: characterization of engineering geological properties and vegetational features of hillslope deposits in stable and unstable areas (i.e., shallow landslides), assessment of the correlations between structural roots cell components and shallow landslides occurrence through the application of spectroscopic techniques, implementation and comparison of shallow landslide susceptibility modelling by means of two different physically-based methods (pSHALSTAB and SlideforMAP susceptibility models) which differ in the implementation of the vegetation contribution. The study areas are both the Garfagnana and Cardoso basins (Northern Apennines, Italy). The Garfagnana study area extends for about 240 km2 along the Serchio River valley, parallel to the eastern margin of the Apuan Alps and the Northern Apennines Ridge, reaching the maximum altitude of about 1985 m a. s. l., with an average slope of 25°. The Cardoso study area (≃ 13 km2) is a small mountain sub-basin of the Versilia river located in the southern Apuan Alps, reaching the maximum altitude of 1858 m a. s. l., with an average slope of 38°. Fieldwork and laboratory tasks aimed at mapping engineering geology and vegetational characters of hillslope deposits were carried out for a set of hundreds of measurement sites, with the acquisition of hillslope deposit depth, geotechnical horizons, unit weight, above-ground and below-ground vegetation features as well as soil and roots samples for laboratory analysis. Root samples were analyzed through the implementation of different spectroscopic techniques (Raman, Fourier Transform Infrared-FTIR and elemental Laser Induced Breakdown Spectroscopy-LIBS). In order to obtain the map distribution of engineering geology parameters, a spatial analysis has been performed by clustering morphometric variables stratified as a function of bedrock lithological units. As for the below-ground vegetation results, the application of the Root Bundle Model (RBMw) showed that the lateral root reinforcement slightly increases with increasing stem size and slightly decreases with increasing distances from the stem. According to the RBMw results, most of the basal root reinforcement concentrates in the shallower slope deposit layers and decreases sharply with increasing depth. Differences in terms of root density (RAR) moving from unstable (i.e., shallow landslides) to more stable locations were observed. In order to implement the lateral root reinforcement into the infinite slope model and limit equilibrium approach the new parameter “equivalent root cohesion” has been defined. This parameter considering both the lateral root depth and the extension of the shallow landslide basal surface, shows magnitudes (few kPa) of the additional root cohesion commonly adopted in the literature. Root density data showed a moderate relationship with the hillslope deposit saturated hydraulic conductivity, with a decrease of root density and saturated hydraulic conductivity as the distance from the tree increases and the hillslope deposit depth increases. Differences in terms of lignin, cellulose and nutrient contents moving from unstable (i.e., shallow landslides) to more stable locations were observed using Raman, FTIR and LIBS techniques. Based on these experimental results spectroscopic portable techniques might be suitable to detect differences in the biochemical composition of wood depending on vegetational parameters, as well as to provide information for shallow landslide prediction. Shallow landslide susceptibility analysis was performed using two different physically based models: pSHALSTAB and SlideforMAP. The comparison between modelled unstable areas and the distribution of observed landslides was quantitatively performed in terms of ROC curve analysis. Performances of both models are comparable or even better than other physically based models described in literature, with AUROC values higher than 0.7 (numeric threshold generally considered as a reasonable prediction). Despite the fact that, especially for the SlideforMAP model, the bedrock geology controls the quality of the output maps (with lower reliability resulting for carbonate bedrock areas), the application of SlideforMAP allowed us to simulate different vegetation scenarios related to the Garfagnana basin and to depict protective and potential forests, suggesting the application of this model as a reliable tool in sustainable forest management and bioengineering studies.