Simulations of landslide hazard scenarios by a geophysical safety factor

摘要

Soil response to rainfall is a complex phenomenon that requires modeling of many sources of heterogeneity, whose variations can be relevant on various timescales and whose precise description requires a large amount of data inputs. Due to the great complexity of the problem, many simplifying assumptions are usually made in modeling landslides triggered by rainfall. As regards rainfall-induced shallow landslides, conventional approaches base slope stability analyses on the infinite slope model combined with hydrological models, which provide the time evolution of groundwater pressure head and volumetric water content. On the other hand, the response of geophysical quantities to water changes depends also on the variations in mechanical and hydrological properties. For this reason, we attempt a different approach to the problem of slope stability assessment by shifting the focus on the analysis of variations in geophysical properties. In this paper, starting from experimental resistivity data acquired in a test area, we perform a series of numerical simulations to study how changes in soil resistivity spatial distributions may affect the size of unstable areas. We use a simple cellular automaton whose states are defined by the values of a local and time-dependent geophysical factor of safety, which depends on soil electrical resistivity and slope inclination. We studied the probability of occurrence of rainfall-induced shallow landslide events by driving the system to instability through a decrease in electrical resistivity values. Numerical simulations are performed by varying number and intensity of the applied perturbations. Hazard scenarios obtained by in situ distributions of resistivity values are compared with those coming from initial random distributed resistivity values. Our results suggest possible critical rates of resistivity changes for triggering instability in the investigated area and point out the crucial role of resistivity variations in prediction of larger events