Regional-scale modelling of liquefaction-induced shallow landslides in unsaturated slopes

Regional modelling of rainfall-induced landslide triggering poses several technical challenges. These events can originate from a number of hydro-mechanical processes, such as soil-strength degradation, development of localized zones of pore-water pressures, liquefaction, among others. At the same time, the interplay between the spatial variability of topographic attributes, soil properties and transient infiltration can lead to a widespread distribution of distinct slope failure mechanisms across the same landscape. To this aim, this contribution describes a simulation platform for the efficient generation of storm-induced, landslide susceptibility maps in which different slope instability mechanisms can be considered. The framework relies on a vectorized finite element (FE) algorithm that performs fully-coupled simulations of transient infiltration in unsaturated soils, while input and output processing stages are linked to a Geographical Information System. To illustrate the capabilities of the proposed framework, the role of several hydro-mechanical processes on the inception of slope instability are first explored (i.e., coupled flow-deformation analyses, constitutive couplings). After this, results of regional-scale simulations are presented, where it is shown that such considerations can affect the computed spatio-temporal patterns of landslide triggering. Lastly, approaches to incorporate uncertainty of input data into landslide susceptibility zonation by using spatially-correlated random fields are discussed. The proposed framework provides an important step towards the development of robust, physically-based models for regional landslide hazard assessment.