A statistical study of precursor activity in rain-induced landslides

Spring-block models were introduced several years ago to model avalanche-like dynamics for catastrophic landslides. This article aims to address precursor phenomena in rain-induced landslides through the enrichment of a two-dimensional spring-block model with displacement gradients and stochasticity. This approach is along the lines of Aifantis and coworkers, who introduced deterministic gradients to stabilize material behavior in the softening regime and later expanded on it by introducing stochastic terms to model the competition between deterministic gradients and random effects. To this end, an appropriate stochastic constitutive relation, introducing an effective dissipation parameter (which is related to structural heterogeneity and deformation processes at the failure plane) is used. It is shown that there is definite precursor activity before rain-induced landslides that can be modeled by a process of material yielding at the failure plane (a “weak” plane at a certain depth, parallel to the surface of the soil or rock mass, over which landslide takes place). This activity is related to the exponent b of the power-law distribution of the slip events within the failure plane. Cellular automaton simulations were used to verify the analytical predictions, which can be useful to geoscientists and engineers, as an early prediction of the initiation of a landslide can minimize its catastrophic results through proper safety and precautionary measures.