A novel prototype Landslide Early Warning System accounting for variable antecedent soil hydrological status

The hazard assessment of rainfall-triggered shallow landslides commonly relies on the recognition of rainfall thresholds that, coupled with methods for rainfall forecasting, are a main part of Landslide Early Warning Systems. Unfortunately, their reliability is affected by relevant uncertainties mainly linked to the lack of both inventories of past landslides and associated rainfall. Furthermore, commonly used rainfall thresholds do not consider the significant effect of antecedent soil hydrological status on the initiation of rainfall-triggered shallow landslides, which makes them subject to missed alarms and false alarms.

This work poses the basis for the development of a novel prototype Landslide Early Warning System that incorporates antecedent soil hydrological conditions and works with both nowcasted or real-time measured rainfall and soil hydrology monitoring (or occasional measurements). Starting from physics-based Intensity–Duration rainfall thresholds already known in the literature for landslides involving ash-fall pyroclastic soil-mantled slopes of the Campania region, empirical time-decay relationships for forecasting the initial soil water pressure head value were statistically analyzed long-lasting time series of field monitoring. This achievement allowed the reconstruction of a 3D hydrological threshold built on a multivariate regression model of three variables: rainfall intensity, rainfall duration, and initial soil water pressure head.

Statistical analysis of the distribution of time-decay coefficient values reveals varying mean and median values (the latter is less influenced by outliers), particularly with lower values in the wet season, where soil contributes to slower variation over a relatively narrow interval near saturation. Comparison between real (i.e., monitored time series) and modeled data (i.e., using time-decay laws) shows a good match.

The results obtained are intended to be an in-depth study on the importance of the antecedent soil hydrological conditions on critical rainfall triggering debris flows. They can be conceived as generally advancing the definition of Landslide Early Warning Systems based on dynamic hydrologic landslide-predisposing factors, generally applicable in all cases in which the lack of past landslide inventories and/or the unreliability of corresponding rainfall records, as well as different antecedent soil hydrological conditions, prevent the setting of effective empirical rainfall thresholds.