X-SLIP: A SLIP-based multi-approach algorithm to predict the spatial-temporal triggering of rainfall-induced shallow landslides over large areas

This paper describes the implementation of a multi-approach algorithm for the spatial and temporal prediction of soil slips over large areas, based on a simplified physically-based model: SLIP (Shallow Landslide Instability Prediction). It tries to overcome the limits of the widespread methods that use Artificial Intelligence to derive data-driven and strongly site-specific mathematical formulations, commonly not suitable for temporal predic-tion. All the required model parameters are evaluated for pixels of a grid, by applying to easily available ter-ritorial data different approaches: Finite Differences and Lagrange interpolation, Ray Cast Algorithm and Natural Neighbour method. The proposed algorithm is validated by application to two areas of the Emilia-Romagna Region (Italy) of a quite large extension (360 and 420 km2), where the position of soil slips occurred in the past is known. Results are shown to be aligned with others computed through another SLIP-based software. Through the Receiver Operating Characteristic (ROC) curve and the Area Under Curve (AUC), the excellent prediction of the model (AUC > 80%) is also demonstrated. The study remarks the importance of realistic spatial differentiation of soil parameters and modelling of rainfall interception by vegetation.

Automated summary

This paper presents a multi-approach algorithm, based on a simplified physically-based model, for predicting soil slips over large areas. The algorithm overcomes the limitations of AI-driven methods, which are not suitable for temporal prediction. Using different approaches, the algorithm evaluates model parameters for grid pixels based on easily available territorial data. The algorithm is validated in the Emilia-Romagna Region of Italy and demonstrates excellent prediction capabilities with an AUC > 80%.