Reconstruction of the evolution phases of a landslide by using multi‑layer back‑analysis methods

Back analysis is the most common method to study landslide movements after the event, and it allows us to understand how a landslide evolved along the slope. This paper presents the back-analysis of the Pomarico landslide (Basilicata, Italy) that occurred on January 25th, 2019, on the southwestern slope of the Pomarico hill. The landslide, of rotational clayey retrogressive type—planar sliding, evolved in different phases until it caused a paroxysmal movement in the early afternoon on January 29th, 2019. The landslide caused the collapse of a bulkhead (built at the end of the twentieth century) and of some buildings along the village’s main road. In this paper, a multi-layer back-analysis study is presented, based on the limit equilibrium model (LEM), applying the solution proposed by Morgenstern and Price in Geotechnique 15(1):79–93zh, (1965) and implemented in the freeware software SSAP 2010. The analysis allowed the reconstruction of the entire landslide evolution, using geotechnical parameters obtained from both laboratory and in situ tests, and data from the literature. The application of multilayer back-analysis made it possible to avoid the homogenisation of the layers, modelling the event according to the real conditions present on the slope. The use of the SSAP software made it possible to curb the problem related to the theoretical limitation of the shape of the rupture surfaces, by evaluating independently the friction angle locally and by discarding all those surfaces, which, due to this problem, presented a non-reliable factor of safety (FS) value. The modelling revealed a slope that is highly unstable as the height of the water table changes. The FS calculated under water table conditions close to ground level was less than 1 (FS = 0.98), simulating the first landslide movement (November 2018). The subsequent model reconstructed the critical surface responsible for the January 2019 movement and calculated the FS present on the slope (FS = 1.01). Eventually, the paroxysmal event on January 29th, 2019, was modelled, returning an FS of 0.83, and a sliding surface that sets below the bulkhead, causing its failure. Furthermore, the modelling of the slope in the presence of adequate retaining structures demonstrated the (non-) effectiveness of the retaining wall system represented by the bulkhead. The proposed method of analysis suggests further applications in similar complex multi-layer soil-structure interaction scenarios.