Application of a physically based model to forecast shallow landslides at a regional scale

Teresa Salvatici,Hervè Stevenin,Veronica Tofani,Carlo Tacconi Stefanelli,Elena Benedetta Masi,Nicola Casagli,Filippo Catani,Miriana Petrolo,Guglielmo Rossi,Michele D'Ambrosio,Sara Ratto,Veronica Pazzi,Pietro Vannocci,Ascanio Rosi

doi:10.5194/nhess-18-1919-2018

Abstract

Abstract. In this work, we apply a physically based model, namely the HIRESSS (HIgh REsolution Slope Stability Simulator) model, to forecast the occurrence of shallow landslides at the regional scale. HIRESSS is a physically based distributed slope stability simulator for analyzing shallow landslide triggering conditions during a rainfall event. The modeling software is made up of two parts: hydrological and geotechnical. The hydrological model is based on an analytical solution from an approximated form of the Richards equation, while the geotechnical stability model is based on an infinite slope model that takes the unsaturated soil condition into account. The test area is a portion of the Aosta Valley region, located in the northwest of the Alpine mountain chain. The geomorphology of the region is characterized by steep slopes with elevations ranging from 400 m a.s.l. on the Dora Baltea River's floodplain to 4810 m a.s.l. at Mont Blanc. In the study area, the mean annual precipitation is about 800–900 mm. These features make the territory very prone to landslides, mainly shallow rapid landslides and rockfalls. In order to apply the model and to increase its reliability, an in-depth study of the geotechnical and hydrological properties of hillslopes controlling shallow landslide formation was conducted. In particular, two campaigns of on site measurements and laboratory experiments were performed using 12 survey points. The data collected contributed to the generation of an input map of parameters for the HIRESSS model. In order to consider the effect of vegetation on slope stability, the soil reinforcement due to the presence of roots was also taken into account; this was done based on vegetation maps and literature values of root cohesion. The model was applied using back analysis for two past events that affected the Aosta Valley region between 2008 and 2009, triggering several fast shallow landslides. The validation of the results, carried out using a database of past landslides, provided good results and a good prediction accuracy for the HIRESSS model from both a temporal and spatial point of view.

Highlights

Landslide prediction at a regional scale can be performed following two approaches: (a) using rainfall thresholds based on the statistical analysis of rainfall and landslides, and (b) using physically based deterministic models
While the first approach is currently extensively used at regional scales (Aleotti, 2004; Cannon et al, 2011; Martelloni et al, 2012; Rosi et al, 2012; Lagomarsino et al, 2013), the latter is more frequently applied at slope or catchment scales (Dietrich and Montgomery, 1998; Pack et al, 2001; Baum et al, 2002, 2010; Lu and Godt, 2008; Simoni et al, 2008; Ren et al, 2010; Arnone et al, 2011; Salciarini et al, 2012, 2017; Park et al, 2013; Rossi et al, 2013)
The model was applied in back analysis to two past rainfall events that triggered several shallow landslides in the study areas between 2008 and 2009

Summary

Introduction

Landslide prediction at a regional scale can be performed following two approaches: (a) using rainfall thresholds based on the statistical analysis of rainfall and landslides, and (b) using physically based deterministic models. The poor knowledge of hydrological and geotechnical parameters’ spatial distributions, caused by the extreme heterogeneity and inherent variability of soil at large scales (Mercogliano et al, 2013; Tofani et al, 2017), means that the application of physically based models is generally avoided at regional scales. Physically based models allow for the spatial and temporal prediction of the occurrence of landslides with high accu-. T. Salvatici et al.: Application of a physically based model to forecast shallow landslides at a regional scale racy, producing accurate hazard maps that can be of help for landslide risk assessment and management

Objectives

Methods

Results

Discussion

Conclusion