Using predisposition factors for landslide hazard assessment: a new GIS-based methodology

Switzerland is exceptionally subjected to landslides; indeed, about 10% of its area is considered as unstable. Making this observation, its Department of the Environment (BAFU) introduced in 1997 a method to realize landslide hazard map. It is routinely used but, like most of the methods applied in Europe to map unstable areas, it is mainly based on the signs of previous or current phenomena (geomorphologic mapping, archive consultation, etc.) even though instabilities can appear where there is nothing to show that they existed earlier. Furthermore, the transcription from the geomorphologic map to the hazard map can vary according to the geologist or the geographer who realizes it: this method is affected by a certain lack of transparency and is not really efficient to forecast landslides. The aim of this project is to introduce the bedrock of a new method for landslide hazard mapping; based on instability predisposition assessment, it involves the designation of main factors for landslide susceptibility, their integration in a GIS to calculate a landslide predisposition index and the implementation of new methods to evaluate these factors; to be competitive, these process will have to be both cheap and quick. After showing that cohesion and hydraulic conductivity of loose materials were strongly linked to their granulometry and plasticity index, we implemented two new field tests, one based on teledetection and one coupled sedimentometric and blue methylen tests to evaluate these parameters. The hydraulic conductivity of fractured rocks was obtained from the analysis of their geometrical properties (fractures density, aperture size and orientation). The other factors were extracted from DEM and hydrologic mapping. We added a last factor related to the predisposition of the geotype (new classification for geologic formations, based on genetic standards for loose material and on lithologic standards for hard rock) to slope instability process: the latter enabled us to integrate attributes proper to each geotypes (over-consolidation for ground moraines, stratifications for glaciolacustrine deposits, etc...) and which would be long and complex to integrate to a GIS. Afterward, we implemented an ArcGis® toolbox allowing to lead to a landslide susceptibility index. Finally, we applied this methodology (from field survey to GIS operations) to ten sites in different contexts in Switzerland. This new methodology can be considered as a cheap and efficient way to forecast landslides.