Early warning systems and time series modelling : A new challenge for landslide risk prevention

International audienceThis paper deals with the application of statistical methods for the analysis of a multi-parameter data set recorded at an active landslide, the Ruines de Séchilienne (France). Here a monitoring system was set up on the western edge of the existing large active zone. It is based on an integrated technological platform, which combines microseismic, geotechnical, hydrogeological and meteorological measurements, together with three-dimensional displacement measurements. The work, still in progress, illustrates difficulties in modelling the times series due to the ongoing self-accelerating deformation processes of the landslide. In-depth processing and analysis of multi-parameter time series remains a tricky task when using only data-based statistical methods. Additional investigations are being made to set-up statistical analysis methods suitable for automated use on a routine basis