Hydrogeological threshold using effective rainfall and support vector machine (SVM) applied to a deep seated unstable slope (Séchilienne, French Alps)

Aurélien Vallet (1) , Catherine Bertrand (1) , Davit Varron (2) , Jacques Mudry (1)

1 LCE - Laboratoire Chrono-environnement (UMR 6249)
2 LMB - Laboratoire de Mathématiques de Besançon (UMR 6623)

Rainfall threshold is a widely used method for estimating minimum critical rainfall amount which can yield to slope failure. Literature reviews shows that most of the threshold studies are subjective and not optimal. In addition, for this study effective rainfall was considered for threshold definition. Support vector machine (SVM) and automatic events identification were used in order to establish an optimal and objective threshold for Séchilienne landslide. The method has been designed in order to be easily incorporated in a landslide warning system. Hydrogeological threshold results show similar performance and configuration for effective rainfall and raw rainfall. Accuracy of the two thresholds is comparable and the index combination antecedent/precedent is pretty much identical, 9D/4D and 10D/5D for effective and raw rainfall respectively. Accuracy of both thresholds defined for Séchilienne landslide (> 86%), makes it appropriate to be integrated in a landslide warning system.

Mots clés: hydrogeological threshold; support vector machine; landslide

Domaines: Hydrologie; Milieux et Changements globaux; Pétrographie

Aurélien Vallet, Catherine Bertrand, Davit Varron, Jacques Mudry. Hydrogeological threshold using effective rainfall and support vector machine (SVM) applied to a deep seated unstable slope (Séchilienne, French Alps). Journée Aléa Gravitaire, Sep 2013, Grenoble, France. pp.1-6. ⟨hal-01062389v2⟩