Mean landslide geometries inferred from a global database of earthquake- and non-earthquake-triggered landslides

Abstract

Ranging in size from very small to tremendous, landslides often cause loss of life and damage to infrastructure, property and the environment. They are triggered by a variety and combinations of causes among which the role of water and seismic shaking have the most serious consequences. In this regard, seismic wave amplification due to topography as well as to the impedance contrast between the landslide mass and its underlying bedrock are of particular interest. Therefore, high resolution reconstruction of the lateral confinement of the landslide mass and the exact measurement of the mechanical properties are a necessity. A global chronological database was created to study and compare 2D and 3D geometries of landslides, i.e. of landslides properly sliding on a rupture surface. It contains 277 seismically and non-seismically induced landslides whose rupture masses were measured in all available details allowing for statistical analyses of their shapes and to create numerical models thereupon based. Detailed studies reveal that values of distinct geometrical parameters have different statistical behaviors. As for dimension related parameters, occurrence frequencies follow decreasing exponential distributions and mean values progressively increase with landslide magnitude. In contrast, occurrence frequencies of shape-related parameters follow normal distributions and mean values are constant throughout different landslide magnitudes. Dimensions and shapes of landslides are thus to be regarded in a precise and distinctive manner when analyzing seismically induced slope displacements.