COSEISMIC LANDSLIDE REACTIVATION CHARACTERISTICS DETERMINED FROM DYNAMIC RING-SHEAR TESTING

Large earthquakes often cause widespread landsliding that alters landscapes and presents significant hazards to human safety and the built environment. For example, the 2008, Mw 7.9 Wenchuan, China earthquake triggered more than 56,000 landslides that killed about 20,000 people. Predicting the occurrence and nature of coseismic landslides remains elusive largely because limitations on laboratory apparatus and a lack of instrumental field observations have precluded understanding the basic response of geologic materials to seismically induced shearing. Coastal Oregon, USA is a region of numerous landslides and great subduction-zone earthquakes that recur every 300-500 yrs, the most recent of which occurred during January 1700. Reactivation of existing landslides during future great earthquakes could threaten human safety because many of these slides potentially impact tsunami evacuation and emergency response routes.

To better understand the coseismic shear behavior of landslides, we used specialized ring-shear apparatus capable of dynamic force application and pore-water pressure control to study the potential coseismic response of two landslides typical of coastal Oregon. We subjected specimens obtained from the basal shear zone of these landslides to dynamic stresses appropriate for the slides, their geological settings, and a great subduction-zone earthquake. These are the first tests of their kind performed to study coseismic landslide reactivation. We observed non-linear shear strength variation with shear rate and cumulative displacement. The strength characteristics resulted in some specimens failing catastrophically (i.e., infinite displacement at existing gravitational stresses) whereas others experienced only a few centimeters of displacement. Our findings suggest that assumed material strengths and methods typically used for predicting coseismic landslide displacement require modification; new material strength models are needed and the concept of threshold ground accelerations above which sliding occurs and below which stability exists is much too simplistic. We conclude that many existing landslides in coastal Oregon similar to those we studied will likely move several meters or more during future great earthquakes.