Regional-Scale Coseismic Landslide Hazard Modeling and Consequence Analysis

Thesis (Ph.D.)--University of Washington, 2017Earthquake-induced, or coseismic, landslides occur in great number during moderate to large (>M5) earthquakes. These landslides typically occur across regions spanning hundreds to thousands of square kilometers. The widespread geographic distribution of coseismic landslides makes them, by definition, regional-scale events. Current regional-scale landslide hazard analyses are typically based solely on the analysis of shallow translational landslides, neglecting many mechanistically-diverse types of landslides triggered by earthquakes that have historically driven coseismic-losses. Existing landslide hazard models are then difficult to translate into coseismic landslide risk, the likelihood of loss of life or property due to such a landslide, as the modeled landslide is not representative of many observed landslides, and virtually no quantitative data on the consequences and vulnerability of structures are available for landslides. Here a new multi-modal framework for coseismic landslide hazard analysis is presented. Coseismic landslide hazard is developed from multiple, distinct, types of landslides to explicitly account for differences in failure mechanism, susceptible terrain, and consequences of each landslide. This work improves regional-scale coseismic landslide hazard and consequence analyses by developing mode-based hazard analyses to explicitly account for mechanistic differences in coseismic landslides and developing unique mode-specific landslide consequence data. These advances in modeling and knowledge of coseismic landslide consequences, paired with an assessment of the effect of anthropogenic modification on landslide susceptibility, provide a greatly improved framework in which coseismic landslide risk can be quantitatively assessed. Improved quantitative risk analyses will better enable mitigation strategies and decisions to be made to protect individuals, communities, and the built environment from future coseismic landslide losses. The multi-modal framework was developed to compute coseismic landslide hazard for Lebanon considering four landslide modes with unique source regions, failure mechanisms, and implications to human and economic risk. Coseismic landslide hazard was computed for 10% in 50-year probabilistic seismic hazard peak ground accelerations, as well as historic scenario earthquakes, to assess the model’s ability to match field observations of historic landslide activity and mapped coseismic landslides. High-quality empirical data on the physical consequences of coseismic rock impacts to residential dwellings is presented from the 2010-2011 Canterbury earthquake sequence. These quantitative measures of rock impacts are a critical missing piece in the analysis of quantitative risk assessment, and provide a foundation for life-safety decision-making anywhere residential homes are exposed to falling rock. Landslides triggered by the 2011 M9.0 Tohoku, Japan earthquake were mapped and analyzed to assess the role of topographic, climatic, anthropogenic, and ground motion parameters in the initiation of different landslide failure modes as well as the consequences to the built environment caused by each landslide type. A physically-based framework, based on a common simplified landslide initiation model, was found to describe the observed concentrations of landslides triggered by the Tohoku earthquake, whereas commonly-held explanatory metrics such as peak ground acceleration did not. Anthropogenic modification was found to significantly alter expected material strengths, and landslide susceptibility to failure, for debris slides triggered by the Tohoku earthquake. 42% of all Tohoku-triggered landslides impacted the built environment, destroying roadways and structures, and causing significant economic losses. 67% of all Tohoku-triggered landslides occurred in anthropogenically modified hillslopes. These data suggest many of the risks associated with large magnitude subduction zone earthquake-induced landslides may be of our own design, threatening many aggressively developed regions with decreased slope resistance to landslides and increased potential for loss. Probabilistic coseismic landslide hazard analyses are presented for Seattle, Washington based on a suite of 30 synthetic M9.0 Cascadia Subduction Zone earthquakes, the 2001 M6.8 Nisqually earthquake, and a scenario M7.0 synthetic Seattle Fault earthquake. A simplified 475-year return period probabilistic coseismic landslide hazard analysis, combining multiple sources of seismicity, was then developed for Seattle. The multi-modal coseismic landslide hazard analysis, combined with probabilistic estimates of material strength and ground shaking and empirical ground saturation data, is presented as a state-of-the-art regional-scale coseismic landslide hazard analysis, and as a basis for future analyses assessing Cascadia Subduction Zone coseismic landslide hazard across the Pacific Northwest