Landslides in Cascadia: Using geochronometry and spatial analysis to understand the timing, triggering and spatial distribution of slope failures in the Pacific Northwest United States
MetadataShow full item record
Landslides kill hundreds to thousands of people every year, cause billions of dollars in infrastructure damage, and act as important drivers of landscape evolution. In the Pacific Northwest USA, landslides routinely block roads and railways and periodically destroy homes, as recently evinced by the catastrophic 2014 Oso Landslide, which killed 43 people. Ongoing mapping efforts, aided by the ever-growing availability of bare-earth lidar elevation data, have identified tens of thousands of landslides in Washington and Oregon States. Little is known about the timing of these slope failures, and without age constraints, it is impossible to assess recurrence frequency or understand past landslide triggers. In Chapter 2, I address this problem by developing a landslide dating technique which uses surface roughness measured from lidar data as a proxy for landslide age. Unlike other landslide dating methods such as radiocarbon, exposure dating, or dendrochronology, the surface-roughness dating technique can be practically applied on a regional scale and offers an important tool for estimating landslide recurrence interval and assessing changes in landslide frequency across space. I apply this roughness dating technique to a small study area directly surrounding the 2014 Oso Landslide and show that landslides with similar destructive capacity as the Oso Landslide have been occurring within the North Fork Stillaguamish Valley every few hundred years for the last few thousand years. Importantly, this suggests that the Oso Landslide was not an anomalous event. In Chapter 3, I apply this technique on a much larger scale, estimating the ages of 9,938 manually mapped landslides across ~10,000 km2 of similar bedrock geology in the central Oregon Coast Range, an area thought to experience violent ground shaking during Cascadia Subduction Zone earthquakes. By analyzing observed and simulated landslide frequency, I find little evidence for widespread coseismic deep-seated landsliding in this region and instead postulate that most deep-seated landslides in the inventory were triggered by rainfall. In the final research chapter, I analyze spatiotemporal patterns in deep-seated landslide frequency in the Oregon Coast Range. I find that landslide frequency decreased in the rain shadow of the Oregon Coast Range during both the early and late Holocene, times when the paleobotanical record indicates climate drying, while landslide frequency remained high on the wetter west side of the range.