USING LIDAR-BASED ROUGHNESS DATING TO PARSE RAINFALL VERSUS COSEISMIC LANDSLIDE TRIGGERING ALONG THE CASCADIA SUBDUCTION ZONE (Invited Presentation)

The Pacific Northwest United States (PNW) is subject to two primary landslide triggers: frequent winter storms and infrequent high magnitude earthquakes. The increasing availability of lidar bare-earth imagery has revealed many thousands of landslides beneath the thick canopy of trees which blanket much of the region. For all its benefit in landslide identification, the applications of lidar data extend beyond improved mapping. Lidar can also be used to quantitatively analyze morphology and extract critical information about landslide timing. Previous work has shown that landslide surface roughness measured using lidar decreases with time at a predictable rate, and can therefore be used to estimate landslide age on a regional scale. This approach allows for better constraints on when and where landslides have occurred in the past, which may help in parsing the relative importance of rainfall versus seismic triggering and could lead to improved landslide susceptibility modeling. These are especially important topics in the PNW, where a wet climate is compounded by magnitude 8-9 earthquakes along the Cascadia Subduction Zone every 300-500 years, and as recently as 1700 CE. Here, we use surface roughness dating to estimate the ages of ~10,000 manually mapped deep-seated landslides in the Tyee Formation of the central Oregon Coast Range, where predicted peak ground accelerations are expected to exceed 0.6g during future megathrust earthquakes. We first interrogate the landslide age-frequency data and estimate bounds on the number of landslides within our study area likely triggered during the 1700 CE earthquake. We then examine spatial patterns in landslides with ages around the time of the 1700 CE earthquake and compare these patterns with predicted susceptibility using traditional Newmark sliding block analyses. Our results show that coseismic landslides account for less than half of all recent (<1,000 ybp) deep-seated landslides, and in many places are anticorrelated with predicted susceptibility. Instead, rainfall driven landslides comprise the majo­rity of mapped slope failures. These findings suggest that even within the same rock type, variations in strength, structure, or rainfall contribute more towards driving deep-seated landslides than peak ground acceleration in the PNW.