No abstract

The coastal Pacific Northwest USA hosts thousands of deep-seated landslides. Historic landslides have primarily been triggered by rainfall, but the region is also prone to large earthquakes on the 1100-km-long Cascadia Subduction Zone megathrust. Little is known about the number of landslides triggered by these earthquakes because the last magnitude 9 rupture occurred in 1700 CE. Here, we map 9938 deep-seated bedrock landslides in the Oregon Coast Range and use surface roughness dating to estimate that past earthquakes triggered fewer than half of the landslides in the past 1000 years. We find landslide frequency increases with mean annual precipitation but not with modeled peak ground acceleration or proximity to the megathrust. Our results agree with findings about other recent subduction zone earthquakes where relatively few deep-seated landslides were mapped and suggest that despite proximity to the megathrust, most deep-seated landslides in the Oregon Coast Range were triggered by rainfall.

Documenting spatial and temporal patterns of past landsliding is a challenging step in quantifying the effect of landslides on landscape evolution. While landslide inventories can map spatial distributions, lack of dateable material, landslide reactivations, or time, access, and cost constraints generally limit dating large numbers of landslides to analyze temporal patterns. Here we quantify the record of the Holocene history of deep‐seated landsliding along a 25 km stretch of the North Fork Stillaguamish River valley, Washington State, USA, including the 2014 Oso landslide, which killed 43 people. We estimate the ages of more than 200 deep‐seated landslides in glacial sediment by defining an empirical relationship between landslide deposit age from radiocarbon dating and landslide deposit surface roughness. We show that roughness systematically decreases with age as a function of topographic wavelength, consistent with models of disturbance‐driven soil transport. The age‐roughness model predicts a peak in landslide frequency at ~1000 calibrated (cal) years B.P., with very few landslide deposits older than 7000 cal years B.P. or younger than 100 cal years B.P., likely reflecting a combination of preservation bias and a complex history of changing climate, base level, and seismic shaking in the study area. Most recent landslides have occurred where channels actively interact with the toes of hillslopes composed of glacial sediments, suggesting that lateral channel migration is a primary control on the location of large deep‐seated landslides in the valley.