A Four-Year Catalog of Cascadia Earthquakes Using the Cascadia Initiative Amphibious Array and Subspace Detection Algorithms

Abstract:

The Cascadia subduction zone has been a key target area in the NSF MARGINS and GeoPRISMS programs because its mysteries are important for understanding processes and hazards associated with this megathrust fault. Evidence from the geologic record of earthquake shaking and tsunami along the northwestern coast of North America highlight possible hazards, yet a lack of seismicity recorded with most modern seismograph systems has limited our understanding of when and where future large earthquakes may occur. Because the seismogenic portion of the Cascadia subduction zone is largely offshore, land-based seismic networks can miss small magnitude earthquake fault slip. The Cascadia Initiative onshore-offshore seismic deployment was thus a key instrumentation component of the MARGINS and GeoPRISMS programs. This seismic network, operating between July 2011-October 2015, has been used for a variety of structural imaging and earthquake source studies of the margin. Previous work using traditional STA/LTA detection using this data identified a small set of earthquakes on the megathrust (Stone et al. 2018). Our work leverages this catalog, as well as other possible megathrust events in catalogs from permanent land stations, in a subspace detection algorithm to find many more earthquakes within several clusters across the Cascadia subduction zone. We present a final catalog of the new detections, located using regionally appropriate velocity models from the literature. We highlight over 1600 newly detected earthquakes (duration magnitude 1.2-4.6) at subducted seamounts offshore central Oregon, and we delineate specific patches offshore northern California, central Oregon, and southern Washington to Vancouver Island. We evaluate these new event locations in the context of paleo-rupture segments and estimated asperities from the most recent great earthquake in 1700. We then describe along-strike variations in earthquake stress drop, determined using coda spectral ratios, for subsets of these events that correlate to observed geodetic locking variations.