Assessing Margin‐Wide Rupture Behaviors Along the Cascadia Megathrust With 3‐D Dynamic Rupture Simulations

From California to British Columbia, the Pacific Northwest coast bears an omnipresent earthquake and tsunami hazard from the Cascadia subduction zone. Multiple lines of evidence suggests that magnitude eight and greater megathrust earthquakes have occurred ‐ the most recent being 321 years ago (i.e., 1700 A.D.). Outstanding questions for the next great megathrust event include where it will initiate, what conditions are favorable for rupture to span the convergent margin, and how much slip may be expected. We develop the first 3‐D fully dynamic rupture simulations for the Cascadia subduction zone that are driven by fault stress, strength and friction to address these questions. The initial dynamic stress drop distribution in our simulations is constrained by geodetic coupling models, with segment locations taken from geologic analyses. We document the sensitivity of nucleation location and stress drop to the final seismic moment and coseismic subsidence amplitudes. We find that the final earthquake size strongly depends on the amount of slip deficit in the central Cascadia region, which is inferred to be creeping interseismically, for a given initiation location in southern or northern Cascadia. Several simulations are also presented here that can closely approximate recorded coastal subsidence from the 1700 A.D. event without invoking localized high‐stress asperities along the down‐dip locked region of the megathrust. These results can be used to inform earthquake and tsunami hazards for not only Cascadia, but other subduction zones that have limited seismic observations but a wealth of geodetic inference. Plain Language Summary The largest earthquakes on Earth occur along faults that develop between two tectonic plates that come into contact. Termed megathrust earthquakes, these catastrophic events are responsible for generating both strong ground‐shaking and tsunamis. The Cascadia megathrust fault straddles the Pacific coastline of North America and from evidence in both the United States and Japan, we know this fault last slipped in 1700 A.D. We have combined models of strain buildup (geodetic coupling models) with state‐of‐the‐art 3‐D computer simulations to understand the potential hazard of a future earthquake in Cascadia and show what factors might lead to the fault slipping its entire length. We compare our simulations to geologic measurements of permanent ground movement from 1700 A.D. Our results demonstrate that no matter where the earthquake i