Assessing the potential tsunami hazard of poorly-constrained active crustal faults in the northern Cascadia forearc

Probabilistic tsunami hazard assessment of seismic sources requires constraints on earthquake rupture area, rupture style, and frequency, as well as high-resolution bathymetry and topography to enable accurate tsunami modelling. For the Cascadia subduction zone megathrust, source characteristics are relatively well constrained by multiple datasets, although uncertainties remain, e.g., regarding the potential involvement of secondary faults, and event recurrence is approximated from on- and offshore paleoseismic data. Modelling and paleo-event data show that the greatest wave heights from megathrust tsunamis impact the outer coastlines and inlets from northern California to British Columbia. Within the Juan de Fuca and Georgia Straits between Vancouver Island and mainland North America, smaller waves can be expected from megathrust sources. However, this area of the Cascadia forearc hosts a number of other potential tsunami sources in the form of onshore-offshore crustal faults that accommodate dominantly right-lateral transpression on the northern limb of the Olympic orocline. Several of these faults have only recently been identified as Holocene-active or even mapped for the first time, and their tsunami hazard has yet to be assessed. Although crustal earthquakes have much lower magnitudes and longer recurrence intervals than great megathrust earthquakes, they may generate locally large tsunamis with little warning time between the onset of seismic shaking and the arrival of tsunami waves. We focus on improving hazard constraints for several crustal faults in the southern Strait of Georgia, including the Birch Bay, Sandy Point, and Skipjack Island fault zones. Challenges include a general lack of significant historical activity, limited paleoseismic and paleotsunami data, and poorly constrained sub-seafloor structure. We explore additional uncertainty that stems from the potential for complex ruptures on multiple fault strands and for significant local tsunamigenic contributions from coseismically-triggered landslides. Such challenges preclude quantitative probabilistic hazard assessment, but informed estimates and modelling of tsunamigenic rupture scenarios can provide useful hazard information for vulnerable communities.