This study was presented at the Seismological Society of America’s 2010 Annual Meeting in Portland, Oregon.
Robert C Witter, Oregon Dept of Geology and Mineral Industries, Newport Coastal Field Office, PO Box 1033, Newport, OR 97365,
Yinglong Zhang, OGI School of Science and Engineering, Oregon Health & Science University, Portland, OR 97291 1000;
Chris Goldfinger, COAS, Oregon State University, Corvallis, OR 97331;
George R Priest, Oregon Dept of Geology and Mineral Industries, Newport Coastal Field Office, PO Box 1033, Newport, OR 97365. email@example.com
Kelin Wang, Geological Survey of Canada, Pacific Geoscience Centre, 9860 West Saanich Road, Sidney, BC V8L 4B2, Canada.
Marine and coastal paleoseismic evidence for Cascadia subduction earthquakes imply a range of rupture scenarios that provide model inputs for tsunami simulations. 41 turbidites from submarine channels along the length of the margin define a mean Holocene recurrence of ~530 yr for ruptures ≥800-km-long and ~240 yr for southern Cascadia earthquakes that ruptured 3 shorter segments. Coastal paleoseismic records spanning the past ~7000 yr include 13 tsunami deposits archived in Bradley Lake in southern Oregon. We test the smallest Cascadia tsunami scenarios capable of reaching the lake for consistency with paleoseismic data.
Earthquake scenarios employ either: 1) regional rupture with slip distribution symmetrically tapering to zero up and down dip; or 2) regional rupture diverting slip onto an offshore splay fault. Maximum slip in each scenario varies as the product of selected recurrence intervals and the convergence rate. Using the hydrodynamic model SELFE, we ran >50 tsunami simulations on numerical grids that reflect inferred changes in coastal paleotopography.
Simulating the 1700 tsunami requires earthquake slip equivalent to ≥400 yr of convergence using the regional symmetric slip model. Augmenting uplift with a splay fault reduces the recurrence time to 360 yr – still longer than the 174 to 341 yr range of paleoseismic intervals that correlate with tsunami deposits in the lake. Earlier tsunamis, likely smaller than the 1700 wave, probably followed the largest earthquakes when the shoreline migrated to its most landward position due to subsidence and coastal erosion. Tsunami simulations with these conditions require a minimum recurrence of 280 yr. Other factors like seafloor acceleration or extreme tides may account for the smallest Cascadia tsunamis that reached the lake. Alternatively, these small events may release stored strain from previous earthquake cycles.
Figure 1. Index maps. a, Plate-tectonic setting of the northwestern United States. Bold arrows show plate convergence at 40 mm yr-1. b, Holocene sand dunes ornament broad Pleistocene marine terraces along the coast near Bradley Lake (map at left). Evidence of historical change in shoreline position includes historical photographs (at right), a wave eroded sea cliff mantled by dune sand and the shoreline in 1925 mapped by early coast surveyors (ref).
Figure 2. Correlation of radiocarbon data from coastal paleoseismic sites in south-central Oregon and offshore turbidite sequences from Hydrate Ridge and Rogue Channel. Stratigraphic correlations of turbidites (ref Goldfinger et al., 2010) differentiate long (>500 km) ruptures (bold dashes) of most of the fault from shorter (<500 km) segment breaks (thin dashes) along the southern margin. Up arrows denote maximum age estimates on detrital samples. Symbol width represents relative size of earthquake inferred from deposit thickness or amount of subsidence evident in change in microfossil assemblages.
Figure 3. Cascadia earthquake scenarios. a, Vertical deformation produced by three scenarios at the latitude of Bradley Lake. Rectangles span the range of subsidence inferred from fossil diatom assemblages (refs Kelsey et al., 2002; Witter et al., 2003). b, Profile of fault slip distribution at the latitude of Bradley Lake. Red line shows slip truncated where the splay fault intersects the sea floor. Bold black line delineates slip patch on plate interface.
Figure S1. Stratigraphic evidence of disturbances in Bradley Lake sedimentary record caused by great Cascadia earthquakes and their tsunamis (Kelsey et al., 2005). a, Conceptual model of tsunami inundation in a coastal lake that leaves a sand deposit in the sedimentary record. b, Debris layers that interrupt lake record reflect earthquake shaking that destabilizes basin walls but reflect no tsunami inundation.
Figure S2. Maps and profiles of numerical grids used in tsunami simulations. Grid 5 reflects the modern topography derived from lidar data (Oregon Lidar Consortium, 2009) and includes a straight outlet channel reconstructed from historical maps and photos (refs). Grid 6 depicts the inferred landscape in 1700 based on the position of the 1925 shoreline. Grid 7 represents the most landward possible position of the shoreline based on the presence of a paleo-seacliff buried by sand dunes. Profiles (lower right) along the outlet channel show the progressive landward shift in shoreline in Grids 5−7, respectively.