Although widespread agreement that the Cascadia subduction zone produces great earthquakes of magnitude 8 to 9 was reached decades ago, debate continues about the rupture lengths, magnitudes, and frequency of megathrust earthquakes recorded by wetland stratigraphy fringing Cascadia’s estuaries. Correlation of such coastal earthquake evidence along the subduction zone has largely relied on relative position in a stratigraphic sequence and maximum-limiting 14C ages with errors of decades to hundreds of years. Offshore, a 10,000-year record of turbidites in marine cores is interpreted as an archive of strong shaking from great earthquakes, with an average frequency of about 500 years in northern Cascadia versus 200-300 years in southern Oregon and northern California. Onshore, fewer events marked by sharp (<3 mm) peat-mud (mud-over-peat) contacts in tidal wetland stratigraphic sequences have been widely inferred to record sudden relative sea-level rise due to coseismic subsidence during megathrust earthquakes: 4-7 sharp subsidence contacts in 3500 years at estuaries in northern Oregon and southern Washington (500-800 year average recurrence), and 9-12 sharp subsidence contacts in over 6000 years in sequences in central and southern Oregon (500-900 year average recurrence). Improved understanding of the onshore and offshore records is critical to the assessment of earthquake hazard in western North America and of tsunami hazard in the Pacific basin. However, because dating the turbidite record is inherently much less precise than are age models for subsidence events in the most thoroughly studied tidal wetland sequences, accurate reconstruction of the times of Cascadia’s great earthquakes depends on the ages from the onshore record.
Although methods to reduce uncertainty in the limits of resolution of tidal stratigraphy for recording earthquakes of a particular magnitude, and ways to distinguish earthquake subsidence stratigraphic contacts from non-seismic contacts, have been discussed for decades (e.g., Nelson, 1992; Atwater, 1992; Darienzo et al., 1994; Nelson et al., 1996a; Atwater and Hemphill-Haley, 1997; Witter et al., 2001; Kelsey et al., 2002; Nelson et al., 2006; Graehl et al., 2014; Milker et al., 2016), consensus about the threshold of resolution (minimum identifiable evidence of an earthquake) of tidal stratigraphy and, therefore, the completeness of Cascadia’s coastal record of great earthquakes, remains elusive. Although the most distinct, widespread contacts likely record close to a meter of coastal subsidence during the greatest megathrust earthquakes (e.g., M8.8-M9), other contacts may record <0.5 m of subsidence onshore of patches of low stress release on the megathrust during great earthquakes, during lesser megathrust earthquakes (e.g., M8.2-8.6), or from localized subsidence near upper-plate faults that slip during or independently of megathrust earthquakes (Nelson et al., 1996b; Wang et al., 2013; Kemp et al., 2018).
At the Siuslaw River estuary in central Oregon (lat. 43.97°) a stratigraphy of 9-12 peat-mud contacts, similar to those described from many Cascadia estuaries, may record a greater number of earthquakes during the past 2000 years than at any other of the tens of tidal wetland sites to the north and south. Here, as well as at many tens of other Cascadia tidal wetland sites, peat-mud contacts mark the tops of couplets of tidal flat and low marsh mud gradually shoaling upward into middle and high marsh peat. Along core transects across an 800-m-wide, island marsh in the Siuslaw River, we traced the 9 most continuous of 12-15 peaty beds dating from the past 2000 years for 250-500 m, but we had difficulty correlating the 3-6 intervening beds >50-100 m. We attribute the sharper, more extensive upper contacts on peaty beds—two capped by sandy beds probably deposited by tsunamis—to sudden coseismic subsidence of middle and high marshes, but origins for other upper and lower contacts boundi