The wetland sediments fringing estuaries at the Cascadia subduction zone harbor a unique record of plate-boundary earthquakes during the past 5,000 years. However, the precision of past measurements of land-level changes at Cascadia is low (errors of greater than plus or minus 0.5 meter), the measurements are spatially limited, and they span only fractions of complete cycles of subduction zone strain accumulation and release. This makes past measurements insufficient for determining which hypotheses of plate-boundary deformation are most valid. This deficiency will be redressed by applying recently developed statistical transfer functions to microfossils, such as foraminifera and diatoms, collected from Cascadia estuarine sediments. Similar studies of sea-level change on other continents have obtained an unprecedented vertical resolution of plus or minus 0.1-0.3 meter. Seasonal and spatial differences in modern foraminfera and diatoms from Cascadia estuaries are assessed and the results are used to improve our microfossil-based transfer functions. Transfer functions are applied to fossil foraminiferal and diatom data at several estuaries to reconstruct land-level changes spanning the four most recent great earthquake cycles in the central Cascadia subduction zone. Detailed lithostratigraphical and biostratigraphical descriptions and radiocarbon dating of estuarine sediment allow correlation of microfossil reconstructions among estuaries and help reconstruct a history of plate-boundary deformation in Oregon. The improved vertical resolution of our reconstructions will: (1) yield more precise measures of Cascadia deformation over time periods of 10 to 100 years; (2) provide critical tests for competing hypotheses of coastal uplift versus subsidence just prior to great earthquakes; (3) help constrain regional models of Cascadia plate-boundary deformation; and (4) directly test hypotheses of slip-predictable, time-predictable, and slip-time-unpredictable strain accumulation proposed for other subduction zones.
Powerful magnitude 8 and 9 earthquakes and associated tsunamis during the last decade in Indonesia, Chile, and Japan, are a stark reminder of the threat that the Cascadia subduction zone poses to the population of northern California, Oregon, Washington, and British Columbia. No written evidence documents great megathrust earthquakes and their tsunami in northern California, Oregon, or Washington. With the exception of historical documents that describe a tsunami in Japan generated by the most recent Cascadia great earthquake (26 January AD 1700), records from submerged coastal forests and wetlands provide the only evidence for large earthquakes on this coastline. Our study builds on previous research that identified the timings of past earthquakes and broadly estimated the amount of coastal subsidence. Previous research did not have the resolution to compare the variations in subsidence during great earthquakes both among sites (spatial variability) and among different earthquakes (temporal variability). We used statistical techniques to produce estimates of earthquake subsidence using microscopic fauna called foraminifera that are found in marine, estuarine, and tidal marsh environments. Our estimates of coastal subsidence are precise enough to address issues of temporal and spatial variability. We collected over 150 modern samples of foraminifera from different elevations in tidal flat and marsh environments. We developed models that related the different species of foraminifera found in these samples to elevation. These models can estimate the elevation of a sample taken from a core of sediment documenting previous earthquakes with a resolution of ± 0.2 m. We tested the models by simulating the subsidence associated with a great earthquake in southern Oregon. We did this by transplanting a block of high marsh sediment into the tidal flat; a vertical change of ~0.6 m. Our models reconstructed the elevation change within 0.03 m of the actual value. This gives us high confidence in our ability to reconstruct the amount of subsidence during past great earthquakes. Most previous Cascadia paleoseismology studies have implied that all great earthquakes prior to the AD1700 were of similar size and magnitude. Using our validated methods, we reconstructed the amount of coastal subsidence at a site near Siletz Bay, Oregon. Four earthquakes were identified beneath coastal marshes on Salishan Spit. While three of the earthquakes produced subsidence of 0.5 to 0.8 m, the penultimate earthquake between 800 and 900 years ago, produced less than 0.3 m. This suggests that the lengths and widths of past great earthquake ruptures at Cascadia have varied significantly, as they have for similar historical great earthquakes at other subduction zones. Although so far based only on work at one site, this finding has great significance for assessing the hazard from great earthquakes in western North America.. Realistic models of the earthquake rupture during the AD1700 earthquake also have the potential to make future seismic and tsunami hazard assessments more accurate. We used our more precise methods to reconstruct the amount of subsidence during the AD 1700 earthquake at 17 sites along the Cascadia coastline. We demonstrated that coastal subsidence along the Cascadia coast varies spatially during the same earthquake. If areas of lesser and greater subsidence are persistent through multiple earthquake cycles, then more complex models of earthquake rupture are needed to adequately assess the hazard associated with future earthquakes. We have used stratigraphic and sedimentologic evidence of Cascadia earthquakes and tsunamis to validate computer models earthquakes and tsunamis used to assess tsunami hazard to coastal communities. At Cannon Beach, Oregon we applied a simple sediment transport model to estimate the flow speed of the AD1700 Cascadia tsunami from the thickness and particle size distribution of sandy sediment deposited by the tsunami waves. The model suggests the 1700 tsunami flowed inland at rates between 6.5–7.6 m/s. In southern Oregon, we validated computer tsunami simulations using tsunami deposits in a coastal lake. The tsunami simulations that matched evidence in the lake helped constrain the size of the source earthquakes that produced the tsunamis. Results from both studies were incorporated into a regional tsunami hazard assessment that provides the foundation for tsunami evacuation maps and tsunami disaster planning. We have published 14 peer reviewed publications, 2 book chapters and numerous conference abstracts with more to follow. The project has supported graduate and undergraduate students, and postdoctoral fellows. Undergraduates at the University of Pennsylvania, Sarah Brody, Kendra McCoy, and Will Kearney worked on their senior theses for this project and are now in the graduate programs at Duke, Rutgers, and Boston University, respectively. Pei-Ling Wang successfully defended her Masters at the University of Victoria and is now at National Taiwan University. Andrea Hawkes and Simon Engelhart were graduate and/or postdoctoral fellows at the University of Pennsylvania and have now obtained tenure track Assistant Professor positions at the University of North Carolina and University of Rhode Island, respectively.
