INTER- AND INTRA-SITE VARIABILITY OF COSEISMIC SUBSIDENCE ESTIMATES FROM NORTHERN HUMBOLDT BAY, CALIFORNIA

Estimates of coastal vertical deformation during megathrust earthquakes can be derived from microfossil analysis of wetland stratigraphic sequences. Along the coasts of the Cascadia subduction zone, quantitative relative sea-level reconstructions based on foraminiferal transfer functions yield precise (<±0.3m) estimates of coseismic vertical deformation. We extend application of this approach to the southern Cascadia margin to evaluate within-site and within-bay depositional variability from megathrust earthquakes and test the reproducibility of a validated foraminiferal transfer function at northern Humboldt Bay, California (~44.8°N, -124.2°W). We examine four abrupt mud-over-peat (coseismic subsidence) contacts along a 6-km transect at Jacoby Creek, McDaniel Creek, and Mad River. Our quantitative reconstructions of relative sea-level rise across each contact using a foraminiferal transfer function give subsidence estimates to ±0.26 m. To assess the reproducibility of our estimates, we analyzed 26 sediment cores containing the four mud-over-peat contacts: nine for the AD 1700 contact (average of 0.37 m subsidence), five for the ~870 cal yr BP contact (average of 0.33 m), six for the ~1125 cal yr BP contact (average of 0.43 m), and six for the ~1600 cal yr BP contact. The estimates for the 1600 cal yr BP contact are only minimums because the contact formed above the upper limit of foraminiferal habitation (the largest is ≥0.56 m). Differences in inter-site subsidence estimates range from 0.16 m for the ~1125 cal yr BP contact to 0.41 m for the AD 1700 contact. The variability of intra-site subsidence estimates ranges from a minimum of 0.07 m for the ~870 cal yr BP contact at McDaniel Creek to a maximum of 0.31 m for the AD 1700 contact at Mad River. Future work will investigate whether the use of a new Bayesian transfer function reduces the variability within and among sites. Based on our results using the current non-Bayesian transfer function, we recommend replicate reconstructions for each contact in each estuary to ensure confidence in estimates of coseismic subsidence.