Improving Estimates of Coseismic Subsidence from southern Cascadia Subduction Zone Earthquakes at northern Humboldt Bay, California

Abstract:

Geological estimates of subsidence from past earthquakes help to constrain Cascadia subduction zone (CSZ) earthquake rupture models. To improve subsidence estimates for past earthquakes along the southern CSZ, we apply transfer function analysis on microfossils from 3 intertidal marshes in northern Humboldt Bay, California, ~60 km north of the Mendocino Triple Junction. The transfer function method uses elevation-dependent intertidal foraminiferal and diatom assemblages to reconstruct relative sea-level (RSL) change indicated by shifts in microfossil assemblages. We interpret stratigraphic evidence associated with sudden shifts in microfossils to reflect sudden RSL rise due to subsidence during past CSZ earthquakes. Laterally extensive (>5 km) and sharp mud-over-peat contacts beneath marshes at Jacoby Creek, Mad River Slough, and McDaniel Slough demonstrate widespread earthquake subsidence in northern Humboldt Bay. C-14 ages of plant macrofossils taken from above and below three contacts that correlate across all three sites, provide estimates of the times of subsidence at ~250 yr BP, ~1300 yr BP and ~1700 yr BP. Two further contacts observed at only two sites provide evidence for subsidence during possible CSZ earthquakes at ~900 yr BP and ~1100 yr BP. Our study contributes 20 AMS radiocarbon ages, of identifiable plant macrofossils, that improve estimates of the timing of past earthquakes along the southern CSZ. We anticipate that our results will provide more accurate and precise reconstructions of RSL change induced by southern CSZ earthquakes. Prior to our work, studies in northern Humboldt Bay provided subsidence estimates with vertical uncertainties >±0.5 m; too imprecise to adequately constrain earthquake rupture models. Our method, applied recently in coastal Oregon, has shown that subsidence during past CSZ earthquakes can be reconstructed with a precision of ±0.3m and substantially improves constraints on rupture models used for seismic hazard assessment.