Holocene Seismicity in Puget Sound: A Statistical Perspective

Richard Styron

I presented work done with Brian Sherrod (USGS Seattle) on paleoseismologic data in the Puget Sound region of Washington at AGU. In this work, I take ~30 paleoearthquakes and estimate their magnitudes with new methods that incorporate both rupture length and measured offsets, and then calculate recurrence interval probability distributions and use survival analysis statistical methods to estimate time-dependent earthquake hazards.

The results, briefy:

  • Earthquakes on upper-plate faults in the Puget Sound region are M 6.4-7.5

  • Modal (most-likely) recurrence interval for earthquakes in the Puget Lowland are very short (decades), and a few centuries for individual fault zones, though the means are much longer.

  • These short modal recurrence times indicate elevated time-dependent earthquake hazard following an earthquake. Earthquakes are clustered in time.

I'm writing the magnitude estimation stuff up for Geophysical Research Letters right now. I'll hopefully have a manuscript submitted by early January.

The work evolved a bit since August, when I wrote the abstract. That title and abstract are:

Earthquake clustering and recurrence intervals in the Puget Sound region, Washington: A statistical perspective

Richard Styron and Brian Sherrod


While active upper-crustal faulting in the Cascadian forearc has long been recognized, analysis of LiDAR topographic data has allowed for the construction of a fairly complete record of surface faulting of the ~16 ka glacial landscape of the Puget Lowland (WA). We have recently compiled all of the existing age data and refined the estimates of timing for each event in OxCal through synthesis of the paleoearthquake sequences at 68 trenches. We then performed a Monte Carlo analysis of the earthquake history by creating millions of synthetic earthquake histories from direct sampling of the earthquake timing probability density functions; in this way we are able to explore the dataset while accounting for the uncertainty in each event. A striking feature of the earthquake history is the highly irregular strain release over the Holocene; the region seems to undergo long periods of seismic quiescence punctuated by episodes of more intense earthquake activity. These temporal clusters of earthquakes involve faults throughout the Puget Lowland and are not restricted to certain faults or sub-regions. The regional earthquake recurrence interval probability shows a mode at 20-50 years, which then declines monotonically to ~5,000 years. This very short mode is a sign of earthquake clustering, which may reflect static or dynamic earthquake triggering by other events. Furthermore, it strongly suggests that the earthquake hazard following a moderate to large earthquake in the region will remain elevated for decades afterward. We are now focusing on the probabilistic reconstruction of the rupture extents and magnitudes of all the paleoearthquakes in the dataset through analysis of LiDAR- and trench-derived displacement measurements and fault characteristics. This will allow us to better understand regional strain rates and earthquake frequency-magnitude relationships, and allow for physical as well as statistical modeling of earthquake clustering.