Removing Bias and Retaining Resolution for Paleoearthquake Dates Corrected by Inherited Age

Katherine M. Scharer, Glenn P. Biasi, Devin McPhillips, & Matthew E. Kirby

Submitted September 8, 2024, SCEC Contribution #13725, 2024 SCEC Annual Meeting Poster #095

Paleoseismic event chronologies are typically developed from radiocarbon dates on detrital organic material. These dates include bias known as the inherited age, which encompasses the time between carbon fixation in the plant and deposition of the dated material at the paleoseismic site. Scharer et al. (2024) developed distributions for the inherited age of radiocarbon dates in Transverse Range watersheds using: (1) recent fires where the date of the fire and the date of deposition are known, and (2) modeled ages from well-dated Holocene deposits. Both distributions have a central peak with a median inherited age of ~90 years, meaning paleoearthquake estimates derived from these dates are similarly too old. Unfortunately, the inherited age distributions are quite broad, with 95% range spanning six centuries. Thus, while the bias in the event ages can be corrected, the broad uncertainty of the inherited age distribution contributes to reduced precision in the resulting dates.

We address this challenge with a new approach to correct for inherited age based on order statistics. Order statistics describe a distribution of the first, second, and nth sample sorted by age given a parent distribution (i.e. the inherited age distribution). The order statistical approach conforms with expectations that the youngest sample in a layer is most likely picked from the early part of the inherited distribution, and older samples from older parts of the inherited curve. For example, given five samples, the order statistics distribution of the youngest sample would concentrate in the youngest ~20% of the inherited age curve, but allow that the sample could have come from other parts of the curve at lesser probabilities.

We benchmark an order-statistical approach to dated layer using log-normal distribution where expectations are well understood. We shift each date by the mean of the order-statistics distribution and stack the resultant dates to develop a single layer date. The ordering distributions add information because they provide an average inherited correction for each sample separately, correcting the youngest sample the least and the oldest sample the most. We show how application of order statistics corrections can address the inherited age bias while limiting loss in precision for the Pallett Creek paleoseismic site on the San Andreas Fault.

Key Words
paleoseismology

Citation
Scharer, K. M., Biasi, G. P., McPhillips, D., & Kirby, M. E. (2024, 09). Removing Bias and Retaining Resolution for Paleoearthquake Dates Corrected by Inherited Age. Poster Presentation at 2024 SCEC Annual Meeting.

Related Projects & Working Groups
Earthquake Geology