Insights into the site-to-site correlation of paleoseismic data

Integration of paleoseismic data from multiple sites is important to assess the past fault rupture scenarios and determine an earthquake chronology for the entire fault system. However, the current methods used to combine paleoseismic data are diverse and lack theoretical foundations from a mathematical perspective. We present a method to evaluate and integrate paleoseismic event data from multiple sites into a single earthquake time history. We apply this method to the central-eastern fault sections of the Altyn Tagh Fault using data from ten fault trenches. Applying a Bayesian approach we constructed time-stratigraphic models that yield the probability density functions corresponding to the age of individual earthquakes at each site. Then, our method to integrate these data consists of two main steps: 1) we constructed a rupture pool with all the modeled event ages, and we evaluated the overlapping degree between the site PDFs; 2) For sufficiently contemporary PDFs we combine them by computing the weighted-mean method which emphasizes the overlap in the site earthquake times. The weighted-mean method yields smaller earthquake-time uncertainties compared to the rupture-mean approach and is consistent with the earthquake rupture assumptions behind the integration of paleoseismic data and the probability theory of density functions. This approach helps to clarify the timing and rupture extent of past earthquakes along central-eastern ATF and is essential to improve the earthquake probability assessment for the region.