Non-ergodic Ground-Motion Models for Crustal Earthquakes in Georgia

Abstract

A sparse data set consisting of 33 local strong-motion recordings from earthquakes with magnitudes between 4.0 and 7.1 and distances between 8 and 190 km is used to develop an initial non-ergodic ground-motion model for the Georgia Republic for spectral acceleration at T = 0.2 s. The non-ergodic adjustments capture the local differences in the linear site effects and linear path effects. The non-ergodic GMMs have reduced aleatory variability along with shifts (increases or decreases) to the median ground motion for a given site/source pair. For sites in central Georgia, there is some local data to constrain the shift in the median ground motion from the base ergodic GMM. For sites far from the available local ground-motion data, such as the Enguri dam site, there is larger epistemic uncertainty in the shifts to median. In this case, the reduced aleatory variability will be offset by the larger epistemic uncertainty. Although there is large uncertainty, the advantage of using the non-ergodic approach in regions with sparse local ground-motion data is that the large epistemic uncertainty range is properly quantified and it shows the potential change to the hazard that can be expected as local ground-motion data are collected.