With an increasing number of strong‐motion records over the past decade, it has become clear that there are significant differences in ground‐motion scaling even within relatively small regions such as California. These differences are not modeled in typical ground‐motion models (GMMs) that are based on the ergodic assumption. By including systematic source, path, and site effects in fully nonergodic GMMs, it is possible to reduce the value of the aleatory variability by about 30%–40%; however, to use this reduced aleatory variability, it is important to account for the epistemic uncertainty in the nonergodic terms.
We build a nonergodic GMM for California that explicitly accounts for nonergodic source, path, and site effects by combining the results and methods outlined by Dawood and Rodriguez‐Marek (2013) and Landwehr et al. (2016). The deviation of the systematic terms form the ergodic mean is constrained by observed ground‐motion data. The uncertainty of the systematic terms is tracked and explicitly taken into account. The uncertainty becomes small in the vicinity of observed data and large in regions that lack data. We calculate seismic hazard for three different sites in California using the nonergodic GMM. The results show that the mean hazard can change significantly for sites that have data in the vicinity compared to an ergodic hazard analysis, whereas the epistemic uncertainty associated with hazard results increases for sites with no observed data as compared to the ergodic approach. This presents the first fully nonergodic seismic hazard analysis with the systematic effects based on observed ground‐motion and fully including their uncertainty.