Modeling the Spatial Correlation of Systematic Source Effects in Non-Ergodic Ground Motion Models for the Ridgecrest Area

The current state of the practice in probabilistic seismic hazard analysis (PSHA) employs ergodic ground motion models (GMMs) to describe the probabilistic distribution of an intensity measure, which assumes that the ground motion variability observed in a global database is the same as the variability in ground motion at a single site-source combination. However, the fast-growing empirical ground motion databases indicate significant regional differences in ground motions due to repeatable and systematic source, path, and site effects. These systematic effects, which are spatially correlated, are not consistent with the ergodic assumption, promoting the transition to non-ergodic GMMs for PSHA. In this study, we use Gaussian processes to model the spatial correlation structures of systematic source effects for the Ridgecrest area, which provides insights into non-ergodic PSHA. We consider two different models for the spatial correlation of the systematic and repeatable source effects for the peak ground acceleration: an isotropic stationary model and an anisotropic non-stationary model. In the isotropic stationary model, the spatial correlation of the source effects depends only on the separating distances of two seismic sources, which is commonly used in current non-ergodic GMMs. The new anisotropic non-stationary model proposed in this study considers the fault geometries of the earthquakes and the separating distance of the seismic sources. Our results show that the spatial correlation model provides information on the spatial variation of the repeatable source effects induced by complex physical processes and reduces the associated aleatory variability. We also find that the spatial correlation of systematic source effects in the Ridgecrest area is best characterized by the anisotropic non-stationary model, which ensures better extrapolation of the source effects to regions without significant data. The developed spatial correlation model provides insights in the context of nonergodic-based PSHA considering multiple seismic sources. In addition, the proposed model improves the epistemic uncertainty treatment of alternative spatial correlation structures for repeatable source effects.