Advancing the Use of Simulated Ground Motions in Physics-Based Probabilistic Seismic Hazard Analysis from Research to Practice

Abstract

Physics-based probabilistic seismic hazard analysis (PSHA), in which 3D numerical simulations replace traditional ground-motion models, has been under development for several decades. The development of high-performance computing capabilities and techniques has enabled significant advancements, such as the inclusion of complex near-fault source and path effects, which allow the extrapolation of ground motion to that of rare events at close distances; however, a gap remains between research and practice. This article compares and contrasts traditional PSHA with current physics-based PSHA and shows that method aleatory variability, which is apparent aleatory variability due to simplifications in the simulation method, is missing from current physics-based PSHA calculations. We highlight the need for proper separation and quantification of apparent aleatory variability and epistemic uncertainty, as well as validation of the distribution of the simulated ground motions for a given scenario. We discuss how the standard deviation of the misfits from simulation validations often includes both method aleatory variability and parametric epistemic uncertainty in the median, which is from scientific uncertainty in simulation inputs that are not part of the traditional hazard calculation as part of the source model. The limitations of the simulation method are often difficult to estimate directly from the validation because it is challenging to separate its effects from those of incorrect inputs. We provide an overview of simulation simplifications that lead to method aleatory variability, methods of incorporating method aleatory variability, and a list of recommendations for advancing physics-based PSHA from research to practice.