Framework for Aleatory Variability and Epistemic Uncertainty for the Ground-Motion Characterization Based on the Level of Simplification

ABSTRACT Aleatory variability and epistemic uncertainty are commonly used concepts in probabilistic seismic hazard analysis (PSHA); however, separating uncertainty into aleatory variability and epistemic uncertainty is often seen as arbitrary. As part of the ground-motion characterization, we present a clarifying framework for defining aleatory variability and epistemic uncertainty for ground-motion models (GMMs). Aleatory variability is mainly due to unmodeled physical behaviors affecting ground motion. In contrast, epistemic uncertainty refers to the scientific uncertainty that the earthquake effects included in the model are modeled correctly. What is treated as aleatory variability and epistemic uncertainty depends on the level of model simplification. Simple models have larger aleatory variability and smaller epistemic uncertainty than complex models that model more physical behaviors. The framework has two parts: the method component, related to the algorithm and basic formulation for computing the ground motion, and the parametric component, which captures the effect of inputs to the GMM that are not included in the hazard integral. Each part has three components: aleatory variability, epistemic uncertainty in the median ground motion, and epistemic uncertainty in the size of the aleatory variability. The six terms provide a framework to ensure that all parts of the aleatory variability and epistemic uncertainty are included once and only once in the hazard calculation. The framework is especially beneficial as a guide for incorporating more complex GMMs into PSHA; it clarifies the separation of aleatory variability and epistemic uncertainty for nonergodic GMMs and numerical simulations.