Classification of Aleatory Variability and Epistemic Uncertainty for Probabilistic Seismic Hazard Analyses
Description:
Aleatory variability and epistemic uncertainty are used in Probabilistic Seismic Hazard Analysis (PSHA) to differentiate between aspects of the source models and ground-motion models that are treated as alternative credible models (epistemic uncertainty) and those that are treated as randomness (aleatory variability). The concepts of aleatory variability and epistemic uncertainty have been widely used in PSHA since the 1980s, however, confusion remains concerning their distinction in specific applications. One cause of confusion is that the classification of aleatory or epistemic terms is not absolute and depends on the parameterization of the models used. In general, models with a greater degree of simplification will have larger aleatory variability. This non-uniqueness is often a source of concern for earthquake scientists and has contributed to the belief that the distinction between aleatory variability and epistemic uncertainty is arbitrary; however, once a model is selected with a certain level of simplification, there is no ambiguity for the classification into aleatory and epistemic terms. We present a classifying framework for the source and ground-motion model parameters into the following aleatory and epistemic components: aleatory standard deviation about the median, epistemic uncertainty of the model for the median, and epistemic uncertainty of the size of the aleatory standard deviation. For each parameter, the terms are separated into two parts: modeling and parametric. The modeling parts are based on comparisons to data, and the parametric parts are based on modeling a range of inputs into the model. This framework provides clarification for checking that all parts of aleatory variability and epistemic uncertainty are systematically categorized and included in the models without double counting.
Session: Site-specific Modeling of Seismic Ground Response: Are We Quantitative Enough to Predict? [Poster]
Type: Poster
Date: 4/19/2023
Presentation Time: 08:00 AM (local time)
Presenting Author: Irene Y. Liou
Student Presenter: Yes
Invited Presentation:
Authors
Irene Liou Presenting Author Corresponding Author iyliou@ucdavis.edu University of California, Davis |
Norman Abrahamson abrahamson@berkeley.edu University of California, Davis |
|
|
|
|
|
|
|
Classification of Aleatory Variability and Epistemic Uncertainty for Probabilistic Seismic Hazard Analyses
Category
General Session