← Back to Sessions

Aleatory Variability and Epistemic Uncertainty from Physics-based Ground-motion Simulations as part of Probabilistic Seismic Hazard Analysis

Description: 

Modern seismic hazard studies require quantifying aleatory variability and epistemic uncertainty of ground-motion estimates. For numerical simulations of ground motions, the six-component framework of Liou and Abrahamson (2024) is used to identify all parts of the aleatory variability and epistemic uncertainty. The six components are grouped into the method category and the parametric category. The method category refers to the algorithm and basic formulation of the simulations, and the parametric component refers to the inputs to the simulation. Each category has three parts: aleatory variability, epistemic uncertainty in the median, and epistemic uncertainty in the aleatory variability. All six components are needed for probabilistic seismic hazard analysis (PSHA) applications. For current sets of ground-motion simulations, some of the six components of the epistemic uncertainty and aleatory variability are missing, limiting the applicability of physics-based PSHA to research studies. The missing components need to be addressed before physics-based simulations are widely used in PSHA. For 3-D simulations, the key missing component is the parametric epistemic uncertainty of the median due to the uncertainty in the 3-D velocity models. Ideally, multiple 3-D velocity models would be used; however, this involves a significant effort. As an alternative, the standard deviation of the misfit between observations and predictions for past earthquakes can be partitioned into the method aleatory variability due to limitations and simplifications in the simulation method and parametric epistemic uncertainty due to the errors in the 3-D velocity model. Though the partitioning for a specific simulation set is unknown, generic percentages can be used to split the standard deviation from the validation into the method aleatory and parametric epistemic components. We provide an example of estimating the six components of aleatory variability and epistemic uncertainty using some assumed values and identify what is needed from 3-D simulations to move physics-based PSHA from research to engineering practice.

Session: Challenges and Opportunities in Constraining Ground-motion Models from Physics-based Ground-motion Simulations - II

Type: Oral

Date: 4/17/2025

Presentation Time: 11:00 AM (local time)

Presenting Author: Irene

Student Presenter: No

Invited Presentation: 

Poster Number:

Authors