Methods for Unbiased Ground-Motion Intensity Conversion Equations and Implications for Hazard Map Assessment in California
Description:
Ground-motion intensity conversion equations (GMICE) are used to describe the relationship between ground-motion amplitude and shaking intensity. GMICE are crucial for ShakeMap, ShakeAlert, and comparisons of results of probabilistic seismic hazard analysis (typically given in ground motion) to observations of historical shaking characterized by the intensity. While GMICE are particularly useful when they can be used to convert directly from ground motion to intensity, without the inclusion of the earthquake magnitude or distance to the site, we show that this approach leads to biased intensity estimates for PGM values above or below the median for the scenario.
The current approach to developing GMICE is to conduct an initial regression for intensity as a function of peak ground motion (PGM), commonly acceleration or velocity. Some GMICE include magnitude and distance effects using residuals in subsequent regressions, but these terms are only added into the initial equation that predicts intensity as a function of PGM. With this approach, there is an implicit assumption that the PGM and intensity are causal and fully correlated, which is not the case. We show that the current approach to GMICE development overestimates the variability of predicted intensity because residuals of the intensity from an intensity prediction equation (IPE) and the residuals of the PGM from a ground-motion model (GMM) are only partially correlated. We explore these effects and their causes using simulated data and GMICE from California based on the Worden et al. (2012) GMICE based on PGA. These results can be generalized to other GMICE, as the formulation of the Worden et al. (2012) model is consistent with standard practices used to develop GMICE, and intensity scales have been shown to be approximately interchangeable. This issue appears to be a major factor in explaining why published hazard maps for California and other areas appear to overpredict shaking relative to historical data.
Session: Advances in Probabilistic Seismic Hazard Analysis and Applications [Poster]
Type: Poster
Date: 4/20/2023
Presentation Time: 08:00 AM (local time)
Presenting Author: Molly Gallahue
Student Presenter: Yes
Invited Presentation:
Authors
Molly Gallahue Presenting Author Corresponding Author mollygallahue2023@u.northwestern.edu Northwestern University |
Norman Abrahamson abrahamson@berkeley.edu University of California, Berkeley |
|
|
|
|
|
|
|
Methods for Unbiased Ground-Motion Intensity Conversion Equations and Implications for Hazard Map Assessment in California
Category
Advances in Probabilistic Seismic Hazard Analysis and Applications