Ground Motion Model for Predicting Significant Duration Constrained by Seismological Simulations
Description:
We developed a GMM for predicting significant durations based on the normalized Arias intensity.
Using seismological simulations through a 3D finite difference code (SW4), we showed that a GMM based on an additive behavior between the source, path and site terms provides better physics fundamentals on the behavior and makes us more confident in extrapolating the data than multiplicative-behaved models. Our data distribution exploration led us to define a site term with a log-normally distributed random effect, with a larger heteroskedasticity on small values of VS30, i.e., the variability is larger for softer sites than for stiffer sites. Then, we constrained the source term based on stochastic finite source simulations. We found a coupling scaling pattern between the source term with the rupture dimensions (i.e., magnitude) and the source-site distance. This aspect of the model shows that not only is the time window where the energy is released from the source important but there is also an important contribution from the relative ground motion amplitude. Furthermore, distance from the site to different rupture sub-segments influences the duration differently because of the relative amplitude-distance attenuation. To take advantage of all the benefits of working with normally distributed data, we applied a power-transformation to the model and to the final predictive GMM to ensure it follows a normal distribution. In conclusion, the significant duration process has a skewed distribution, being in between the skewness of the log-normal and the symmetry of the normal distribution. The complexities of the model formulation (event and site terms) require more advanced statistical tools than used for standard ground motion models development (i.e., random effect with least squares).
Session: Site-specific Modeling of Seismic Ground Response: Are We Quantitative Enough to Predict?
Type: Oral
Date: 4/19/2023
Presentation Time: 08:30 AM (local time)
Presenting Author: Camilo Ignacio Pinilla-Ramos
Student Presenter: Yes
Invited Presentation:
Authors
Camilo Ignacio Pinilla-Ramos Presenting Author Corresponding Author camilo.pinilla@berkeley.edu University of California, Berkeley |
Norman Abrahamson abrahamson@berkeley.edu University of California, Berkeley |
Robert Kayen rkayen@berkeley.edu U.S. Geological Survey |
Van-Bang Phung phungvb@earth.sinica.edu.tw Institute of Earth Sciences, Academia Sinica |
Pablo Castellanos-Nash pcastellanos@udec.cl Universidad de Concepción |
|
|
|
|
Ground Motion Model for Predicting Significant Duration Constrained by Seismological Simulations
Category
Site-specific Modeling of Seismic Ground Response: Are We Quantitative Enough to Predict?