Poster #221, Ground Motions

Region-specific Fourier-based site amplification modeling

Jeff Bayless, & Jonathan P. Stewart
Poster Image: 

Poster Presentation

2020 SCEC Annual Meeting, Poster #221, SCEC Contribution #10394 VIEW PDF
This SCEC study addresses the need for improved nonlinear site amplification models on the SCEC BBP. We infer site response in northern and southern California from earthquake recordings and develop appropriate models for frequency-dependent amplification of horizontal-component Fourier amplitude ordinates as a function of site parameter Vs30 and the peak ground acceleration on reference rock. The bulk of this study follows the conceptual framework of Seyhan and Stewart (2014), in which a semi-empirical site amplification model was developed for response spectra (PSA). We take similar steps using Fourier amplitude ordinates.

The proposed model supplements the available options...
on the SCEC BBP for site amplification.
This model incorporates several updates from our previous work: including refining the analysis regions and expanding the database of recorded ground motions with those recorded since the conclusion of the NGA West2 project; these extra data have been compiled by Wang and Stewart (2019), Nweke et al. (2018) and Buckreis et al. (2019). The linear portion of the model has southern- and northern- California region-specific terms. The nonlinear portion of the model is developed from a combined database, including data from outside of California. We work with Fourier amplitude spectra (FAS) residuals from the Bayless and Abrahamson (2019; BA19 hereafter) ergodic FAS GMPE.

Overall, the linear Vs30 scaling models compare favorably with other empirically-based models (e.g. Bora et al. 2019; Seyhan and Stewart, 2014); especially at low frequencies. The Hashash et al. (2019) model is most different because it features peaks in amplification over distinct frequency bands which correspond to resonance with the modal frequencies of the Vs profile. The data analysis implies considerable regional variations in the linear Vs30 scaling. Based on our results, the Los Angeles region has stronger linear Vs30 scaling of FAS than the Bay Area region at low frequencies.