Challenges in Predicting Seismic Site Response with 1D Analyses: Conclusions from 114 KiK‐net Vertical Seismometer Arrays

Research Article| September 04, 2018 Challenges in Predicting Seismic Site Response with 1D Analyses: Conclusions from 114 KiK‐net Vertical Seismometer Arrays James Kaklamanos; James Kaklamanos aDepartment of Civil Engineering, Merrimack College, 315 Turnpike Street, North Andover, Massachusetts 01845, KaklamanosJ@merrimack.edu Search for other works by this author on: GSW Google Scholar Brendon A. Bradley Brendon A. Bradley bDepartment of Civil and Natural Resources Engineering, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand, brendon.bradley@canterbury.ac.nz Search for other works by this author on: GSW Google Scholar Bulletin of the Seismological Society of America (2018) 108 (5A): 2816–2838. https://doi.org/10.1785/0120180062 Article history first online: 04 Sep 2018 Cite View This Citation Add to Citation Manager Share Icon Share Facebook Twitter LinkedIn MailTo Tools Icon Tools Get Permissions Search Site Citation James Kaklamanos, Brendon A. Bradley; Challenges in Predicting Seismic Site Response with 1D Analyses: Conclusions from 114 KiK‐net Vertical Seismometer Arrays. Bulletin of the Seismological Society of America 2018;; 108 (5A): 2816–2838. doi: https://doi.org/10.1785/0120180062 Download citation file: Ris (Zotero) Refmanager EasyBib Bookends Mendeley Papers EndNote RefWorks BibTex toolbar search Search Dropdown Menu toolbar search search input Search input auto suggest filter your search All ContentBy SocietyBulletin of the Seismological Society of America Search Advanced Search Abstract Nonlinear (NL) soil behavior often exhibits a strong influence on surficial ground motions, and seismic site‐response models attempt to quantify these effects. However, site‐response models exhibit large uncertainties and can on occasion poorly replicate observed ground motions. In this study, NL total‐stress site‐response model predictions for 5626 ground motions at 114 KiK‐net vertical seismometer arrays are calculated and compared with observed ground motions and predictions from linear (L) and equivalent‐linear (EQL) analyses. Using this large database of 1D site‐response model predictions, a variety of statistical analyses are performed to quantify model bias and precision, and these statistical analyses are paired with physical insights into site and ground‐motion behavior.As expected, there are deviations between the L, EQL, and NL site‐response constitutive models at large shear strains. When using cumulative Arias intensity to assess the temporal evolution of model uncertainty, the EQL model is shown to have excessive bias early in the ground‐motion record, but this bias is obscured when the entire record is considered. Another less intuitive result is that, on average, across the entire dataset, all models (L, EQL, and NL) tend to underpredict high‐frequency ground motions in the aggregate. A number of physical hypotheses are explored to provide insights into these persistent biases. The use of a depth‐dependent shear‐wave velocity gradient, in particular, has a significant impact on the model bias, even more so than changing the constitutive model for dynamic soil behavior. Using an unprecedented number of sites and ground motions, the results of this study provide insights for the present limitations of, and potential improvements to, 1D site‐response model predictions. You do not have access to this content, please speak to your institutional administrator if you feel you should have access.