Evaluating nonlinear effective stress site response analyses using records from the Canterbury earthquake sequence

• Strong motion records from the Canterbury earthquake sequence are examined.
• Perform 1D nonlinear effective stress site response analyses at liquefiable sites.
• Surface motions are deconvolved for use as input motions for site response analyses.
• Key aspects of recorded motions are generally captured by the analyses.
• Observed bias is due to pore water pressure model in analyses and input motions.

A widely used one-dimensional nonlinear effective stress site response analysis program is used to model the response of potentially liquefiable soils during strong shaking. Ground motion records from six events of the 2010–2011 Canterbury earthquake sequence and the extensive site investigation data that have been obtained for the Christchurch area provide the basis for the analyses. The results of the analyses depend significantly on the input motions and soil profile characterization, so these important aspects are examined. Deconvolved Riccarton Gravel input motions were generated, because recorded rock or firm layer motions were not available. Nonlinear effective stress seismic site response analyses are shown to capture key aspects of the observed soil response through the comparison of acceleration response spectra of calculated surface motions to those of recorded surface motions; however, equivalent-linear and total stress nonlinear analyses capture these aspects as well. Biases in the computed motions compared to recorded motions were realized for some cases but they can be attributed primarily to the uncertainty in the development of the input motions used in the analyses.