Sensitivity of Site Response Analyses to Input Motion Selection Protocols

Input (bedrock) ground motions are a critical component of site-specific evaluations of earthquake effects. In this study, we investigate the sensitivities of site-response predictions to input motion selection protocols for different types of site conditions. Analyses are performed for alternative profiles (soft and stiff) in two sedimentary basins, Seattle, Washington, and Boston, Massachusetts, to evaluate the influence of different tectonic, geologic, and geotechnical conditions. Using the results of probabilistic seismic hazard analyses at each site, we develop suites of input motions using alternative target spectrum definitions for various magnitude-distance combinations, perform one-dimensional site response analyses, and evaluate the resulting sensitivities of the site response predictions to the various input motion selection protocols. When the soil behavior is strongly nonlinear at soft sites, we find that the influence of input motion selection protocols on site-response uncertainty decreases significantly, and that nonlinear soil behavior can suppress the variability in surface ground motions. Analyses using alternative profiles indicate that predicted ground motions at stiffer sites are more greatly influenced by input motion selection protocols relative to softer sites where significant nonlinear behavior may occur.