Ergodic site response models are used to estimate the difference between ground motion intensity measures at reference rock sites and intensity measures for other site conditions. Such models are typically conditioned on the time-averaged shear wave velocity in the upper 30 m of the site (VS30) and are available for pseudo-spectral accelerations for oscillator periods between 0.01 sec (approximately PGA) and 10 sec. What makes a site response model â€˜ergodicâ€™ is that it is derived using a large database and/or a large inventory of simulation results, which are fit using various regression equations. The resulting predictions represent medians across many sites with different characteristics, but conditioned on VS30 and perhaps additional secondary parameters. Because such models do not include other site-specific characteristics, the resulting ground motions have large aleatory variability. Past and current practice is that ergodic site response models derived for conditions in the western US are applied nation-wide (e.g., site factors in ASCE 7/16 and earlier editions). Recent research conducted as part of Next-Generation Attenuation projects demonstrates significant regional variations in site response, with weaker scaling of ground motions with VS30 in the Pacific Northwest and central/eastern portions of the US than in the western US. Such regional effects are being incorporated into a pending version of the US Geological Survey national seismic hazard maps and future versions of ASCE 7.
This presentation will describe the site response physics underlying ergodic models, how such models are derived, and their application in seismic hazard analyses. Trends in ergodic and non-ergodic (site-specific) site response model development, and the benefits of each type of modeling approach, will be explained.