The response of soils at shallow depth to strong shaking is a complex phenomenon (12). Amplitude builds as the waves slow (another consequence of energy conservation), so seismic shaking is typically amplified in the soft soils and unconsolidated sediments near the ground surface, where the wave speed can be much lower than in hard rock. For this reason, the average shear velocity in the upper 30 meters or so has become the primary basis for the National Earthquake Hazard Reduction Program (NEHRP) site classification used in many building codes, including the 1997 Uniform Building Code (UBC), 2000 International Building Code (IBC), and 2000 American Society of Civil Engineers (ASCE) Standards 7-98 (13). The high amplitude predicted by the linear wave theory is thought to be reduced by the nonlinear response of the unsaturated near-surface layers. Laboratory tests clearly demonstrate nonlinear strain behavior in soils under dynamic loading, but the importance of nonlinearity during actual earthquakes continues to be debated. Using available ground-motion data to differentiate nonlinear strain behavior from other wave propagation effects has usually been difficult. For example, interpretations of the data collected in Mexico City from the 1985 Michoacan earthquake reached conflicting conclusions on the importance of the nonlinearity of the city’s soft clay deposits (14). On the other hand, direct evidence of significant nonlinear soil response was clearly observed in the motions recorded by surface and subsurface (borehole) instruments at saturated sandy sites that liquefied during the 1987 Superstition Hills, California, and the 1995 Hyogo-ken Nanbu earthquakes (15). Aside from these extreme cases where the soil failed, indirect evidence of nonlinear site response on soils that remained stable during strong shaking is becoming more apparent with the greater number of seismograms being recorded in strong-motion arrays throughout the world (16). However, more of these data are clearly needed to better understand and predict this phenomenon.
Another interesting aspect of seismic shaking is that it can vary substantially from one tectonic setting to another. For example, the motion from similar-sized earthquakes is observed to be stronger in the central and eastern United States than west of the Rocky Mountains. Felt areas and areas of specific intensity (isoseismals) are also larger for earthquakes in the central and eastern United States compared to those of earthquakes with similar magnitudes in the western United States. Earthquakes in the older, stronger regions of the continent generally have greater stress drops and therefore radiate more high-frequency energy for a given amount of fault slip; moreover, their seismic waves propagate with less attenuation compared to earthquakes in plate boundary deformation zones. The attenuation difference is probably attributable to lower temperature, reduced scattering, and more continuous waveguide for crustal shear en-