multipathing along sharp geologic boundaries at basin edges, as well as by amplification due to near-site properties. Although near-site effects such as liquefaction can be strongly nonlinear, most aspects of seismic-wave propagation are linear phenomena described by well-understood physics. Therefore, if the seismic source could be specified precisely and the wave velocities, density, and intrinsic attenuation were sufficiently well known, it would be possible to predict the time history of strong motions by a numerical calculation. The research goal is to use this physics-based approach to go beyond empirical attenuation relationships in characterizing strong ground motions and their secondary effects.
Goal: Predict the strong ground motions caused by earthquakes and the nonlinear responses of surface layers to these motions—including fault rupture, landsliding, and liquefaction—with enough spatial and temporal detail to assess seismic risk accurately.
Research in this field should focus on urban areas where the consequences of large earthquakes are most severe (Box 6.4). In particular, past earthquakes have demonstrated that areas of damage are often localized in highly populated sedimentary basins near active faults. Site-specific information about the time histories of shaking will be needed for performance-based design of structures in such settings. The challenge of urban hazard mapping is to predict ground-motion effects over an extended region with an acceptable level of reliability. Ground-motion maps for the 1989 Loma Prieta and 1994 Northridge events demonstrated that while the major urban regions of California were sufficiently instrumented to determine a first-order distribution of ground motions, the networks were not dense enough to provide a direct correlation of local damage patterns with ground-motion levels. Ground-motion simulations can potentially be used to interpolate the recorded data for more detailed seismic zonation, provided that the subsurface structure is adequately characterized.
Plausible objectives for the next 10 years are (1) to determine the structure of high-risk areas well enough to model the surface motions from a specified seismic source at all frequencies up to at least 1 hertz and (2) to formulate useful, consistent, stochastic representations of surface motions up to at least 10 hertz. At present, computer simulations in areas where the three-dimensional structure of the crust is best known, such as the Los Angeles region, can model the peak amplitudes only below about 0.3 hertz (Figure 6.5). To extend these calculations to the higher frequencies needed for engineering applications, a much greater volume of seismological data will be needed to map the three-dimensional structure of the crust. Future densification of urban seismic networks should strive for