ing rupture, and the permanent subsidence and uplift. Strong ground motion may, in turn, cause ground failure—slumps, landslides, liquefaction, and lateral spread—depending on shaking intensity (usually stronger nearer the source) and local site conditions. If it occurs offshore, fault displacement can generate tsunamis capable of inundating nearby and distant shorelines. Ground failure and tsunamis are examples of secondary hazards (2).
Tectonic earthquakes are spontaneous releases of tectonic stress that produce macroscopic, permanent displacements across fault surfaces (ruptures) and within the rock mass around faults (co-seismic deformations). Most fault ruptures are confined to buried regions of the crust where brittle behavior allows stick-slip instabilities to nucleate (e.g., between 2 and 20 kilometers deep in most continental deformation zones). Such ruptures propagate to the surface only in larger earthquakes. When this happens, however, almost any structure built across the rupture path will be deformed by the severe strains characteristic of primary ground failure (Figure 3.1). Predicting the magnitude and extent of fault rupture is therefore a major issue in seismic hazard analysis.
Ruptures tend to occur along faults that have produced large earthquakes in the past, so a map of active faults is a first-order representation of the rupture hazard. The average amount of co-seismic slip increases systematically with earthquake magnitude (3), and the maximum displacement tends to occur toward the middle of the rupturing segment. These behaviors can be used to quantify the hazard along well-defined active faults. For example, where the Hollywood subway crosses the Hollywood fault in Los Angeles, California, the maximum expected slip is estimated to be 1 to 2 meters. In anticipation, the Metropolitan Transportation Authority overbored the subway tunnel to allow the tracks to be realigned after such an earthquake.
Mapped faults are often categorized as active and inactive, but doing so is problematic because the maximum magnitude, frequency of rupture, and other measures of activity can be highly variable among faults in the same tectonic province. Even within a single zone, the distribution of recent faulting can be considerably more complex than the simple traces that represent active faults on small-scale geologic maps. Detailed mapping reveals a wide range of features, such as segmentation, stepovers, and faulting at conjugate angles, often with self-similar scaling (Figure 3.2). The faulting patterns observed in large earthquakes show similar complexity, which can vary rapidly along strike. In some places, the rupture may be a single, clean break, while elsewhere it may occupy a zone tens or hundreds