sions with lower standard deviations and better fit were obtained by Slemmons (1982b) and using more sophisticated magnitude values and statistical methods by Bonilla et al. (1984). Where there is good field expression of the length and/or maximum displacement from geomorphic expression of fault scarps, or where these parameters can be measured from soil-stratigraphic relations, it is possible to infer the approximate magnitude of paleoseismic events. These correlations between fault parameters and earthquake magnitude have been made for siting and engineering design of vital structures in many parts of the world for diverse tectonic settings, including extensional areas such as the Basin and Range province (Wallace, 1977, 1978; Cluff et al., 1980; Schwartz and Coppersmith, 1984) and regions of strike-slip faulting (Sieh, 1984; Sieh and Jahns, 1984) and for thrust faulting in regions with compressional tectonics (Woodward-Clyde Consultants, 1984). An application to intraplate locations is shown by the left-oblique faulting of the Meers Fault zone in Oklahoma. There a 26- to 38-km-long fault scarp can be mapped. Using correlations of length to MS magnitude, the scarp can be inferred to have formed during prehistoric earthquakes of between 6.5 to 7.5 magnitude (Slemmons et al., 1985). This paleoseismic evidence is especially important since no historical earthquakes have been recorded in the area. Elsewhere in this zone a general alignment of epicenters of small earthquakes has been noted (Gordon, 1985).
In summary, active fault parameters such as length and displacement can be used to estimate earthquake magnitudes through the regression formula presented by Slemmons (1982b) or Bonilla et al. (1984), and the parameters can be used directly in a moment magnitude calculation. An example of these correlations is shown in Figure 3.4.
The segmentation of fault systems involves the identification of individual fault segments that appear to have continuity, character, and orientation; these suggest that a segment will rupture as a unit (Slemmons, 1982b). Individual fault segments have different characteristics relative to adjacent segments or are separated from adjacent segments by identifiable discontinuities.
Figure 3.5 illustrates the concept that fault zones rupture in segments with an example from Turkey. During the period 1939 to 1967, the North Anatolian Fault system ruptured as segments and not as a single through-going event.
The delineation of segments involves the identification of discontinuities in the fault system. Discontinui
ties can be divided into two categories, geometric and inhomogeneous; these categories are borrowed from seismologists who have used these terms for asperities and barriers (Aki, 1984). Examples of geometric discontinuities include fault intersections, such as branch faults or cross-fault terminations; fault-zone features, such as en echelon segments, separations, and changes in attitude; and fault terminations. Inhomogeneous discontinuities include variations in fault width, local stress regimes, and rates of displacement.
Segall and Pollard (1980), acknowledging that fault traces consist of numerous discrete segments, have developed a two-dimensional mathematical solution for nonintersecting cracks. Their solution describes the mechanical behavior of left-stepping versus right-stepping en echelon cracks in a right lateral stress regime. Segall and Pollard state, “for right lateral shear and left-stepping cracks, normal tractions on the overlapping crack ends increase and inhibit frictional sliding, whereas for right-stepping cracks, normal tractions decrease and facilitate sliding.” Bakun et al. (1980) studied the seismicity and behavior of the San Andreas Fault zone in central California where strain release is characterized by creep and moderate earthquakes. In their analysis, they modeled fault segments in an en echelon fashion (Figure 3.6). Behavior of these segments was as would be pre-