dicted by Segall and Pollard’s model, strengthening the idea that fault systems behave in a segmented manner.
If the discontinuities are large enough, segmentation of a fault system may be intuitively easy (e.g., a 10-km en echelon step appears to be a substantial discontinuity with a good chance of stopping a rupture). In less clear cases of discontinuities, several lines of geologic, seismologic, and geometric evidence must be gathered to suggest or substantiate the existence of the discontinuity. An understanding and delineation of the entire earthquake history of a fault can be used as relatively strong evidence of segmentation. Once convinced that a particular segmentation is reasonable, various approaches including correlation with faulting parameters can be used to estimate earthquake magnitudes. Schwartz and Coppersmith (1984; Chapter 14, this volume) suggested examples of fault zones for which the segmentation model may be applicable.
Earthquake recurrence intervals can vary markedly from fault to fault. Historical seismicity of the Parkfield segment of the San Andreas Fault system suggests a recurrence rate of 21±4 yr (Bakun and McEvilly, 1984), whereas soils and trenching data suggest to Machette (1978) that the County Dump Fault in New Mexico has a recurrence interval of 90,000 to 190,000 yr. As pointed out by Wallace (1970) and Schwartz and Coppersmith (1984; Chapter 14, this volume), the slip rate of a fault directly affects recurrence rates.
Various models have been used to explain earthquake recurrence, including the time-predictable (Shimazaki and Nakata, 1980; Bufe et al., 1977), slip-predictable (Shimazaki and Nakata, 1980), and the periodical model (Bakun and McEvilly, 1984).
Wallace (1970) discussed recurrence for the San Andreas Fault. Subsequent refinements in dating, exploratory trenching, geomorphic expression, and low-Sun-angle aerial reconnaissance or photography have greatly expanded knowledge of faulting recurrence, paleoseismicity, and scarp morphologic change. These new methods are critical to active fault evaluations and timing and probability analyses.
Studies determining paleoseismic history have been conducted recently at Pallett Creek (Sieh, 1984) and Wallace Creek (Sieh and Jahns, 1984) and at Cajon Pass (Weldon and Sieh, 1985) along the San Andreas Fault zone. Cross-cutting relationships and radiocarbon dates