Field outcrop studies and detailed interpretation of 3-D seismic data indicate that faults in nature and at various scales (e.g., at the scale of the upper continental crust or that of the sedimentary cover) typically have sinuous traces and that long faults are composite zones made of coalesced, shorter fault segments. During the early stages of extension of a brittle layer, short fault segments form (Fig. 1). Their location and distribution are either random or may be clustered by the presence of preexisting zones of weakness. With continued extension, each short fault segment propagates along strike (Fig. 1). Analytical and numerical simulations of crack growth and propagation (Pollard and Aydin, 1984; Olson and Pollard, 1991; Pollard, 2000) have shown that the tips of a propagating crack or fault change the local stress regime. Studies have also shown that when the tips of two cracks or faults propagate close to one another, the cracks or faults start to influence each other's growth (Willemse, 1997). Thus, normal faults do not propagate along strike forever but tend to terminate along strike by linking with other faults, thereby forming relay zones connecting the two faults (e.g., Gibbs, 1984; Pollard and Aydin, 1984; Larsen, 1988; Morley et al., 1990; Peacock and Sanderson, 1991; Zhang et al., 1991; Trudgill and Cartwright, 1994; Dawers and Anders, 1995; Hooper and More, 1995; Davies et al., 1997; Koledoye et al., 2000). Likewise, in areas subjected to a strike-slip regime, strike-slip faults are rarely made of a single, straight and continuous plane. The main fault is commonly made of segments that are laterally offset. Depending on the fault's sense of slip and on the type of offset, linkage between such fault segments can lead to the formation of local transtensional (e.g., pull-apart basins), or transpressional (e.g., restraining bends) zones.

Figure 1. Nucleation of randomly distributed faults

Nucleation of randomly distributed faults

Cartoon illustrating the nucleation of randomly distributed faults followed by their propagation along strike.

Whether regional tectonics involves strike-slip or extension, along-strike propagation of laterally offset faults leads to the formation of short transition zones or relay zones, in which the strains and displacements associated with one fault are transmitted to another fault. Because the stress regime in such relay zones can greatly vary in space, the trends and dips of fault planes and of strata in relay zones, pull-apart basins, and restraining bends can greatly vary within short distances. Field mapping of outcrop data usually does not provide much information about the third dimension (depth). Although 3-D seismic data may cover the three dimensions, their clarity is damaged by the fact that pull-apart basins and relay zones often comprise steeply-dipping reflectors (fault planes and stratigraphic horizons) that lower seismic resolution. Furthermore, neither outcrop nor 3-D seismic analysis can directly provide reliable information about the evolution of such structures through time, which is the reason why a significant number of researchers have turned to forward modeling to analyze and predict the evolution of relay zones and pull-apart basins through time. Because 3-D numerical modeling is still technically challenging, most researchers have relied on experimental (physical) modeling. Such physical experiments have focused on strike-slip (e.g., Faugère and Brun, 1984; Sims et al., 1999) or on extensional (Vendeville, 1991; Le Calvez and Vendeville, 1996) regimes.