Pilotaje

Faults typically grow through radial propagation and/or segment linkage. Lateral growth on isolated faults can occur as a result of both coseismic and aseismic release (e.g.

Walsh and Watterson, 1987), increasing both displacement and length proportionally. However, faults often interact and link to form larger complex systems. In this scenario,

growth will not only be governed by seismic activity and aseismic creep, but also by the kinematics involved when the stress regimes from adjacent faults interact (e.g. Pollard and Segall, 1987; Burgmann et al., 1994; Peacock, 2002).

Studies incorporating both geological data and numerical modelling have indicated that there is a length / displacement relationship:

dmax = cLn

where dmax = maximum cumulative displacement, c = constant based on rock properties, L

= maximum linear fault length and n = exponent value ranging from 0.5 – 2 (Young-Seog and Sanderson, 2005). On isolated faults, maximum displacement will occur at the centre of the fault, with a linear tapering of displacement to zero at the fault tips (e.g. Young- Seog and Sanderson, 2005). When faults begin to interact and link, models predict that displacement maxima move towards the point of interaction, and displacement rapidly drops off towards the fault tips, due to the interaction between adjacent stress fields (Peacock and Sanderson, 1996). Figure 1.7 illustrates the two methods of fault growth and the associated displacement profiles.

Trudgill and Cartwright (1994) and Cartwright et al. (1996) examined aerial photographs and structural maps of Canyonlands, Utah, mapping fault segments ranging from ~300 m - >10 km in length. Segment boundaries were recognisable by changes in strike or an abrupt offset along strike. An example displacement profile appeared to indicate a single fault structure, with maximum displacement located towards the centre of the fault system. However, locations of high variability within this overall profile (Figure 1.8) were found to coincide with breached relay structures (Trudgill and Cartwright, 1994; Cartwright et al., 1996).

Figure 1.7: Illustration of two methods of fault growth and the resulting length / displacement profiles. 1) Fault growth by radial propagation. Dashed red lines indicate the location of maximum displacement. It can be seen that the profile maintains a bell-shaped curve, with maximum displacement always located at the centre of the fault. 2) Fault growth by segment linkage. As the faults grow, initially by radial propagation, they begin to overlap and interact. This causes maximum displacement to move towards the point of interaction. Finally the segments link, forming a profile that is indicative of a single fault, but with displacement lows indicating the positions of relic segment boundaries. (After Cartwright et al., 1995).

Figure 1.8: Illustration from

Cartwright et al. (1996) showing the displacement profile along a segmented fault system.

A) Fault segment geometry

B) Displacement profile. Dotted red line shows a typical bell- shaped profile of an isolated fault.

Once through – going linkage has occurred, the displacement profile will retain minima that indicating the positions of relic segment boundaries (e.g. Peacock and Sanderson, 1991; Cartwright et al., 1995; McLeod et al., 2000). However, at some locations, the transfer of displacement between segments was taken up on subsidiary faults that ran parallel to the main faults (Trudgill and Cartwright, 1994), producing a displacement high at segment boundaries instead of a minimum.

Goldsworthy and Jackson (2001) state that sub-parallel faults may show displacement variations that indicate strain is migrating between faults on either a permanent basis, or episodically, on timescales that are relatively rapid. This suggests that strain can be transferred periodically between adjacent, sub-parallel faults if a segment boundary on one is acting as an asperity, in addition to longer term permanent partitioning of strain onto more established through-going faults.

Accurate analysis of fault growth and displacement profiles can be affected by: resolution and censoring (e.g. Walsh and Watterson, 1988); extent and contribution from damage zones surrounding the main fault (Peacock and Sanderson, 1991; Knott et al., 1996; Peacock, 2002;); drag and rotation of sediments in proximity to the fault (e.g. Trudgill and Cartwright, 1994; Mansfield and Cartwright, 1996); and the maximum length of the surface trace of fault displacement measured (e.g. Walsh and Watterson, 1988), as this can often be shorter than the true length at depth on a developing system. These factors will be addressed, wherever possible, throughout the error and displacement analysis conducted in this study (Chapters 4 and 6).

Field studies often show a wide scatter in length / displacement data points. Cartwright et al. (1995) believe that some of the scatter can be explained by the cycle of fault interaction, growth, and finally linkage (Figure 1.9). As two segments link, the increase in length can cause under-displacement relative to the length / displacement ratio. Displacement is accrued until the ratio is restored, and the fault begins to behave as an isolated structure again, growing laterally as well as accruing vertical displacement. This periodic deviation from the idealized linear length / displacement relationship creates a step-like pattern (Figure 1.9).

Figure 1.9: Illustration of the step-like displacement profile produced by segment linkage, where L = fault length and D = fault displacement. The dashed red line indicates idealized fault growth, showing a linear relationship between length and displacement. Linkage between segments causes deviations from this line due to periodic under-displacement relative to length. The degree of deviation is dependant on the length and number of segments involved. (After Cartwright et al., 1995).

Another hypothesis to explain scatter in field data is put forward by Peacock and Sanderson (1991). Fault segment linkage will increase a fault’s ability to release strain, and so in order to maintain the regional strain rate, the displacement rate on other faults may slow or cease completely. This adjustment may take time to occur, and so pre and post-linkage length / displacement ratios may be variable (Peacock and Sanderson, 1991), departing from the idealized length / displacement ratio predicted by numerical modeling.

Periods of quiescence, earthquake clustering and the chosen period of observation may mean that the slip rates estimated for a segmented system may not match the rates required to create the long-term fault morphology (Cowie and Roberts, 2001). Likewise, longer periods of observation may not reveal short term variability in displacement rates, which can identify the initiation of fault interaction and linkage (e.g. Mansfield and Cartwright, 2001).

This study provides a unique opportunity to the study short-term variability of a young fault system developing within an established continental rift. Longer-term fault activity has been studied on neighbouring faults, such as the Eliki Fault (Armijo et al., 1996; Micarelli et al., 2003; McNeill and Collier, 2004; McNeill et al., 2005b), but this is the first opportunity to address very short timescale variations that can then be compared to the long-term fault behaviour of the established faults.