Deeper Fault Geometry and the Low-Angle-Normal-Fault Enigma….127

Although been mapped in the brittle crust of many extending tectonic provinces and accommodate significant crustal extension, activity especially seismic slips on low-angle normal faults has long been questioned (Collettini, 2011, and references therein). In a simple brittle extensional regime with vertical maximum principle stress (Anderson, 1951) containing isotropic and fluid saturated (hydrostatic) rock mass, according to frictional sliding failure criterion pre-existing planar discontinuity dipping lower than 30° is unable to be reactivated unless the pore fluid pressure exceeds the minimum principle stress; similar situation applies for normal fault initiation that based on Coulomb shear fracturing failure criterion fault planes dip less than 30° cannot be formed (Sibson, 1985, 2000; Collettini and Sibson, 2001). Another way to keep slips on the low-dipping planes is to radically reduce its frictional coefficient by the massive presence of phyllosilicates. However significant fluid overpressure in extensional regimes is thought to difficult to be maintained as tensile environment promotes permeability and is vulnerable to hydrofracturing, meaning the high-pressure fluid is easily drained (Sibson, 2000). Presence of massive phyllosilicate on the fault zone results from intense fluid-assisted diffusion mass transfer or the incorporation of particular lithology (e.g. serpinitinite), while the velocity-strengthening behavior of the phyllosilicates permits only aseismic slips on the fault as creeping (e.g. Moore and Rymer, 2007; Collettini et al., 2009). For the Hsinchuang Thrust and the main decollement in northern Taiwan both the strong mineralization of phyllosilicates and significant fluid overpressure are not observed.

What adds to the dilemma is the absence of moderate to large (M>5.5) normal faulting earthquakes occurring on rupture planes dipping less than 30° globally

128

(Jackson and White, 1989; Collettini and Sibson, 2001), rendering further doubts on the seismogenic potential of low-angle normal fault systems.

5.5.2.2 Half-space elastic dislocation modelling of fault plane models with double-ramp geometry

Based upon the above arguments, a set of experiment is designed to explore if the now deep-seated normal faults can be involved in the Shanchiao Fault. The shallow part of the fault plane is listric as previously constrained, and a deeper steeper fault patch is introduced resulting in a ramp-flat-ramp geometry with two bends (Fig. 5-9).

Geometry of the shallow listric part is taken from one of the good fits (Fig. 5-8) with parameters a, b and n (Fig. 5-6B) as 75°, 15° and 3, respectively. The deeper fault patch as the second ramp is set to be dipping at 60°, the optimal angle for normal fault initiation and slip. The experiments are aimed at resolving the depth of the flat-ramp junction where the fault is bend downward (‘m’ in Fig. 5-9). The modeling outcome is presented in Fig. 5-10. When the junction depth is shallower than 8 km the double-ramp geometry produces an anticline in the hanging wall (axis at 7 km away from the fault when m=5, and at 13 km when m=7) instead of the roll-over monocline revealed by the horizon depth distribution, and the model results in the anticline area significantly underestimate the tectonic offset. When the bend depth from the middle sub-horizontal patch to the deep ramp patch is 8 to 9 km deep excellent fits with the Jingmei Formation top horizon record are obtained. Compared with the listric good fits, the double-ramp good fits are able to envelope or contain the geologic data of the entire profile with lessened underestimations for data 7 to 11 km away from the fault, and exhibit smoother trends between 5 and 13 km away from the fault. Therefore the double-ramp geometry with the deeper bend at 8 to 9 km depth is a good candidate for the upper crustal configuration of the Shanchiao Fault.

Fig. 5-9. Settings of the double-ramp models in elastic half-space dislocation modeling.

129

Fig. 5-10. Modeling outcomes of the double-ramp models. When m=8 to 9 km deep, excellent fits with the Jingmei Formation top horizon record are obtained with lessened underestimations for data 7 to 11 km away from the fault and smoother trends between 5 and 13 km away from the fault, compared with the listric good fits.

130

5.5.2.3 The role of pre-orogen rift normal faults on post-collisional extension tectonics

The deeper steep fault patch in the double-ramp geometry illustrated above is a possible manifestation of the pre-orogen normal fault buried under the orogenic wedge. Although this deep patch might be a newly-formed structure during the formation of the Shanchiao Fault in the post-orogenic processes, the fact that the Shanchiao Fault evidently re-slipped the pre-existing thrust décollement and the widespread presence of rift normal faults on the Chinese Continental Margin (Teng and Lin, 2004) strongly advocates its reactivation origin. The rift normal faults were mainly active from Eocene to late Oligocene or early Miocene and extend from the continental margin sediments into the basement till probably mid-crustal depth (Teng et al., 1992; Lin et al., 2003). When incorporated into the collision zone, the pre-orogen normal faults are mostly concealed beneath the wedge as the main detachment in the Western Foothill fold-thrust belt is located in Miocene and late Oligocene strata (Suppe, 1980), though their reverse faulting reactivation are also postulated (e.g. Mouthereau and Lacombe, 2006). For northern Taiwan the detachment level for the Hsinchuang Thrust and regional décollement is considered to be in the Wuchishan Formation deposited in late Oligocene. Thus the rift-related normal faults contained within the continental margin basement and lower sediments were underthrust below the frontal part of the wedge during the convergence and are difficult to be delineated from surface geological data. The modeling results thereby suggest that in northern Taiwan around the Taipei region pre-orogen rift normal fault is present under the fold-thrust pile and is currently re-slipped by the post-orogenic / post-collisional extensional faulting. To summarize, the modeling results indicate the Shanchiao Fault has reactivated both the Hsinchuang Thrust at intermediate depth and a rift normal fault further deep (Fig. 5-11).

A further supporting evidence for the deep-steepening of the Shanchiao Fault plane comes from detailed relocation and focal mechanism determination of a recent M_W 3.8 earthquake happened on 23 October 2004. The epicenter is located slightly east of the eastern border of the Taipei Basin within the foothills, at depth estimated between 7 and 10 km (Lin, 2005; Chen et al., 2010b; Lee et al., 2010). With improved inversion methods, the earthquake rupture source parameters from the determined focal mechanism point to a normal fault striking northeast and dipping about 60° to southeast (Chen et al., 2010b; Lee et al., 2010). The fault plane dip estimates are in agreement with the proposed double-ramp geometry, and the rupture was likely to have occurred on the upper boundary of the lower ramp fault patch (Fig. 5-11).

The continuing deviations between geological data and modeling outcomes from 7

131

to 12 km on the Wuku-Sanchung-Taipei profile, though narrowed down in the double-ramp models, implies more complexities exist about the deeper fault plane geometry or other geological agents affecting the Jingmei Formation top horizon depth variation in the region, and is subject to further investigation.