The dynamic propagation of an earthquake rupture will generate inelastic deformation within its surrounding medium, culminating in the development of a fault damage zone. These are heavily fractured volumes of rock that flank the fault core, where the majority of displacement has been accommodated. Using the example of the Alpine Fault, I assess the mechanical and chemical processes associated with fault damage, which strongly condition the short and long term evolution of a fault.
Approximately 70% of the oblique-dextral motion between the Australian and Pacific plates on the South Island of New Zealand is localised onto the Alpine Fault. A continuous record of its damage zone extending <30 m above its principal slip zones (PSZs) is provided by core recovered during the first phase of the Deep Fault Drilling Project (DFDP-1). A combination of visual core descriptions, X-ray and neutron tomographic core scanning, and X-ray diffraction demonstrate that the damage zone is epitomised by gouge-filled ‘phyllosilicate-enriched’ fractures. These contain a soft fine-grained fill, which have a relatively low density but are hydrogen-rich. Their bulk composition reflects both wear of the surrounding rock and phyllosilicate mineralisation during hydrothermal alteration. Fracture density in DFDP-1 core and field transects does not increase with proximity to the PSZs, but does systematically vary with lithology.
By reorienting core sections with respect to geographically referenced borehole televiewer logs of the DFDP-1 boreholes, the true orientations of 637 fractures was obtained. Combined with field observations, these results indicate that damage zone fractures occupy a wide range of orientations. This reflects variable stress states adjacent to the Alpine Fault, which are generated by a fault trace that is non-planar in the near-surface, and non-optimally orientated with respect to the regional stresses at depth. The mechanical anisotropy of the foliated mylonites that host the damage zone cannot strongly influence fracturing.
In field transects, broadly-oriented fractures are confined to within 50-150 m of the Alpine Fault, and this is the best estimate of damage zone width along its central section. This width is comparable to elsewhere along-strike. Therefore, the Alpine Fault is considered to be embedded within a tabular localised damage zone, as documented in other structurally mature crustal-scale faults.
Petrological and scanning electron microscopy demonstrate that gouge-filled fractures extend to the micron scale where calcite, chlorite, K-feldspar and muscovite veins are also present. These veins and fractures form in a cyclic manner that may be operative throughout the Alpine Fault’s seismogenic zone. The documentation of some veins that cross-cut cataclasite textures requires that strain localisation from the 10-20 m thick Alpine Fault cataclasites to its 10-50 cm thick PSZ gouges must occur at depths <2-5 km.
This study has sampled the Alpine Fault damage zone late in its cycle of stress accumulation. Within this period, micro and macro-scale fracture healing has occurred at distances <25 m of the PSZs. A combination of fluid over-pressures and unhealed damage elsewhere permit a Low Velocity Zone around the Alpine Fault.
Understanding the dynamics of shallow earthquake rupture and coseismic slip in carbonate-dominated fault zones is exceptionally important in many areas worldwide. Although constraints are available from rock mechanics experiments and paleoseismological investigations of surface breaks, little is currently known about the dynamic processes that occur in narrow fault slipping zones in the near-surface (<3 km) environment.
The Tre Monti Fault, central Italy (exhumed from depths <2 km) is an active carbonate-dominated fault zone that has probably hosted large (Mw >6) earthquakes. In previous studies of the principal slip zone (PSZ) of the Tre Monti Fault, peculiar aggregate- type grains were recognised and termed “Clast-Cortex Aggregates” (CCAs). Such grains contain a central clast (often angular) of host rock or reworked cataclasite surrounded by an outer, often laminated cortex containing much finergrained material. CCAs were only identified in the high-strain PSZ of the Tre Monti Fault (i.e. they are restricted to an ultracataclastic layer <2 cm thick) and as such they potentially record the effects of localized dynamic slip processes during the seismic cycle. However, because of their monomineralic composition, fine grain size and extremely low porosity, previous attempts to image CCAs using standard SEM-based techniques were not successful.
Using high-resolution Electron Backscatter Diffraction (EBSD) analysis of a single CCA, this thesis presents new observations regarding the microstructure and evolution of CCAs in carbonate-dominated fault zones. Results indicate that: 1) the diameter of individual grains comprising the CCA outer cortex and surrounding matrix of the PSZ is on the order of 1-5 ?m; 2) mean grain size decreases progressively from the inner laminations towards the outer laminations; 3) in all locations around the CCA, and in the surrounding matrix of the PSZ, individual grains are markedly elongate (mean aspect ratio between 1.5 and 2.1) and are aligned at c. 70°-90° to the principal slip surface, and; 4) there is no significant crystallographic preferred ordinations of grains within the laminations of the outer cortex.
These new results support a model in which CCAs form by accretion of grains on to the outside of a central clast rotating clast within the cataclastic PSZ at shallow depths (<2 km). The PSZ is progressively crushed during slip events decreasing the grain size and resulting in grain size grading within the laminations of the CCAs. The formation of CCAs may be related to coseismic slip, however the data does not allow tight constraints on slip rates during formation. Following the formation of the CCAs, compaction within the fault zone, perhaps accommodated by pressure solution processes, results in individual grains within the CCA and surrounding matrix developing a systematic elongate shape. such distributed compaction may correspond to the shallow interseismic creep that has been observed at shallow depths following major recent earthquakes in carbonate-dominated fault zones.
At Mt. Raddle in the Olivine Range, South Westland, pristine mantle harzburgite of the Dun Mountain Ophiolite Belt (DMOB) shows a complete transition through to the serpentinite mélange of the Livingstone Fault, which juxtaposes the ultramafics against the mainly quartzofeldspathic Caples Terrane. The mélange comprises a ~100 m wide zone of scaly serpentinite with embedded pods of massive serpentinite and serpentinised harzburgite. The orientation of the scaly fabric, the long axes of entrained pods and the strike of fractures within the pods are all sub-parallel to the mélange boundaries. As the western mélange boundary is approached, progressive hydration of pristine harzburgite in the DMOB occurs by serpentinisation of olivine and pyroxene, and the development of Fe-chromite from Cr-spinel. Incipient hydration may be recorded by overgrowths of secondary olivine around pyroxene.
A talc-tremolite-clinopyroxene metasomatic reaction zone developed at the serpentinite mélange-Caples Terrane contact. Petrographic and microstructural observations indicate that talc was formed as the initial reaction product. Later production of tremolite and clinopyroxene occurred by replacement and overgrowth of talc, to the extent that talc now comprises <5 vol% of samples derived from the reaction zone. Broadly synchronous growth of tremolite and clinopyroxene indicates that the composition of fluids in the reaction zone was spatially and temporally variable.143Nd/144Nd and 87Sr/86Sr isotope ratios show that fluids interacting with the Caples Terrane were important in controlling metasomatic reactions and are a likely source for Ca in the reaction zone. Isotope profiles indicate that extensive fluid circulation occurred across the full width of the serpentinite mélange. In contrast, serpentinised harzburgites in the DMOB retain primitive (mantle) isotopic ratios indicating that fluids in the mélange did not penetrate significantly into the serpentinising harzburgite.
Early formation of frictionally-weak talc is interpreted to have localised strain in the metasomatic reaction zone. Subsequent formation of tremolite and clinopyroxene, which are stronger and more frictionally unstable than talc, may have promoted strengthening of the reaction zone and a rheological transition towards unstable rupture. The lithologies and sequential metasomatic reactions described here, together with broad estimates of P-T conditions, suggest that the Livingstone Fault at Mt. Raddle provides a superbly exposed example of some of the physio-chemical processes thought to be important along active serpentinite-bearing shear zones, including the San Andreas Fault and subduction zone megathrusts.
Large normal faults are often reactivated as high-angle reverse faults during compressional basin inversion. Although a common situation worldwide, basin inversion is poorly understood from a mechanical perspective, as high-angle reverse faults are severely misoriented for reactivation (frictional ‘lock-up’ should occur at dips of c. 60°). Prevailing models of failure on high-angle reverse faults rely on fluid overpressure, however such models are calculated on the assumption of Byerlee-type friction (friction coefficient of 0.6 – 0.85).
The Moonlight Fault Zone in New Zealand, is a >200 km long Oligocene basin-bounding normal fault that reactivated in the Miocene as a high-angle reverse fault during basin inversion (present dip angle 65°-77°). Excellent exposures of the fault zone exhumed from c. 4-10 km depth are found in creek sections along the entire strike length. This thesis presents field and microstructural observations concerning the structure and fault rock assemblages found in five creek sections within the MFZ, and aims to provide some of the first observational constraints on the structure and possible mechanical properties of reactivated normal faults.
In the MFZ wall rocks are mainly quartz-albite-muscovite-chlorite schists with a strong foliation that is everywhere sub-parallel to the Moonlight Fault (i.e. dip angle 65°-75°), with deformation varying in response to host rock composition. Where the hanging wall consists of well foliated, intact greenschist or quartzofeldspathic gneiss, pseudotachylyte is present lying largely sub-parallel to the foliation. Where the hanging wall consists of fissile greyschist it is host to foliation-parallel fault breccias. The footwall is mainly greyschist where fault movements resulted in the formation of meso to macro scale folds whose fold axial planes lie parallel to the orientation of the Moonlight Fault. Where folding has not accommodated all reverse-slip, Moonlight Fault-parallel breccias are present.
Although the overall structure of the fault zone changes significantly along strike in response to wall rock composition, the fault core always contains interconnected layers of foliated cataclasite or gouge rich in authigenically-grown chlorite and muscovite which is regionally significant and, critically, interconnected on a regional scale. The fault core is regularly flanked by a zone of breccia which at times shows a strain transition into the <5 metre thick fault core. Microstructural evidence suggests deformation in the fault core was accommodated by a combination of cataclasis, frictional slip along phyllosilicate seams and dissolution-precipitation.
Published frictional strength measurements for chlorite and muscovite (friction coefficient of 0.32-0.38) are used to explore mechanical models of frictional reactivation along high-angle reverse faults. Results show that low-friction fault cores increase the frictional lock-up angle to 71°, allowing for easier reactivation of faults that initially formed at 60°. These results indicate that low frictional strength may play an important role in slip on high-angle reverse faults during basin inversion.
The Moonlight Fault Zone (MFZ) is a large, regionally significant fault in the South Island of New Zealand that acted as a basin?bounding normal fault in the Oligocene and was subsequently reactivated as a high angle reverse fault in the late Miocene. The fault zone is well exposed in and around Fan Creek, near Queenstown. Using detailed field and microstructural observations, this research dissertation documents the structure and fault rock assemblages found along the MFZ in the Fan Creek area, and draws conclusions regarding the exhumation history of the fault zone, deformation processes within the fault rocks, and potential fault weakening mechanisms active during high angle reverse movements. The exposures in Fan Creek consist of four lower greenschist facies metamorphic host rock lithologies and an infaulted package of Oligocene sediments (sandstones and conglomerates). Textural zone IV (Torlesse terrane) phyllites in the hanging wall are juxtaposed against the infaulted Oligocene sediments or textural zone IIB (Caples terrane) phyllites and schists in the footwall. Deformation in the hanging wall is concentrated within foliation?parallel fault breccias that show evidence of multiple cycles of seismic slip and cataclasis alternating with restrengthening and dissolution?precipitation creep processes. The main fault contact is defined by phyllosilicate?rich cataclasites containing a well?defined foliation of aligned and interconnected, platy muscovite and chlorite grains at the schist?sediment faulted contact, and a discontinuous layer of relatively impermeable red fault gouge at the schist?schist faulted contact. Emanating from the main fault plane and tapering away into the immediate hanging wall are large quartz veins, interpreted as the result of build?ups of fluid overpressure in the footwall. Secondary fault strands, separating the metamorphic footwall lithologies, contain complex breccia and vein networks, including quartz veins associated with sulphide mineralization. Brittle shears, some containing green, foliated cataclasites are present in the footwall schists. Results suggest that both elevated fluid pressures and frictionally weak, interconnected phyllosilicates along the main fault trace facilitated Miocene?Recent high angle reverse movements of up to c. 8 km across the MFZ. These processes combined to lower the shear strength of the fault resulting in episodic rupture, the release of fluids into the hanging wall and the onset of fluid?mediated diffusive mass transfer (dissolution-precipitation) processes.
Currently, pseudotachylytes are the only unequivocal indicator of seismic slip in the rock record, but pseudotachylytes do not form in certain litholo- gies (e.g. carbonates) and are mainly restricted to depths of >5 km. Seis- mically active carbonate-bearing areas such as central Italy, Greece and the Himalayan belt would benefit from a better understanding of how ancient seismicity is preserved in the rock record, to better evaluate the future haz- ard. Using a unique dataset of synthetic fault rocks produced in earthquake-like laboratory experiments, this project aimed to quantify the microstructures produced in mixed calcite-dolomite gouges. Specifically, the project focuses on the microstructural evolution of samples deformed under identical seis- mic conditions (Max. slip vel. = 1 msô€€€1, n = 18 MPa) but taken to increasing displacements of 0.03 to 0.4 m (representative of average coseis- mic slip during approx. Mw 4 - 6 earthquakes). Quantitative analysis of SEM images, combined with chemical and opti- cal analysis, shows that grain size and gouge fabric evolve systematically with increasing displacement. The bulk of the microstructural changes oc- cur within the first c. 0.1 m of slip, during a transient strengthening and dynamic weakening phase during which grain comminution occurs by cat- aclasis. The weaker calcite phase wraps around relatively rigid dolomite clasts to define coarse foliations formed entirely by brittle processes. After c. 0.1 m of slip, strain has localised within the gouge layers and tempera- ture sensitive processes (e.g. thermal decomposition, grain-scale plasticity iii and recrystallization) become increasingly important near the localized slip surface. During this phase, cataclastic processes in the bulk gouge layer slow down or cease but the coarse foliations continue to rotate slowly to- wards parallelism with the localized slip surface. The results indicate that relatively small-displacement, seismic slip events in upper-crustal fault rocks could generate foliated fault rocks paired with a sharp principal slip zone on which carbonate thermal decomposition prod- ucts may be present. This may leave a lasting microstructural signature of seismic slip in the rock record.
Earthquakes often rupture through carbonate rocks in the upper crust, therefore it is important to understand the microstructures produced during faulting. This may help to identify the mode in which particular carbonate-‐bearing faults slipped in the past (i.e. seismically during an earthquake or aseismically during creep). This is relevant for earthquake hazard analysis and it is also an important first step in using fault rock microstructures to quantify fault mechanical behaviour (e.g. dynamic stresses, frictional heal anomalies). This study involves a microstructural analysis of calcite fault gouges experimentally deformed under “earthquake-‐like” conditions (slip velocity >1 m s-‐1; displacements <5 m; normal stresses < 30 MPa). The specimens were deformed in the Slow to High Velocity Rotary-‐Shear frictional Apparatus (SHIVA) at the INGV, Rome. Three samples made of pure calcite gouge were analysed in this work. The samples were deformed in a ring-‐shaped (external/internal diameters of 55 mm and 35 mm) gouge sample holder. Scanning Electron Microscope (SEM) images were collected of the three deformed samples using polished thin sections cut approximately parallel to the slip direction and perpendicular to gouge layer boundaries. Using representative SEM images, grain boundaries were manually traced and the resulting images of the calcite aggregates were quantified (e.g. grain sizes, grain orientations, grain locations) using Image SXM software. Three distinct microstructural domains were identified in the deformed calcite gouges; 1) plastically deformed layers of gouge recrystallization, 2) transitional layers with partial recrystallization and 3) brittly deformed layers. The plastically deformed layer becomes thicker with increasing mechanical work (and frictional heating). Grains within the plastically deformed layer evolve to become more elongate and aligned parallel to gouge layer boundaries with increasing mechanical work. Two well-‐defined grain shape-‐ preferred orientations (SPO) are present in the plastically deformed layer. The dominant SPO results from plastic stretching and elongation of grains, while the secondary SPO is due to fracturing of the elongate grains at high angles to their long axes. Grain sizes in the gouge layers generally decrease towards the localized principal slip surfaces, but the data indicate some late stage grain growth in the plastically deformed layers close to principal slip surfaces, possibly reflecting grain annealing immediately following the experimental slip pulses. Overall, the results show that calcite gouge microstructures correlate in a systematic way to both the experimental conditions and the distance from the localized principal slip surface. This is significant because it suggests that experimental and natural microstructures could be reliably compared to understand aspects of the seismic deformation history of natural carbonate-‐bearing fault zones.
The Moonlight Fault Zone (MFZ) is a regionally significant structure in Otago that was reactivated as a high angle reverse fault in the Miocene. This movement exhumed fault rocks from the mid to upper crust which has provided the opportunity to study the structure of the fault zone from this depth, the dominant deformation processes that occurred during faulting and, significantly, the weakening mechanisms that may have facilitated high angle reverse movements. Field and microstructural examination of the Moonlight Fault Zone in the Matukituki Valley revealed that solidified frictional melts (pseudotachylytes) are present within the hanging wall greenschist at least 500 m from the main fault trace. The pseudotachylytes lie parallel or sub-parallel to the steeply-dipping host rock foliation. The presence of pseudotachylytes indicates that the hanging wall was exhumed from at least 5 km depth and that brittle failure within the MFZ occurred, at least in part, by localised seismic slip. Along the main trace of the Moonlight Fault there is a c. 15 m wide zone of deformation that contains a progressive transition from random fabric breccias to well foliated cataclasites ≤1 m from the fault trace. The foliated cataclasites contain microstructural evidence (e.g. dissolution seams enriched in titanite, overgrowths of chlorite in strain-shadows) of fluid-induced dissolution – precipitation reactions associated with diffusive mass transfer. Alteration of load bearing phases such as quartz and feldspar led to the widespread formation of chlorite and muscovite in the main fault. This produced well foliated, interconnected networks of weak phyllosilicate-rich fault rocks. The presence of interconnected phyllosilicates may have lowered the frictional strength of the Moonlight Fault and thus likely contributed towards reactivation of this poorly oriented, high angle reverse fault. The network of foliated phyllosilicates may also have acted as a fluid seal, allowing for build-up in fluid pressure in the footwall and leading to further weakening. The close association between pseudotachylytes and phyllosilicates (containing evidence for dissolution – precipitation) suggests that the MFZ preserves fault rock evidence for both seismic slip and slower aseismic creep.