Measuring Crustal Seismic Anisotropy through Shear Wave Splitting

<p>This thesis involves the study of crustal seismic anisotropy through shear wave splitting. For the past three decades, shear wave splitting (SWS) measurements from crustal earthquakes have been utilized as a technique to characterize seismic anisotropic structures and to infer in situ crustal properties such as the state of the stress and fracture geometry and density. However, the potential of this technique is yet to be realized in part because measurements on local earthquakes are often controlled and/or affected by physical mechanisms and processes which lead to variations in measurements and make interpretation difficult. Many studies have suggested a variety of physical mechanisms that control and/or affect SWS measurements, but few studies have quantitatively tested these suggestions. This thesis seeks to fill this gap by investigating what controls crustal shear-wave splitting (SWS) measurements using empirical and numerical simulation approaches, with the ultimate aim of improving SWS interpretation. For our empirical approach, we used two case studies to investigate what physical processes control seismic anisotropy in the crust at different scales and tectonic settings. In the numerical simulation test, we simulate the propagation of seismic waves in a variety of scenarios. We begin by measuring crustal anisotropy via SWS analysis around central New Zealand, where clusters of closely-spaced earthquakes have occurred. We used over 40,000 crustal earthquakes across 36 stations spanning close to 5.5 years between 2013 and 2018 to generate the largest catalog of high-quality SWS measurements (~102,000) around the Marlborough and Wellington region. The size of our SWS catalog allowed us to perform a detailed systematic analysis to investigate the processes that control crustal anisotropy and we also investigated the spatial and temporal variation of the anisotropic structure around the region. We observed a significant spatial variation of SWS measurements in Central New Zealand. We found that the crustal anisotropy around Central New Zealand is confined to the upper few kilometers of the crust, and is controlled by either one mechanism or a combination of more than one (such as structural, tectonic stresses, and gravitational stresses). The high correspondence between the orientation of the maximum horizontal compressive stress calculated from gravitational potential energy from topography and average fast polarization orientation around the Kaikōura region suggests that gravitationally induced stresses control the crustal anisotropy in the Kaikōura region. We suggest that examining the effect of gravitational stresses on crustal seismic anisotropy should not be neglected in future studies. We also observed no significant temporal changes in the state of anisotropy over the 5.5 year period despite the occurrence of significant seismicity. For the second empirical study, we characterized the anisotropic structure of a fault approaching failure (the Alpine Fault of New Zealand). We performed detailed SWS analysis on local earthquakes that were recorded on a dense array of 159 three-component seismometers with inter-station spacing about 30 m around the Whataroa Valley, New Zealand. The SWS analysis of data from this dense deployment enabled us to map the spatial characteristics of the anisotropic structure and also to investigate the mechanisms that control anisotropy in the Whataroa valley in the vicinity of the Alpine Fault. We observed that the orientation of the fast direction is parallel to the strike of the Alpine Fault trace and the orientations of the regional and borehole foliation planes. We also observed that there was no significant spatial variation of the anisotropic structure as we move across the Alpine Fault trace from the hanging wall to the footwall. We inferred that the geological structures, such as the Alpine Fault fabric and foliations within the valley, are the main mechanisms that control the anisotropic structure in the Whataroa valley. For our numerical simulation approach, we simulate waveforms propagating through an anisotropic media (using both 1-D and 3-D techniques). We simulate a variety of scenarios, to investigate how some of the suggested physical mechanisms affect SWS measurements. We considered (1) the effect on seismic waves caused by scatterers along the waves' propagation path, (2) the effect of the earthquake source mechanism, (3) the effect of incidence angle of the incoming shear wave. We observed that some of these mechanisms (such as the incidence angle of the incoming shear wave and scatterers) significantly affect SWS measurement while others such as earthquake source mechanisms have less effect on SWS measurements. We also observed that the effect of most of these physical mechanisms depends on the wavelength of the propagating shear wave relative to the size of the features. There is a significant effect on SWS measurements if the size of the physical mechanism (such as scatterers) is comparable to the wavelength of the incoming shear wave. With a larger wavelength, the wave treats the feature as a homogeneous medium.</p>

Measuring Crustal Seismic Anisotropy through Shear Wave Splitting

<p>This thesis involves the study of crustal seismic anisotropy through shear wave splitting. For the past three decades, shear wave splitting (SWS) measurements from crustal earthquakes have been utilized as a technique to characterize seismic anisotropic structures and to infer in situ crustal properties such as the state of the stress and fracture geometry and density. However, the potential of this technique is yet to be realized in part because measurements on local earthquakes are often controlled and/or affected by physical mechanisms and processes which lead to variations in measurements and make interpretation difficult. Many studies have suggested a variety of physical mechanisms that control and/or affect SWS measurements, but few studies have quantitatively tested these suggestions. This thesis seeks to fill this gap by investigating what controls crustal shear-wave splitting (SWS) measurements using empirical and numerical simulation approaches, with the ultimate aim of improving SWS interpretation. For our empirical approach, we used two case studies to investigate what physical processes control seismic anisotropy in the crust at different scales and tectonic settings. In the numerical simulation test, we simulate the propagation of seismic waves in a variety of scenarios. We begin by measuring crustal anisotropy via SWS analysis around central New Zealand, where clusters of closely-spaced earthquakes have occurred. We used over 40,000 crustal earthquakes across 36 stations spanning close to 5.5 years between 2013 and 2018 to generate the largest catalog of high-quality SWS measurements (~102,000) around the Marlborough and Wellington region. The size of our SWS catalog allowed us to perform a detailed systematic analysis to investigate the processes that control crustal anisotropy and we also investigated the spatial and temporal variation of the anisotropic structure around the region. We observed a significant spatial variation of SWS measurements in Central New Zealand. We found that the crustal anisotropy around Central New Zealand is confined to the upper few kilometers of the crust, and is controlled by either one mechanism or a combination of more than one (such as structural, tectonic stresses, and gravitational stresses). The high correspondence between the orientation of the maximum horizontal compressive stress calculated from gravitational potential energy from topography and average fast polarization orientation around the Kaikōura region suggests that gravitationally induced stresses control the crustal anisotropy in the Kaikōura region. We suggest that examining the effect of gravitational stresses on crustal seismic anisotropy should not be neglected in future studies. We also observed no significant temporal changes in the state of anisotropy over the 5.5 year period despite the occurrence of significant seismicity. For the second empirical study, we characterized the anisotropic structure of a fault approaching failure (the Alpine Fault of New Zealand). We performed detailed SWS analysis on local earthquakes that were recorded on a dense array of 159 three-component seismometers with inter-station spacing about 30 m around the Whataroa Valley, New Zealand. The SWS analysis of data from this dense deployment enabled us to map the spatial characteristics of the anisotropic structure and also to investigate the mechanisms that control anisotropy in the Whataroa valley in the vicinity of the Alpine Fault. We observed that the orientation of the fast direction is parallel to the strike of the Alpine Fault trace and the orientations of the regional and borehole foliation planes. We also observed that there was no significant spatial variation of the anisotropic structure as we move across the Alpine Fault trace from the hanging wall to the footwall. We inferred that the geological structures, such as the Alpine Fault fabric and foliations within the valley, are the main mechanisms that control the anisotropic structure in the Whataroa valley. For our numerical simulation approach, we simulate waveforms propagating through an anisotropic media (using both 1-D and 3-D techniques). We simulate a variety of scenarios, to investigate how some of the suggested physical mechanisms affect SWS measurements. We considered (1) the effect on seismic waves caused by scatterers along the waves' propagation path, (2) the effect of the earthquake source mechanism, (3) the effect of incidence angle of the incoming shear wave. We observed that some of these mechanisms (such as the incidence angle of the incoming shear wave and scatterers) significantly affect SWS measurement while others such as earthquake source mechanisms have less effect on SWS measurements. We also observed that the effect of most of these physical mechanisms depends on the wavelength of the propagating shear wave relative to the size of the features. There is a significant effect on SWS measurements if the size of the physical mechanism (such as scatterers) is comparable to the wavelength of the incoming shear wave. With a larger wavelength, the wave treats the feature as a homogeneous medium.</p>

Microseismicity, tectonic stress, and exhumation rates near the central Alpine Fault, New Zealand

<p>This thesis documents a detailed examination of the seismic activity and characteristics of crustal deformation along the central Alpine Fault, a major obliquely convergent plate-boundary fault. Paleoseismic evidence has established that the Alpine Fault produces large to great (M7−8) earthquakes every 250−300 years, in a quasi-periodic manner, with the last surface-rupturing earthquake occurring in 1717. This renders the fault late in its typical earthquake cycle, posing substantial seismic risk to southern and central New Zealand. Understanding the seismic and tectonic character of this fault may yield information of both societal and scientific significance regarding seismic hazard and late-interseismic processes leading up to a large earthquake. However, the central Alpine Fault is currently seismically quiescent when compared to adjacent regions, and therefore requires detailed, long-duration observations to study seismotectonic processes. The work in this thesis addresses the need for a greater understanding of along-strike variations in seismic character of the Alpine Fault ahead of an anticipated large earthquake. To achieve observations with high spatial and temporal resolution across the length of the central Alpine Fault, I use 8.5 years of continuous seismic data from the Southern Alps Microearthquake Borehole Array (SAMBA), and data from four other temporary seismic networks and five local GeoNet permanent sites. Incorporating all of these temporary and permanent seismic sites provides us with a dense composite network of seismometers. Without such a dense network, homogeneous examination of the characteristics of low-magnitude seismicity near the Alpine Fault would be impossible. Using this dataset, I have constructed the most extensive microearthquake catalog for the central Alpine Fault region to date, containing 9,111 earthquakes and covering the time between late 2008 and early 2017. To construct this catalog I created an objective workflow to ensure catalog uniformity. Overall, 7,719 earthquakes were successfully relocated with location uncertainties generally ≤ 0.5 km in both the horizontal and vertical directions. The majority of the earthquakes were found to occur southeast of the Alpine Fault (i.e. in the hanging-wall). I observed a lack of seismicity beneath Aoraki/Mount Cook that has previously been shown to be associated with locally high uplift rates (6–10 mm/yr) and high geothermal gradients (∼60◦C/km). Seismogenic cut-off depths were observed to significantly vary along the strike of the Alpine Fault, ranging from 8 km beneath the highest topography to 20 km in the adjacent areas. To quantify the scale of the seismic deformation, a new local magnitude scale was also derived, corrected for geometric spreading, attenuation and site terms based on individually calculated GeoNet moment magnitude (Mw) values. Earthquake local magnitudes range between ML –1.2 and 4.6 and the catalog is complete above ML 1.1. To examine the stress regime near the central Alpine Fault, I built a new data set of 845 focal mechanisms from earthquakes in our catalog. This was achieved by manually determining P wave arrival polarity picks from all earthquakes larger than ML 1.5. In order to determine the orientations and characteristics of the stress parameters, I grouped these focal mechanisms and performed stress inversion calculations that provided an average maximum horizontal compressive stress orientation, SHmax, of 121±11◦ , which is uniform within uncertainty along the length of the central Southern Alps. I observed an average angle of 65◦ between the SHmax and the strike of the Alpine Fault, which is consistent with results from similar previous studies in the northern and southern sections of the Alpine Fault. This implies that the Alpine Fault is misoriented for reactivation, in the prevailing stress field. Using a 1-D steady-state thermal structure model constrained by seismicity and thermochronology data, I investigated the crustal thermal structure and vertical kinematics of the central Southern Alps orogen. The short-term seismicity data and longer-term thermochronology data impose complementary constraints on the model. I observed a large variation in exhumation rate estimates (1–8 mm/yr) along the length of the Alpine Fault, with maximum calculated values observed near Aoraki/Mount Cook. I calculated the temperature at the brittle-ductile transition zone, which ranges from 440 to 457◦C in the different models considered. This temperature is slightly hotter than expected for crust composed by quartz-rich rocks, but consistent with the presence of feldspar-rich mafic rocks in parts of the crust.</p>

Microseismicity, tectonic stress, and exhumation rates near the central Alpine Fault, New Zealand

<p>This thesis documents a detailed examination of the seismic activity and characteristics of crustal deformation along the central Alpine Fault, a major obliquely convergent plate-boundary fault. Paleoseismic evidence has established that the Alpine Fault produces large to great (M7−8) earthquakes every 250−300 years, in a quasi-periodic manner, with the last surface-rupturing earthquake occurring in 1717. This renders the fault late in its typical earthquake cycle, posing substantial seismic risk to southern and central New Zealand. Understanding the seismic and tectonic character of this fault may yield information of both societal and scientific significance regarding seismic hazard and late-interseismic processes leading up to a large earthquake. However, the central Alpine Fault is currently seismically quiescent when compared to adjacent regions, and therefore requires detailed, long-duration observations to study seismotectonic processes. The work in this thesis addresses the need for a greater understanding of along-strike variations in seismic character of the Alpine Fault ahead of an anticipated large earthquake. To achieve observations with high spatial and temporal resolution across the length of the central Alpine Fault, I use 8.5 years of continuous seismic data from the Southern Alps Microearthquake Borehole Array (SAMBA), and data from four other temporary seismic networks and five local GeoNet permanent sites. Incorporating all of these temporary and permanent seismic sites provides us with a dense composite network of seismometers. Without such a dense network, homogeneous examination of the characteristics of low-magnitude seismicity near the Alpine Fault would be impossible. Using this dataset, I have constructed the most extensive microearthquake catalog for the central Alpine Fault region to date, containing 9,111 earthquakes and covering the time between late 2008 and early 2017. To construct this catalog I created an objective workflow to ensure catalog uniformity. Overall, 7,719 earthquakes were successfully relocated with location uncertainties generally ≤ 0.5 km in both the horizontal and vertical directions. The majority of the earthquakes were found to occur southeast of the Alpine Fault (i.e. in the hanging-wall). I observed a lack of seismicity beneath Aoraki/Mount Cook that has previously been shown to be associated with locally high uplift rates (6–10 mm/yr) and high geothermal gradients (∼60◦C/km). Seismogenic cut-off depths were observed to significantly vary along the strike of the Alpine Fault, ranging from 8 km beneath the highest topography to 20 km in the adjacent areas. To quantify the scale of the seismic deformation, a new local magnitude scale was also derived, corrected for geometric spreading, attenuation and site terms based on individually calculated GeoNet moment magnitude (Mw) values. Earthquake local magnitudes range between ML –1.2 and 4.6 and the catalog is complete above ML 1.1. To examine the stress regime near the central Alpine Fault, I built a new data set of 845 focal mechanisms from earthquakes in our catalog. This was achieved by manually determining P wave arrival polarity picks from all earthquakes larger than ML 1.5. In order to determine the orientations and characteristics of the stress parameters, I grouped these focal mechanisms and performed stress inversion calculations that provided an average maximum horizontal compressive stress orientation, SHmax, of 121±11◦ , which is uniform within uncertainty along the length of the central Southern Alps. I observed an average angle of 65◦ between the SHmax and the strike of the Alpine Fault, which is consistent with results from similar previous studies in the northern and southern sections of the Alpine Fault. This implies that the Alpine Fault is misoriented for reactivation, in the prevailing stress field. Using a 1-D steady-state thermal structure model constrained by seismicity and thermochronology data, I investigated the crustal thermal structure and vertical kinematics of the central Southern Alps orogen. The short-term seismicity data and longer-term thermochronology data impose complementary constraints on the model. I observed a large variation in exhumation rate estimates (1–8 mm/yr) along the length of the Alpine Fault, with maximum calculated values observed near Aoraki/Mount Cook. I calculated the temperature at the brittle-ductile transition zone, which ranges from 440 to 457◦C in the different models considered. This temperature is slightly hotter than expected for crust composed by quartz-rich rocks, but consistent with the presence of feldspar-rich mafic rocks in parts of the crust.</p>

Using low-frequency earthquakes to monitor slow tectonic deformation in the central Southern Alps, New Zealand

<p>This thesis involves the study of low-frequency earthquakes (LFEs) in the central Southern Alps. The Alpine Fault is the principal locus of deformation within the Australia–Pacific plate boundary in the South Island of New Zealand and it is late in its typical ∼300-year seismic cycle. Surveying the seismicity associated with slow deformation in the vicinity of the Alpine Fault may provide constraints on the stresses acting on a major transpressive margin prior to an anticipated great (≥M8) earthquake. Here, we use 8 years of data from the Southern Alps Microearthquake Borehole Array (SAMBA) (amongst those, 3 years of data were collected as part of this project) in order to: (1) generate an updated LFE catalogue using an improved matched-filter technique that incorporates phase-weighted stacking; (2) compute LFE focal mechanisms and invert them to infer the crustal stress field on the deep extent of the Alpine Fault; (3) expand the LFE catalogue to cover a wider range of spatial/temporal behaviours; (4) study LFE families’ characteristics to identify periods where slow slip might happen. We first use fourteen primary LFE templates in an iterative matched-filter and stacking routine, which allows the detection of similar signals and produces LFE families sharing common locations. We generate an 8-yr catalogue containing 10,000 LFEs that are combined for each of the 14 LFE families using phase-weighted stacking to produce signals with the highest possible signal-to-noise ratios. We find LFEs to occur almost continuously during the 8-yr study period and we highlight two types of LFE distributions: (1) discrete behaviour with an inter-event time exceeding 2 minutes; (2) burst-like behaviour with an inter-event time below 2 minutes. The discrete events are interpreted as small-scale frequent deformation on the deep extent of the Alpine Fault and the LFE bursts (corresponding in most cases to known episodes of tremor or large regional earthquakes) are interpreted as brief periods of increased slip activity indicative of slow slip. We compute improved non-linear earthquake locations using a 3D velocity model and find LFEs to occur below the seismogenic zone at depths of 17–42 km, on or near the hypothesised deep extent of the Alpine Fault. We then compute the first estimates of LFE focal mechanisms associated with continental faulting. Focal mechanisms, in conjunction with recurrence intervals, are consistent with quasi-continuous shear faulting on the deep extent of the Alpine Fault. We then generate a new catalogue that regroups hundreds of LFE families. This time 638 synthetic LFE waveforms are generated using a 3D grid and used as primary templates in a matched-filter routine. Of those, 529 templates yield enough detections during the first iteration of the matched-filter routine (≥ 500 detections over the 8-yr study period) and are kept for further analysis. We then use the best 25% of correlated events for each LFE family to generate linear stacks which create new LFE templates. From there, we run a second and final iteration of the matched-filter routine with the new LFE templates to obtain our final LFE catalogue. The remaining 529 templates detect between 150 and 1,671 events each totalling 300,996 detections over the 8-yr study period. Of those 529 LFEs, we manage to locate 378 families. Their depths range between 11 and 60 km and LFEs locate mainly in the southern part of the SAMBA network. We finally examine individual LFE family rates and occurrence patterns. They indicate that LFE sources seem to evolve from an episodic or ‘stepped’ to a continuous behaviour with depth. This transition may correspond to an evolution from a stick-slip to a stable-sliding slip regime. Hence, we propose that the distinctive features of LFE occurrence patterns reflect variations in the in-situ stress and frictional conditions at the individual LFE source locations on the Alpine Fault. Finally, we use this new extensive catalogue as a tool for in-depth analyses of the deep central Alpine Fault structure and its slip behaviour. We identify eight episodes of increased LFE activity between 2009 and 2017 and provide time windows for further investigations of tremor and slow slip. We also study the spatial and temporal behaviours of LFEs and find that LFEs with synchronous occurrence patterns tend to be clustered in space. We thus suggest that individual LFE sources form spatially coherent clusters that may represent localised asperities or elastic patches on the deep Alpine Fault interface. We infer that those clusters may have a similar rheological response to tectonic forcing or to potential slow slip events. Eventually, we discover slow (10km/day) and rapid (∼20-25km/h) migrations of LFEs along the Alpine Fault. The slow migration might be controlled by slow slip events themselves while the rapid velocities could be explained by the LFE sources’ intrinsic properties.</p>

Cosmogenic nuclide dating of the sediments of Bulmer Cavern: Implications for the uplift history of southern Northwest Nelson, South Island New Zealand

<p>The landscape of Northwest Nelson shows evidence of significant tectonic activity since the inception of the Austro-Pacific plate boundary in the Eocene. Evidence of subsidence followed by rapid uplift from the Eocene to the late Miocene is preserved in the sedimentary basins of Northwest Nelson. However, the effects of erosion mean there is very little evidence of post-Miocene tectonic activity preserved in the Northwest Nelson area. This is a period of particular interest, because it coincides with the onset of rapid uplift along the Alpine Fault, which is located to the south, and the very sparse published data for this period suggest very low uplift rates compared to other areas close to the Alpine Fault. Cosmogenic nuclide burial dating of sediments preserved in Bulmer Cavern, indicate an uplift rate of 0.13mm/a from the mid-Pliocene to the start of the Pleistocene and 0.067mm/a since the start of the Pleistocene. The Pleistocene uplift rate is similar to other published uplift rates for this period from the northern parts of Northwest Nelson, suggesting that the whole of Northwest Nelson has experienced relative tectonic stability compared to other areas close to the Alpine Fault during this period. The mid-Pliocene uplift rate is possibly the first precisely constrained uplift rate in the area for this period, and suggests that there has been a progressive decrease in uplift rates from much higher rates in the late Miocene.</p>

Cosmogenic nuclide dating of the sediments of Bulmer Cavern: Implications for the uplift history of southern Northwest Nelson, South Island New Zealand

<p>The landscape of Northwest Nelson shows evidence of significant tectonic activity since the inception of the Austro-Pacific plate boundary in the Eocene. Evidence of subsidence followed by rapid uplift from the Eocene to the late Miocene is preserved in the sedimentary basins of Northwest Nelson. However, the effects of erosion mean there is very little evidence of post-Miocene tectonic activity preserved in the Northwest Nelson area. This is a period of particular interest, because it coincides with the onset of rapid uplift along the Alpine Fault, which is located to the south, and the very sparse published data for this period suggest very low uplift rates compared to other areas close to the Alpine Fault. Cosmogenic nuclide burial dating of sediments preserved in Bulmer Cavern, indicate an uplift rate of 0.13mm/a from the mid-Pliocene to the start of the Pleistocene and 0.067mm/a since the start of the Pleistocene. The Pleistocene uplift rate is similar to other published uplift rates for this period from the northern parts of Northwest Nelson, suggesting that the whole of Northwest Nelson has experienced relative tectonic stability compared to other areas close to the Alpine Fault during this period. The mid-Pliocene uplift rate is possibly the first precisely constrained uplift rate in the area for this period, and suggests that there has been a progressive decrease in uplift rates from much higher rates in the late Miocene.</p>

Using low-frequency earthquakes to monitor slow tectonic deformation in the central Southern Alps, New Zealand

<p>This thesis involves the study of low-frequency earthquakes (LFEs) in the central Southern Alps. The Alpine Fault is the principal locus of deformation within the Australia–Pacific plate boundary in the South Island of New Zealand and it is late in its typical ∼300-year seismic cycle. Surveying the seismicity associated with slow deformation in the vicinity of the Alpine Fault may provide constraints on the stresses acting on a major transpressive margin prior to an anticipated great (≥M8) earthquake. Here, we use 8 years of data from the Southern Alps Microearthquake Borehole Array (SAMBA) (amongst those, 3 years of data were collected as part of this project) in order to: (1) generate an updated LFE catalogue using an improved matched-filter technique that incorporates phase-weighted stacking; (2) compute LFE focal mechanisms and invert them to infer the crustal stress field on the deep extent of the Alpine Fault; (3) expand the LFE catalogue to cover a wider range of spatial/temporal behaviours; (4) study LFE families’ characteristics to identify periods where slow slip might happen. We first use fourteen primary LFE templates in an iterative matched-filter and stacking routine, which allows the detection of similar signals and produces LFE families sharing common locations. We generate an 8-yr catalogue containing 10,000 LFEs that are combined for each of the 14 LFE families using phase-weighted stacking to produce signals with the highest possible signal-to-noise ratios. We find LFEs to occur almost continuously during the 8-yr study period and we highlight two types of LFE distributions: (1) discrete behaviour with an inter-event time exceeding 2 minutes; (2) burst-like behaviour with an inter-event time below 2 minutes. The discrete events are interpreted as small-scale frequent deformation on the deep extent of the Alpine Fault and the LFE bursts (corresponding in most cases to known episodes of tremor or large regional earthquakes) are interpreted as brief periods of increased slip activity indicative of slow slip. We compute improved non-linear earthquake locations using a 3D velocity model and find LFEs to occur below the seismogenic zone at depths of 17–42 km, on or near the hypothesised deep extent of the Alpine Fault. We then compute the first estimates of LFE focal mechanisms associated with continental faulting. Focal mechanisms, in conjunction with recurrence intervals, are consistent with quasi-continuous shear faulting on the deep extent of the Alpine Fault. We then generate a new catalogue that regroups hundreds of LFE families. This time 638 synthetic LFE waveforms are generated using a 3D grid and used as primary templates in a matched-filter routine. Of those, 529 templates yield enough detections during the first iteration of the matched-filter routine (≥ 500 detections over the 8-yr study period) and are kept for further analysis. We then use the best 25% of correlated events for each LFE family to generate linear stacks which create new LFE templates. From there, we run a second and final iteration of the matched-filter routine with the new LFE templates to obtain our final LFE catalogue. The remaining 529 templates detect between 150 and 1,671 events each totalling 300,996 detections over the 8-yr study period. Of those 529 LFEs, we manage to locate 378 families. Their depths range between 11 and 60 km and LFEs locate mainly in the southern part of the SAMBA network. We finally examine individual LFE family rates and occurrence patterns. They indicate that LFE sources seem to evolve from an episodic or ‘stepped’ to a continuous behaviour with depth. This transition may correspond to an evolution from a stick-slip to a stable-sliding slip regime. Hence, we propose that the distinctive features of LFE occurrence patterns reflect variations in the in-situ stress and frictional conditions at the individual LFE source locations on the Alpine Fault. Finally, we use this new extensive catalogue as a tool for in-depth analyses of the deep central Alpine Fault structure and its slip behaviour. We identify eight episodes of increased LFE activity between 2009 and 2017 and provide time windows for further investigations of tremor and slow slip. We also study the spatial and temporal behaviours of LFEs and find that LFEs with synchronous occurrence patterns tend to be clustered in space. We thus suggest that individual LFE sources form spatially coherent clusters that may represent localised asperities or elastic patches on the deep Alpine Fault interface. We infer that those clusters may have a similar rheological response to tectonic forcing or to potential slow slip events. Eventually, we discover slow (10km/day) and rapid (∼20-25km/h) migrations of LFEs along the Alpine Fault. The slow migration might be controlled by slow slip events themselves while the rapid velocities could be explained by the LFE sources’ intrinsic properties.</p>

A gravity analysis of the Alpine Fault and the DFDP-2 drill site, Whataroa valley, South Westland, South Island, New Zealand

<p>The second phase of drilling into the Alpine Fault (DFDP-2), was completed in the Whataroa River valley, a former glacial valley located in central Westland, South Island, New Zealand. The site is located next to a steep hillside on the hanging-wall, ~1 km southeast of the mapped surface trace of the Alpine Fault. Projection of the hillside suggests a sediment thickness of 100 ± 40 m at the drill site; however, the sediment thickness was approximately double pre-drill estimates. Additionally, the surface expression and shallow geometry of the Alpine Fault in the Whataroa River valley, is not well-defined due to post-glacial burial of the fault zone. This thesis describes a gravity study designed to better constrain sub-surface structure beneath the DFDP-2 drill site and across the Alpine Fault. During this study, 466 new high-precision gravity observations were collected (standard error = 0.015 mGal) and amalgamated with 134 existing gravity stations, yielding comprehensive coverage of gravity data across the study area. A high density of observations was achieved within pre-determined zones, in addition to regional measurements so that residual gravity anomaly maps could be produced. The maps reveal: a negative residual gravity anomaly interpreted as a dextrally-offset glacial channel at least 350-450 m deep; steep localised gravity gradients near the Alpine Fault and DFDP-2 drill site that are interpreted as faulted and/or eroded boundaries; and a negative gravity anomaly adjacent to the DFDP-2 drill site that is interpreted as the deepest point of an over-deepened glacial lake. Gravity models were used to estimate the bedrock-sediment interface geometry near the DFDP-2 drill site and Alpine Fault. Structural inversion of the density boundary next to the drill site suggests either a moderately-dipping reverse fault or sub-vertical erosional wall exists beneath the hillside. Additional constraints on physical properties from direct density measurements or seismic velocity determinations and direct constraints on sediment thickness and layer geometry from seismic surveys will in future allow this new high-precision gravity dataset to be modelled more effectively.</p>