Improved earthquake detection as a probe for active fault structures in New Zealand's central Southern Alps

<p>This thesis concerns the detection and analysis of micro-seismicity and low-frequency earthquakes in New Zealand's central Southern Alps. We make use of the 6.5 year continuous seismic dataset collected using the Southern Alps Microearthquake Borehole Array (SAMBA), alongside other temporary and permanent seismic deployments nearby. The small station spacing of this deployment allows for high resolution seismic studies near the Alpine Fault, a dextral-transpressive plate boundary fault between the Pacific and Australian plates. Using this dataset we have documented the rst evidence of low-frequency earthquakes on or near the deep extent of the Alpine Fault. By using a network based crosscorrelation detection method we have generated a 3 year catalogue of 14 low-frequency earthquake families. These low-frequency earthquake families locate close to other indicators and models of the deep extent of the Alpine Fault, and we interpret these low-frequency earthquakes to represent shear failure on or near the deep extent of the Alpine Fault. These low-frequency earthquakes highlight a near-continuous background rate of deformation, punctuated by short periods of tremor. We also observe higher rates of low-frequency earthquake generation after large regional earthquakes. The magnitudes of our low-frequency earthquakes range from Mʟ‒0.8‒1.8, and appear to follow an exponential distribution, implying that there might be a characteristic length-scale of failure. We have extended the catalogue of low-frequency earthquake templates using the full 6.5 year dataset and an objective synthetic detection methodology. We developed a new methodology for template detection after other methods failed, or were not feasible. This method employs simple synthetic template events, which, rather than trying to capture all of the complexities of the body waves we try to detect, approximate a simple waveform that does not correlate well with background noise. To undertake this method we have developed a multi-parallel Python package, which is highly portable (we have run this on computers ranging from dual-core, 8GB RAM laptops to a 393 node, 6349 CPU cluster computer) and distributed via an open-source model. This package was run through the 6.5 year dataset on the New Zealand E-Science PAN cluster to e fficiently (<48 hours clock-time) generate a spatially and temporally continuous catalogue of low-frequency earthquake templates. Using this method to detect an initial suite of over 25,000 detections grouped into 600 families we have generated 600 good quality, discrete stacked waveforms for use in further matched-filter detection routines. We have shown that, for templates with both P and S-phase picks, these templates locate near to our previously determined low-frequency earthquake family locations. Using a network matched- filter detection technique we have generated a catalogue of micro-seismicity in a region of low-seismicity near the Whataroa Valley, motivated by the Deep-Fault Drilling Project; Phase-2. We detected 300 earthquakes that include a selection of near-repeating earthquakes. We find that most detected events are not similar enough to be termed repeating. For 106 earthquakes we are able to generate high-precision magnitudes calculated by singular-value decomposition of similar waveforms. We find a high b-value of 1.44 for these earthquakes, with no earthquakes above Mʟ1.6. By generating high precision cross-correlation derived picks for individual detections and employing a double-difference location methodology we show that seismicity does not delineate a single structure; rather we interpret the detected seismicity as temporally-limited earthquake sequences on small asperities adjacent to the Alpine Fault. Focal mechanisms for the best recorded events show dominantly strike-slip mechanisms, with lesser reverse and normal components. During the drilling of the Deep-Fault Drilling Project: Phase-2 borehole we operated a real-time earthquake detection system around the drill-site. This was a multi-national effort involving 16 seismologists in three countries monitoring the automatic detections in shifts. During the 5 month real-time monitoring period we detected and located 493 earthquakes, none of which occurred within 3km of the drill-site, nor required changes to the drilling operations. We undertook this monitoring using open-source software, which employed a standard energy based detection scheme. This thesis has contributed four complementary earthquake catalogues, a further three years of continuous seismic data from the central Southern Alps, and an opensource Python package for detection and analysis of earthquakes using cross-correlation techniques. The characteristics of these catalogues highlight deformation modes on and near one of the world's major strike-slip plate boundaries, both at depth, and at the upper extent of the seismogenic zone.</p>

Improved earthquake detection as a probe for active fault structures in New Zealand's central Southern Alps

<p>This thesis concerns the detection and analysis of micro-seismicity and low-frequency earthquakes in New Zealand's central Southern Alps. We make use of the 6.5 year continuous seismic dataset collected using the Southern Alps Microearthquake Borehole Array (SAMBA), alongside other temporary and permanent seismic deployments nearby. The small station spacing of this deployment allows for high resolution seismic studies near the Alpine Fault, a dextral-transpressive plate boundary fault between the Pacific and Australian plates. Using this dataset we have documented the rst evidence of low-frequency earthquakes on or near the deep extent of the Alpine Fault. By using a network based crosscorrelation detection method we have generated a 3 year catalogue of 14 low-frequency earthquake families. These low-frequency earthquake families locate close to other indicators and models of the deep extent of the Alpine Fault, and we interpret these low-frequency earthquakes to represent shear failure on or near the deep extent of the Alpine Fault. These low-frequency earthquakes highlight a near-continuous background rate of deformation, punctuated by short periods of tremor. We also observe higher rates of low-frequency earthquake generation after large regional earthquakes. The magnitudes of our low-frequency earthquakes range from Mʟ‒0.8‒1.8, and appear to follow an exponential distribution, implying that there might be a characteristic length-scale of failure. We have extended the catalogue of low-frequency earthquake templates using the full 6.5 year dataset and an objective synthetic detection methodology. We developed a new methodology for template detection after other methods failed, or were not feasible. This method employs simple synthetic template events, which, rather than trying to capture all of the complexities of the body waves we try to detect, approximate a simple waveform that does not correlate well with background noise. To undertake this method we have developed a multi-parallel Python package, which is highly portable (we have run this on computers ranging from dual-core, 8GB RAM laptops to a 393 node, 6349 CPU cluster computer) and distributed via an open-source model. This package was run through the 6.5 year dataset on the New Zealand E-Science PAN cluster to e fficiently (<48 hours clock-time) generate a spatially and temporally continuous catalogue of low-frequency earthquake templates. Using this method to detect an initial suite of over 25,000 detections grouped into 600 families we have generated 600 good quality, discrete stacked waveforms for use in further matched-filter detection routines. We have shown that, for templates with both P and S-phase picks, these templates locate near to our previously determined low-frequency earthquake family locations. Using a network matched- filter detection technique we have generated a catalogue of micro-seismicity in a region of low-seismicity near the Whataroa Valley, motivated by the Deep-Fault Drilling Project; Phase-2. We detected 300 earthquakes that include a selection of near-repeating earthquakes. We find that most detected events are not similar enough to be termed repeating. For 106 earthquakes we are able to generate high-precision magnitudes calculated by singular-value decomposition of similar waveforms. We find a high b-value of 1.44 for these earthquakes, with no earthquakes above Mʟ1.6. By generating high precision cross-correlation derived picks for individual detections and employing a double-difference location methodology we show that seismicity does not delineate a single structure; rather we interpret the detected seismicity as temporally-limited earthquake sequences on small asperities adjacent to the Alpine Fault. Focal mechanisms for the best recorded events show dominantly strike-slip mechanisms, with lesser reverse and normal components. During the drilling of the Deep-Fault Drilling Project: Phase-2 borehole we operated a real-time earthquake detection system around the drill-site. This was a multi-national effort involving 16 seismologists in three countries monitoring the automatic detections in shifts. During the 5 month real-time monitoring period we detected and located 493 earthquakes, none of which occurred within 3km of the drill-site, nor required changes to the drilling operations. We undertook this monitoring using open-source software, which employed a standard energy based detection scheme. This thesis has contributed four complementary earthquake catalogues, a further three years of continuous seismic data from the central Southern Alps, and an opensource Python package for detection and analysis of earthquakes using cross-correlation techniques. The characteristics of these catalogues highlight deformation modes on and near one of the world's major strike-slip plate boundaries, both at depth, and at the upper extent of the seismogenic zone.</p>

Microseismicity Associated with Actively Exploited Geothermal Systems: Earthquake Detection and Probabilistic Location at Rotokawa and Statistical Seismic Network Design at Kawerau

<p>Geothermal power has progressively been recognised as an important energy resource due to the depletion of old power sources, and as a more environmentally aware population pushes for an increase in renewable energy sources. Monitoring microseismicity occurring in active geothermal systems is one means of both characterising the system’s fault architecture and characterising fluid/rock interaction in response to production. This study focuses on better understanding seismicity in two active geothermal fields, through the development and implementation of two different algorithms: an automated microearthquake detection algorithm using a matched filter technique (improving earthquake detection), and an optimal seismic network design algorithm (improving earthquake location). Both algorithms have been implemented in codes that are easily adaptable to other data sets. The first of these algorithms has been applied to five months of continuous seismic waveform data spanning a fluid injection operation in the Rotokawa geothermal field. The cross-correlation of 14 high-quality master events with the continuous seismic data yields 2461 newly detected earthquakes spanning the magnitude range M=-0.4 to M=2.6 with a mean magnitude of M=0.47. The earthquakes detected with each master event exhibit high waveform similarity over approximately three orders of magnitude, and appear to follow a Gutenberg-Richter power law with a catalogue completeness down to M~ 0. Hypocentres for these detected events computed using the probabilistic earthquake location algorithm NonLinLoc reveal the dominant locus of seismicity to lie between 1.0–2.5 km depth, a location consistent with that of the Rotokawa Andesite which forms the Rotokawa reservoir. Focal mechanism solutions for the master events are predominantly normal, with half displaying a large strike-slip component, and the stress parameters obtained for this suite of focal mechanisms imply a northeast–southwest oriented maximum horizontal stress: both of these results are consistent with the extensional regime of the TVZ. Seismicity occurring within a 300 m horizontal radius of the injection well’s feed-zones, and extending to 5 km depth, initially exhibits a correlation with injection flow rates with a ~ 2 day lag, and seismicity rates decrease ~ 10 weeks after injection. We surmise that seismicity within the injection region and close to the injection well is likely to be injection-induced, with one portion of the injectate returning to the production region, while the other either migrates southeastward out of the field or remains within the injection region; the origin of seismicity within the production region in relationship to production and injection processes is unclear. The second of these algorithms involves the derivation of a design criterion, which we apply to inform the expansion of the existing seismic monitoring programme at Kawerau geothermal field; we also apply an early version to the short-term/rapid-response network design following the M7.1 September 2010 Darfield earthquake. Unlike previous seismic network design algorithms, the new algorithm incorporates methods for the realistic representation of 3D velocity structures and attenuation models for both P and S travel times, a surface noise model, and the ability to apply complex weighting functions to the earthquake set. The results demonstrate the utility of this algorithm in even simplistic cases, and show how each new parameter incorporated into the design model affects the optimal network design obtained, identifying the need for accurate input data to provide optimal results.</p>

Microseismicity Associated with Actively Exploited Geothermal Systems: Earthquake Detection and Probabilistic Location at Rotokawa and Statistical Seismic Network Design at Kawerau

<p>Geothermal power has progressively been recognised as an important energy resource due to the depletion of old power sources, and as a more environmentally aware population pushes for an increase in renewable energy sources. Monitoring microseismicity occurring in active geothermal systems is one means of both characterising the system’s fault architecture and characterising fluid/rock interaction in response to production. This study focuses on better understanding seismicity in two active geothermal fields, through the development and implementation of two different algorithms: an automated microearthquake detection algorithm using a matched filter technique (improving earthquake detection), and an optimal seismic network design algorithm (improving earthquake location). Both algorithms have been implemented in codes that are easily adaptable to other data sets. The first of these algorithms has been applied to five months of continuous seismic waveform data spanning a fluid injection operation in the Rotokawa geothermal field. The cross-correlation of 14 high-quality master events with the continuous seismic data yields 2461 newly detected earthquakes spanning the magnitude range M=-0.4 to M=2.6 with a mean magnitude of M=0.47. The earthquakes detected with each master event exhibit high waveform similarity over approximately three orders of magnitude, and appear to follow a Gutenberg-Richter power law with a catalogue completeness down to M~ 0. Hypocentres for these detected events computed using the probabilistic earthquake location algorithm NonLinLoc reveal the dominant locus of seismicity to lie between 1.0–2.5 km depth, a location consistent with that of the Rotokawa Andesite which forms the Rotokawa reservoir. Focal mechanism solutions for the master events are predominantly normal, with half displaying a large strike-slip component, and the stress parameters obtained for this suite of focal mechanisms imply a northeast–southwest oriented maximum horizontal stress: both of these results are consistent with the extensional regime of the TVZ. Seismicity occurring within a 300 m horizontal radius of the injection well’s feed-zones, and extending to 5 km depth, initially exhibits a correlation with injection flow rates with a ~ 2 day lag, and seismicity rates decrease ~ 10 weeks after injection. We surmise that seismicity within the injection region and close to the injection well is likely to be injection-induced, with one portion of the injectate returning to the production region, while the other either migrates southeastward out of the field or remains within the injection region; the origin of seismicity within the production region in relationship to production and injection processes is unclear. The second of these algorithms involves the derivation of a design criterion, which we apply to inform the expansion of the existing seismic monitoring programme at Kawerau geothermal field; we also apply an early version to the short-term/rapid-response network design following the M7.1 September 2010 Darfield earthquake. Unlike previous seismic network design algorithms, the new algorithm incorporates methods for the realistic representation of 3D velocity structures and attenuation models for both P and S travel times, a surface noise model, and the ability to apply complex weighting functions to the earthquake set. The results demonstrate the utility of this algorithm in even simplistic cases, and show how each new parameter incorporated into the design model affects the optimal network design obtained, identifying the need for accurate input data to provide optimal results.</p>

Studi Literatur Penggunaan Sensor untuk Sistem Deteksi Gempa

As a country located among three major tectonic plate confluences, namely the Eurasian Plate, the Indo-Australian Plate, and the Pacific Plate, Indonesia is an earthquake-prone area. Information about the earthquake occurrences is distributed by the BMKG through social media, websites and television. However, it is undeniable that there are still people who do not have televisions or smartphones. Therefore, earthquake warning systems began to be widely developed. Furthermore, sensors for earthquake vibration detection have an important role in earthquake warning systems. So in this article, the author has conducted a literature study and analysis of the sensors commonly used in earthquake detection systems. This study aimed to present sensor recommendations to be used as earthquake sensors. The result of the literature study, that the author had done, is the selection of sensors is customized to the needs and budget of the project. A description of the sensor recommendations for each project has also been included in this article.

Deep learning for earthquake detection and location

Understanding the causality between the events leading to fault slip and the earthquake recording is important for seismic design and monitoring of underground structures, bridges and reinforced concrete buildings as well as climate mitigation projects like carbon sequestration and energy technologies like enhanced geothermal systems or oilfield wastewater disposal. The Federal Emergency Management Agency (FEMA) reported in 2017, that earthquake losses in the United States add up to about \$6.1 billion dollars annually. This number only addresses direct economic losses to buildings, and does not cover damage and losses to critical facilities, transportation and utility lifelines or indirect economic losses. A holistic framework to study earthquakes would incorporate seismic wave propagation and pressure perturbations, and have a dialogue with the deep learning framework for earthquake detection and location. In this document, we delve into the deep learning module.

An Earthquake Detection and Location Architecture for Continuous Seismograms: Phase Picking, Association, Location, and Matched Filter (PALM)

We developed an earthquake detection and location architecture for continuous seismograms that incorporates phase picking, phase association, location, and matched-filter techniques (PALM). The PALM architecture incorporates two modules: (1) PAL, the initial detection following picking, association, and location processes, and (2) match, expand, shift, and stack (MESS), a matched-filter detector that augments the template catalog. The effectiveness of PALM is demonstrated in building an early aftershock catalog for the 2019 Ridgecrest, California, earthquake. By comparing with Southern California Seismic Network arrival times, we show that the PAL picker combines the strengths of short-term average/long-term average and the kurtosis picker, realizing robust phase detection and precise picking. Our final MESS catalog is compared with two other matched-filter catalogs by Ross, Idini, et al. (2019) and Shelly (2020). We find that PALM directly recovers unbiased and detailed features in seismicity from continuous seismograms, which can be efficiently implemented to scan continuous waveforms without the need for visual inspection.