UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Slip behaviour of the Queen Charlotte plate boundary before and after the 2012, Mw 7.8 Haida Gwaii earthquake… Hayward, Timothy W. 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_may_hayward_timothy.pdf [ 3.59MB ]
Metadata
JSON: 24-1.0343273.json
JSON-LD: 24-1.0343273-ld.json
RDF/XML (Pretty): 24-1.0343273-rdf.xml
RDF/JSON: 24-1.0343273-rdf.json
Turtle: 24-1.0343273-turtle.txt
N-Triples: 24-1.0343273-rdf-ntriples.txt
Original Record: 24-1.0343273-source.json
Full Text
24-1.0343273-fulltext.txt
Citation
24-1.0343273.ris

Full Text

Slip behaviour of the Queen Charlotte plate boundary beforeand after the 2012, MW 7.8 Haida Gwaii earthquake:evidence from repeating earthquakesbyTimothy W. HaywardB.Sc. (Honours Geophysics), University of Manitoba, 2014A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFMaster of ScienceinTHE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES(Geophysics)The University of British Columbia(Vancouver)March 2017c© Timothy W. Hayward, 2017AbstractAbstractThe Queen Charlotte plate boundary, near Haida Gwaii, B.C., includes the dextral, strike-slip, Queen Charlotte Fault (QCF) and the subduction interface between the downgoing Pacificand overriding North American plates. In this dissertation, we present a comprehensive repeat-ing earthquake catalogue that represents an effective slip meter for both faults in the area. Thecatalogue comprises 730 individual earthquakes (0.3≤MW ≤ 3.5) arranged into 224 repeatingearthquake families on the basis of waveform similarity. We employ and extend existing rela-tionships for repeating earthquake magnitudes and slips to provide cumulative slip histories forthe QCF and subduction interface in 6 adjacent zones within the study area between 52.3◦Nand 53.8◦N. We find evidence for creep on both faults; however, the creep rate is significantlyless than plate motion rates, which suggests partial locking of both faults. The QCF exhibitsthe highest degrees of locking south of 52.8◦N, which indicates that the seismic hazard for amajor strike-slip earthquake is highest in the southern part of the study area. The October 28,2012, MW 7.8 Haida Gwaii thrust earthquake occurred in our study area and had a significanteffect on the plate boundary. The QCF is observed to undergo accelerated, right-lateral slip for1-2 months following the earthquake. The subduction interface exhibits afterslip thrust motionthat persists for the duration of the study period (i.e., 3 years and 2 months after the HaidaGwaii earthquake). Afterslip is greatest on the periphery of the main rupture zone of the HaidaGwaii event.iiPrefacePrefaceThis dissertation represents original intellectual work of the author, Timothy W. Hay-ward, and his supervisor, Professor Michael Bostock. Much of the analysis within this disser-tation was made possible by software originally developed by Professor Michael Bostock andhis research group at the University of British Columbia.iiiTable of ContentsTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.1 Data Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2 Assembling the Repeating Earthquake Catalogue . . . . . . . . . . . . . . . . 52.3 Addressing Catalogue Completeness . . . . . . . . . . . . . . . . . . . . . . . 113 Repeating Earthquake Properties . . . . . . . . . . . . . . . . . . . . . . . . . . 133.1 Magnitude and Moment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133.1.1 Converting Local Magnitude to Moment Magnitude . . . . . . . . . . . 133.1.2 Moments from Singular Value Decomposition . . . . . . . . . . . . . . 143.2 Fault Slip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22ivTable of Contents3.3 Polarity Constraints on Focal Mechanisms . . . . . . . . . . . . . . . . . . . . 233.4 Combining Repeating Earthquake Families into Groups . . . . . . . . . . . . . 263.4.1 Grouping Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . 263.4.2 Final Group Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . 274 Slip History of the Queen Charlotte Plate Boundary . . . . . . . . . . . . . . . . 324.1 Calculating Lower and Upper Limits of Slip . . . . . . . . . . . . . . . . . . . 324.2 Cumulative Slip and Slip Rate Estimates . . . . . . . . . . . . . . . . . . . . . 354.2.1 Slip Rates of the Queen Charlotte Fault . . . . . . . . . . . . . . . . . 364.2.2 Slip Rates of the Subduction Interface . . . . . . . . . . . . . . . . . . 385 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425.1 Interseismic Slip of the Queen Charlotte Plate Boundary . . . . . . . . . . . . 425.1.1 Interseismic Slip: the Queen Charlotte Fault . . . . . . . . . . . . . . . 435.1.2 Interseismic Slip: the Subduction Interface . . . . . . . . . . . . . . . 445.2 Postseismic Slip of the Queen Charlotte Plate Boundary . . . . . . . . . . . . . 465.2.1 Postseismic Slip: the Queen Charlotte Fault . . . . . . . . . . . . . . . 465.2.2 Postseismic Slip: the Subduction Interface . . . . . . . . . . . . . . . . 486 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Appendix A Definition of all Repeating Earthquake Families . . . . . . . . . . . . . 57Appendix B Constraining Repeating Earthquake Magnitudes . . . . . . . . . . . . 90Appendix C Repeating Earthquake Groups . . . . . . . . . . . . . . . . . . . . . . 95vList of TablesList of TablesTable 1 Repeating earthquake families classified by focal mechanism assignmentand fault interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25Table 2 Explanation of the different scenarios considered to determine upper andlower estimates of slip for a given repeating earthquake group . . . . . . . . 35Table 3 Average slip rates for QCF groups during time periods prior to, and after,the Mw 7.8 Haida Gwaii earthquake on Oct. 28, 2012 . . . . . . . . . . . . 36Table 4 Average slip rates for subduction groups during time periods prior to, andafter, the Mw 7.8 Haida Gwaii earthquake on Oct. 28, 2012 . . . . . . . . . 39Table 5 Details of all repeating earthquake families. . . . . . . . . . . . . . . . . . . 57Table 6 Details of all repeating earthquake groups. . . . . . . . . . . . . . . . . . . 97viList of FiguresList of FiguresFigure 1 (a) Map of the study area and (b) Schematic illustrating the major elementsof the Queen Charlotte plate boundary. . . . . . . . . . . . . . . . . . . . . 2Figure 2 Example of the high waveform similarity observed between repeating earth-quakes in a common family. . . . . . . . . . . . . . . . . . . . . . . . . . 6Figure 3 Comparison of correlation coefficients measured in different frequency bands,used to set a baseline with previous studies. . . . . . . . . . . . . . . . . . 8Figure 4 Schematic representation of the UPGMA clustering algorithm employed todefine repeating earthquake families. . . . . . . . . . . . . . . . . . . . . . 10Figure 5 Map of all repeating earthquake families. . . . . . . . . . . . . . . . . . . 12Figure 6 (a) Relationship between ML and log10M0 and (b) relationship between MLand MW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Figure 7 Example of using Singular Value Decomposition to estimate the momentof repeating earthquakes. . . . . . . . . . . . . . . . . . . . . . . . . . . . 16Figure 8 Comparison of magnitude estimates before and after using the SingularValue Decomposition technique. . . . . . . . . . . . . . . . . . . . . . . . 20Figure 9 Final magnitude distribution of the repeating earthquake catalogue. . . . . . 21Figure 10 Examples of slip histories for steady-type and burst-type families. . . . . . 23Figure 11 Examples of how focal mechanisms are assigned to repeating earthquakeson the basis of P-wave polarity. . . . . . . . . . . . . . . . . . . . . . . . . 24viiList of FiguresFigure 12 A comparison of repeating earthquakes families from a common group.All families exhibit high waveform similarity, proximal hypocenters, andrelated slip histories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28Figure 13 (a,b) Locations of repeating earthquake groups and (c,d) Example cross-sections of repeating earthquake groups. . . . . . . . . . . . . . . . . . . . 29Figure 14 Average cumulative slip histories for all QCF-related repeating earthquakegroups. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33Figure 15 Average cumulative slip histories for all subduction-related repeating earth-quake groups. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34Figure 16 Testing the accuracy and robustness of two different absolute moment con-straints. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94Figure 17 Cross-sections of all repeating earthquake groups. . . . . . . . . . . . . . . 96viiiAcknowledgmentsAcknowledgmentsI would first like to acknowledge my supervisor, Professor Michael Bostock, for all hisguidance during my time at UBC. This work would not have been possible without his hardwork and help along the way.I’d also like to thank the members of my supervisory committee, Professor Eldad Haber,Professor Mark Jellinek, and Professor Emeritus Ron Clowes for their helpful comments through-out my Master’s degree. Additionally, I acknowledge my examination committee, ProfessorEmeritus Ron Clowes and Professor Felix Herrmann for their comments on my thesis anddefence.The seismogram data used for this project was acquired from the Earthquakes Canadadatabase. Specifically, I’d like to thank Xiuying Jin and Tim Coˆte´ for their help in retrievingthe data from the repository.Lastly, I would like to acknowledge the other members of the Laboratory for Earth-quake Seismology (LEQS) research group at UBC: Alex Plourde, Genevie`ve Savard, Lindsay(Yuling) Chuang, Takeshi Akuhara, Xuejun Han, and Yariv Hamiel. Thank you for your en-couragement, assistance, and fun times throughout my Master’s at UBC.ixCHAPTER 1. IntroductionChapter 1IntroductionThe Queen Charlotte plate boundary (QCPB) is located off the northwest coast of NorthAmerica, between the Cascadia subduction zone to the south and the Yakutat collisional zoneto the north. The QCPB has hosted Canada’s two largest recorded earthquakes: the 1949, Ms8.1 Queen Charlotte Islands earthquake [Bostwick, 1984; Rogers, 1986; Nishenko and Jacob,1990], and the 2012, MW 7.8 Haida Gwaii earthquake [Lay et al., 2013; Cassidy et al., 2014;Kao et al., 2015]. Farther to the north in Alaska, USA, the plate boundary also spawned theMW 7.5 Craig, Alaska earthquake in 2013 [Aderhold and Abercrombie, 2015; Holtkamp andRuppert, 2015], as well as other large earthquakes in 1958 and 1972 [Nishenko and Jacob,1990]. These earthquakes underline the considerable seismic hazard posed by the QCPB.The QCPB includes the Queen Charlotte Fault (QCF) separating the Pacific and NorthAmerican plates (Figure 1). The system is relatively simple north of Haida Gwaii, whererelative plate motion is almost purely transform and thus a major dextral strike-slip systemis present [e.g., Plafker et al., 1978]. However, in the region near Haida Gwaii, the relativeplate motions include a significant 15-20◦ oblique convergent component [e.g., Hyndman andHamilton, 1993] due to a gradual left-stepping of the QCF. Two possible end-members wereoriginally proposed to explain the mechanism through which this convergence is accommo-dated: (1) the Pacific plate is obliquely subducting beneath the North American plate [e.g.,1CHAPTER 1. IntroductionBritish !Columbia!Pacific !Plate!North !American !Plate!52 !mm/yr!(a)! (b)!Pacific !Plate! North !American !Plate!Sedimentary!Terrace! Queen Charlotte!Fault!Trench!Figure 1: (a) Map of the study area (red, dashed box) including the islands of Haida Gwaii.Also shown are the relative Pacific-North American plate motion [white arrow; Nykolaishen et al.,2015], the Queen Charlotte Fault (white line), background seismicity during the study period (graydots), the location of the 2012 Haida Gwaii Earthquake (lower hemisphere focal mechanism), andthe seismograph stations used in this study. Station DIB is marked by a white star. Bathymetry iscoloured from 3 km water depth (dark blue) to 0 (white). Inset map shows the location of HaidaGwaii (dashed rectangle) within the rest of British Columbia, Canada. (b) Schematic illustratingthe major elements of the Queen Charlotte plate boundary (after Yorath and Hyndman [1983]).Relative plate motions are shown as black arrows. The Pacific Plate obliquely subducts beneaththe North American Plate resulting in both right-lateral strike-slip and under-thrust motions.2CHAPTER 1. IntroductionHyndman and Ellis, 1981; Hyndman et al., 1982; Mackie et al., 1989; Smith et al., 2003;Bustin et al., 2007], or (2) the convergent component is accounted for by internal deformationand crustal shortening within the two plates [e.g., Dehler and Clowes, 1988; Rohr et al., 2000].The oblique subduction hypothesis is now widely accepted (Figure 1b) in large part due to the2012 Haida Gwaii earthquake [e.g., Hyndman, 2015]. This MW 7.8 earthquake, unlike otherlarge earthquakes along the QCPB, exhibited a thrust mechanism with a sense of motion that islargely margin-perpendicular (Figure 1; Lay et al. 2013; Cassidy et al. 2014). Kao et al. [2015]suggest that the sense of motion may have been closer to parallel with relative plate motions,but nonetheless attribute the event to oblique subduction of the Pacific plate beneath the NorthAmerican plate. The QCPB, therefore, exhibits slip partitioning, with strike-slip motions takenup along the QCF and convergent motions taken up on the subduction interface [Lay et al.,2013; Hyndman, 2015; Kao et al., 2015]. Although the 2012 earthquake launched a numberof geophysical studies of the QCPB, the region remains understudied relative to other majortranscurrent plate boundaries such as the San Andreas, Alpine, and Anatolian Faults. Our ob-jective in this study is to supply new constraints on the slip budget of the QCPB through theanalysis of repeating earthquakes. Previous studies have analyzed the microseismicity of theQCPB [e.g., Hyndman and Ellis, 1981; Be´rube´ et al., 1989], but the current work representsthe first study to utilize repeating earthquakes in the area.Repeating earthquakes refer to earthquakes with highly similar locations and rupturecharacteristics that demonstrate repeated slip of a common fault patch [e.g., Nadeau and John-son, 1998; Igarashi et al., 2003]. Repeating earthquakes have been observed in a variety oftectonic environments around the world, perhaps most notably along the San Andreas Faultby Nadeau et al. [1995], Nadeau and Johnson [1998], and others. Following the work ofNadeau and Johnson [1998], important empirical and theoretical relations between repeatingearthquakes and fault slip/moment have been derived and applied in numerous tectonic envi-ronments [e.g., Chen et al., 2007; Kato et al., 2016; Uchida et al., 2016]. In this study, our goalis to create a catalogue of repeating earthquakes along the QCPB and gain new insight into the3CHAPTER 1. Introductionslip dynamics and segmentation along the QCPB. Our first step is to cluster earthquakes intorepeating earthquake “families” and generate estimates of earthquake moments/slips. Then,following Matsubara et al. [2005] and Uchida et al. [2016], we combine families into proximalgroups and analyze the average slip of these groups through time. We consider slip along theQCF and subduction fault segments separately, and pay particular attention to changes in theslip patterns before and after the 2012 Haida Gwaii earthquake.4CHAPTER 2. DataChapter 2Data2.1 Data SelectionSeismogram data for this study were obtained from the Canadian National SeismographNetwork [Earthquakes Canada, 2016] as 24-hour waveform records in the period between Jan-uary 1, 2005 and December 31, 2015. The seismograph stations and study area are shown inFigure 1a. Six stations on the island of Haida Gwaii are employed for this study; however,due to lower signal-to-noise ratio (SNR) and intermittent coverage at the majority of stations,station DIB proves to be most reliable (data available for 99% of the study period) and is thefocus of much of the analysis. The study area (∼22,000 km2) is centered on DIB and includesall points within ∼100 km of DIB, due to the diminishing SNR levels of small earthquakesat greater distances. The hypocenter of the 2012 Haida Gwaii earthquake is located in thesouthern part of the study area (Figure 1a).2.2 Assembling the Repeating Earthquake CatalogueIn this study, we use the term repeating earthquake “family” to refer to a set of discreteearthquakes with highly similar location and focal mechanism, and that we infer to representrepeated slip of the same fault patch. Families are identified as sets of waveforms with highcorrelation coefficients across all events (Figure 2).5CHAPTER 2. DataTime (s)0 1 2 3 4 5 6 7 8 9 10130227  23:24:28 (M 2.5)130928  11:06:34 (M 1.9)140416  04:41:47 (M 1.7)140614  07:09:35 (M 1.4)141019  06:06:36 (M 2.1)151214  11:56:14 (M 1.6)Figure 2: Vertical component waveforms, bandpass filtered between 2-14 Hz, for family 186.Event times are given on left in YYMMDD HH:MM:SS format, moment magnitudes are given onright in parentheses, and the seismograms have been normalized by their maximum amplitudes toenable visual comparison. Both P waves (∼0.75 s) and S waves (∼7 s) display high waveformsimilarity.To create a catalogue of repeating earthquake families we start with the Geological Sur-vey of Canada earthquakes list of∼7300 earthquakes (M≥ 0) in the study area recorded duringthe 2005-2015 study period. Manually selected event windows start immediately prior to the Pwave arrival and end when the S wave coda envelope amplitude is 10% that of the maximumamplitude. This window is chosen to ensure that waveform correlations register primarily sig-nal with minimal contamination by noise. The correlation between every pair of earthquakewaveforms is then calculated to create a ∼[7300 × 7300] matrix of pairwise correlation coef-ficients (CC).Two important aspects of all repeating earthquake studies are the choice of CC thresholdand the clustering algorithm used to define families. Changing the threshold or the clustering6CHAPTER 2. Dataalgorithm can lead to different family definitions and different numbers of earthquakes iden-tified as repeats. In addressing the CC threshold, we considered a passband of 2-14 Hz forall our analyses and set the CC threshold at 0.90. The pass band is relatively conservative, asmany previous studies have employed a narrower passband of ∼1-8 Hz and a CC threshold of∼0.95 [e.g., Igarashi et al., 2003; Matsubara et al., 2005]. We choose a low corner of 2 Hzinstead of 1 Hz due to poorer SNR at frequencies near 1 Hz in the vicinity of station DIB. Thehigh corner of 14 Hz was chosen to compensate for diminished spatial resolution due to lim-ited station coverage. Because much of our analysis relies on the single (3-component) stationDIB, there are fewer constraints on waveform similarity than in previous studies of repeatingearthquakes where larger numbers of stations are available. The higher frequency waveformsignature provides a more unique “finger-print” for each family than would be achieved usinglower frequencies alone. To assess the effect of this modified frequency passband, we considerthe differences in CC when measured at different frequencies. In order to facilitate statisticalanalysis, we apply the Fisher Transformation [Fisher, 1921] to all CC values to yield the Fishercorrelation coefficient (FCC):FCC =12ln[1+CC1−CC]. (1)The FCC exists on the interval (-∞, ∞) and is more nearly normally distributed than the CC,which is bounded on the interval [-1,1] [Fisher, 1921]. To establish a baseline with previousstudies that employed larger numbers of stations, we compare the FCC measured in both the2-8 Hz and 2-14 Hz passbands. A more ideal comparison would use a 1-8 Hz passband (insteadof 2-8 Hz) as the benchmark, but unfortunately, due to poor SNR for frequencies near 1 Hz,this proves to be difficult. We compute pairwise FCCs measured at both 2-8 Hz and 2-14 Hz,and fit the data using a total least squares approach (Fig. 3a; Van Huffel and Vandewalle 1991):FCC2−14Hz = (0.8266)FCC2−8Hz−0.0103 . (2)7CHAPTER 2. Data FCC2-8 Hz1 1.5 2 2.5 FCC2-14 Hz11.522.5y = xy = 0.8266*x - 0.0103 FCC2-8 Hz -  FCC2-14 Hz-0.4 0 0.4 0.8Count0200400(b)!(a)!Figure 3: (a) Crossplot of all pairwise Fisher correlation coefficients (FCC) between events at 2-8Hz and 2-14 Hz. Only points with FCC2−8Hz > 1.4722 (CC2−8Hz > 0.90) are used to calculatethe total least squares best fit line (solid blue line). The 1-to-1 line (dashed red line) is shown forreference. (b) Histogram of the difference between FCC for all pairwise event comparisons in thetwo passbands of interest. The distribution is approximately normal with a mean of 0.4.The errors to the total least squares line are approximately normally distributed in the FCC do-main with a mean of 0 and a standard deviation of 0.13. Only points with FCC2−8Hz > 1.4722(CC2−8Hz > 0.90) are used to calculate the best-fitting line (2) because we are only interestedin the scaling for waveforms exhibiting high correlation. As expected, the FCC is lower inthe wider passband of 2-14 Hz than in the narrower 2-8 Hz passband. From (2), a value ofFCC2−8Hz = 1.8318 (CC2−8Hz = 0.95) corresponds to FCC2−14Hz = 1.5039 (CC2−14Hz =0.9058). In Figure 3b the difference between FCC2−8Hz and FCC2−14Hz is shown. The distri-bution is approximately normal with a mean of ∼0.4. A difference in FCC of 0.4 correspondsto a difference in CC of ∼0.06-0.03 (in the vicinity of CC=[0.90,0.95]). From this analysis,we establish that a threshold of CC2−14Hz = 0.90 corresponds approximately to the CC2−8Hzmeasure of 0.95 that is more typical of previous studies on repeating earthquakes. The CCthreshold of 0.90 is the same as that employed by Schmittbuhl et al. [2016], and similar to thatof Chen et al. [2008].We define the repeating earthquake families by implementing a clustering algorithm8CHAPTER 2. Databased on the Unweighted Pair Group Method with Arithmetic Mean (UPGMA; Romesburg2004). Many previous repeating earthquake studies have used “chain-like” methods, whereinif event A correlates highly with B, and B correlates highly with C, then it is assumed that A andC also belong together, regardless of their correlation [e.g., Igarashi et al., 2003; Rubinstein andEllsworth, 2010; Uchida and Matsuzawa, 2013]. However, in this study, possibly due to thelimited station availability, we find that chain-like algorithms lead to large families with eventsthat clearly do not belong in the same repeating earthquake family, and so a more sophisticatedclustering algorithm is needed.The UPGMA algorithm is an agglomerative, hierarchical process, and as such, eachevent begins as its own cluster, and clusters are subsequently combined until a desired CCthreshold has been reached. Consider a pairwise CC matrix with 5 events labelled A, B, C,D, and E as in Figure 4. The clustering process performs the following steps: (1) group thetwo events with the highest CC (A and D) into a new cluster (AD), (2) recalculate the CC ofthis new cluster with all other items (B,C, and E) as the average of the CCs with the cluster’spreviously separate members, (3) locate the new highest CC in the matrix (whether that bebetween clusters, individuals, or a combination of the two) and repeat steps 1 and 2 until thedesired CC threshold is reached. The algorithm is unweighted because whenever correlationsto a new cluster are calculated, each original CC value is weighted equally (i.e., correlationsmade earlier are weighted equally to those made later). For example, at the first iteration inFigure 4, the new CC values are recalculated simply by:CCAD↔B =CCA↔B+CCD↔B2, (3)where, e.g., CCA↔B is defined as the CC between events A and B. At the third iteration inFigure 4, the new CCs must be calculated more carefully to account for the different numberof events in the clusters being combined:CCADB↔EC =2 [CCAD↔EC]+1 [CCB↔EC]3. (4)9CHAPTER 2. DataA B C D E A 0.96 0.82 0.98 0.84 B 0.96 0.88 0.94 0.90 C 0.82 0.88 0.92 0.96 D 0.98 0.94 0.92 0.88 E 0.84 0.90 0.96 0.88 AD B C E AD 0.95* 0.87* 0.86* B 0.95* 0.88 0.90 C 0.87* 0.88 0.96 E 0.86* 0.90 0.96 AD EC B AD 0.865* 0.95 EC 0.865* 0.89* B 0.95 0.89* ADB EC ADB 0.873* EC 0.873* 1st Iteration 2nd Iteration 3rd Iteration 1 A D B E C 1.00 0.98 0.96 Correlation Coefficient 0.90 0.88 2 3 4 0.86 (b)!(a)!Figure 4: (a) Schematic representation of the UPGMA clustering algorithm employed to definethe repeating earthquake families. Events are named A, B,..., E and their pair-wise correlationcoefficients are given as matrix entries. The highest correlations in the current matrix are indicatedby a red circle and the coefficients that were re-calculated during the previous iteration are markedby an asterisk. Assuming a correlation threshold of 0.90, the algorithm would halt at iteration 3(due to the highest correlation being less than the threshold) and two earthquake families wouldbe defined (ADB and EC). (b) The same toy problem as in (a), but illustrated as a dendrogram.Iteration numbers are given in circles, with solid circles indicating steps that were completed anddashed circles indicating steps that were not completed due to the correlation threshold (dashedgreen line) being too low.This algorithm results in clusters in which the average CC of each member event withall other members in the family is greater than the desired CC threshold. We run the UPGMAalgorithm with a CC threshold of 0.90 on the [7300 × 7300] matrix of all pairwise CCs.This defines 224 families comprising at least 2 earthquakes each, for a total of 494 individual10CHAPTER 2. Datarepeating earthquakes that are thought to be associated with the QCPB. The location of eachfamily is defined as the average location of all repeating earthquakes within the family.2.3 Addressing Catalogue CompletenessAs previously mentioned, Haida Gwaii was not consistently instrumented during thestudy period of 2005-2015. As a consequence, many events, including repeating earthquakes,are likely missing from the Geological Survey of Canada’s earthquakes catalogue. To addressthis issue, we employ a matched filter to search for additional repeating events [Gibbons andRingdal, 2006]. Each repeating earthquake family is assigned a representative event that hasthe highest average correlation with other members of the family. This event is then used asa template to search the DIB data set for additional (non-catalogued) events with CC greaterthan 0.90. In doing so, we supplement the 224 repeating earthquake families with an additional236 events, bringing the total number of earthquakes in the repeating earthquake catalogue to730 (Appendix A). These 224 families, and associated 730 earthquakes, are also manuallyinspected to ensure that the repeating earthquake catalogue is robust. Note that in addition tothe 224 families associated directly with the QCPB, 92 other families were also identified in thestudy area by the UPGMA and matched filter processes. However, through visual inspectionof these 92 families’ locations and/or polarities inconsistent with expected focal mechanisms,they are inferred to occur on subsidiary structures and thus are rejected from further analysis.All families, both related and unrelated to the QCPB, are shown in Figure 5.11CHAPTER 2. DataFigure 5: Map of all repeating earthquake families defined in this study, both related (coloured cir-cles) and unrelated (gray squares) to the Queen Charlotte plate boundary. The circles are colouredby the number of events in the family. The red and black dashed line is the QCF, and bathymetryis coloured with the same scale as in Figure 1a.12CHAPTER 3. Repeating Earthquake PropertiesChapter 3Repeating Earthquake PropertiesThe repeating earthquake catalogue resulting from UPGMA clustering (Section 2.2) andmatched filter processing (Section 2.3) affords a comprehensive list of families and events;however, information on magnitudes and focal mechanisms is incomplete. In this Chapter, wecalculate moments (M0), moment magnitudes (MW ), slips, and focal mechanisms for each ofthe repeating earthquakes in the catalogue, and assemble the repeating earthquake families intogroups along the QCPB.3.1 Magnitude and Moment3.1.1 Converting Local Magnitude to Moment MagnitudeDue to the low magnitudes of most earthquakes in our catalogue, many were originallycalculated using the local magnitude (ML) scale from which we must estimate M0 and slip. Wefirst convert the ML values to MW using the relation of Shearer et al. [2006], which considersthe scaling between ML, MW , and M0. Moment magnitude is defined by [Hanks and Kanamori,1979]:MW =23[log10 M0−9.1] , (5)13CHAPTER 3. Repeating Earthquake Propertieswith M0 measured in N·m. However, in their study of California earthquakes, Shearer et al.[2006] demonstrate a linear relationship between ML and log10M0 with slope 0.96 over therange 1 ≤ ML ≤ 3 (Figure 6a), consistent with the expected slope of ∼1.0 for earthquakesexhibiting self-similarity [Shearer, 2009]. This observation for ML deviates from that for MW ,which exhibits a slope of 2/3 as per (5). The larger scaling factor for ML varies across studies,but is consistently greater than 2/3 [e.g., Bakun, 1984; Abercrombie, 1996; Ben-Zion and Zhu,2002]. This scaling only applies to ML values below a certain magnitude threshold, as onlymagnitudes of small earthquakes will be underestimated by the local magnitude calculation[Figure 6; Shearer et al., 2006]. We adopt the same ML = 3.0 threshold as Shearer et al. [2006],such that for ML ≥ 3.0 the ML and MW scales are considered equivalent. For ML < 3.0, ascaling factor of 0.96 is applied to calculate MW (Figure 6b):MW =[(23)( 10.96)]ML+0.917 : ML < 3.0ML : ML ≥ 3.0. (6)From these estimates of MW , it is straight-forward to calculate M0 from (5).3.1.2 Moments from Singular Value DecompositionThe MW and M0 values that we determine from (5, 6) are a good starting point; how-ever, they are not sufficiently accurate to achieve meaningful estimates of fault slip. Due to thelimited station coverage and inaccuracies involved in the conversion from ML to MW , a bettermethod to estimate earthquake moment is required. Moreover, the earthquakes identified viathe matched filter processing (Section 2.3) are without initial magnitude estimates. We followRubinstein and Ellsworth [2010] and apply Singular Value Decomposition (SVD) to both esti-mate the moments of the events detected by the matched filter and better constrain the momentsof the earthquakes with magnitude estimates from (6).The SVD method takes advantages of the waveform similarity within a family to pro-vide better estimates of relative earthquake moments than conventional magnitude estimation14CHAPTER 3. Repeating Earthquake PropertiesMw0 1 2 3 4 5ML012345 Mw  = M LSee Equation 41!2!3!4!5!-1! 0! 1! 2!M L!Relative log10M0!M L      (0.96) log10M0!8	  (a)! (b)!6!Figure 6: (a) Relationship between ML and log10M0 (after Shearer et al. [2006]). The best fit line(blue line) has a slope of 0.96, and is valid for ML < 3.0. Above this threshold (black, dashed line),the data do not follow the same linear relationship. The second population of points to the right,with higher log10M0, results from incorrect recording of network gain, and are not considered. (b)Relationship used in the present study to relate MW to ML (red line) based on (a). See text and (6).techniques. This technique is particularly favourable when dealing with small earthquakes(which exhibit approximately constant durations when recorded at frequencies significantly be-low their corner frequencies) and when station coverage is limited [Rubinstein and Ellsworth,2010], and thus is well-suited to our study. For each repeating earthquake family, consider amatrix, R, whose rows are composed of the aligned time series of all waveforms constitutingthat family, station, and component. Using SVD, R can be decomposed asR = USVT (7)where U and V are matrices with columns of input and output basis vectors, respectively, and Sis a diagonal matrix containing singular values (si) along the main diagonal. The importance ofindividual output basis vectors (columns ofV, vi) in describingR is expressed by the associatedsi, with higher values of si indicating a more important output basis vector. The input basisvectors (columns of U, ui) describe the weights applied to each scaled output vector (sivi) to15CHAPTER 3. Repeating Earthquake Propertiesreconstruct the rows of the original data matrix, R.In the case of repeating earthquakes, R is composed of highly similar rows, so mostof the data can be explained by the first output basis vector (i.e., s1  s2,...,n ; Figure 7).For every earthquake we calculate the `2-norm of the residual resulting from subtracting theweighted first basis output vector from the original waveform (Figure 7). The average residual,across all earthquakes of all families, has just 20.9% of the energy of the original waveform,which compares well with the value of 18.6% quoted by Rubinstein and Ellsworth [2010]. Inconsidering only the first basis output vector, we assume that all other output basis vectorsTime (s)0 1 2 3 4 5 6 7 8 9 10131004  01:52:03 (1.00  M 2.9) [0.04]140820  18:30:52 (0.11  M 2.3) [0.13]141210  22:10:44 (0.07  M 2.1) [0.33]150121  00:04:11 (0.34  M 2.6) [0.28]151211  04:18:19 (0.14  M 2.3) [0.29]Figure 7: Vertical component waveforms recorded at station DIB (dashed black line) and SVDreconstruction using the weighted first basis output vector (solid red line) for family 183. Theoriginal and reconstructed waveforms show nearly perfect overlap and the residual waveforms(i.e., original minus reconstructed; solid blue line) are small. Dates of each event are given on theleft in YYMMDD HH:MM:SS format. Moment weights relative to the first event are shown inparentheses with the final magnitude estimate (see text for full explanation). The `2-norm of theresidual relative to the `2-norm of the original seismogram is shown in square brackets on the right.Plotted amplitudes within one event (original, reconstruction, and residual) are true, but scales varybetween the five different events.16CHAPTER 3. Repeating Earthquake Propertiesrepresent noise. It is likely that in addition to noise, subtle changes in source properties and/ormaterial properties along the earthquake ray path constitute the lower level output basis vectors.However, following Rubinstein and Ellsworth [2010], we ignore these other vectors due to theirrelatively small amplitudes and because the first output basis vector effectively constrains theamplitude of the repeating earthquakes. The first column of U (ui1) thus contains the relativemoment weights for all earthquakes in a given repeating earthquake family. This allows for allpairwise comparisons of relative moments to be made within a family, such asM0A−wABM0B = 0 , (8)where M0A and M0B are the absolute moments of event A and event B, respectively, and wAB isthe moment ratio of A relative to B. The SVD is performed separately on the three componentwaveforms from station DIB, and the ratio wAB (8) is taken as the average result across the threecomponents. The way in which we utilize these pairwise M0 comparisons differs from that ofRubinstein and Ellsworth [2010], as follows. We first consider all pairwise M0 comparisonsfor a given family and assemble a matrix equation. For a family with n earthquakes there aren(n−1)2 pairwise comparisons, and thusn(n−1)2 rows in the system. For example, for a familywith three earthquakes (A,B, and C), the system can be written as1 −wAB 01 0 −wAC0 1 −wCBM0AM0BM0C=000 . (9)However, in (9), the rank of the left-hand matrix is 2, whereas the number of rows is 3 (gen-erally the rank is (n− 1) and the number of rows is n(n−1)2 ). The redundant equations in thissystem are perfectly consistent because they come from the weighting of a single output basisvector, and thus we may choose the first (n− 1) equations (ordered systematically as in (9))and discard the remaining equations. This corresponds to retaining all comparisons to the first17CHAPTER 3. Repeating Earthquake Propertiesearthquake, which results in a system of the form 1 −wAB 01 0 −wACM0AM0BM0C= 00 . (10)In general, for a family with n earthquakes, the system is written as1 −w12 0 0 · · · 01 0 −w13 0 · · · 0............ . . ....1 0 0 0 · · · −w1nM01M02M03...M0n=00...0. (11)Note that the left-hand matrix in (11) has size [(n−1)×n] and rank (n−1). The equations inthis system enforce precise relative moments between all members of a repeating earthquakefamily. To constrain absolute moments, we utilize the original M0 estimates derived fromSection 3.1.1. There are numerous options regarding how to employ the original magnitudeestimates. A single earthquake (e.g., the one with largest magnitude) could be used, in combi-nation with weights wi j (11), to define M0 for all other earthquakes within the family. However,to avoid relying on a single measurement and instead take advantage of all available informa-tion, we prefer to use a combination of all catalogued absolute moment estimates, which wedenote by Mcat0 . One simple way to accomplish this is to constrain the sum of M0 in the solutionto be equal to the sum of Mcat0 , but this approach is susceptible to errors in magnitude estimatesfor the largest events. We suggest a more intricate combination of original recorded moments,18CHAPTER 3. Repeating Earthquake Propertieswhich is added as an additional row to (11), producing the system1 −w12 0 0 · · · 01 0 −w13 0 · · · 0............ . . ....1 0 0 0 · · · −w1n∏j 6=1w1 j ∏j 6=2w1 j ∏j 6=3w1 j ∏j 6=4w1 j · · · ∏j 6=nw1 jM01M02M03...M0n=00...0∑i(Mcat0i ∏j 6=iw1 j),(12)or, more succinctly,Gm = d . (13)We choose the Mcat0 constraint in (12) because it is orthogonal to the rest of the system and sobetter honours the scaling constraints in (11), which are anticipated to be more accurate androbust than constraints from absolute moments. A full derivation and justification of the finalequation in (12) is available in Appendix B. By representing the system in this way, we satisfyall relative M0 information from the SVD analysis in the first (n− 1) rows, and constrain allMcat0 information with a single equation (i.e., the nth row). Recall that in the present study,matched filter detections have no associated Mcat0 values. Consequently, the nth row in G maybe missing entries, such that ∏j 6=iw1 j is replaced by 0. Importantly, even in the case of missingMcat0 values, G has size [n×n] and rank n. These properties of G allow us to solve (13) for thetrue moments (m) directly:m = G−1d . (14)In the present study, the condition number of G for all repeating earthquake families is suffi-ciently low to use (14) as an accurate and stable solution.As mentioned previously, this method produces M0 estimates for earthquakes that hadno prior measure of M0, and also improves the accuracy of M0 estimates for all repeatingearthquakes in our catalogue [Rubinstein and Ellsworth, 2010]. Predictably, larger earthquakes19CHAPTER 3. Repeating Earthquake Propertieshave greater control on the SVD than smaller earthquakes (Figure 8). We find that, followingthe SVD and inversion processes, earthquakes with original MW > 2.5 rarely change by morethan 0.2 magnitude units, whereas the magnitudes of smaller earthquakes can change by upto ∼0.6 (Figure 8). We also note that there appears to be a small gross bias by the procedureto decrease the magnitudes compared to those from the original recorded magnitudes. Thisis observed as a “fat-tail” on the negative side of the distribution in Figure 8, and may resultfrom significant power of the original waveforms being present in the lower-level basis vectorsthat are discarded in the SVD analysis. Rubinstein and Ellsworth [2010] report that this effectis weakly magnitude-dependent, whereby smaller earthquakes are more affected than largerearthquakes. This leads to a weak magnitude-dependent bias: smaller earthquakes are morelikely to be underestimated by this procedure, and larger earthquakes are more likely to beoverestimated. However, our results agree with those of Rubinstein and Ellsworth [2010] inthat this magnitude-dependent bias is very weak and can be safely ignored.Mw, orig.1 1.5 2 2.5 3 3.5Mw, final -  Mw, orig.-0.6-0.4-0.200.20.40.6Count10 0 10 1 10 2Mw, orig.1 1.5 2 2.5 3 3.5Mw, final -  Mw, orig.-0.6-0.4-0.200.20.40.6Count10 0 10 1 10 2Figure 8: Comparison of MW estimates before and after SVD analysis and inversion for earth-quake moment. Only events that had initial M estimates are shown (494 total events). Mw,orig.values are those after correcting from ML (see Section 3.1.1). The cross-plot shows that changes tosmaller events are more extreme than those to larger events because larger magnitude events tendto control the SVD analysis. The histogram to the right (red axes), which uses the same y-axis asthe crossplot, shows that there is a slight bias in the analysis to smaller magnitudes.20CHAPTER 3. Repeating Earthquake PropertiesThe final magnitude distribution of the repeating earthquakes is in the range 0.3≤MW ≤3.5, and the majority of the events fall near MW 2.0 (Figure 9). As expected, matched filtering(Section 2.3) identifies a higher proportion of low magnitude earthquakes than the originalclustering process (Figure 9), because these low magnitude events are more likely to be missedby standard detection routines. However, note that the matched filtering does contribute asubstantial number of higher magnitude events (Figure 9b), and thus contributes significantlyto catalogue completeness and our efforts to accurately quantify the slip budgets of the QCPB.The original search for repeating earthquakes was performed over the magnitude range 0.0 ≤M ≤ 7.0, but no large earthquakes in the study area satisfied the waveform similarity criteriadescribed above to be included in repeating earthquake families.0 1 2 3 4CatalogueEvents0751500 1 2 3 4Matched FilterEvents02550Mw0 1 2 3 4AllEvents0100200(b)!(a)!(c)!Figure 9: Magnitude histograms for repeating earthquakes (a) from the original GSC earthquakelist, (b) added by the matched filter process, and (c) from the combination of the two (i.e., theentire repeating earthquake catalogue of the present study). The matched filter identifies a higherproportion of smaller magnitude events, but still contributes a significant number of larger earth-quakes to the repeating earthquake catalogue. Note the different event scales (y-axes) for each ofthe histograms.21CHAPTER 3. Repeating Earthquake Properties3.2 Fault SlipThe major goal of this study is to gain an understanding of the slip dynamics of theQCPB. Accordingly, the slip for each individual repeating earthquake must be estimated. Todo this, we employ a relation relating M0 to fault slip (d) for repeating earthquakes from Nadeauand Johnson [1998]:log10 d = (0.17) log10 M0−2.36 , (15)where d is measured in cm, and M0 is measured in dyne·cm. This relationship was originallydeveloped by Nadeau and Johnson [1998] for the San Andreas Fault, but has been later shownto apply to repeating earthquakes in numerous tectonic environments, including the northeast-ern Japan subduction zone [Igarashi et al., 2003] and the Chihshang Fault in Taiwan [Chenet al., 2008]. The relationship between moment and slip (15) has subsequently been appliedto repeating earthquakes on both subduction and strike-slip faults world-wide [e.g., Matsubaraet al., 2005; Kato et al., 2016; Uchida et al., 2016]. Assuming it is generally valid, we employ(15) for all repeating earthquakes in our study, including both those on the subduction interfaceand those on the strike-slip QCF.The slip characteristics of all repeating earthquake families in the present study may bedivided into two main groups: “burst”-type, and “steady”-type families. Burst-type familieshave all their events clustered closely in time and therefore do not generally persist for a signifi-cant portion of the study period, however some families exhibit multiple burst episodes (Figure10). Conversely, steady-type families tend to have more regularly spaced repeating earthquakesover a longer time period (Figure 10). A similar phenomenon has been observed by previousstudies in the northeastern Japan subduction zone [Igarashi et al., 2003] and North AnatolianFault [Schmittbuhl et al., 2016]. Burst-type events are expected to occur during nucleation andafter-shock sequences of large earthquakes, whereas the steady-type families are inferred to re-sult from constant tectonic loading on the fault [Igarashi et al., 2003; Schmittbuhl et al., 2016].Although the slip patterns, and the underlying stress-forcing regimes, may be very different for22CHAPTER 3. Repeating Earthquake Properties2013 2014 2015 2016Cumulative Slip[m]00.20.40.6 Family 186Time2013 2014 2015 2016Cumulative Slip[m]00.20.40.60.81Family 223(b)!(a)!Figure 10: Examples of slip histories for (a) a steady-type repeating earthquake family whereisolated earthquakes occur in regular recurrence times, and (b) a burst-type repeating earthquakefamily where earthquakes are clustered in time. The burst-type family 223 consists of two burstsof activity in mid-2014 and late-2015. Individual earthquakes are shown by pink circles.these two family types, both types contribute significantly to the overall slip budget of the plateboundary and are thus both important in the analysis.3.3 Polarity Constraints on Focal MechanismsDue to limited station coverage and low magnitudes of the repeating earthquakes inthis study, we are unable to solve for reliable focal mechanisms. However, it is important todistinguish, where possible, between repeating earthquakes on the subduction and strike-slipportions of the QCPB (Figure 1b). To accomplish this, we manually inspect the P-wave po-larities of all repeating earthquake families and determine if they are consistent with dextralstrike-slip motion along the QCF, thrust motion along the subduction interface, or neither (Fig-ure 11). Although repeating earthquakes can be observed along smaller faults, most repeatingearthquakes (in particular, steady-type families) occur on major plate boundaries because a23CHAPTER 3. Repeating Earthquake Propertiesnear-constant tectonic force can generate repeated slips of the same fault patch [e.g., Igarashiet al., 2003; Chen et al., 2007]. With this in mind, repeating earthquake families that are classi-fied as subduction-type in this study are those with polarities that are consistent with subductionmotion and inconsistent with dextral strike-slip motion for all available event/station record-ings. A similar criterion is applied for families that are classified as QCF-type. Examples ofthis classification process are illustrated in Figure 11.ZEN05/05/24  13:53:10  [002]MOBCZEN12/10/24  01:54:08  [049]MOBCTime (s)0 1 2 3ZBNBTime (s)0 1 2 3ZBNB(a)! (b)!05/05/24  !13:53:10  ![002]!12/10/24  !01:54:08  ![049]!MOBC!BNB!Figure 11: Example of how focal mechanisms are assigned to repeating earthquake families inthe present study. (a) Two nearby events and their assumed upper hemisphere focal mechanismsare shown relative to stations MOBC and BNB (yellow triangles) and the QCF (black dashedline). Dates are given in YY/MM/DD HH:MM:SS format with family ID given in brackets. (b)Observed P-waves for the two events recorded at MOBC (top panel) and BNB (bottom panel).Both subduction thrust and dextral strike-slip earthquakes from the given area exhibit negativepolarity at MOBC. However, at BNB the strike-slip event (positive polarity) can be distinguishedfrom the subduction event (negative polarity). See text for full explanation of process.The hypocenter of a family is also considered when classifying it as subduction- or QCF-type. However, the locations are generally used as supporting, rather than defining, evidencebecause the hypocenters being used are those from the original GSC earthquakes catalogue, andas such may be subject to appreciable errors, especially for smaller earthquakes. For example,many earthquakes in the repeating earthquake catalogue do not possess depth estimates, and sousing depth of the earthquake to distinguish between the QCF and subduction interface is not24CHAPTER 3. Repeating Earthquake Propertiespossible. In general, if multiple earthquakes consistently show that a family’s location cannotbe attributed to either the QCF or subduction interfaces then the hypocenter information isconsidered, but in many cases it is the waveform and polarity that is used as the principal lineof evidence in defining the focal mechanism of a family.In some cases, due to station geometry, it is not possible to distinguish between polar-ities for strike-slip and thrust motions. In these cases, the family is classified as “unclear”and is considered in the slip calculations for both the subduction and QCF interfaces to placelower and upper limits on slip estimates. The distribution of families based on focal mech-anism and associated fault interface is shown in Table 1. As mentioned in Section 2.3, theprocess of inspecting polarities reveals 92 additional families that, when considering polaritiesand hypocenter locations, are classified neither as subduction nor QCF earthquakes, and thusare rejected from further analysis (Figure 5). The majority of these additional families arecomposed of only two events and do not contribute significantly to the overall moment budgetof the repeating earthquakes in this study.Table 1: Repeating earthquake families classified by focal mechanism assignment and fault in-terface. Families in the“unclear” category are those whose polarities cannot be unambiguouslyattributed to either thrust or strike-slip motions. These unclear families are considered in the slipcalculations for both subduction and QCF interfaces to create upper and lower estimates of slip.Type Number ofFamiliesTotal Number ofRepeating EarthquakesTotal CumulativeMoment Release [N·m]Subduction Thrust 87 320 7.9043×1014QCF Strike-Slip 76 240 1.2020×1015Unclear 61 170 2.2586×101425CHAPTER 3. Repeating Earthquake Properties3.4 Combining Repeating Earthquake Families into Groups3.4.1 Grouping MethodologyPrevious studies have demonstrated that repeating earthquakes are effective slip metersfor both the seismic fault patch that ruptures during the earthquake and surrounding areas thatslip aseismically. This theory is supported by both observational field studies [e.g., Nadeauand Johnson, 1998; Chen et al., 2007; Kato et al., 2016; Schmittbuhl et al., 2016; Uchida et al.,2016] as well as modelling and lab studies [e.g., Beeler et al., 2001; Chen and Lapusta, 2009].The measurements of slip resulting from repeating earthquakes differ from those obtained byGPS techniques in that the repeating earthquake method samples slip on the fault surface di-rectly, whereas GPS measures slip at the earth’s surface and requires an inverse problem toresolve slip of the fault at depth. With this in mind, our main objective in the present studyis to determine the slip characteristics of the entire QCPB using the point measurements ofslip that we obtain from each repeating earthquake family. We follow the work of Matsub-ara et al. [2005] and Uchida et al. [2016] by arranging the repeating earthquake families intogroups that are related both spatially and by focal mechanism assignment. That is, we divideall repeating earthquake families that represent dextral strike-slip motion along the QCF intospatially-related groups along the fault. The same is done for the subduction interface. Wethen average the slip histories for all repeating earthquake families in a given group to recoverthe average slip on the fault, in that area, through time. In theory all repeating earthquakefamilies in a given group should have nearly identical slip histories because they all representslip of a similar area of the fault. But, in practice we find that although families in a commongroup often have related slip histories, there are also discrepancies. The grouping and averag-ing process allows for the removal of inconsistencies and spurious signals, which is requiredto investigate the underlying slip signal of the fault interface. To clarify, in the present study, arepeating earthquake “family” is composed of individual earthquakes that represent slip of thesame fault patch, whereas a “group” refers to a collection of families that have the same focal26CHAPTER 3. Repeating Earthquake Propertiesmechanism and are located near each other.To define groups, we search for families that have similar waveforms and thus similar fo-cal mechanisms. The same UPGMA clustering used to define the repeating earthquake families(Section 2.2) is applied to the first basis output vectors of each family from the SVD analysis(Section 3.1.2). The output basis vectors represent the common elements of each earthquakein a family, and are good approximations of the family’s average waveform. Families resultingfrom the same sense of slip in nearby areas yield similar P and S waves, but different S minusP lag times (Figure 12). Because S waves generally exhibit larger amplitudes than P waves, theCC value, and thus the grouping process, is controlled by the similarity of the S waves. Figure12 illustrates an example of many families in an area exhibiting similar waveforms (and thusrepresenting similar focal mechanisms), despite the fact that they originate from different, butnearby, locations on the fault interface. The example shown in Figure 12 also illustrates thatthe families in a group (group 5S in Figure 12) typically exhibit common temporal behaviorwhereby one family is more likely to slip when a nearby family slips. This temporal similarityis not ubiquitous across all families of all groups, but is observed in many cases.The UPGMA algorithm is used to define initial groups of subduction- and QCF-typegroups along the QCPB, and then manual inspection of all families in associated areas is per-formed to assign all families to groups. Recall from Section 3.3 that in some cases a familycannot be unambiguously assigned to correspond to either dextral strike-slip motion along theQCF or thrust motion on the subduction interface. In these ambiguous cases, the family isclassified as having an unclear focal mechanism (Table 1) and is considered in both the nearestsubduction- and QCF-type groups, which allows us to place upper and lower limits on groupslip.3.4.2 Final Group DefinitionsThe final group definitions are shown in Figure 13. Because groups are determined basedon observed clustering of the families, there is variation between groups in regards to map area27CHAPTER 3. Repeating Earthquake PropertiesNorth069200183186218EastTime (s)0 0.25 0.5 0.75 1VerticalNorth069200183186218EastTime (s)6.4 6.65 6.9 7.15 7.4Vertical2 kmQCFTime (yr)2006 2016Cumulative Slip (cm)050100150200250069200183186218North069200183186218EastTime (s)0 0.25 0.5 0.75 1VerticalNorth069200183186218EastTime (s)6.4 6.65 6.9 7.15 7.4Vertical2 kmQCFTime (yr)2006 2016Cumulative Slip (cm)050100150200250069200183186218NorthEasti  (s). . .VerticalNorthEasti  (s). . . . .Vertical kQCFi  (yr)Cumulative Slip (cm)54°N!53°N!133°W! 134°W!DIB!(50 km)!Mw 7.8!(a)! (b)! (c)!(d)!North069200183186218EastTime (s)0 0.25 0.5 0.75 1VerticalNorth069200183186218EastTime (s)6.4 6.65 6.9 7.15 7.4Vertical2 kmQCFTime (yr)2006 2016Cumulative Slip (cm)050100150200250069200183186218Figure 12: A selection of families from group 5S. (a) P waves and (b) S waves recorded at stationDIB. The waveforms are the first output basis vector for the family from the SVD analysis (Section3.1.2) and represent family average waveforms. P waves have been amplified relative to S waves tofacilitate visual comparison. Family labels are given on north component (top panel) and orderingis consistent for other components (middle and bottom panels). Both P waves and S waves showhigh waveform similarity between families, despite the varying P to S lag times (representingdifferent hypocenter locations). (c) Locations of the families. Map location is shown in inset map(white dashed box). (d) Slip history for the families, showing a temporal connection betweenfamilies. Individual repeating earthquakes are shown by squares, and the time of the 2012 MW 7.8Haida Gwaii earthquake is given as a yellow line.28CHAPTER 3. Repeating Earthquake PropertiesDistance Along Profile (km)0 5 10 15 20Depth (km)20100 Distance Along Profile (km)0 5 10 15Depth (km)25201510Distance Along Profile (km)0 5 10 15 20Depth (km)20100 Distance Along Profile (km)0 5 10 15Depth (km)25201510SW! NE!NW! SE!(b)!(a)![5S]![6Q]!5S!4S!3S!2S!1S!6S!5Q!4Q!3Q!2Q!1Q!6Q!(c)! (d)!Figure 13: Groupings of (a) QCF strike-slip type and (b) subduction thrust type repeating earth-quake families. Individual families (coloured circles) are coloured by group. Families with am-biguous polarities (gray squares) are shown in both (a) and (b). Also shown in (a) and (b) is thehypocenter (lower-hemisphere focal mechanism), rupture zone (dashed, purple line), and area ofmaximum slip (filled, purple polygon) for the 2012 Haida Gwaii earthquake [Hyndman, 2015;Nykolaishen et al., 2015]. (c) Example of a cross-section of a QCF group (6Q), with profile takenalong the strike of the QCF. (d) Example of a cross-section of a subduction group (5S), with profiletaken perpendicular to the strike of the QCF (i.e., parallel to plate subduction). The location of thesurface trace of the QCF is shown with a blue, dashed line. In cross-sections, filled circles arefamily average locations, empty circles are individual repeating earthquake locations, and there isa horizontal exaggeration of 2. Note that some families do not have depth estimates, and are absentfrom the cross-sections. A complete set of cross-sections is provided in Appendix C.29CHAPTER 3. Repeating Earthquake Propertiesand number of families. Groups represent map areas ranging between 600-1500 km2 (Figure13a,b), consist of 9-32 families, and 24-112 individual repeating earthquakes (excluding “un-clear” groups). For a full definition of each group, see Appendix C. Groups are labelled as1Q, 1S, 1U, 2Q, 2S, 2U, etc., where the leading number corresponds to the area of the QCPBand the trailing letter represents the strike-slip QCF (“Q”), the subduction interface (“S”), orunclear (“U”) slip-types. Note that groups 1S and 3Q are empty because no families in the cor-responding areas exhibit waveforms that are unambiguously due to subduction or strike-slipmotions, respectively. However, the unclear families in these areas (i.e., those in groups 1Uand 3U) may be due to either sense of slip. Consequently, although there are no well-definedgroups 1S and 3Q, upper- and lower-limit slip estimates for these areas are still provided basedon the unclear families in the corresponding areas.Although it is difficult to rely on precise hypocenter information of the repeating earth-quakes, due in large part to their small magnitudes and limited station coverage, it is nonethe-less insightful to present depth cross-sections of all repeating earthquake groups. Cross-sectionsfor groups 6Q and 5S are shown in Figure 13c,d, and the remainder are provided in AppendixC. Many families are missing catalogue depths and so are omitted from the cross-section anal-ysis. For groups classified as QCF-type, we create profiles parallel to the QCF surface trace. Inthese groups, families appear uniformly scattered in depth in the range ∼5-20 km (Figure 13c;Appendix C). The distribution of these families’ hypocenters is consistent with the inferencethat they represent strike-slip motion on the vertical QCF. For groups classified as subduction-type, the cross-section profiles are created perpendicular to the QCF so as to be parallel withthe dip of the subducting Pacific plate. The locations of families in groups 3S, 4S, 5S, and 6Sdefine a dipping planar feature (Figure 13d; Appendix C), supporting the inference that theserepeating earthquake families occur on the plate interface between the subducting Pacific plateand the overriding North American plate. The dip angles measured from groups 3S, 4S, 5S,and 6S are ∼32◦, 32◦, 31◦, and 24◦ respectively. These dip angles for the subducting plateare consistent with the value of 28 ± 5◦ estimated from receiver function analysis by Bustin30CHAPTER 3. Repeating Earthquake Propertieset al. [2007]. Other studies employing receiver functions and thermal modelling have reporteda shallower dip angle of∼20◦ [Smith et al., 2003; Gosselin et al., 2015]. Note that hypocentersfor group 2S show too much scatter to permit an accurate measure of dip, and group 1S has nofamilies to analyze (Appendix C).31CHAPTER 4. Slip History of the Queen Charlotte Plate BoundaryChapter 4Slip History of the Queen Charlotte PlateBoundaryFollowing the definitions of repeating earthquake groups (Section 3.4; Figure 13), weare able to investigate the slip histories of both the QCF and subduction interface, which to-gether constitute the QCPB. For a given group, we consider the cumulative slip histories of allassociated families (e.g., Figures 10 and 12d), and compute the mean to provide the averageslip history for the group (Figures 14 and 15).4.1 Calculating Lower and Upper Limits of SlipRecall that in addition to the QCF (“Q”) and subduction (“S”) groups, there are alsogroups composed of families with ambiguous polarities that are classified as unclear (“U”)groups. The unclear families are utilized to provide upper and lower limits for slip estimation.For a given Q- or S-group, every combination of including and excluding unclear families inthe calculation of the mean slip history is considered. For example, if an area has three unclearfamilies nearby (UA, UB, and UC), then there are a total of 8 scenarios to consider (Table 2). Inthe present study, the unclear group with the highest number of families (group 4U) includes16 families, which leads to 65,536 scenarios to consider when determining the upper and lower32CHAPTER 4. Slip History of the Queen Charlotte Plate Boundary(b)!(a)!(d)!(c)!(f)!(e)!Figure 14: Average cumulative slip histories for all QCF-related repeating earthquake groups.Group IDs are given in brackets in the top left of each plot, and their locations are shown in Figure13a. Main group estimates (dark blue lines) are calculated as the average cumulative slip for allrepeating earthquake families in the group. Lower and upper limits (light blue lines and shadedareas) are calculated through inclusion of families with ambiguous polarities (see text). Group 3Qhas no definitive families, and thus has only an upper maximum for slip. The time of the 2012, MW7.8 Haida Gwaii earthquake is shown with a vertical yellow line. Also shown is the rate of relativeplate motion that acts parallel to the QCF (∼4.90 cm/yr).33CHAPTER 4. Slip History of the Queen Charlotte Plate Boundary(b)!(a)!(d)!(c)!(f)!(e)!Figure 15: Average cumulative slip histories for all subduction-related repeating earthquakegroups. Group IDs are given in brackets in the top left of each plot, and their locations are shownin Figure 13b. Main group estimates (dark blue lines) are calculated as the average cumulative slipfor all repeating earthquake families in the group. Lower and upper limits (light blue lines andshaded areas) are calculated through inclusion of families with ambiguous polarities (see text).Group 1S has no definitive families, and thus has only an upper maximum for slip. The time of the2012, MW 7.8 Haida Gwaii earthquake is shown with a vertical yellow line. Also shown is the rateof relative plate motion that acts parallel to subduction (∼1.75 cm/yr).34CHAPTER 4. Slip History of the Queen Charlotte Plate BoundaryTable 2: Explanation of the different scenarios considered to determine upper and lower estimatesof slip for a given group. Every combination of including and excluding “unclear” families isconsidered, and then maxima and minima are calculated to determine upper and lower limits.In this example there are n families that constitute the main group in question (M1,2,...,n), and 3families in the associated “unclear” group (UA, UB, and UC).Scenario Families Considered inMean Slip Calculation1 M1,2,...,n2 M1,2,...,n, UA3 M1,2,...,n, UB4 M1,2,...,n, UC5 M1,2,...,n, UA, UB6 M1,2,...,n, UA, UC7 M1,2,...,n, UB, UC8 M1,2,...,n, UA, UB, UClimits of slip. For each scenario, the mean slip history is calculated for all families involved.Then, for every point in time, the minimum and maximum cumulative slip is chosen from thelist of all scenarios to yield lower and upper limits on cumulative slip at that point in time.4.2 Cumulative Slip and Slip Rate EstimatesThe final cumulative slip of each group is presented, along with the lower and upperlimits as defined above (Section 4.1), in Figure 14 (all QCF-related groups) and Figure 15 (allsubduction-related groups). Note that groups 3Q (Figure 14c) and 1S (Figure 15a) are poorlyconstrained because they possess no definitive families, and thus are characterized solely by anupper limit of slip based on unclear families in their respective areas. Aside from these emptygroups, 4Q (Figure 14d) and 4S (Figure 15d) exhibit the most uncertainty because there aremore unclear families in that area than anywhere else along the boundary.Also included in Figures 14 and 15 are the relative plate motion rates between the Pacificand North American plates in the direction that is applicable to the given group. The total sliprate vector between the Pacific and North American plates is ∼5.2 cm/yr, and is characterizedby 4.8 - 5.0 cm/yr parallel to the QCF and 1.5 - 2.0 cm/yr perpendicular to the QCF [e.g.,Hyndman and Hamilton, 1993; Mazzotti et al., 2003; Hyndman, 2015; Nykolaishen et al.,35CHAPTER 4. Slip History of the Queen Charlotte Plate Boundary2015]. Including these slip rates in Figures 14 and 15 allows us to determine if the observedslip rates, as measured by repeating earthquakes, are lower than, consistent with, or in excessof the expected slip rates from large scale plate motions.One important issue to address is whether significant changes in the slip behavior of theQCPB occur after the 2012, MW 7.8 Haida Gwaii subduction earthquake (hereafter HGEQ).For this reason, we employ cumulative slip curves in Figures 14 and 15 to determine slip ratesfor all repeating earthquake groups both before and after the HGEQ. Note that the cumulativeslip measurements in Figure 15 do not include slip directly due to the HGEQ, estimated to beon average 3.3 m and at maximum 7.7 m [Lay et al., 2013], or large aftershocks thereof [Kaoet al., 2015]. Unlike the main HGEQ event and its large aftershocks, the repeating earthquakesrepresent steady slip of the fault interface. We investigate the repeating earthquake groupsrelated to the QCF and subduction interface separately.4.2.1 Slip Rates of the Queen Charlotte FaultWe first consider the QCF, whose slip history is represented by repeating earthquakegroups 1Q, 2Q, ..., 6Q (Figures 13a and 14). The average slip rates for these repeating earth-quake groups, both before and after the HGEQ, are shown in Table 3. To determine the “maingroup” estimates we simply consider the difference in cumulative slip of the main group inTable 3: Average slip rates for QCF-related groups (see Figure 14) during time periods prior to,and after, the Mw 7.8 Haida Gwaii earthquake on Oct. 28, 2012. Note that slip rates followingthe Haida Gwaii earthquake include (if applicable) accelerated slip immediately following theearthquake. See text for explanation of how slip rates are calculated.Group2005/01/01 - 2012/10/28Slip Rate (cm/yr)2012/10/28 - 2015/12/31Slip Rate (cm/yr)LowerLimitMainGroupUpperLimitLowerLimitMainGroupUpperLimit1Q 1.84 2.04 2.08 0.11 0.49 1.152Q 2.31 2.50 2.62 0.55 1.07 1.173Q - - 4.28 - - -4Q 2.11 2.42 2.75 1.31 3.36 3.365Q 0.53 0.59 1.30 4.60 5.20 5.636Q 0.68 0.71 1.18 6.21 7.76 7.8236CHAPTER 4. Slip History of the Queen Charlotte Plate Boundaryquestion (Figure 14) during the time period of interest, and divide by the length of the timeperiod. The upper and lower slip rate limits in Table 4 are calculated by using the main groupcumulative slip value at the beginning of the time period, and either the upper or lower limitof cumulative slip (Figure 14) at the end of the time period. Note that slip rates in Table 3 in-corporate short periods of fast slip for some families (e.g., immediately following the HGEQ;Figure 14d,e,f), and thus may not be representative for the entire duration of the reported timeperiod. Recall that the component of Pacific - North American plate motion acting in the di-rection parallel to the QCF is ∼4.8 - 5.0 cm/yr; this provides a natural comparison for the sliprates determined from repeating earthquakes.Groups 1Q and 2Q represent the northernmost section of the QCF, which is also the areaof the QCF that is farthest from the hypocenter of the HGEQ (Figure 13a). The slip behaviourof these two groups is very similar (Figure 14a,b). Average slip rates of groups 1Q and 2Q priorto the HGEQ are 2.04 and 2.50 cm/yr, respectively, and after the HGEQ these rates decreaseto 0.49 and 1.07 cm/yr, respectively (Table 3). However, pre-HGEQ slip rates are inflated dueto short periods of fast slip. Group 1Q slips at ∼6.9 cm/yr between July, 2006 and September,2007, and group 2Q exhibits very fast slip in June, 2009. Note that the large transient increasein slip for group 2Q in June, 2009 is not due to a single, large earthquake contaminating thegroup average; rather, eight separate earthquakes contribute to this increase. If short periodsof fast slip are ignored, the slip rates of these two groups remain nearly constant through time,and there appears to be no influence from the HGEQ on slip of the QCF in this region. Overall,the slip rate of these groups is much lower than the large scale plate motion in the directionparallel to the QCF (∼4.8 - 5.0 cm/yr).Group 3Q has no definitive families, and thus only an upper maximum of cumulativeslip based on unclear families within its area. However, it is worth noting that the upper limiton slip rate for the time period prior to the HGEQ (4.28 cm/yr; Table 3) is similar to the platemotion rate (∼4.8 - 5.0 cm/yr). We cannot establish upper/lower limit slip rate estimates forpost-HGEQ times, because there are no repeating earthquakes in groups 3Q or 3U after the37CHAPTER 4. Slip History of the Queen Charlotte Plate BoundaryHGEQ.Group 4Q straddles the northern limit of the rupture area of the HGEQ (Figure 13a), andshows interesting slip behavior. The group exhibits no slip (or very low slip rate if the upperlimit is used) between January, 2005 and June, 2008. Between June, 2008 and the HGEQ, theslip rate is 4.3 cm/yr, which is similar to the plate motion rate of ∼4.8 - 5.0 cm/yr. The post-HGEQ slip rate is 3.36 cm/yr, and includes a burst of high-slip activity for 1 month followingthe HGEQ.Both groups 5Q and 6Q fall within the rupture zone of the HGEQ. Group 5Q is on theperiphery of the rupture zone, whereas 6Q is near the zone of maximum slip and contains theepicenter of the event (Figure 13a). The average pre-HGEQ slip rates are similar at 0.59 and0.71 cm/yr for groups 5Q and 6Q, respectively (Table 3), and both lie well below the platemotion parallel to the QCF (∼4.8 - 5.0 cm/yr). The pre-HGEQ slip rate of group 6Q is moreconstant through time, whereas all of the slip for group 5Q occurs in 2010 (Figure 14d,e).However, if we consider the upper limit of cumulative slip for 5Q (Figure 14d), the true pre-HGEQ slip of the QCF in the region of 5Q may in fact be more regular. The HGEQ appearsto have triggered activity for both groups, with very high slip rates for 2 months following theearthquake. Thereafter, the average slip rates of groups 5Q and 6Q return to values comparableto those prior to the HGEQ. Accelerated slip is again observed for both groups at the beginningof 2015, persisting for ∼6-10 months. The post-HGEQ cumulative slip of group 6S is veryclose to the upper limit of slip (Figure 14f), so if a significant number of unclear families(i.e., those in group 6U) are in fact related to the QCF, the actual slip rate may be slower thancurrently reported (Table 3).4.2.2 Slip Rates of the Subduction InterfaceNext, we investigate the repeating earthquake groups that represent thrust motions on thesubduction interface of the QCPB: 1S, 2S, ..., 6S (Figures 13b and 15). Based on the cumulativeslip curves in Figure 15, we calculate main group, lower limit, and upper limit estimates of slip38CHAPTER 4. Slip History of the Queen Charlotte Plate Boundaryrate, both before and after the HGEQ (Table 4). These slip rates are determined as in Section4.2.1, with one exception: the upper limit slip rate for group 1S, for both pre- and post-HGEQtimes, is calculated using only the upper limit of cumulative slip from Figure 15a, because thereis no main group estimate of slip. In the case of the subduction interface, we are interested incomparing our slip rate values with the component of plate motion that is perpendicular to theQCF (i.e., parallel to subduction). Recall that this margin-perpendicular motion of the Pacificand North American plates is ∼1.5 - 2.0 cm/yr.Table 4: Average slip rates for subduction-related groups (see Figure 15) during time periods priorto, and after, the Mw 7.8 Haida Gwaii earthquake on Oct. 28, 2012. Note that slip rates followingthe Haida Gwaii earthquake include (if applicable) accelerated slip immediately following theearthquake. See text for explanation of how slip rates are calculated.Group2005/01/01 - 2012/10/28Slip Rate (cm/yr)2012/10/28 - 2015/12/31Slip Rate (cm/yr)LowerLimitMainGroupUpperLimitLowerLimitMainGroupUpperLimit1S - - 2.44 - - 3.452S 2.07 2.14 2.26 0.83 2.81 1.203S 2.22 2.31 2.45 3.48 4.17 4.214S 0.90 0.97 1.88 4.56 5.81 5.915S 0.48 0.51 1.05 7.24 8.41 8.446S 1.25 1.35 1.82 1.85 2.65 3.25The slip rate of the subduction interface in the vicinity of group 1S (i.e., most northerlysection of the study area) is poorly constrained due to the absence of repeating earthquakefamilies in this group (Figure 15a; Appendix C). Nonetheless, we calculate the upper limits ofslip rate for the presumed subduction interface in this area to be 2.44 and 3.45 cm/yr for pre-and post-HGEQ times, respectively (Table 4).Groups 2S and 3S both represent slip along the subduction interface north of the mainrupture zone of the HGEQ (Figure 13b). The slip histories of these two groups are very similar(Figure 15b,c), with pre-HGEQ slip rates calculated to be 2.14 and 2.31 cm/yr for 2S and 3S,respectively (Table 4). These pre-HGEQ slip rates are slightly higher than, but comparable to,the estimated convergence rate of the Pacific and North American plates (∼1.5 - 2.0 cm/yr).39CHAPTER 4. Slip History of the Queen Charlotte Plate BoundaryThe HGEQ does not appear to have had a significant effect on the slip rates of groups 2S and3S, as no immediate acceleration in slip rate is observed for either group. However, the post-HGEQ average slip rates for these groups are higher than pre-HGEQ rates. The post-HGEQaverage slip rates for 2S and 3S are measured to be 2.81 and 4.17 cm/yr, respectively (Table4). These higher values are due, in large part, to short episodes of fast slip in December, 2015(Group 2S) and March - April, 2015 (Group 3S).The slip rates of groups 4S, 5S, and 6S are expected to be more affected by the HGEQthan those of 2S and 3S due to the proximity of these groups to the earthquake. Group 6S iswithin the main HGEQ rupture zone and contains the hypocenter of the event, 5S is withinthe northern section of the rupture zone, and 4S straddles the northern limit of rupture (Figure13b). All three of these groups exhibit slip rates that are lower than the plate motion rate ofconvergence (∼1.5 - 2.0 cm/yr) for pre-HGEQ times: 0.97 , 0.51, and 1.35 cm/yr for groups4S, 5S, and 6S, respectively (Table 4). Note that the pre-HGEQ cumulative slip curves forgroups 4S and 5S are near the calculated lower limits and far from the upper limits of slip(Figure 15d,e). Accordingly, the actual pre-HGEQ slip rates for groups 4S and 5S may beslightly higher than those reported in Table 4. All of groups 4S, 5S, and 6S experience sliprate increases at the onset of the HGEQ (Figure 15d,e,f). The post-HGEQ slip rates for groups4S, 5S, and 6S are 5.81, 8.41, and 2.65 cm/yr, respectively (Table 4). All three of these groupsexhibit slip rates that are below the plate motion convergence rate prior to the HGEQ, andgreater than the plate motion convergence rate after the HGEQ. Importantly, the post-HGEQincreases in slip rates are relatively steady for the time period beginning immediately after theHGEQ through the end of the study period (December, 2015).Overall, the cumulative slip histories of subduction groups (with the possible exceptionof group 3S) are better behaved than the corresponding QCF groups. That is, the QCF groupsdisplay more intermittent slip rates (Figure 14), whereas the subduction groups tend to slipmore steadily (Figure 15). An important example of this difference in slip behavior is theobserved effect of the HGEQ on nearby groups. Groups 4S, 5S, and 6S display an abrupt40CHAPTER 4. Slip History of the Queen Charlotte Plate Boundaryincrease in slip rate at the onset of the HGEQ, and this slip rate is relatively steady until the endof the study period. Conversely, although QCF groups in the same area (i.e., 4Q, 5Q, and 6Q)do exhibit accelerated slip immediately following the HGEQ, it only lasts for a short period of1 - 2 months (Figure 14d,e,f). The implications of these observations are discussed in Chapter5.41CHAPTER 5. DiscussionChapter 5Discussion5.1 Interseismic Slip of the Queen Charlotte PlateBoundaryThe repeating earthquake catalogue and the associated cumulative slip curves (Figures14 and 15) provide insight into the slip dynamics of both the QCF and subduction interface inthe study area. In this Section we discuss the implications of our results for the interseismicperiod (i.e., prior to the HGEQ). Because the HGEQ occurred on the subduction interface andthe QCF has not experienced a large earthquake in this area since 1949, the entire study periodmay be considered to be interseismic with regards to the QCF. Nonetheless, we do observe aresponse of the QCF to the HGEQ (Section 4.2.1). For simplicity, we will refer to all timesprior to the HGEQ as interseismic, and all times after the HGEQ as postseismic, for both thesubduction interface and the QCF. Note that the 2013, MW 7.5 Craig earthquake occurred onthe QCF ∼200 km north of our study area [Aderhold and Abercrombie, 2015; Holtkamp andRuppert, 2015]), but we do not observe any response to this earthquake and thus do not considerit in our analysis.42CHAPTER 5. Discussion5.1.1 Interseismic Slip: the Queen Charlotte FaultThe slip rate measurements from our repeating earthquake catalogue indicate that theQCF exhibits interseismic slip throughout the entire study area, and is therefore not completelylocked. However, the slip rates that we observe are substantially lower than the plate motionrate parallel to the QCF. Recall that the total plate motion vector is ∼5.2 cm/yr, and that 4.8- 5.0 cm/yr acts parallel to the QCF, whereas the pre-HGEQ slip rates that we measure arebetween ∼0.6 - 2.5 cm/yr (Table 3). The difference in plate motion and actual slip on thefault requires the QCF to be partially locked and stress to be accumulating. This inference isconsistent with the results of Mazzotti et al. [2003], who show that the model with the best fit toGPS data in the area is one with a locked QCF between 0 and 14 km depth, steady aseismic slipbelow 20 km, and a transition between the two zones. Many of the QCF repeating earthquakefamilies are located in the 14-20 km depth range; however, we do also observe shallowerfamilies (e.g., Figure 13c), and these shallow families do not exhibit different slip behaviourthan deep families. This may indicate some local fault creep in the otherwise locked shallowportion of the QCF, however it may also be due to errors in catalogue depths. Wang et al. [2015]report that the QCF is expected to intersect the subduction interface at 15-20 km depth, whichis consistent with both the model of Mazzotti et al. [2003] and the location of the repeatingearthquakes in the present study.The QCF groups’ slip rates decrease from north (2.04 and 2.50 cm/yr for groups 1Q and2Q, respectively) to south (0.59 and 0.71 cm/yr for groups 5Q and 6Q, respectively). Thisdecrease implies a higher degree of locking in the south than in the north. The QCF is closerto parallel with plate motions in the northern part of the study area (Figure 1a) and thereforeaccommodates a higher proportion of the plate motion, but this difference cannot account forthe large difference in slip rates that we observe. This variation in slip rate further suggeststhat southern regions of the study area (i.e., groups 5Q and 6Q) are more likely to rupture ina major strike-slip earthquake in the future, due to larger stress accumulation. The occurrenceof Canada’s largest recorded earthquake, the 1949, Ms 8.1 Queen Charlotte Islands earthquake,43CHAPTER 5. Discussionprovides support for this interpretation. This right-lateral, strike-slip event occurred on theQCF within our study area; its epicenter was located at 53.62◦N, 133.27◦W [Bostwick, 1984;Rogers, 1986; Nishenko and Jacob, 1990; Earthquakes Canada, 2016]. The extent of the rup-ture zone is not known precisely. However, slip from the 1949 earthquake was greatest inthe northern region of the present study area (the epicenter is in zone 1 of Figure 13a), anddecreased to the south with its southern limit near ∼52.5◦N [Bostwick, 1984; Rogers, 1986;Nishenko and Jacob, 1990]. The smaller (or, potentially, complete lack of) slip in the south dur-ing the 1949 earthquake also suggests that the QCF has higher seismic potential in the southernpart of the study area than in the north.5.1.2 Interseismic Slip: the Subduction InterfaceLike the QCF, the subduction interface exhibits non-zero pre-HGEQ slip rates through-out the study area, which suggests that the subduction thrust is not fully locked. Mazzottiet al. [2003] model the subduction interface in the same way they do the QCF: a locked zonebetween 0 and 14 km depth, a transition zone between 14 and 20 km, and a stable sliding,aseismic zone below 20 km. Our slip rates are generally consistent with the Mazzotti et al.[2003] model, however their model includes annual thrust motions of 0.6 - 1.0 cm/yr, and ourpre-HGEQ rates are higher in some places (0.5 - 2.3 cm/yr; Table 3). The depths of repeatingearthquakes are consistent with the Mazzotti et al. [2003] model, which requires a locked faultbetween 0 and 14 km. Nearly all subduction families in the present study occur on a dippingplane, inferred to be the subduction interface (Figure 13d), and are located below 14 km. Theonly shallower family is family 187 (group 5S; Figure 13d), which is assigned an (average)depth of 11.5 km due to one of its two constituent events being catalogued at 1.0 km depth.Whether the catalogue depth is incorrect or the family should be excluded from the study is notclear. However, the inclusion of this family does not significantly alter the slip rate of group5S and so we choose to retain it. A non-zero slip rate on the subduction interface is also con-sistent with Wang et al. [2015], who report that the subduction interface is expected to undergo44CHAPTER 5. Discussionfault creep landward of the QCF. Note, however, that our repeating earthquakes include thoseseaward of, albeit within ∼10 km of, the QCF and so suggest some creep behavior on the faultinterface in a region that is modelled by Wang et al. [2015] to be nearly fully locked. We cannotrule out that this discrepancy is due to errors in the locations of the repeating earthquakes.The northern subduction groups (2S and 3S) exhibit higher rates of pre-HGEQ slip thanthe southern subduction groups (4S, 5S, and 6S; Table 4). In fact, the slip rates of 2S and 3S(2.14 and 2.31 cm/yr, respectively) are slightly higher than the plate motion convergence rate(1.5 - 2.0 cm/yr), indicating that slip on the fault in the areas of 2S and 3S is keeping up withplate motion with very little accumulated stress. Furthermore, this observation is consistentwith the reported northern extent of the HGEQ rupture zone that lies south of groups 2S and 3S(Figure 13b; Hyndman 2015; Nykolaishen et al. 2015). Because these areas did not accumulatesignificant stress prior to the HGEQ, they would not be expected to experience significant slipduring the HGEQ.In contrast, the pre-HGEQ slip rates of the southern subduction groups 4S, 5S, and 6S(Table 4) are lower than the rate of convergence from plate motions (∼1.5 - 2.0 cm/yr). As forQCF slip rates in these zones (Section 5.1.1), these values indicate loading of the subductionfault and stress accumulation. Note, however, that Mazzotti et al. [2003] report 0.3 - 0.7 cm/yrof shortening within the North American plate, so not all of the ∼1.5 - 2.0 cm/yr of platemotion convergence is necessarily translated into stress accumulation on the subduction thrust.Nonetheless, the location and extent of the rupture zone of the HGEQ (Figure 13b; Hyndman2015; Nykolaishen et al. 2015) is generally consistent with our results because those zonesexhibiting stress accumulation also experienced slip during the HGEQ. However, we note thatthe pre-HGEQ slip rate of 6S (1.35 cm/yr) is higher than that of 4S and 5S (0.97 and 0.51cm/yr, respectively). Accordingly, one might expect that 5S accumulated the most stress in theinterseismic period and should therefore slip more than 6S during the HGEQ, but this is notthe case (Figure 13b). The reason for this inconsistency is not known at this time.Lastly, there are no definitive subduction families in group 1S, which represents the45CHAPTER 5. Discussionmost northerly section of the study area. The lack of subduction earthquakes in 1S may bemeaningful, because immediately south of this area the strike of the QCF changes and becomesmore closely parallel to relative plate motion (Figures 1a and 13b). This change in strike of theQCPB results in less convergence between the Pacific and North American plates, and thereforesubduction is likely not expressed in the same way as it is farther south, although underthrustingis nonetheless expected in this area [Hyndman, 2015]. This change in subduction behaviouris consistent with Tre´hu et al. [2015], who report an abrupt change in mechanical propertiesof the QCPB at 53.2◦N. Alternatively, earthquake detection thresholds might be higher in thisregion due to its greater distance from network stations to the south.5.2 Postseismic Slip of the Queen Charlotte Plate BoundaryThe HGEQ is the second largest earthquake ever recorded in Canada and offers an excel-lent opportunity to study the effects that large earthquakes have on their tectonic environments.This particular earthquake is of interest because it occurred within the subduction componentof the very complex QCPB. Thus, it affords insight into the responses of both the strike-slipQCF and the subduction interface to a major subduction earthquake.5.2.1 Postseismic Slip: the Queen Charlotte FaultThe observed post-HGEQ slip rate responses of the QCF (Figure 14) display variationfrom north to south across the study area. To the north, groups 1Q and 2Q (Figure 13a) exhibitno change to their slip patterns due to the HGEQ. Whether or not group 3Q falls into thisclassification as well is difficult to determine because of its large uncertainties. In contrast, thesouthern part of the study area (i.e., groups 4Q, 5Q, and 6Q) exhibits clear slip accelerationin response to the HGEQ. This effect is strongest in group 6Q, which falls within the zone ofmaximum rupture for the HGEQ, and decreases to the north for groups 5Q and 4Q. However,these slip accelerations are short-lived (1-2 months following the HGEQ), after which pre-HGEQ slip rates are re-established. At this time, we are unaware of the cause for additional46CHAPTER 5. Discussionslip acceleration initiated at the beginning of 2015 for groups 5Q and 6Q (Figure 14e,f). Onepossible cause is the occurrence of a MW 6.3 earthquake that occurred at 51.4◦N, 131.1◦W onApril 24, 2015 [Earthquakes Canada, 2016], but whether this magnitude of earthquake couldsignificantly alter the slip of the QCF this far north is unclear. In summary, our results indicatethat north of∼53.0◦N the HGEQ did not significantly alter the QCF slip pattern, but that southof ∼53.0◦N (to at least ∼52.3◦N) the QCF experienced short-term accelerated slip due to theHGEQ and returned to pre-HGEQ slip rates after 1-2 months.Our results are consistent with the findings from a number of other studies. Hobbs et al.[2015] investigate the Coulomb stress changes due to the HGEQ along the QCF, and reportthat the Coulomb stress change is strongly positive (i.e., promoting fault slip) in the regions ofgroups 4Q, 5Q, and 6Q, but is weakly negative (i.e., inhibiting fault slip) in the northern sectionof our study area. From a conceptual standpoint, Hobbs et al. [2015] explain that enhancedslip of the QCF near the HGEQ rupture zone occurs because movement of the hanging wallduring the thrust event acts to locally unclamp the QCF and promote movement. Additionally,Nykolaishen et al. [2015] report that GPS motions near ∼52.5◦N, 131.8◦W (i.e., within group6Q of the present study; Figure 13a) are consistent with induced, deep, aseismic slip along theQCF for up to 1 year following the HGEQ. Lastly, the aftershock sequence of the HGEQ hasbeen studied by Farahbod and Kao [2015] and Kao et al. [2015]. Although the majority ofaftershocks represent normal faulting, due mainly to extension of the downgoing plate updipof the rupture zone, there are also right-lateral strike-slip events observed. Between 52.5◦Nand 53◦N these strike-slip events are located near the trace of the QCF and are interpreted torepresent motion of QCF related to the HGEQ [Farahbod and Kao, 2015; Kao et al., 2015]. Ourresults from repeating earthquakes, in conjunction with the studies mentioned above, suggestthat the QCF participated in the HGEQ sequence south of ∼53◦N, but was unaffected to thenorth of this region. It is important to note that although some elastic strain along the QCF wasreleased during the HGEQ, the QCF remains partially locked, and thus the risk of future largestrike-slip earthquakes in the area (especially south of∼53◦N) remains high [e.g., Hobbs et al.,47CHAPTER 5. Discussion2015; Kao et al., 2015].5.2.2 Postseismic Slip: the Subduction InterfaceThe subduction interface, like the QCF, exhibits post-HGEQ slip rates that vary fromnorth to south. North of ∼53◦N, groups 2S and 3S show no observable change in slip rate dueto the HGEQ (Figures 13b and 15b,c). Matsubara et al. [2005] make a similar observation forthe 2003 Off-Tokachi subduction earthquake in Japan, where repeating earthquake slip ratesaway from the main rupture zone do not significantly change. However, south of ∼53◦N,groups 4S, 5S, and 6S all exhibit accelerated slip immediately following the HGEQ, and theseslip rates exceed the convergence rate between the Pacific and North American plates (Figure15d,e,f). Importantly, this accelerated slip persists until the end of the study period (December,2015), 3 years and 2 months after the HGEQ. Recall that the QCF also exhibits accelerated slipdue to the HGEQ, but that this effect only lasted for 1-2 months. The accelerated subductionslip is strongest for groups 5S (post-HGEQ slip rate of 8.41 cm/yr) and 4S (5.81 cm/yr), andweaker for group 6S (2.65 cm/yr; Table 4).This slip on the subduction thrust represents afterslip to the HGEQ. Note that, consistentwith the afterslip model of Wang et al. [2012], afterslip is strongest in areas surrounding themain rupture zone of the HGEQ (i.e., groups 4S and 5S), and weaker within the zone of max-imum slip (i.e., group 6S; Figure 13b). In fact, the locations of repeating earthquake familieswithin group 6S all lie outside the zone of largest slip from the HGEQ (Figure 13b). Similarresults are reported by Igarashi et al. [2003], who find that repeating earthquakes tend to belocated near, but not within, zones of maximum rupture from large subduction earthquakesbecause these regions exhibit the highest degrees of aseismic slip.Wang et al. [2012] model the duration of afterslip for MW 8.0-8.4 subduction earth-quakes, and demonstrate that afterslip may persist for 3-7 years. The HGEQ is a slightlysmaller event (MW 7.8) and thus afterslip would be expected to occur over a shorter time pe-riod. However, other factors, including convergence rate and plate geometry, also affect the48CHAPTER 5. Discussionduration of afterslip [Wang et al., 2012]. Our results indicate that afterslip is present until atleast the end of the study period (December, 2015), approximately 3 years and 2 months afterthe HGEQ. Slip histories for groups 4S and 6S indicate that the rate of afterslip may be de-creasing, consistent with Matsubara et al. [2005] and Wang et al. [2012]; however, this trendis not observed for group 5S. Afterslip estimates made here are also generally consistent withGPS measurements of Nykolaishen et al. [2015], which indicate that the northern periphery ofthe rupture zone (i.e., groups 4S and 5S) exhibited postseismic thrust motion for the durationof their study period (October 28, 2012 - December 31, 2013). Note that afterslip has alsobeen observed using repeating earthquakes in other subduction zones around the world (e.g.,the northeastern Japan subduction zone by Igarashi et al. [2003] and Matsubara et al. [2005]).49CHAPTER 6. ConclusionsChapter 6ConclusionsWe have demonstrated the utility of repeating earthquakes to study the slip dynamicsof the Queen Charlotte plate boundary (QCPB), which includes both the strike-slip QueenCharlotte Fault (QCF) and the subduction interface between the Pacific and North Americanplates. Repeating earthquake families are identified through a combination of cluster analysisand matched filtering, both based on waveform similarity. The repeating earthquake cataloguethat we present here includes 224 families, with 730 individual earthquakes. The subductioninterface is represented by 87 families (320 earthquakes) and the QCF by 76 families (240earthquakes). There are also 61 families (170 earthquakes) whose polarities preclude assign-ment of a fault of origin. These ambiguous families are incorporated within our analysis to setmaximum and minimum bounds on slip for both fault structures.We extend the algorithm of Rubinstein and Ellsworth [2010] that employs Singular ValueDecomposition (SVD) and absolute moment estimates, to determine accurate magnitude es-timates for repeating earthquakes. This process exploits the inherent waveform similarityobserved for repeating earthquakes to provide better magnitude estimates than conventionalmethods.The repeating earthquake families are arranged into three sets: those consistent with anorigin on the QCF, those consistent with an origin on the subduction thrust, and those with an50CHAPTER 6. Conclusionsunclear origin. These sets are each further divided based on location into 6 adjacent groupsalong the QCPB. The average cumulative slip history for both the QCF and subduction inter-face is then determined for each of the 6 zones.We find evidence of fault creep on both the QCF and the subduction interface throughoutthe study area. However, with the exception of the northernmost subduction interface, all creeprates are less than the relative plate motion rates between the Pacific and North Americanplates, indicating partial locking of both the QCF and subduction thrust. The QCF exhibitshigher degrees of locking and stress-loading in the southern region of the study area (∼52.3◦N- 52.8◦N) than in the north (∼52.8◦N - 53.8◦N), which supports the notion that the seismichazard along the QCF is highest in the south.The 2012, MW 7.8 Haida Gwaii earthquake (HGEQ) produced changes in slip behaviouralong both the QCF and subduction interface. Near the main rupture, the QCF exhibits short-term (1-2 month) accelerated slip in a right-lateral strike-slip sense following the HGEQ. How-ever, most of the QCF elastic-strain was not released during the HGEQ and the seismic risk inthis region remains high. In contrast, the subduction thrust has undergone prolonged aseismicafterslip for at least 3 years following the HGEQ. This afterslip is greatest on the periphery ofthe main HGEQ rupture.51ReferencesReferencesAbercrombie, R. (1996). The magnitude-frequency distribution of earthquakes recorded withdeep seismometers at Cajon Pass, southern California. Tectonophysics, 261:1–7.Aderhold, K. and Abercrombie, R. (2015). Seismic rupture on an oceanic-continental plateboundary: strike-slip earthquakes along the Queen Charlotte-Fairweather Fault. Bulletin ofthe Seismological Society of America, 105(2B):1129–1142.Bakun, W. (1984). Seismic moments, local magnitudes, and coda-duration magnitudesfor earthquakes in central California. Bulletin of the Seismological Society of America,74(2):439–458.Beeler, N., Lockner, D., and Hickman, S. (2001). A simple stick-slip and creep-slip model forrepeating earthquakes and its implication for microearthquakes at Parkfield. Bulletin of theSeismological Society of America, 91(6):1797–1804.Ben-Zion, Y. and Zhu, L. (2002). Potency-magnitude scaling relations for souther Californiaearthquakes with 1.0< mL < 7.0. Geophysics Journal International, 148:F1–F5.Be´rube´, J., Rogers, G., Ellis, R., and Hasselgren, F. (1989). A microseismicity survey of theQueen Charlotte Islands region. Canadian Journal of Earth Science, 26:2556–2566.Bostwick, T. (1984). A re-examination of the August 22, 1949 Queen Charlotte earthquake.Master’s thesis, University of British Columbia, Vancouver, Canada.Bustin, A., Hyndman, R., Kao, H., and Cassidy, J. (2007). Evidence for underthrusting beneaththe Queen Charlotte Margin, British Columbia, from teleseismic receiver function analysis.Geophysics Journal International, 171(3):1198–1211.Cassidy, J., Rogers, G., and Hyndman, R. (2014). An overview of the 28 October 2012 Mw7.7 earthquake in Haida Gwaii, Canada: a tsunamigenic thrust event along a predominantlystrike-slip margin. Pure and Applied Geophysics, 171:3457–3465. doi: 10.1007/s00024-014-0775-1.Chen, K., Nadeau, R., and Rau, R. (2007). Towards a universal rule on the recurrenceinterval scaling of repeating earthquakes? Geophysical Research Letters, 34. doi:10.1029/2007GL030554.52ReferencesChen, K., Nadeau, R., and Rau, R. (2008). Characteristic repeating earthquakes in an arc-continent collision boundary zone: The Chihshang fault of eastern Taiwan. Earth and Plan-etary Science Letters, 276:262–272.Chen, T. and Lapusta, N. (2009). Scaling of small repeating earthquakes explained by inter-action of seismic and aseismic slip in a rate and state fault model. Journal of GeophysicalResearch, 114. doi: 10.1029/2008JB005749.Dehler, S. and Clowes, R. (1988). The Queen Charlotte Islands refraction project. Part I. TheQueen Charlotte Fault Zone. Canadian Journal of Earth Science, 25:1857–1870.Earthquakes Canada (2016). Continuous waveform achive. Natural Resources Canada - Geo-logical Survey of Canada, AutoDRM@seismo.NRCan.gc.ca.Farahbod, A. and Kao, H. (2015). Spatiotemporal distribution of event during the first weekof the 2012 Haida Gwaii aftershock sequence. Bulletin of the Seismological Society ofAmerica, 105(2B):1231–1240.Fisher, R. (1921). On the probable error of a coefficient of correlation deduced from a smallsample. Metron, 1:3–32.Gibbons, S. and Ringdal, F. (2006). The detection of low magnitude seismic events usingarray-based waveform correlation. Geophysics Journal International, 165:149–166.Gosselin, J., Cassidy, J., and Dosso, S. (2015). Shear-wave velocity structure in the vicinity ofthe 2012 Mw 7.8 Haida Gwaii earthquake from receiver function inversion. Bulletin of theSeismological Society of America, 105(2B):1106–1113.Hanks, T. and Kanamori, H. (1979). A moment magnitude scale. Journal of GeophysicalResearch, 84(B5):2348–2350.Hobbs, T., Cassidy, J., Dosso, S., and Brillon, C. (2015). Coulomb stress changes followingthe Mw 7.8 Haida Gwaii, Canada, earthquake: Implications for seismic hazard. Bulletin ofthe Seismological Society of America, 105(2B):1253–1264.Holtkamp, S. and Ruppert, N. (2015). A high resolution aftershock catalog of the magnitude7.5 Craig, Alaska, earthquake on 5 January 2013. Bulletin of the Seismological Society ofAmerica, 105(2B):1143–1152.Hyndman, R. (2015). Tectonics and structure of the Queen Charlotte Fault Zone, HaidaGwaii, and large thrust earthquakes. Bulletin of the Seismological Society of America,105(2B):1058–1075.Hyndman, R. and Ellis, R. (1981). Queen Charlotte fault zone: microearthquakes from atemporary array of land stations and ocean bottom seismographs. Canadian Journal ofEarth Science, 18:776–788.Hyndman, R. and Hamilton, T. (1993). Queen Charlotte area Cenozoic tectonics and volcan-ism and their association with relative plate motions along the northeastern Pacific margin.Journal of Geophysical Research, 98(B8):14257–14277.53ReferencesHyndman, R., Lewis, T., J.A., W., Burgess, M., Chapman, D., and Yamano, M. (1982). QueenCharlotte fault zone: heat flow measurements. Canadian Journal of Earth Science, 19:1657–1669.Igarashi, T., Matsuzawa, T., and Hasegawa, A. (2003). Repeating earthquakes and interplateaseismic slip in the northeastern Japan subduction zone. Journal of Geophysical Research,108(B5):2249. doi: 10.1029/2002JB001920.Kao, H., Shan, S., and Farahbod, A. (2015). Source characteristics of the 2012 Haida Gwaiiearthquake sequence. Bulletin of the Seismological Society of America, 105(2B):1206–1218.Kato, A., Fukuda, J., T., K., and S., N. (2016). Accelerated nucleation of the 2014 Iquique,Chile Mw earthquake. Scientific Reports, 6. doi: 10.1038/srep24792.Lay, T., Ye, L., Kanamori, H., Yamazaki, Y., Cheung, K., Kwong, K., and Koper, K. (2013).The October 28, 2012 Mw7.8 Haida Gwaii underthrusting earthquake and tsunami: slip par-tioning along the Queen Charlotte Fault transpressional plate boundary. Earth and PlanetaryScience Letters, 375:57–70.Mackie, D., Clowes, R., Dehler, S., Ellis, R., and Morel-A`-l’Huissier, P. (1989). The QueenCharlotte Islands refraction project. Part II. Structural model for transition from Pacific plateto North American plate. Canadian Journal of Earth Sciences, 26(9):1713–1725.Matsubara, M., Yagi, Y., and Obara, K. (2005). Plate boundary slip associated with the 2003Off-Tokachi earthquake based on small repeating earthquake data. Geophysical ResearchLetters, 32. doi: 10.1029/2004GL022310.Mazzotti, S., Hyndman, R., Fluck, P., Smith, A., and Schmidt, M. (2003). Distribution of thePacific/North America motion in the Queen Charlotte Islands - S. Alaska plate boundaryzone. Geophysical Research Letters, 30. doi: 10.1029/2003GL017586.Nadeau, R., Foxall, W., and McEvilly, T. (1995). Clustering and periodic recurrence of mi-croearthquakes on the San Andreas fault and Parkfield, California. Science, 267:503–507.Nadeau, R. and Johnson, L. (1998). Seismological studies at Parkfield VI: moment releaserates and estimates of course parameters for small repeating earthquakes. Bulletin of theSeismological Society of America, 88(3):790–814.Nishenko, S. and Jacob, K. (1990). Seismic potential of the Queen Charlotte-Alaska-Aleutianseismic zone. Journal of Geophysical Research, 95(B3):2511–2532.Nykolaishen, L., Dragert, H., Wang, K., James, T., and Schmidt, M. (2015). Gps observationsof crustal deformation associated with the 2012 Mw 7.8 Haida Gwaii earthquake. Bulletin ofthe Seismological Society of America, 105(2B):1241–1252.Plafker, G., Hudson, T., and Bruns, T. (1978). Late Quaternary offsets along the Fairweatherfault and crustal plate interactions in southern Alaska. Canadian Journal of Earth Science,15:805–816.54ReferencesRogers, G. (1986). Seismic gaps along the Queen Charlotte Fault. Earthquake PredictionResearch, 4:1–11.Rohr, K., Scheidhauer, M., and Tre´hu, A. (2000). Transpression between two warmmafic plates: the Queen Charlotte Fault revisited. Journal of Geophysical Research,105(B4):8147–8172.Romesburg, H. (2004). Cluster Analysis for Researchers. Lulu Press, North Carolina, USA.Rubinstein, J. and Ellsworth, W. (2010). Precise estimation of repeating earthquake mo-ment: example from Parkfield, California. Bulletin of the Seismological Society of America,100(5A):1952–1961.Schmittbuhl, J., Karabulut, H., O., L., and Bouchon, M. (2016). Long-lasting seismic re-peaters in the Central Basin of the Main Marmara Fault. Geophysical Research Letters,43(18):9527–9534.Shearer, P. (2009). Introduction to Seismology, 2nd Edition. Cambridge University Press, NewYork, USA.Shearer, P., Prieto, G., and Hauksson, E. (2006). Comprehensive analysis of earthquakesource spectra in southern California. Journal of Geophysical Research, 111. doi:10.1029/2005JB003979.Smith, A., R.D., H., Cassidy, J., and Wang, K. (2003). Stucture, seismicity, and ther-mal regime of the Queen Charlotte Transform Margin. Journal of Geophysical Research,108(B11):2156–2202.Tre´hu, A., Scheidhauer, M., Rohr, K., Tikoff, B., Walton, M., Gulick, S., and Roland, E. (2015).An abrupt transition in the mechanical response of the upper crust to transpression along theQueen Charlotte Fault. Bulletin of the Seismological Society of America, 105(2B):1114–1128.Uchida, N., Iinuma, T., Nadeau, R., Bu¨rgmann, R., and Hino, R. (2016). Periodic slow sliptriggers megathrust zone earthquakes in northeastern Japan. Science, 351(6272):488–492.Uchida, N. and Matsuzawa, T. (2013). Pre- and postseismic slow slip surrounding the 2011Tohoku-oki earthquake rupture. Earth and Planetary Science Letters, 374:81–91.Van Huffel, S. and Vandewalle, J. (1991). The total least squares problem: computationalaspects and analysis. Society for Industrial and Applied Mathematics, Philadelphia, USA.Wang, K., He, J., Schulzeck, F., Hyndman, R., and Riedel, M. (2015). Thermal condition ofthe 27 October 2012 Mw 7.8 Haida Gwaii subduction earthquake at the obliquely convergentQueen Charlotte Margin. Bulletin of the Seismological Society of America, 105(2B):1290–1300.Wang, K., Hu, Y., and He, J. (2012). Deformation cycles of the subduction earthquakes in aviscoelastic Earth. Nature, 484:327–332.55ReferencesYorath, C. and Hyndman, R. (1983). Subsidence and thermal history of Queen Charlotte Basin.Canadian Journal of Earth Science, 20:135–159.56Appendix A. Definition of all Repeating Earthquake FamiliesAppendix ADefinition of all Repeating EarthquakeFamiliesIn Chapter 2 we explain how the repeating earthquake catalogue is created, and howindividual earthquakes are clustered into families. The repeating earthquakes within a givenfamily represent slip of the same fault patch, and are identified on the basis high waveformsimilarity. Table 5 provides a definition for each of the 224 families identified in this study.Table 5: Details of the repeating earthquake families used in the present study. Family IDs arearbitrarily assigned, but consistent throughout the study. Times are given as seconds into the day(i.e., seconds after midnight). Reported magnitudes are those after the SVD analysis (see Section3.1.2). Events with detection method noted as “GSC” originated from the GSC earthquakes list,whereas those noted as “MF” were identified with the matched filter processing (see Section 2.3).Events from the matched filter processing do not have reported latitudes, longitudes, or depths, andonly a small selection of GSC events have reported depths.FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod001 05/02/21 33489.000 2.512 10.41 53.705 -133.422 - GSC12/09/11 44831.000 2.456 10.08 53.768 -133.321 - GSC57Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod002 05/05/24 49059.000 1.740 6.62 52.570 -132.044 - GSC05/05/24 49990.000 1.695 6.45 52.571 -131.969 - GSC05/06/13 77741.550 1.107 4.56 - - - MF003 05/06/04 19565.000 1.483 5.69 52.426 -131.806 - GSC05/06/04 20172.000 1.678 6.38 52.426 -131.810 - GSC004 05/06/07 58977.000 1.831 6.98 52.603 -131.972 - GSC05/06/09 34844.000 2.311 9.26 52.589 -132.041 - GSC005 05/06/19 17467.000 1.591 6.06 53.276 -133.171 - GSC05/06/19 19124.000 1.566 5.97 53.272 -133.176 - GSC15/08/06 5687.000 2.444 10.01 53.307 -133.075 27.6 GSC15/08/06 7143.150 1.403 5.43 - - - MF006 05/06/19 52179.000 1.666 6.34 53.288 -133.137 - GSC05/06/19 57809.000 1.572 5.99 53.288 -133.138 - GSC15/11/08 70518.650 2.066 8.01 - - - MF15/11/08 71272.850 1.645 6.26 - - - MF007 05/06/19 59638.000 1.559 5.95 53.288 -133.136 - GSC05/06/19 70764.000 2.236 8.85 53.288 -133.142 - GSC008 05/06/19 24362.000 1.522 5.82 53.269 -133.173 - GSC05/06/19 62948.000 1.498 5.74 53.275 -133.168 - GSC009 05/07/13 22392.000 1.685 6.41 52.535 -131.871 - GSC05/07/13 22965.000 1.695 6.45 52.537 -131.873 - GSC010 05/05/09 42773.000 1.402 5.43 53.224 -132.810 - GSC05/07/13 79083.000 2.166 8.5 53.269 -132.731 - GSC58Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod011 05/07/18 19061.000 1.577 6.02 53.496 -133.170 - GSC05/07/18 21424.000 1.579 6.02 53.497 -133.166 - GSC012 05/08/31 27136.000 2.399 9.74 53.842 -133.448 - GSC08/05/23 24048.000 2.364 9.55 53.827 -133.399 - GSC14/02/23 37344.000 2.436 9.96 53.806 -133.415 14.7 GSC013 05/09/13 22775.000 1.733 6.59 52.519 -131.907 - GSC05/09/13 26703.000 1.703 6.48 52.517 -131.923 - GSC014 05/11/07 26402.000 1.622 6.18 53.264 -133.135 - GSC15/08/10 3752.000 1.861 7.1 53.235 -133.190 10.1 GSC015 05/11/22 83229.000 2.444 10.01 52.880 -132.418 - GSC11/04/26 74573.000 1.958 2 52.868 -132.442 - GSC016 05/12/15 67295.000 1.373 5.33 53.095 -132.788 - GSC11/07/19 3861.000 1.368 5.32 53.096 -132.759 - GSC11/07/19 4016.875 0.330 2.89 - - - MF017 05/12/16 15753.000 1.829 6.97 53.268 -132.740 - GSC05/12/16 16274.000 1.551 5.92 53.276 -132.713 - GSC018 05/12/27 63683.000 1.981 7.62 52.629 -132.130 - GSC09/12/15 83606.000 1.704 6.48 52.640 -132.139 - GSC019 06/01/03 46357.000 1.603 6.11 52.631 -132.319 - GSC11/05/14 22033.000 1.213 4.86 52.663 -132.263 - GSC020 06/01/04 67679.000 1.915 7.33 52.620 -132.200 - GSC10/09/02 28750.000 1.935 7.42 52.625 -132.219 - GSC59Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod021 06/01/17 4119.000 2.028 7.84 53.068 -132.682 - GSC08/07/13 24070.750 1.388 5.38 - - - MF13/07/20 78373.950 1.043 4.4 - - - MF15/02/21 16249.000 1.125 2.34 53.084 -132.658 23.5 GSC022 06/01/26 56308.000 1.661 6.32 53.229 -133.095 - GSC06/01/26 83409.000 2.029 7.84 53.212 -133.115 - GSC023 06/01/29 66298.000 1.434 5.53 53.236 -132.764 - GSC06/02/15 46384.000 1.232 4.91 53.225 -132.772 - GSC024 05/08/01 13529.525 0.718 3.63 - - - MF05/08/07 14801.000 1.406 5.44 53.076 -132.643 - GSC06/03/13 413.000 1.399 5.42 53.070 -132.671 - GSC13/07/20 78245.850 1.172 4.74 - - - MF025 06/03/29 57036.000 1.988 7.65 52.613 -132.154 - GSC10/01/29 51543.000 2.040 7.89 52.625 -132.109 - GSC026 06/04/01 55540.000 1.347 5.26 53.255 -133.155 - GSC06/04/02 32335.000 1.466 5.63 53.231 -133.217 - GSC15/09/25 63610.250 1.236 4.92 - - - MF027 06/05/11 66846.000 1.771 6.74 52.982 -132.595 - GSC06/05/11 69930.000 1.610 6.13 52.978 -132.597 - GSC06/05/11 70071.000 1.840 7.02 52.979 -132.592 - GSC028 06/05/11 71118.525 0.988 4.26 - - - MF06/05/11 72474.000 1.767 6.72 52.977 -132.594 - GSC06/05/12 5728.000 1.729 6.57 52.979 -132.593 - GSC06/05/12 8904.300 0.928 4.11 - - - MF60Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod029 06/05/10 45051.000 1.528 5.84 52.672 -132.378 - GSC06/05/12 38244.000 1.619 6.16 52.681 -132.357 - GSC030 06/06/03 57696.000 2.382 9.65 53.188 -132.873 - GSC13/07/02 73224.600 1.972 7.58 - - - MF15/08/12 86233.000 2.129 8.32 53.178 -132.951 4.9 GSC031 06/08/21 67004.000 1.551 5.92 52.921 -132.542 - GSC06/08/21 67209.000 1.281 5.06 52.904 -132.542 - GSC10/10/20 63790.000 1.434 5.53 52.932 -132.534 - GSC032 06/09/20 22644.000 1.969 7.57 53.278 -132.817 - GSC06/09/20 24230.000 1.946 7.47 53.276 -132.821 - GSC07/02/24 19573.550 0.842 3.91 - - - MF033 06/09/26 34078.000 1.540 5.88 52.510 -131.927 - GSC06/09/26 41210.000 1.475 5.66 52.511 -131.917 - GSC034 06/10/08 73108.450 0.871 3.97 - - - MF06/10/09 3878.000 1.766 6.72 53.275 -132.811 - GSC06/10/10 31052.000 1.460 5.62 53.280 -132.813 - GSC07/05/08 85939.000 1.747 6.65 53.278 -132.810 - GSC07/05/09 3607.525 1.059 4.44 - - - MF07/05/09 19777.975 0.816 3.85 - - - MF035 06/11/02 45038.000 2.334 9.38 53.511 -133.209 - GSC06/11/03 4864.000 1.821 6.94 53.478 -133.322 - GSC036 06/10/08 73826.000 2.182 8.58 53.213 -132.993 - GSC06/11/03 85387.000 1.786 6.8 53.208 -132.973 - GSC61Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod037 05/08/14 50529.000 1.765 6.72 53.284 -132.807 - GSC05/08/27 75644.575 0.999 4.28 - - - MF06/11/08 19678.000 1.385 5.37 53.255 -132.770 - GSC07/03/05 30784.450 1.249 4.96 - - - MF038 07/01/23 389.000 1.629 6.2 52.557 -131.845 - GSC07/01/23 13061.000 1.438 5.54 52.556 -131.840 - GSC039 07/01/23 20503.000 2.282 9.1 53.501 -133.351 - GSC07/01/23 22133.000 2.141 8.38 53.505 -133.358 - GSC040 07/02/14 9702.000 2.069 8.03 52.763 -132.288 - GSC07/02/14 11263.000 1.606 6.12 52.765 -132.294 - GSC041 07/03/31 42445.000 1.433 5.53 53.278 -132.764 - GSC07/06/28 74874.000 1.579 6.02 53.265 -132.803 - GSC042 07/04/03 76763.000 1.589 6.06 53.190 -132.886 - GSC07/04/03 77816.000 1.674 6.36 53.200 -132.876 - GSC07/04/04 1301.750 1.003 4.29 - - - MF07/04/04 4376.000 1.483 5.69 53.196 -132.902 - GSC043 07/04/01 82444.000 2.379 9.63 53.512 -133.192 - GSC07/04/12 55566.000 1.712 6.51 53.526 -133.166 - GSC044 05/04/30 84401.550 1.915 7.33 - - - MF07/04/13 70167.000 1.853 7.07 52.614 -132.195 - GSC09/12/21 52220.000 1.602 6.1 52.619 -132.175 - GSC045 07/04/30 38616.000 2.143 8.38 52.781 -132.364 - GSC07/04/30 39401.000 2.126 8.3 52.804 -132.366 - GSC62Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod046 07/05/21 2307.700 1.366 5.31 - - - MF07/05/21 23855.000 1.838 7.01 52.996 -132.764 - GSC07/05/21 24097.000 1.710 6.5 52.981 -132.822 - GSC047 07/06/14 8133.000 1.831 6.98 53.332 -133.165 - GSC07/06/14 12019.000 1.962 7.54 53.326 -133.181 - GSC15/04/02 40944.350 2.042 7.9 - - - MF048 07/06/15 16275.000 1.820 6.94 52.809 -132.323 - GSC07/08/01 72479.000 0.806 3.82 52.811 -132.309 - GSC049 07/06/27 32518.000 1.953 7.5 52.682 -132.017 - GSC07/12/04 11290.000 1.833 6.99 52.657 -132.064 - GSC12/10/24 6848.175 1.756 6.68 - - - MF050 07/07/03 6462.000 1.635 6.22 53.208 -132.686 - GSC07/07/03 22257.000 1.739 6.61 53.195 -132.706 - GSC07/07/03 54475.975 1.699 6.46 - - - MF07/07/03 81962.550 1.154 4.69 - - - MF051 07/07/03 12735.000 2.238 8.87 53.189 -132.722 - GSC07/07/03 12826.000 1.458 5.61 53.207 -132.708 - GSC07/07/03 13301.000 1.429 5.51 53.209 -132.676 - GSC07/07/03 41951.725 1.346 5.25 - - - MF63Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod052 07/07/13 60027.075 0.775 3.76 - - - MF07/07/13 68073.850 0.932 4.12 - - - MF07/07/13 71479.000 1.028 4.36 53.215 -132.641 - GSC07/07/14 2447.775 0.802 3.82 - - - MF07/07/14 13916.000 1.488 5.71 53.208 -132.691 - GSC07/07/15 27368.100 0.323 2.88 - - - MF053 07/07/21 24643.000 1.456 5.6 53.240 -132.721 - GSC07/07/21 26172.000 1.414 5.47 53.229 -132.765 - GSC07/07/21 26458.000 0.592 3.37 - - - MF11/08/03 66792.325 0.780 3.77 - - - MF054 07/07/26 44113.000 1.614 6.15 53.275 -132.745 - GSC07/07/26 79981.700 1.364 5.31 - - - MF14/09/23 85395.000 2.375 9.61 53.273 -132.748 19.9 GSC055 07/07/31 31990.000 1.754 6.67 52.600 -131.961 - GSC07/07/31 46190.000 1.112 4.58 52.609 -131.918 - GSC056 07/08/28 3057.000 1.804 6.87 53.310 -133.064 - GSC07/08/28 84168.000 2.031 7.85 53.308 -133.069 - GSC057 07/09/09 32695.000 2.048 7.93 53.506 -133.245 - GSC07/09/09 55924.000 1.510 5.78 53.520 -133.207 - GSC058 07/01/25 27047.000 2.107 8.21 52.907 -132.416 - GSC07/09/12 36824.000 1.614 6.15 52.904 -132.520 - GSC07/09/12 40335.775 1.321 5.18 - - - MF059 07/10/03 79203.000 1.977 7.6 53.188 -132.814 - GSC07/10/03 84601.000 2.100 8.17 53.186 -132.811 - GSC64Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod060 07/10/12 46652.000 2.036 7.87 52.877 -132.511 - GSC07/10/12 49593.000 1.798 6.85 52.877 -132.497 - GSC061 07/03/16 76059.000 2.008 7.74 53.448 -132.485 - GSC07/06/26 75615.450 2.010 7.75 - - - MF07/11/21 11859.000 2.213 8.74 53.347 -132.911 - GSC062 07/11/23 49075.000 1.933 7.41 52.610 -132.024 - GSC07/12/02 72645.000 2.025 7.82 52.658 -131.978 - GSC063 07/11/20 3011.000 1.452 5.59 52.907 -132.558 - GSC07/11/29 84765.775 1.519 5.81 - - - MF07/11/29 85293.000 2.041 7.9 52.907 -132.557 - GSC07/12/24 83691.450 2.042 7.9 - - - MF064 08/01/23 66075.000 1.858 7.09 52.895 -132.534 - GSC08/02/06 22917.000 2.162 8.48 52.910 -132.483 - GSC065 08/02/07 25474.000 2.126 8.3 52.892 -132.524 - GSC08/02/07 26150.000 1.874 7.16 52.891 -132.541 - GSC066 08/03/22 33482.000 2.756 12.02 53.269 -133.102 - GSC08/03/22 33571.000 2.844 12.66 53.286 -133.061 - GSC067 08/03/22 78527.000 2.545 10.61 53.276 -133.095 - GSC08/03/23 23121.000 2.077 8.07 53.300 -133.056 - GSC068 08/03/22 78407.000 2.642 11.24 53.282 -133.074 - GSC08/03/22 85185.250 1.958 7.52 - - - MF08/03/23 10399.000 2.481 10.22 53.327 -132.943 - GSC65Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod069 08/05/25 31228.000 1.813 6.91 52.748 -132.346 - GSC12/12/23 16072.550 1.668 6.34 - - - MF14/03/26 25087.575 1.754 6.67 - - - MF14/07/16 36462.750 1.801 6.86 - - - MF14/07/21 30693.975 1.943 7.46 - - - MF15/03/07 67589.000 1.692 6.43 52.754 -132.296 17.0 GSC15/05/02 72061.000 2.133 8.34 52.752 -132.297 21.5 GSC070 08/06/08 40003.000 2.315 9.28 52.841 -132.454 - GSC08/06/08 78082.000 2.041 7.9 52.839 -132.465 - GSC071 08/06/08 50009.775 1.333 5.21 - - - MF08/06/09 21321.000 1.511 5.79 52.842 -132.445 - GSC08/06/09 24221.000 2.375 9.61 52.833 -132.454 13.0 GSC072 08/06/17 1104.000 1.919 7.35 52.584 -132.015 - GSC08/06/17 1333.000 1.813 6.91 52.577 -132.065 - GSC08/06/17 44056.000 2.233 8.84 52.575 -132.048 - GSC08/06/17 57264.450 1.919 7.35 - - - MF08/06/17 57493.150 1.813 6.91 - - - MF073 08/08/10 4216.000 1.840 7.02 53.526 -133.092 - GSC08/08/10 4646.000 1.702 6.47 53.521 -133.108 - GSC074 08/08/09 35194.000 2.167 8.5 53.241 -133.070 - GSC08/08/10 24340.000 1.680 6.39 53.259 -133.015 - GSC075 06/04/29 56404.000 1.925 7.38 52.686 -132.277 - GSC08/10/03 60290.000 2.002 7.72 52.689 -132.288 - GSC12/01/12 83021.375 1.882 7.19 - - - MF66Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod076 08/06/07 59154.000 1.964 7.55 53.509 -133.180 - GSC08/11/05 38913.000 2.026 7.83 53.508 -133.190 - GSC15/10/06 33374.000 1.850 7.06 53.529 -133.178 14.3 GSC077 08/12/08 1985.000 1.854 7.08 52.674 -132.062 - GSC08/12/08 9067.000 1.975 7.6 52.664 -132.075 - GSC078 08/12/08 76707.000 2.185 8.6 53.205 -133.016 - GSC08/12/09 1184.000 1.624 6.18 53.214 -132.983 - GSC08/12/09 76364.000 1.624 6.18 - - - MF08/12/09 78338.000 1.624 6.18 - - - MF079 08/12/05 71382.050 1.673 6.36 - - - MF08/12/10 32837.000 2.398 9.74 52.907 -132.536 - GSC08/12/10 34487.000 2.102 8.19 52.906 -132.561 - GSC080 07/09/09 32923.000 1.457 5.61 53.498 -133.271 - GSC07/09/09 34017.000 1.562 5.96 53.505 -133.263 - GSC08/12/19 48991.000 2.251 8.93 53.505 -133.234 - GSC081 06/02/09 62221.000 1.645 6.26 52.609 -132.116 - GSC09/01/23 6258.000 1.844 7.04 52.618 -132.128 - GSC082 07/04/04 6847.000 1.934 7.41 52.721 -132.272 - GSC09/01/31 25687.000 2.302 9.21 52.714 -132.278 - GSC09/01/31 59707.000 2.302 9.21 - - - MF083 09/02/19 34291.000 2.653 11.31 52.500 -131.910 - GSC09/02/19 39480.000 2.172 8.53 52.492 -131.895 - GSC084 06/07/05 42259.000 1.680 6.39 52.686 -132.000 - GSC09/02/23 65293.000 1.764 6.71 52.608 -132.135 - GSC67Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod085 09/03/02 6919.000 1.900 7.27 52.904 -132.554 - GSC09/03/02 7282.900 1.320 5.17 - - - MF09/03/02 9880.000 2.237 8.86 52.902 -132.548 - GSC086 09/05/04 8609.000 1.948 7.48 52.905 -132.543 - GSC09/05/04 10613.000 1.786 6.8 52.910 -132.558 - GSC10/07/16 77187.000 1.967 7.56 52.899 -132.531 - GSC087 09/04/27 31617.000 1.893 7.24 52.884 -132.400 - GSC09/05/02 14346.000 1.902 7.28 52.885 -132.407 - GSC09/05/12 62395.000 2.006 7.74 52.883 -132.406 - GSC09/06/25 75719.775 1.649 6.27 - - - MF088 09/06/05 1112.000 2.173 8.53 53.241 -132.893 - GSC09/06/05 2301.625 1.054 4.42 - - - MF09/06/05 2984.800 1.148 4.67 - - - MF09/06/05 3108.000 1.563 5.96 53.278 -132.823 - GSC09/06/05 33198.050 1.084 4.5 - - - MF09/06/05 34000.700 1.153 4.69 - - - MF09/06/05 40213.000 1.546 5.91 53.262 -132.876 - GSC09/06/05 60212.775 2.173 8.53 - - - MF09/06/05 61401.625 1.054 4.42 - - - MF09/06/05 62084.800 1.148 4.67 - - - MF09/06/05 62208.000 1.563 5.96 - - - MF09/06/07 67768.375 1.285 5.07 - - - MF09/06/10 8441.900 0.892 4.02 - - - MF68Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod089 09/06/05 1545.100 1.623 6.18 - - - MF09/06/05 1685.050 1.752 6.66 - - - MF09/06/05 1963.000 2.023 7.81 53.240 -132.869 - GSC09/06/05 3668.800 1.512 5.79 - - - MF09/06/05 29664.000 1.839 7.01 53.268 -132.844 - GSC09/06/05 60645.100 1.623 6.18 - - - MF09/06/05 60785.050 1.752 6.66 - - - MF09/06/05 61063.100 2.023 7.81 - - - MF09/06/05 62768.800 1.512 5.79 - - - MF090 09/06/05 32781.000 1.648 6.27 53.264 -132.856 - GSC09/06/05 32990.000 1.958 7.52 53.273 -132.846 - GSC09/06/05 35207.000 1.709 6.5 53.262 -132.865 - GSC09/06/05 50350.825 1.346 5.25 - - - MF09/06/05 55881.000 1.950 7.49 53.279 -132.832 - GSC09/07/02 38491.975 1.281 5.05 - - - MF091 09/06/05 47374.000 1.809 6.89 53.280 -132.819 - GSC09/06/05 47784.000 1.326 5.19 53.263 -132.864 - GSC09/06/05 53158.000 1.753 6.67 53.221 -132.843 - GSC092 09/06/12 33929.000 1.769 6.73 52.452 -131.849 - GSC09/06/12 34120.000 2.096 8.16 52.437 -131.874 - GSC09/06/12 34762.000 1.821 6.94 52.458 -131.835 - GSC09/06/12 35769.000 2.373 9.6 52.459 -131.818 - GSC69Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod093 09/08/13 44465.000 1.249 4.96 52.898 -132.548 - GSC09/08/13 60729.000 2.307 9.23 52.892 -132.571 - GSC09/08/13 61188.000 1.724 6.56 52.896 -132.557 - GSC094 09/08/13 60902.000 1.584 6.04 52.900 -132.544 - GSC09/08/13 64371.000 1.574 6 52.907 -132.535 - GSC095 09/08/11 60241.000 1.357 5.28 53.158 -132.523 - GSC09/08/11 67960.550 0.574 3.34 - - - MF09/08/14 47621.075 0.869 3.97 - - - MF09/08/15 54617.000 1.715 6.52 53.132 -132.598 - GSC09/08/15 74186.000 1.355 5.28 53.122 -132.631 - GSC09/08/16 44799.350 0.482 3.16 - - - MF09/08/19 62491.000 1.390 5.39 53.163 -132.643 - GSC096 09/02/05 52777.000 2.165 8.49 52.861 -132.419 - GSC09/09/02 11035.000 2.045 7.91 52.847 -132.389 - GSC09/09/02 39239.425 1.627 6.19 - - - MF097 09/10/08 30262.000 2.179 8.56 53.300 -133.085 - GSC09/10/08 34186.000 1.455 5.6 53.298 -133.084 - GSC098 09/10/09 68423.000 2.017 7.79 52.709 -132.265 - GSC09/10/09 72093.000 1.801 6.86 52.706 -132.282 - GSC70Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod099 09/10/27 14905.000 1.970 7.57 53.008 -132.657 - GSC09/10/30 1873.000 2.106 8.2 52.996 -132.667 - GSC09/11/02 62097.000 1.450 5.58 53.005 -132.676 - GSC13/07/25 8826.600 1.153 4.69 - - - MF15/03/01 31491.250 1.374 5.34 - - - MF15/03/02 43975.850 1.390 5.39 - - - MF100 09/11/01 50063.375 1.822 6.94 - - - MF09/11/01 50186.000 2.132 8.33 53.302 -133.211 - GSC09/11/01 50705.100 1.658 6.31 - - - MF09/11/01 53558.000 1.909 7.31 53.297 -133.260 - GSC09/11/01 55626.050 1.849 7.05 - - - MF101 05/11/04 13185.000 1.594 6.07 53.008 -132.648 - GSC09/12/06 59539.000 2.033 7.86 53.008 -132.635 - GSC14/03/30 30288.025 1.952 7.49 - - - MF14/05/29 34695.100 1.290 5.08 - - - MF102 09/12/13 14554.000 1.433 5.53 52.877 -132.506 - GSC09/12/13 27477.000 1.800 6.85 52.869 -132.514 - GSC09/12/13 33198.725 1.320 5.17 - - - MF09/12/21 35560.750 1.653 6.29 - - - MF103 09/12/13 12583.000 1.288 5.08 52.544 -131.972 - GSC09/12/13 14890.000 1.382 5.36 52.541 -131.995 - GSC09/12/13 17161.000 1.609 6.13 52.544 -131.987 - GSC71Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod104 09/12/17 16940.000 2.172 8.53 52.707 -132.431 - GSC09/12/17 28933.000 2.013 7.77 52.710 -132.448 - GSC09/12/17 63374.150 2.172 8.53 - - - MF09/12/17 75367.000 2.013 7.77 - - - MF105 09/12/28 48377.000 1.881 7.19 53.113 -132.840 - GSC09/12/28 49340.000 1.963 7.54 53.108 -132.830 - GSC106 05/06/26 27367.000 1.611 6.14 52.620 -132.119 - GSC10/01/15 66627.000 1.948 7.48 52.611 -132.149 - GSC107 08/04/03 10778.000 1.905 7.29 53.528 -133.123 - GSC10/02/09 76406.000 1.785 6.79 53.561 -132.991 - GSC108 10/02/22 36564.000 1.912 7.32 53.215 -132.995 - GSC10/02/22 39608.000 1.608 6.12 53.243 -132.961 - GSC109 10/03/12 82203.000 2.397 9.73 52.583 -132.057 - GSC12/12/03 67616.000 2.124 8.29 52.580 -132.068 16.2 GSC15/04/30 46945.675 1.740 6.62 - - - MF110 10/03/14 24499.000 2.263 9 52.500 -131.869 - GSC10/03/14 24594.000 2.092 8.14 52.523 -131.838 - GSC111 10/02/19 42043.000 1.693 6.44 53.432 -132.894 - GSC10/04/13 57305.000 2.058 7.98 53.480 -132.895 - GSC112 10/04/21 59327.000 2.078 8.07 52.855 -132.194 - GSC10/04/21 77856.000 1.714 6.52 52.800 -132.312 - GSC113 08/02/07 69412.000 2.259 8.97 52.616 -132.104 - GSC10/05/02 15856.000 2.160 8.47 52.673 -132.021 - GSC10/05/21 78589.000 2.109 8.22 52.608 -132.127 - GSC72Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod114 10/06/05 17114.000 1.690 6.43 52.728 -132.309 - GSC10/06/05 17199.000 2.446 10.02 52.727 -132.313 - GSC115 10/06/07 26048.000 1.671 6.36 52.826 -132.441 - GSC10/09/21 36456.000 1.647 6.27 52.825 -132.443 - GSC116 10/06/09 12307.000 1.765 6.71 53.222 -132.786 - GSC10/06/09 22799.000 1.340 5.23 53.229 -132.780 - GSC10/06/10 18015.000 1.670 6.35 53.271 -132.594 - GSC117 08/10/21 17322.000 2.150 8.42 52.849 -132.408 - GSC08/10/21 42347.775 2.150 8.42 - - - MF10/06/22 21667.000 1.854 7.07 52.837 -132.445 - GSC118 10/07/21 30738.000 1.923 7.37 52.429 -131.750 - GSC10/07/21 31047.000 1.836 7 52.414 -131.786 - GSC119 10/07/30 46938.000 1.773 6.75 52.459 -131.871 - GSC10/07/30 50894.000 1.895 7.25 52.467 -131.871 - GSC10/07/30 68192.000 1.516 5.8 52.473 -131.763 - GSC10/07/30 86036.000 1.804 6.87 52.466 -131.856 - GSC10/07/31 5607.000 1.952 7.5 52.463 -131.824 - GSC120 05/06/18 50326.800 1.061 4.44 - - - MF10/08/15 7079.000 1.375 5.34 53.287 -133.016 - GSC10/08/20 29039.025 1.022 4.34 - - - MF11/04/21 13584.000 1.774 6.75 53.359 -132.989 - GSC121 07/09/18 52351.000 2.042 7.9 53.546 -133.108 - GSC10/08/17 33675.000 2.079 8.07 53.449 -133.269 - GSC13/10/24 77047.150 2.192 8.63 - - - MF73Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod122 10/06/17 83585.000 1.810 6.89 52.807 -132.071 - GSC10/08/19 14092.000 1.881 7.19 52.732 -132.250 - GSC123 06/06/01 48500.000 2.148 8.41 52.931 -132.525 - GSC07/07/28 32459.000 1.553 5.93 - - - MF10/08/27 11571.000 2.330 9.36 52.904 -132.606 - GSC124 10/09/13 47777.000 2.931 13.31 52.844 -132.516 - GSC10/09/13 48471.000 1.156 4.7 52.845 -132.513 - GSC125 10/09/14 12905.000 1.649 6.27 53.556 -133.153 - GSC10/09/14 13152.000 1.546 5.91 53.569 -133.105 - GSC126 10/09/13 55428.000 1.300 5.11 53.082 -132.685 - GSC10/09/14 48695.000 1.760 6.7 53.076 -132.684 - GSC13/08/09 80857.775 1.133 4.63 - - - MF127 10/09/17 81295.000 1.141 4.66 53.075 -132.653 - GSC13/06/27 6143.575 0.900 4.04 - - - MF13/06/27 23483.575 0.900 4.04 - - - MF13/07/10 43128.000 2.167 8.5 53.068 -132.673 22.4 GSC13/07/10 44286.150 1.005 4.3 - - - MF13/07/13 67471.800 0.817 3.85 - - - MF13/08/28 29042.425 1.777 6.76 - - - MF128 06/11/24 61655.000 2.196 8.65 53.874 -133.268 - GSC10/09/20 28543.000 2.014 7.77 53.882 -133.449 - GSC13/01/13 71200.800 1.475 5.66 - - - MF14/06/08 81067.275 2.034 7.86 - - - MF74Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod129 10/09/20 73231.000 1.872 7.15 53.250 -133.055 - GSC10/09/20 73759.000 1.817 6.92 53.248 -133.051 - GSC10/09/21 38896.475 1.353 5.27 - - - MF130 10/09/22 14992.000 1.488 5.71 53.274 -132.843 - GSC10/09/22 49092.000 1.530 5.85 53.284 -132.827 - GSC131 10/10/28 25319.000 1.867 7.13 52.604 -131.987 - GSC10/10/28 40326.000 2.244 8.9 52.601 -132.000 - GSC132 10/12/02 65705.000 1.818 6.93 53.309 -132.720 - GSC10/12/04 28107.000 1.718 6.53 53.265 -132.846 - GSC133 07/06/17 41414.000 1.999 7.71 52.642 -132.171 - GSC11/01/25 44676.000 2.076 8.06 52.614 -132.147 - GSC134 09/08/09 75005.125 0.862 3.95 - - - MF09/08/11 11672.300 0.956 4.18 - - - MF09/08/13 18423.375 0.800 3.81 - - - MF09/08/19 63343.600 1.356 5.28 - - - MF09/08/21 78080.000 1.783 6.79 53.149 -132.634 - GSC09/10/14 32586.725 1.705 6.48 - - - MF09/11/25 39328.800 1.018 4.33 - - - MF10/07/13 59248.725 1.302 5.12 - - - MF10/07/27 79697.375 1.335 5.22 - - - MF10/08/16 55482.600 0.988 4.26 - - - MF10/08/26 84030.025 1.145 4.67 - - - MF11/03/11 58716.000 1.812 6.9 53.179 -132.621 - GSC11/03/14 962.000 1.821 6.94 53.185 -132.512 - GSC75Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod135 11/03/15 78185.000 2.516 10.44 53.335 -133.176 - GSC11/03/15 80338.000 1.572 6 53.335 -133.186 - GSC136 11/03/29 82856.000 1.811 6.9 52.851 -132.569 - GSC11/03/30 18494.000 1.890 7.23 52.849 -132.574 - GSC137 06/08/03 45850.275 1.172 4.74 - - - MF11/03/30 69416.000 2.653 11.31 53.189 -132.987 - GSC11/03/30 69617.000 1.569 5.99 53.196 -132.979 - GSC138 11/04/15 47426.000 1.490 5.71 53.123 -132.625 - GSC11/04/21 67655.000 1.803 6.87 53.130 -132.599 - GSC11/04/23 38537.175 1.529 5.85 - - - MF139 11/05/22 41069.000 1.754 6.67 52.904 -132.565 16.7 GSC11/05/23 50684.000 1.435 5.53 52.921 -132.552 - GSC11/05/27 76679.000 1.530 5.85 52.915 -132.562 - GSC140 11/05/12 15040.750 1.085 4.5 - - - MF11/05/12 60154.750 1.085 4.5 - - - MF11/06/06 59318.000 1.271 5.02 53.167 -132.609 - GSC11/06/07 53273.800 0.809 3.83 - - - MF11/09/07 67313.000 2.306 9.23 53.155 -132.506 - GSC11/09/08 73858.150 0.995 4.27 - - - MF141 11/06/09 42531.000 1.672 6.36 52.828 -132.378 - GSC11/06/09 43116.000 1.711 6.51 52.819 -132.419 - GSC13/10/23 61371.500 2.492 10.29 - - - MF76Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod142 11/06/09 42859.000 1.491 5.72 52.832 -132.401 - GSC11/06/09 43371.000 1.343 5.24 52.820 -132.444 - GSC11/06/09 43773.000 1.907 7.3 52.826 -132.409 - GSC11/06/09 44026.000 1.560 5.95 52.842 -132.401 - GSC11/06/09 44322.000 1.588 6.05 52.842 -132.390 - GSC11/06/09 44538.000 1.724 6.56 52.821 -132.406 - GSC11/06/09 45446.000 2.376 9.61 52.816 -132.417 - GSC11/06/09 47002.000 1.540 5.88 52.820 -132.398 - GSC11/06/09 47767.000 1.414 5.47 52.843 -132.393 - GSC11/06/09 48094.000 1.360 5.3 52.833 -132.404 - GSC11/06/09 53094.000 1.362 5.3 52.839 -132.405 - GSC11/06/09 57714.000 2.001 7.71 52.834 -132.393 - GSC143 11/07/19 40277.000 2.043 7.91 53.250 -133.044 - GSC11/07/19 41389.075 1.073 4.47 - - - MF11/07/27 16169.000 1.642 6.25 53.258 -133.053 - GSC144 11/08/12 36880.000 1.770 6.73 53.532 -133.224 - GSC11/08/12 39112.000 1.726 6.56 53.535 -133.167 - GSC145 11/08/24 27885.000 2.351 9.47 53.233 -133.058 - GSC12/12/25 19292.000 2.999 13.86 53.241 -133.029 21.8 GSC146 11/08/31 41248.000 1.782 6.78 52.595 -132.065 - GSC11/08/31 47928.000 1.578 6.02 52.629 -131.997 - GSC147 06/07/12 81325.000 1.949 7.48 53.393 -133.240 - GSC11/09/16 81465.000 1.869 7.14 53.526 -133.114 - GSC77Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod148 11/11/10 44205.000 3.198 15.58 53.112 -132.762 - GSC11/11/11 57922.000 2.547 10.63 53.104 -132.798 - GSC15/07/13 60803.225 1.929 7.39 - - - MF149 06/02/22 13686.000 1.958 7.52 52.712 -132.263 - GSC11/12/21 30892.000 1.860 7.1 52.757 -132.238 - GSC150 12/01/10 59648.000 1.764 6.71 52.909 -132.495 - GSC12/01/12 12015.175 1.839 7.02 - - - MF12/01/12 36086.000 2.258 8.97 52.896 -132.540 - GSC151 12/08/08 14773.000 1.342 5.24 53.229 -132.797 - GSC12/08/08 18997.400 1.206 4.84 - - - MF12/08/08 28217.000 2.167 8.5 53.175 -132.733 - GSC12/08/08 31126.350 1.179 4.76 - - - MF12/08/08 62394.325 0.688 3.57 - - - MF14/10/26 62916.250 0.888 4.01 - - - MF152 12/09/05 28934.000 2.028 7.84 53.337 -133.009 - GSC12/09/05 29138.000 1.889 7.22 53.326 -132.971 - GSC12/09/05 29268.250 1.513 5.79 - - - MF12/09/05 30347.250 1.646 6.26 - - - MF153 12/10/21 3432.000 2.932 13.33 53.344 -132.258 - GSC12/10/21 46082.000 2.362 9.53 53.262 -132.942 26.7 GSC154 12/10/29 76631.000 3.434 17.9 52.564 -131.978 2.0 GSC13/01/12 27292.000 3.407 17.61 52.543 -131.917 10.2 GSC155 12/10/31 26095.000 2.097 8.16 52.894 -132.433 20.0 GSC12/11/05 84316.175 2.437 9.96 - - - MF78Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod156 12/10/29 38138.775 2.069 8.03 - - - MF12/10/31 82035.000 2.167 8.5 52.873 -132.480 15.0 GSC12/11/10 71620.075 2.193 8.63 - - - MF157 12/11/01 51168.000 2.726 11.81 52.478 -131.932 7.0 GSC12/11/04 85097.225 2.576 10.81 - - - MF12/11/16 45273.000 2.595 10.93 52.529 -131.937 5.0 GSC12/12/06 42500.700 2.560 10.71 - - - MF158 12/11/11 65316.000 2.545 10.62 52.533 -131.910 7.7 GSC12/11/16 82755.000 2.346 9.44 52.513 -131.974 2.0 GSC159 12/11/16 11102.000 2.345 9.44 52.587 -132.027 18.6 GSC13/10/05 45317.100 2.312 9.26 - - - MF15/03/16 1097.000 2.401 9.75 52.577 -132.063 17.9 GSC160 12/11/16 55238.000 2.645 11.26 52.545 -131.900 19.0 GSC12/12/12 29430.500 2.573 10.79 - - - MF13/01/11 38249.650 2.360 9.52 - - - MF14/02/13 41950.000 2.727 11.81 52.535 -131.912 11.5 GSC161 12/11/11 65119.100 2.112 8.23 - - - MF12/11/16 84939.000 2.141 8.38 52.628 -131.831 5.0 GSC12/11/23 1390.000 2.411 9.82 52.591 -131.832 4.2 GSC13/02/23 36117.200 2.605 11 - - - MF162 12/10/31 28469.000 2.439 9.97 52.868 -131.891 2.0 GSC12/11/17 26397.000 2.359 9.52 52.741 -132.302 1.0 GSC163 12/11/21 12585.000 2.123 8.29 52.478 -131.932 6.0 GSC12/11/21 14243.000 2.093 8.14 52.545 -131.900 8.0 GSC79Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod164 12/11/27 35447.000 2.310 9.25 53.115 -132.302 2.0 GSC12/11/27 36334.000 2.276 9.07 52.830 -132.465 1.0 GSC165 12/11/28 71041.000 2.468 10.15 52.801 -132.256 19.4 GSC13/01/02 72427.325 2.508 10.39 - - - MF13/02/20 86267.700 2.505 10.37 - - - MF13/08/13 13334.000 2.574 10.8 52.790 -132.243 21.5 GSC15/02/14 15519.000 2.230 8.82 52.788 -132.249 20.0 GSC166 12/11/30 18965.000 2.776 12.16 52.588 -132.020 5.2 GSC12/12/05 52698.000 2.835 12.59 52.581 -131.985 6.0 GSC167 12/12/02 27255.000 2.405 9.78 52.631 -131.946 21.0 GSC13/06/23 35293.175 1.610 6.13 - - - MF15/03/30 8004.000 2.213 8.74 52.582 -132.054 15.6 GSC168 12/12/06 29565.000 2.179 8.57 52.731 -132.399 11.3 GSC12/12/07 3706.000 2.354 9.49 52.760 -132.381 30.0 GSC169 12/11/02 46874.600 2.581 10.84 - - - MF12/11/11 10458.900 1.732 6.59 - - - MF12/11/17 321.000 2.404 9.78 52.846 -132.429 12.0 GSC12/12/08 79.000 2.254 8.95 52.997 -132.328 2.0 GSC170 12/12/06 23761.000 2.507 10.38 52.865 -132.507 23.0 GSC12/12/08 668.000 2.132 8.33 52.881 -132.471 20.0 GSC13/10/18 37489.000 3.080 14.54 - - - MF13/10/18 38668.775 1.962 7.54 - - - MF15/12/15 24167.350 1.946 7.47 - - - MF80Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod171 12/12/19 13353.000 2.791 12.27 52.809 -132.320 22.1 GSC15/07/07 55909.000 2.155 8.44 52.776 -132.349 18.5 GSC172 12/12/22 31396.000 2.159 8.47 52.847 -132.504 14.8 GSC12/12/22 35765.000 2.086 8.11 52.852 -132.471 14.5 GSC13/09/17 22878.825 1.680 6.39 - - - MF14/03/03 12973.850 2.463 10.12 - - - MF173 12/11/26 17401.000 2.323 9.32 52.476 -131.893 0.0 GSC12/12/22 78338.000 2.366 9.56 52.498 -131.874 - GSC13/01/23 25491.025 2.039 7.89 - - - MF13/02/01 52885.675 1.892 7.23 - - - MF13/08/12 65827.000 1.672 6.36 - - - MF15/05/29 628.000 1.885 7.21 - - - MF174 12/11/27 22745.000 2.629 11.16 52.422 -131.858 0.1 GSC13/07/25 57865.000 2.612 11.04 52.468 -131.833 8.7 GSC175 13/07/30 31694.000 2.713 11.72 52.768 -132.233 21.4 GSC14/09/18 49631.000 2.545 10.62 52.812 -132.150 23.3 GSC176 13/04/18 6196.000 2.662 11.37 52.519 -131.939 9.1 GSC13/11/11 69951.000 2.796 12.3 52.502 -131.944 11.0 GSC177 12/10/30 62753.000 2.787 12.23 52.596 -131.905 2.0 GSC12/11/13 38056.525 2.798 12.32 - - - MF13/01/11 30432.825 2.793 12.28 - - - MF14/02/13 15552.000 2.797 12.31 52.509 -131.911 8.5 GSC81Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod178 12/12/23 40951.000 2.114 8.24 52.842 -132.390 20.0 GSC14/03/21 7524.000 2.258 8.97 52.854 -132.366 20.7 GSC14/05/02 6650.850 1.460 5.61 - - - MF15/10/24 55417.100 1.788 6.81 - - - MF179 14/05/17 83126.425 2.064 8 - - - MF14/05/17 83296.675 1.583 6.04 - - - MF14/05/18 2114.000 2.237 8.86 53.288 -133.033 26.3 GSC15/12/30 8017.575 1.738 6.61 - - - MF15/12/30 27717.000 3.100 14.7 53.290 -133.046 21.5 GSC180 08/04/22 39459.000 2.024 7.82 53.828 -133.434 - GSC14/07/02 38314.000 2.172 8.53 53.825 -133.397 - GSC181 13/07/31 38368.750 2.523 10.48 - - - MF14/08/08 47183.000 1.971 7.58 52.895 -132.248 24.0 GSC15/02/10 10125.000 1.576 6.01 52.842 -132.377 15.1 GSC182 13/05/17 27711.275 2.371 9.59 - - - MF13/10/27 50162.725 2.031 7.85 - - - MF14/08/13 63792.000 1.978 7.61 52.806 -132.181 22.9 GSC14/10/16 38503.050 1.639 6.24 - - - MF15/09/30 45828.775 1.413 5.46 - - - MF15/10/03 79413.000 2.013 7.77 52.769 -132.244 21.1 GSC82Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod183 13/10/04 6723.050 2.894 13.03 - - - MF14/08/20 66652.000 2.260 8.98 52.775 -132.238 22.4 GSC14/12/10 79843.525 2.121 8.27 - - - MF15/01/21 250.525 2.570 10.77 - - - MF15/12/11 15499.000 2.314 9.27 52.746 -132.303 15.7 GSC184 09/04/15 45028.000 1.453 5.59 53.511 -133.185 - GSC14/09/09 71996.000 2.187 8.6 53.516 -133.159 24.2 GSC185 13/07/03 1073.150 1.879 7.18 - - - MF14/09/23 15841.000 2.097 8.16 52.620 -132.017 20.1 GSC186 13/02/27 84268.175 2.460 10.1 - - - MF13/09/28 39994.475 1.864 7.12 - - - MF14/04/16 16906.825 1.653 6.29 - - - MF14/06/14 25774.900 1.361 5.3 - - - MF14/10/19 21996.000 2.127 8.31 52.740 -132.312 16.0 GSC15/12/14 42974.000 1.647 6.27 52.739 -132.328 13.4 GSC187 12/11/16 62449.000 2.448 10.03 52.776 -132.344 1.0 GSC14/12/17 14108.000 2.369 9.57 52.771 -132.293 22.0 GSC188 13/03/13 32041.450 1.992 7.67 - - - MF13/11/22 14121.000 2.253 8.94 52.680 -132.133 18.2 GSC15/01/23 3289.000 2.243 8.89 52.667 -132.143 18.2 GSC189 14/10/05 12292.000 2.526 10.5 52.867 -132.469 16.8 GSC15/01/23 13348.000 2.349 9.46 52.849 -132.523 13.1 GSC83Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod190 12/11/04 2100.575 2.533 10.54 - - - MF12/11/11 74810.700 2.043 7.91 - - - MF12/11/17 40356.100 2.581 10.85 - - - MF12/12/25 27032.075 2.596 10.94 - - - MF13/01/28 73757.525 2.432 9.94 - - - MF15/01/15 29010.000 2.730 11.83 52.572 -131.962 10.7 GSC15/01/26 23913.000 2.486 10.25 52.582 -131.942 15.0 GSC191 12/10/31 28153.000 2.915 13.19 52.599 -132.020 6.0 GSC13/01/11 22672.225 2.941 13.39 - - - MF15/01/26 31057.000 2.914 13.18 52.576 -131.875 13.9 GSC192 12/11/25 67580.000 2.536 10.56 52.660 -131.757 17.0 GSC13/04/12 78043.125 2.609 11.02 - - - MF13/07/30 53000.525 2.244 8.89 - - - MF14/02/12 23755.375 2.318 9.29 - - - MF15/01/26 67245.000 2.405 9.78 52.546 -131.868 15.0 GSC193 12/11/01 77207.000 2.628 11.15 52.564 -131.978 2.0 GSC13/01/11 20060.625 2.276 9.07 - - - MF15/01/28 14422.000 2.698 11.61 52.569 -131.869 16.2 GSC194 15/02/21 15986.000 1.722 6.55 53.094 -132.725 23.8 GSC15/02/21 49121.000 1.466 5.63 53.064 -132.740 17.8 GSC15/02/21 49354.000 2.044 7.91 53.098 -132.680 25.9 GSC195 15/02/03 1800.000 1.552 5.92 52.818 -132.411 15.6 GSC15/02/24 27073.000 2.244 8.89 52.820 -132.378 22.2 GSC84Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod196 09/05/06 47626.000 1.387 5.38 53.308 -133.065 - GSC15/03/03 7936.000 2.047 7.93 53.320 -133.075 13.8 GSC197 13/07/23 32677.975 1.274 5.03 - - - MF13/09/01 10073.150 1.682 6.4 - - - MF13/11/25 43723.550 1.450 5.58 - - - MF13/11/25 44815.900 2.006 7.74 - - - MF14/02/10 10074.675 1.182 4.77 - - - MF14/06/17 25145.000 2.451 10.05 52.864 -132.450 21.8 GSC15/03/14 18582.000 1.719 6.54 52.861 -132.471 14.7 GSC15/11/10 39105.775 1.934 7.42 - - - MF198 09/12/05 39856.000 2.289 9.14 53.379 -133.153 - GSC15/03/17 4584.000 2.082 8.09 53.376 -133.179 16.6 GSC199 15/01/27 10139.275 1.631 6.21 - - - MF15/03/13 11031.000 2.023 7.81 52.744 -132.285 13.3 GSC15/03/24 74446.000 2.088 8.12 52.751 -132.272 19.6 GSC200 13/04/30 47537.550 1.897 7.26 - - - MF13/10/10 4641.675 1.940 7.44 - - - MF13/12/06 60003.475 2.162 8.48 - - - MF14/06/27 31969.000 2.074 8.05 52.750 -132.281 16.9 GSC14/07/16 33893.150 2.003 7.72 - - - MF14/07/16 36185.225 1.445 5.56 - - - MF15/03/24 84700.000 1.990 7.67 52.742 -132.311 14.7 GSC15/11/02 38728.000 1.988 7.65 52.757 -132.263 20.6 GSC15/11/09 84785.350 1.362 5.3 - - - MF85Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod201 12/11/19 82753.000 1.927 7.39 52.775 -132.354 20.9 GSC15/04/04 51839.000 2.105 8.2 52.780 -132.336 19.2 GSC202 15/04/05 69585.000 1.741 6.62 52.469 -131.858 15.3 GSC15/04/05 75294.000 2.654 11.32 52.473 -131.829 18.7 GSC203 15/03/17 43881.000 1.970 7.57 52.962 -132.626 16.6 GSC15/04/16 51616.325 1.443 5.56 - - - MF15/04/16 81145.000 2.116 8.25 52.965 -132.636 17.5 GSC15/04/16 81885.850 1.246 4.95 - - - MF15/04/16 82399.650 1.375 5.34 - - - MF15/04/16 84862.200 1.294 5.09 - - - MF15/04/17 11242.000 2.247 8.91 52.988 -132.576 25.2 GSC15/04/17 37790.200 1.184 4.78 - - - MF15/04/17 68526.550 1.016 4.33 - - - MF15/04/19 4666.450 0.997 4.28 - - - MF15/04/20 6200.225 1.256 4.98 - - - MF15/04/20 27594.000 1.207 4.84 - - - MF15/04/20 59155.600 1.653 6.29 - - - MF15/04/22 46860.900 0.935 4.13 - - - MF15/05/12 22550.375 0.876 3.98 - - - MF204 15/05/23 42477.000 1.759 6.69 52.705 -132.327 13.7 GSC15/06/22 28103.000 1.727 6.57 52.714 -132.281 21.5 GSC15/07/02 5277.025 1.100 4.54 - - - MF15/08/09 61967.000 1.815 6.92 52.715 -132.301 18.7 GSC86Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod205 14/04/04 77222.000 2.028 7.84 52.776 -132.246 21.5 GSC15/05/24 71018.000 1.406 5.44 - - - MF15/06/08 79245.000 1.750 6.66 52.752 -132.286 17.2 GSC206 14/08/04 30712.000 1.920 7.35 52.812 -132.259 23.1 GSC15/06/22 79783.000 2.046 7.92 52.767 -132.339 16.9 GSC207 13/07/16 45174.700 2.183 8.58 - - - MF13/09/19 31817.000 2.279 9.08 52.785 -132.316 16.4 GSC15/06/26 56470.000 2.058 7.98 52.791 -132.312 17.7 GSC208 13/06/06 33949.825 1.969 7.57 - - - MF13/06/19 39716.050 1.365 5.31 - - - MF13/11/07 75968.475 2.048 7.93 - - - MF14/09/03 3868.975 1.635 6.22 - - - MF15/07/10 59602.000 1.584 6.04 52.479 -131.933 15.8 GSC15/07/23 34530.000 1.781 6.78 52.507 -131.862 19.5 GSC15/07/23 34882.675 1.281 5.05 - - - MF209 15/07/12 2939.000 2.027 7.83 52.473 -131.825 13.9 GSC15/07/12 5111.575 1.594 6.07 - - - MF15/07/12 11822.100 1.574 6 - - - MF15/07/12 12290.000 2.086 8.11 52.483 -131.823 14.2 GSC210 15/08/01 13779.000 2.446 10.02 52.692 -132.292 25.9 GSC15/08/03 64999.000 1.696 6.45 52.705 -132.286 15.8 GSC211 05/06/19 29605.000 1.537 5.87 53.280 -133.154 - GSC15/08/06 6102.000 2.038 7.88 53.293 -133.141 18.9 GSC87Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod212 15/07/14 17323.000 2.054 7.96 52.519 -131.902 10.5 GSC15/08/14 15911.000 1.993 7.68 52.600 -131.793 13.4 GSC15/09/06 61717.075 1.644 6.26 - - - MF213 10/01/24 41380.000 1.955 7.51 52.747 -132.276 - GSC13/01/17 6863.475 2.020 7.8 - - - MF13/10/10 54376.925 1.775 6.76 - - - MF15/09/14 3391.000 2.068 8.02 52.739 -132.300 16.3 GSC214 08/07/22 711.000 2.443 10 53.499 -133.227 - GSC15/09/15 24402.000 1.992 7.67 53.520 -133.210 10.7 GSC215 13/06/11 19567.050 1.867 7.13 - - - MF15/09/18 7259.000 2.145 8.4 52.594 -132.012 6.9 GSC15/09/18 7630.000 2.018 7.79 52.614 -131.963 10.5 GSC216 15/09/15 11164.000 1.651 6.28 52.573 -131.986 7.2 GSC15/09/18 51721.000 2.186 8.6 52.561 -132.021 7.8 GSC217 15/11/18 47603.000 2.237 8.86 52.749 -132.297 18.1 GSC15/11/19 65640.000 2.461 10.11 52.751 -132.299 21.1 GSC218 12/11/27 1783.000 1.969 7.57 52.739 -132.270 19.2 GSC12/12/01 59142.275 1.968 7.57 - - - MF13/04/21 55768.150 1.745 6.64 - - - MF15/12/22 44894.000 2.391 9.7 52.772 -132.161 23.4 GSC219 15/12/25 37622.000 2.439 9.97 53.176 -132.805 21.8 GSC15/12/25 37767.000 2.145 8.39 53.153 -132.845 18.6 GSC88Appendix A. Definition of all Repeating Earthquake FamiliesTable 5: (continued)FamilyIDDate(YY/MM/DD)Time(seconds)Magnitude(Mw)Slip(cm)Latitude(dec. deg.)Longitude(dec. deg.)Depth(km)DetectionMethod220 15/12/28 74094.000 2.218 8.76 53.068 -132.740 24.0 GSC15/12/29 17614.000 1.863 7.11 53.051 -132.824 15.9 GSC15/12/29 19531.000 1.779 6.77 53.047 -132.787 14.2 GSC221 15/12/29 41033.000 2.279 9.08 53.253 -133.135 10.9 GSC15/12/29 51744.000 2.352 9.48 53.289 -133.087 18.7 GSC15/12/30 8659.500 1.884 7.2 - - - MF222 15/12/30 4878.425 2.465 10.13 - - - MF15/12/30 5093.000 2.454 10.07 53.272 -133.073 25.1 GSC15/12/30 7377.000 2.261 8.99 53.284 -133.078 20.3 GSC15/12/30 13420.000 2.599 10.96 53.282 -133.083 20.1 GSC223 14/05/18 6438.175 1.931 7.4 - - - MF14/05/18 14768.525 2.206 8.7 - - - MF14/05/18 15699.000 1.442 5.56 - - - MF14/05/18 15810.650 1.823 6.95 - - - MF14/05/18 16291.200 1.782 6.78 - - - MF15/12/30 7762.300 1.758 6.69 - - - MF15/12/30 12490.625 1.845 7.04 - - - MF15/12/30 16165.000 2.155 8.45 53.271 -133.062 22.2 GSC15/12/30 20171.000 2.242 8.89 53.267 -133.083 18.1 GSC15/12/30 29911.275 1.869 7.14 - - - MF15/12/31 55520.725 1.894 7.24 - - - MF224 15/12/30 23759.000 1.806 6.88 53.258 -133.098 17.8 GSC15/12/30 24389.000 2.051 7.94 53.338 -132.879 28.1 GSC15/12/30 28964.000 3.100 14.71 53.254 -133.119 22.8 GSC89Appendix B. Constraining Repeating Earthquake MagnitudesAppendix BConstraining Repeating EarthquakeMagnitudesIn Section 3.1.2, we show how to arrive at a system of equations that honours all relative M0weights between repeating earthquakes in a given family. For a family with n earthquakes, the systemis written as 1 −w12 0 0 · · · 01 0 −w13 0 · · · 0............. . ....1 0 0 0 · · · −w1nM01M02M03...M0n=00...0, (16)where M0i is the absolute M0 of event i, and w1i is the M0 ratio of event 1 relative to event i. This system(16) enforces precise relative M0 between all members of a repeating earthquake family.To constrain absolute M0, we employ the original MW and M0 estimates from Section 3.1.1.Consider a three-earthquake family (A, B, and C) with catalogued absolute moment estimates Mcat0A ,Mcat0B , and Mcat0C . Initially, we consider the following system that utilizes the initial Mcat0 estimates, :1 0 00 1 00 0 1M0AM0BM0C=Mcat0AMcat0BMcat0C . (17)90Appendix B. Constraining Repeating Earthquake MagnitudesThis system can be rearranged to resemble that of (16) through decomposition into its relative andabsolute M0 information:1 −wAB 01 0 −wACwABwAC wAC wABM0AM0BM0C=00wABwACMcat0A +wACMcat0B +wABMcat0C , (18)which can be written generally for a family with n earthquakes as1 −w12 0 0 · · · 01 0 −w13 0 · · · 0............. . ....1 0 0 0 · · · −w1n∏j 6=1w1 j ∏j 6=2w1 j ∏j 6=3w1 j ∏j 6=4w1 j · · · ∏j 6=nw1 jM01M02M03...M0n=00...0∑i(Mcat0i ∏j 6=iw1 j), (19)Gm = d . (20)Note that w11 = 1, and that the weights wi j in (19) will not generally be the same as those in (16)because the absolute Mcat0 information will not be completely consistent with the relative M0 informationsupplied through the SVD analysis. Because we expect the the relative information from the SVDanalysis to be more robust than that from the catalogue, we adopt the weight equations from (16), anddiscard those from (19). In (19,20), G has size [n×n] and rank n, which allows us to solve for the truemoments (m) directly:m = G−1d . (21)We confirm that (21) produces stable and accurate results for m by checking that the condition numberof G is sufficiently low for all repeating earthquake families.The equation we use to constrain absolute M0 (nth row of G) is not the only available option.The simplest option is to consider a single Mcat0i (e.g., the largest) and define all absolute M0 within thefamily based on this one value and the weights in (16). This simple method relies strongly on a singleMcat0i value, discards other useful Mcat0 j estimates, and is therefore prone to errors in measurement. A91Appendix B. Constraining Repeating Earthquake Magnitudesbetter method, which utilizes all Mcat0i values, is to constrain the sum of M0 in the solution to be equal tothe sum of Mcat0i . If using this method, the nth row of the system in (19) is defined as[1 1 · · · 1]M01M02...M0n=[∑iMcat0i]. (22)We test the accuracy and robustness of replacing the final equation in (19) with (22) versus the originalsystem. A synthetic family of 10 earthquakes is generated with the MW of each event assigned randomlyfrom a uniform distribution in the range 1.0 ≤ MW ≤ 2.5 and M0 weights, wi j, calculated as in (16).Then, a randomly selected earthquake from the list has its MW changed by ± 1.0, which representsthe case when an original recorded MW is overestimated/underestimated by standard detection methods.Following this perturbation, we solve for m from (19, 21) and measure the `2-norm of the error betweenactual M0 for all earthquakes in the family (i.e., before perturbation) and those produced from solvingthe system. We then solve (21) again, but substitute (22) for the nth row of the system. This process isrepeated 100,000 times for a magnitude perturbation of +1.0 (Figure 16a), and an additional 100,000times for a magnitude perturbation of -1.0 (Figure 16b). When using (19) the error (‖error⊥‖ in Figure16) is lower 83% of the time, compared to the error when substituting (22) for the nth row (‖errorΣ‖ inFigure 16). Upon inspection, the large negative outliers in Figure 16 are due to trials where both formsof the solution are very inaccurate, and thus neither result is desirable in these small number of cases.Overall, the system in (19) appears to be more accurate and robust than if a simple sum of moments isused as the constraint, as in (22).Additional inspection of (19) reveals that the first (n− 1) rows are identical to those in (16),and the nth row is orthogonal to all other rows. In the present study, G has size [n× n] and rank n,and thus can be solved exactly by (21). In this case, orthogonality of the nth row with respect to therest of the system is not necessary, because all equations are satisfied exactly by the solution. However,future applications may involve additional stations and/or separate M0 ratio measurements for individualcomponents. In these situations, G would be a “tall” matrix (i.e., more equations than unknowns), andthe system would be overdetermined. It is desirable to have the absolute M0 constraint orthogonal to the92Appendix B. Constraining Repeating Earthquake Magnitudesrest of the system for the overdetermined problem, because it allows for the absolute M0 constraint tobe satisfied without affecting the relative moment weights that are recovered independently. Because ofthis orthogonality, as well as the robustness/accuracy demonstrated above, we choose (19) as the systemto solve. As such, we satisfy all relative M0 information from the SVD analysis in the first (n−1) rows,and constrain all absolute M0 information with a single, orthogonal equation in the nth row.93Appendix B. Constraining Repeating Earthquake Magnitudes||error ||  -  ||error || 10 13-9 -6 -3 0 3Count10 40123Positive Outlier||error ||  -  ||error || 10 12-3 -2 -1 0 1Count10 40123Negative Outlier||error ||    ||error || 10 13-9 -6 -3 0 3Count10 40123Positive Outlier||error ||    ||error || 10 12-3 -2 -1 0 1Count10 40123Negative Outlier||error ||  -  ||error || 10 13-9 -6 -3 0 3Count10 40123Positive Outlier||error ||  -  ||error || 10 12-3 -2 -1 0 1Count10 40123Negative Outlier||error ||  -  ||error || 10 13-9 -6 -3 0 3Count10 40123Positive Outlier||error ||  -  ||error || 10 12-3 -2 -1 0 1Count10 40123Negative OutlierCountCount||e ror ||  -  ||e ror || 10 13-9 -6 -3 0 3Count10 40123Positive Outlier||e ror ||  -  ||e ror || 10 12-3 -2 -1 0 1Count10 40123Negative Outlierr r 39 6 3 3CountPositi  tli r||error ||  -  ||error || 10 12-3 -2 -1 0 1Count10 40123Negative Outlier||error ||  -  ||error || 10 13-9 -6 -3 0 3Count10 40123Positive Outlier||error ||  -  ||error || 10 12-3 -2 -1 0 1Count10 40123Negative Outlier(b)!(a)!Figure 16: Histograms of error differences in measured M0, relating to the tests of robustnessand accuracy for (19) and (22). (a) A random event is given a positive M0 perturbation, and (b) anegative M0 perturbation. Both (a) and (b) result from 100,000 individual trials. The `2-norm of theerror in M0 (measured in Nm) when using the orthogonal absolute M0 constraint in (19) is given by‖error⊥‖, and the same error measurement but when using the sum of M0 equation in (22) is givenby ‖errorΣ‖. The orthogonal equation is demonstrated to be more resistant to outliers, performingbetter than the simple sum of M0 equation in 83% of trials. See text for full methodology of thistest.94Appendix C. Repeating Earthquake GroupsAppendix CRepeating Earthquake GroupsIn Section 3.4 we explain how repeating earthquake families are divided into groups based ontheir assigned focal mechanisms and locations along the Queen Charlotte plate boundary. Figure 17provides a cross-section for each repeating earthquake group. Note that plots for 3Q and 1S are emptyas there are no families in these groups. For other groups, not all families are visible within the plotsbecause families without depth estimates cannot be plotted. Table 6 provides details for each group.95Appendix C. Repeating Earthquake Groups0 10 20 30Depth (km)3020100 0 5 10 15Depth (km)3020100 10 20 30Depth (km)3020100 0 5 10 15Depth (km)3020100 10 20 30Depth (km)3020100 0 5 10 15Depth (km)3020100 10 20 30Depth (km)3020100 0 5 10 15Depth (km)3020100 10 20 30Depth (km)3020100 0 5 10 15Depth (km)302010Distance (km)0 10 20 30Depth (km)3020100 Distance (km)0 5 10 15Depth (km)3020100 10 20 30Depth (km)3020100 0 5 10 15Depth (km)3020100 10 20 30Depth (km)3020100 0 5 10 15Depth (km)3020100 10 20 30Depth (km)3020100 0 5 10 15Depth (km)3020100 10 20 30Depth (km)3020100 0 5 10 15Depth (km)3020100 10 20 30Depth (km)3020100 0 5 10 15Depth (km)302010Distance (km)0 10 20 30Depth (km)3020100 Distance (km)0 5 10 15Depth (km)302010(b)!(a)!SW! NE!NW! SE![1S]![1Q]![2S]![3S]![4S]![5S]![6S]![2Q]![3Q]![4Q]![5Q]![6Q]!Figure 17: Cross-sections of all (a) QCF-related repeating earthquake groups, with profiles takenparallel to the surface trace of the QCF and horizontal exaggeration of 3.3, and (b) subduction-related repeating earthquake groups, with profiles taken perpendicular to the surface trace of theQCF (i.e., parallel to plate subduction) and horizontal exaggeration of 4.4. Group names aregiven in brackets in the top-right corner of each plot. Filled circles are family average locations,and empty circles are individual repeating earthquake locations. Note that some families do nothave depth estimates, and are absent from the cross-sections. Also note that horizontal distancereference points are arbitrarily chosen for each profile, and do not relate between different plots.96Appendix C. Repeating Earthquake GroupsTable 6: Details of the groupings of repeating earthquake families used in the present study. Strike-slip groups are those assumed to account for dextral strike-slip motion along the QCF, subductiongroups are those assumed to exhibit thrust motion on the subduction interface, and unclear groupsare those whose slip-types cannot be determined (see Section 3.4). For details of specific repeatingearthquake families, see Table 5.GroupNameSlipTypeCenterLocationNumber ofFamiliesNumber ofEarthquakesList ofFamilies1Q Strike-slip 53.7◦N, 133.2◦W 11 24 001 035 039043 057 073080 111 121147 1842Q Strike-slip 53.3◦N, 132.9◦W 18 63 005 006 007008 017 032034 037 054089 090 091116 130 132145 152 2113Q Strike-slip 53.1◦N, 132.65◦W 0 04Q Strike-slip 52.9◦N, 132.4◦W 9 35 070 071 117124 141 142156 169 1815Q Strike-slip 52.75◦N, 132.25◦W 10 26 112 114 122162 165 168201 204 2052106Q Strike-slip 52.5◦N, 131.9◦W 28 92 002 004 018055 072 077103 113 154157 158 160161 163 166173 174 176177 190 191192 193 202208 209 21521697Appendix C. Repeating Earthquake GroupsTable 6: (continued)GroupNameSlipTypeCenterLocationNumber ofFamiliesNumber ofEarthquakesList ofFamilies1S Subduction 53.7◦N, 133.2◦W 0 02S Subduction 53.3◦N, 132.9◦W 32 112 022 023 030036 041 042047 053 059061 066 067068 074 078088 097 100108 120 135137 143 151179 196 198219 221 222223 2243S Subduction 53.1◦N, 132.65◦W 14 74 016 027 028099 105 126127 134 138140 148 194203 2204S Subduction 52.9◦N, 132.4◦W 12 40 086 093 094115 123 155170 172 178189 195 1975S Subduction 52.75◦N, 132.25◦W 16 62 040 048 069075 171 175182 183 186187 199 200206 207 2132186S Subduction 52.5◦N, 131.9◦W 13 32 020 025 044049 081 084106 109 133159 167 18518898Appendix C. Repeating Earthquake GroupsTable 6: (continued)GroupNameSlipTypeCenterLocationNumber ofFamiliesNumber ofEarthquakesList ofFamilies1U Unclear 53.7◦N, 133.2◦W 9 22 011 012 076107 125 128144 180 2142U Unclear 53.3◦N, 132.9◦W 9 28 010 014 026050 051 052056 129 1533U Unclear 53.1◦N, 132.65◦W 5 22 021 024 046095 1014U Unclear 52.9◦N, 132.4◦W 16 45 015 031 058060 063 064065 079 085087 096 102136 139 1501645U Unclear 52.75◦N, 132.25◦W 7 17 029 045 082098 104 1492176U Unclear 52.5◦N, 131.9◦W 15 36 003 009 013019 033 038062 083 092110 118 119131 146 21299

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0343273/manifest

Comment

Related Items