AN ABSTRACT OF THE THESIS OF R. Axel Fabrititus for the degree of Master of Science in Geophysics presented on May 2, 1995. Title: Shear-Wave Anisotropy Across the Cascadia Subduction Zone From a Linear Seismograph Array. Abstract anDroved: Redacted for Privacy Measurements of shear-wave splitting in the SKS phase have been made at broad band digital seismic stations of the TORTISS array in Oregon for events during 1993 and the first quarter of 1994. The average observed splitting parameters, the fast polarization direction () 700 and the time delay (&) 1.61 sec, are well above average. Worldwide the largest observed splitting measurements are of the order of 2 sec time delay. According to gross anisotropic characteristics, the array can be divided in three regions. The first region extending over the Coast Range and has average splitting values of [72°, 1.34 secj, the second region extends over the Willamette Valley and the Cascades shows an average splitting of [66°, 1.74 sec], and the third region, east of the Cascades, has an average splitting of [790, 1.77 sec]. The splitting measurements within each of the first two regions show little systematic variation. In the third region, a systematic rotation of 4 from about 70° to 87° from west to east is observed. The observed shear-wave splitting is assumed to be caused by strain induced lattice preferred orientation of mantle minerals, mainly olivine. The general trend of the and öt estimates is most likely due to the absolute plate motion of the Juan de Fuca plate and the North American plate at the eastern end of the array. Variations in the -& estimates across the TORTISS array can be explained by variations in crustal anisotropy; however, only east of the Cascades (third region) the contribution of cru staT anisotropy could be resolved. A change in dip of the descending slab at about 50 km from the coast might have a contribution to these variations as well. Splitting measurements for events from 2300 to 280° back-azimuth differ from those from 300° to 350°. These differences can be best explained by deep seated anisotropy along the raypaths of the northern and southern event group, anywhere between the CMB and the crust underneath the array. ©Copyright by R. Axel Fabritius May 2, 1995 All Rights Reserved Shear-Wave Anisotropy Across the Cascadia Subduction Zone From a Linear Seismograph Array R. Axel Fabritius A THESIS submitted to Oregon State University in partial fulfillment of the requirements for the degree of Masters of Science Completed May 2, 1995 Commencement June 1995 Masters of Science thesis of R. Axel Fabritius presented on May 2, 1995 APPROVED: Redacted for Privacy MajoProfessor, representing Geophysics Redacted for Privacy Dean of College of Oceanic and Atmcphenl Sciences Redacted for Privacy Dean of I understand that my thesis will become part of the permanent collection of Oregon State University libraries. My signature below authorizes release of my thesis to any reader upon request. Redacted for Privacy R. Axel Fabritius, Author ACKNOWLEDGEMENT First of all I would like to thank my major professor John Nábëlek for providing me with the opportunity to work on the TORTISS and INDEPTH II projects. They will be an unforgettable memory. John's advise throughout my graduate study was always helpful and his comments and reviews were to the point. I especially enjoyed his easy-going "non-academic" side while spending time with him travelling, skiing, or digging. Thanks are also due to Drs. Y. John Chen, Ronald Guenther, R. Bruce Rettig, and Anne M. Trehu for serving on my committee and making helpful comments and suggestions regarding my thesis. Drs. Chen, Guenther and Trehu have also contributed to my graduate education. I would also like to thank Drs. Bob Duncan, Verne Kuim, Shaul Levi, and Clayton Paulson who have all contributed to my graduate education. I really enjoyed the atmosphere among my fellow geophysics students and no, it has nothing to do with the fact that most of them were german. Thanks to Alex, Beate, Bernd, Bird, Christof, Ganyuan, Giubiao, Jochen, Kelly, Kurt, Maren, Rugang, Thorsten and Xiao-Qing and all the others who just passed through. I hope they will not forget to slack off once in a while, even if I am not around to encourage them to do so. I especially like to thank our geophysics grandfather, Jochen, for endless advice, discussions, and his patience in all the study related matters and for helping me get through my downs and times of frustration. My thanks are extended to Xiao-Qing for fruitful discussions and for showing me how some things on the computer can be done easier in a different way. Further thanks to Christof who taught me my first steps on the computer; I credit Christof for the fact that I am now able to distinguish a PC from a SUN workstation. Additionally I would like to thank Steve Azevedo for introducing me into the mystery of pearl and for always being there to help me in "logistic matters". Field work with Steve was always fun. Many thanks to my dear housemates, past and present, GUnter, Jurgen, Andreas and Diana for all their support and all the good times we had together. I will always remember the spontaneous Skat nights with Günter and Jurgen, usually right before exams, and the "quick" games of pooi with Andreas and Diana. Additional thanks to Diana for reviewing my thesis. I especially want to thank my dear M-Friends. Having them around during the last years made it easy to forget about school and relax. in particular I like to thank Marc for the great ski trips, Maurice for showing me the "irish way", and finally Marcelino for introducing me to the "positive vibrations" (thanks man), for all the stuff he has done for me, and for spending some quality time with me in front of the Bean having café. At this point I also would like to thank with all my espresso-shaken heart the beanery team for serving me the thesis essential caffeine. Thank you to Eric Sandvol for providing me with "the" energy-minimization program, which speeded up my calculations, let's say, quite a bit. Last, but certainly not least, I would like to thank my entire family and especially my parents and my brother. Their endless love and support provided the basis for my successful studies. This work was supported by the National Science Foundation grant EAR-9207 181. TABLE OF CONTENTS 1.0 INTRODUCTION 1 2.0 GEOLOGICAL AND TECTONIC SETTING 5 3.0 EXPERIMENT SETUP 9 12 4.0 THEORY 4.1 Anisotropy and shear-wave splitting 4.2 Methods to investigate shear-wave splitting 12 16 Pariclemotionanalysis Waveform cross-correlation 16 17 18 4.2.1 4.2.2 4.2.3 Tangential energy minimization 5.0 ANALYTICAL PROCEDURES AND DATA 21 6.0 RESULTS 49 7.0 DISCUSSION 59 7.1 Localization of anisotropy 59 7.1.1 Thickness of the anisotropic layer 7.1.2 Depth of the anisotropic layer 7.2 Relation of anisotropy to tectonic processes 7.3 Slab effect 7.4 Interpretation summary 7.4.1 Coast Range 7.4.2 Willamette Valley and Cascades 7.4.3 East of the Cascades 8.0 59 60 62 64 66 67 67 68 CONCLUSION 70 BIBLIOGRAPHY 72 APPENDICES 75 LIST OF FIGURES Page Figure 1.1 Compressional and shear-wave velocites in a monociystal of olivine. 4 2.1 Location map of the Juan the Fuca, Pacific, and North America Plates system. 7 2.2 Geomorphic divisions of Oregon. 8 3.1 Topographic map showing exact location of TORTISS stations. 10 4.1.1 A rectilinearly polarized shear-wave S entering an anisotropic medium will split into two rectilinear orthogonal phases S1 and S2. 13 4.1.2 The SKS wave results from a P to SV conversion at the core mantle boundary. 14 4.1.3 Conventions used to derive the relations giving the radial and transverse projections of two orthogonal S waves s1(t) and s2(t) issued from an incident SKS wave s(t). 14 4.2.3.1 Energy minimization method using the Philippine Is!. event (baz 300.8°) recorded at station A24. 19 5. 1 The S-legs underneath the stations of the SKS raypaths form cones underneath the stations, if earthquakes from all back-azimuth are recorded. 24 5.2 Distribution of the events used in this study with respect to Corvallis 25 5.3-13 The following figures show seismograms of all events used in this study. 26 6.1 White vectors show the average of splitting measurements at every station. x 52 6.2 54 Vectors showing the average splitting for the Coast Range (first region, black vectors), Willamette Valley and Cascades (second region, dark grey vectors) and east of the Cascades (third region, light grey vectors). 6.3 Comparison of average splitting from northern (white vectors) southern (black vectors) events. 55 6.4 Vectors showing splitting for different frequency bands for the Philippine Isl. event. 57 7.3.1 The horizontal plane projections of vectors in the dipping planes. 65 LIST OF TABLES Page Table 5.1 Listof events. 23 6.1 Cross-correlation measurements for event: W. CAROLINE ISL. 93/09/26 03:31:19 Mw=6.4 dist=88 .93 baz=283 .4 (1 s-30s) 50 6.2 Weighted average for all events 51 LIST OF APPENDIX FIGURES Page Figure A. 1.1 Anisotropy vector map for the Philippine Isi. event. 77 A. 1.2 Contour plot of energy on the corrected transverse component for all ®, &) pairs for the Philippine Is!. event. 77 A.2. 1 Anisotropy vector map for the Taiwan event. 83 A.2.2 Contour plot of energy on the corrected transverse component for all ®, &) pairs for the Taiwan event. 83 A.4. 1 Anisotropy vector map for the Hindu Kush event. 91 A.4.2 Contour plot of energy on the corrected transverse component for all (4, öt) pairs for the Hindu Kush event. 91 A.5. 1 Anisotropy vector map for the New Ireland event. 98 A.5.2 Contour plot of energy on the corrected transverse component for all (, &) pairs for the New Ireland event. 98 A.6. I Anisotropy vector map for the W. Caroline Isi. event. 105 A.6. 1 Contour plot of energy on the corrected transverse component for all ®, &) pairs for the W. Caroline Is!. event. 105 A.7. 1 Contour plot of energy on the corrected transverse component for 112 all (, öt) pairs for the Xinjiang event. A.8. 1 Anisotropy vector map for the Eastern New Guinea event. 118 A.8.2 Contour plot of energy on the corrected transverse component for all (4, &) pairs for the Eastern New Guinea event. 118 A.9. 1 Contour plot of energy on the corrected transverse component for all (4, &) pairs for the 94 Fiji Isl. event. 125 LIST OF APPENDIX TABLES Page Table A. 1.1 Splitting measurements for the Philippine Isi. event. 76 A.2. 1 Splitting measurements for the Taiwan event. 82 A.3. 1 Splitting measurements for the 93 Fiji Isi. event. 89 A.4. 1 Splitting measurements for the Hindu Kush event. 90 A.5. 1 Splitting measurements for the New Ireland event. 97 A.6. 1 Splitting measurements for the W. Caroline Is!. event. 104 A.7. 1 Splitting measurements for the Xinjiang event. ill A.8. 1 Splitting measurements for the Eastern New Guinea event. 117 A.9. I Splitting measurements for the 94 Fiji Isi. event. 124 1.0 INTRODUCTION Seismic anisotropy, the phenomenon of a medium showing different velocities for different propagation directions through the medium, is a material property influenced by the alignment of minerals. An observational evidence of seismic anisotropy is shear-wave splitting. If a linearly polarized shear wave propagates through an anisotropic medium it splits into two orthogonal components propagating at different velocities. Typically shear wave splitting is characterized by two parameters: the fast (or slow) polarization direction and the delay time between fast and slow polarization directions. Shear-wave splitting can be interpreted in terms of tectonic/geologic processes if a chain of relations between splitting and anisotropy, anisotropy and strain, and between strain and tectonic/geologic processes are considered (for a detailed discussion see Silver and Chan, 1991). Previous studies provide evidence that mantle anisotropy is caused by strain induced lattice preferred orientation of upper mantle minerals (e.g., Nicolas and Christensen, 1987; Silver and Chan, 1991). The most abundant mineral in the upper mantle is olivine. For a shear-wave propagating along the b-axis or the c-axis of a olivine crystal the anisotropy is about 10%, whereas for propagation along the a-axis there is practically no anisotropy (figure 1.1). Three principal hypotheses are considered for the origin of the mineral orienting strain. The first hypothesis is that strain related to the absolute plate motion (APM), orients the azimuth of the fast polarization direction parallel to the APM (Leven et al. 1981, Vinnik et al. 1992). The general applicability of this hypothesis is opposed by Helffrich et al. (1994) who found almost perpendicular same plate. values over 100 km distances within the 2 The second hypothesis is that present crustal stress reflects lithospheric stress that eventually produces strain induced anisotropy. The conceptual difference to the first hypothesis is that it does not invoke a particular physical process and rather states that the origin for crustal stress is also causing anisotropy. In active tectonic areas the dominant tendency is found that the maximum horizontal stress direction is perpendicular to . (e.g., Helffrich et aL, 1994; Silver and Chan, 1993). The third hypothesis is that strain due to the last significant episode of internal deformation of the continental lithosphere is responsible for the alignment of (Silver and Chan, 1991; McNamara et al., personal communication, 1994; Helffrich et al., 1994). This is interpreted as "fossil" anisotropy in stable continental areas and as reflection of presentday tectonism in active regions. Crustal anisotropy can have a variety of causes. One possibility is that crustal anisotropy is a result of preferred alignment of vertically parallel microcracks (Crampin, 1984; Crampin et al., 1984). Another possible origin of crustal anisotropy are foliated felsic (e.g., gneises, schists) and mafic (e.g., amphibole-bearing gabbros) rocks (Barruol and Mainprice, 1993). Because the thickness of the crust is generally not more than 40 km, the crustal contribution to the observed SKS delay time is rather small (i.e., less than 0.3 sec). Recent studies conducted over short spatial scales indicate that shear-wave splitting may vary over short distances (Barruol and Souriau, 1995; Helffrich et al., 1994; Makeyewa et al., 1992; McNamara et al. 199?; Savage and Silver, 1993). Furthermore, Savage and Silver (1993) found that dividing the contributions to anisotropy into two layers would satisfy their observations. They attribute the main part of the observed anisotropy to shallow geologic/tectonic origin, and the smaller part to a lower layer influenced by APM. Makeyewa et al. (1992) suggest the abrupt change in the fast polarization direction over a short distance underneath the Tien Shan collisional belt could be due to a small-scale thermal convection associated with a hot upper mantle. 3 Several suggestions have been made to explain the orientation of in subduction zones: "frozen-in" anisotropy within the subducted plate resulting from mineral alignment prior to subduction (Savage and Silver, 1993), shear strain associated with the subduction (Shih et al., 1991), or a combination of both (Vinnik and Kind, 1993). Bostock and Cassidy (1995) suggest that in the southern Canadian Cordillera the mantle flow associated with the motion of the Juan de Fuca plate, its subduction along the coast, and associated backarc asthenospheric flow align the fast polarization axis parallel to the flow. For the northern Canadian Cordillera, however, they suggest that orogen parallel shear deformation aligns the fast axis parallel to the orogen. In this thesis shear wave splitting in the Cascadia subduction zone is investigated using data from the TORTISS (The ORegon Teleseismic Imaging of the Subducting Slab) experiment, conducted in Oregon during 1993 and 1994. The main purpose of this experiment was to image the geometry of the subducting Juan de Fuca plate using teleseismic broadband data. In the case of shear wave splitting, this is the only study from such large and dense linear array of broadband seismometers. This data set enables the study of variations in seismic anisotropy over an entire subduction complex and helps to determine the likely cause for the fast polarization direction in the Cascadia Subduction Zone. 8.43 798 4.87 4.88 OO1 8.32 '4.42 5.33 4.63 4.64 54LaJ 8.66 t_ 9.81tt89I 4.87 I 4.89 b 4.42 o1It4 110 I 5.53 \,.:_-j_- 4.66 8.83' Figure 1.1 Compressional and shear-wave velocities in a monocrystal of olivine. The main axes of the crystal are a [100], b [010], and c [001]. (From Babuska, 1991). 5 2.0 GEOLOGICAL AND TECTONIC SETTING The TORTISS array was designed to image the Juan de Fuca (JdF) plate beneath Oregon in order to obtain a detailed 2D-picture of the Cascadia Subduction Zone. The JdF plate, as well as the Explorer plate to the north and the Gorda plate to the south, are small remnants of the Farallon plate, which once covered much of the eastern Pacific basin (Atwater, 1970) (figure 2.1). The JdF ridge and Blanco fracture zone constitute the JdF plate's northwestern and southwestern boundary, respectively. The eastern boundary is formed by the North American plate, where the JdF plate is being subducted at the Cascadia subduction zone. According to Wilson (1991), the JdF plate is moving in a direction of N65°E at a rate of 34.4 mm/year relative to the North American plate and in a direction of N67.5°E at a rate of 13.8 mm/year relative to a fixed hot spot coordinate system (values calculated for Corvallis). The TORTISS array extends over 3 major geologic provinces. The first province is the Coast Ranges of western Oregon and the Willamette Valley, the subduction zone's forarc. The second province is the Cascades, the volcanic arc. This province can be divided in two subprovinces, the Oligocene and Miocene volcanic rocks of the Western Cascades and the Pliocene to Holocene volcanic rocks of the High Cascades (Sherrod and Smith, 1989). The third province, the backarc, is represented by the High Lava Plains and the Blue Mountains (figure 2.2). In contrast to Washington, Oregon lacks an active Wadatti-Benioff zone, and therefore little is known about the geometry of the JdF plate beneath Oregon. The most recent geophysical studies on imaging the subducting slab are the tomography studies by VanDecar (1991), Harris et al. (1991) and Rasmussen and Humphreys (1988) and a receiver function study by Owens et al. (1988). For western Washington, Owens et al. (1988) suggest a shallowly (200) dipping plate. East of the Cascades, VanDecar (1991) suggests a slab dipping at about 600 and extending to a depth of about 400 km for central Washington. However, he finds no evidence of deep (below of about 200 km) slab material beneath southern Washington and northern Oregon. For southern Oregon, Harris et at. (1991) suggest a steeply dipping (65°) slab beneath the Cascade Range to extending to at least 200-km depth. 7 1200 1250 1300 LOTTE 'CHAR IS BRITISH A COLUMBIA 2\ Explorer 50° Q Sovonco,_ Z. P/ate -\ /Voo/*o , \ Fault 0 Cobb 4. 0? Seomounts WASHINGTON 1 k Q Jl.JQfl de 1 Columb'° R. 450 ; Fgicc..4 0 P/ate O I OREGON I' I Cape I BIanco -. U U) P/ole PA C/F/C Mendoc,no \cape Mendocin -1 00 'sCALIFORNIA PLATE Andreas \SanFault San Fanc Figure 2.1 Location map of the Juan the Fuca, Pacific, and North America Plates system. Dashed line along the continental margin is the boundary between North America and Juan the Fuca Plates. Stars indicate the location of the TORTISS array. (Modified from Duncan and Kuim, 1989). GEOMORPHIC DIVISIONS of OREGON , / I, I * I * * * * * * NWny U a. I ' 0 I- 0) u o o fr4 ' PLAINS OWYHEE UPLAND c.ct / KLAMATH A%N MOUNTAINS - RANGE I $L.t,y . ' Ufl Figure 2.2 Geomorphic divisions of Oregon. Stars indicate the location of the TORTISS array. (Modified from Baldwin, 1976). 3.0 EXPERIMENT SETUP During the 1993-1994 TORTISS experiment a total of 69 sites was occupied by portable seismic stations (figure 3.1). The highest number of seismometers was deployed during the time span of April 1993 until the beginning of December 1993. Fortyfour portable broadband 3-component seismometers were available for this study (27 Guralp CMG-3ESP and 17 Streckeisen STS-2). All stations were equipped with REFTEK data loggers, timed either with GPS or Omega synchronized clocks. The recording was continuous with 20 samples per second. At the beginning of the experiment, the average station spacing was 4 km in the western and 8 km in the eastern part of the array. As the experiment progressed, some stations were moved from the western towards the eastern part of the array in order to obtain a 4-km spacing over the entire length of the array. From January 1994 until March 1994, the array was reduced to 15 Streckeisen STS-2 instruments and relocated eastward, with a station spacing of 8 km. The TORTISS experiment recorded about 130 earthquakes of magnitude 6 and greater. 10 Figure 3.1 Topographic map showing exact location of TORTISS stations. Between station AOl and A445 station spacing is approximately 4 km and between A445 and A5 15 station spacing is approximately 8 km. OG6L- 09009 006009009- 009- 0 09L 009 0gt 009 006 090 006 099 L 099L 009L 008 L 096 L 00 6 0966 OOt'Z 0999 MJ 12 4.0 THEORY 4.1 Anisotropy and shear-wave splitting Anisotropy in the general sense is defined as a variation of a physical property depending on the direction in which it is measured. In seismology the physical property of concern is the medium velocity. In this thesis the term "seismic anisotropy" will be used to describe large-scale (compared to the wavelength of a seismic wave) velocity anisotropy. The degree of velocity anisotropy of a medium can be described by the coefficient of anisotropy k: k= [(Vmax Vmin)/vmeanl X 100% (Birch, 1960) v = velocity. One manifestation of seismic anisotropy is shear-wave splitting, or birefringence in analogy to optics. A rectilinearly polarized shear-wave entering a homogeneous anisotropic medium will split into two rectilinear orthogonal phases. These two phases will travel at different velocities. Assuming the anisotropic region is homogeneous over a significant portion of the ray path, a difference in the travel time of the two phases will be measurable (figure 4.1.1). Using real data, the most suitable phase to detect shear-wave splitting on a lithospheric scale is the SKS phase. This phase leaves the source as S wave, converts into a P wave at the core-mantle boundary (CMB), propagates through the liquid outer core, then changes back into a S wave at its emergence from the core back into the mantle (figure 4.1.2). Because SKS results from a P to S conversion at the CMB, it would be rectilinearly polarized in the vertical plane of propagation for a spherically isotropic Earth, i.e., it would have no transverse component. However, SKS is often 13 Figure 4.1.1 A rectilinearly polarized shear-wave (S) entering an anisotropic medium will split into two rectilinear orthogonal phases S1 and S2. These two phases will travel at different velocities. (From Babuska, 1991). 14 Figure 4.1.2 The SKS wave results from a P to SV conversion at the core mantle boundary. (From Babuska, 1991). Transverse J epkenter p.. Axis of symmetry SKS ray Figure 4.1.3 Conventions used to derive the relations giving the radial and transverse projections of two orthogonal S waves s1(t) and s2(t) issued from an incident SKS wave s(t). The incident SKS wave s(t) is polarized in the vertical plane and is split into two S waves while travelling through an anisotropic layer represented by a material with hexagonal symmetry and a horizontal axis of symmetry. (From Babuska, 1991). 15 observed having a transverse component. Since the SKS wave left the core radially polarized, the source of the observed transverse energy must be on the receiver side of the raypath. Anisotropy and aspherical structure are possible explanation, which can be distinguished with the help of particle motion analysis. Anisotropy will produce an elliptical particle motion, whereas particle motion caused by an aspherical structure will remain in most cases predominantly rectilinear. Making the assumption that the incidence angle of the ray is nearly vertical, one can write: (4.1) i (t)=s(t)cos(p s2(t)=s(t-&)sin( where: s(t) is the radially polarized SKS waveform in a spherical isotropic Earth s1(t), s2(t) - are the projections of the ground motions onto the fast and slow S-wave polarization directions is the azimuth of the fast S-wave polarization direction with respect to the radial direction is the time difference between the fast and the slow arriving components The rotation of the two split components s 1(t) and s2(t) onto a radial and transverse coordinate system (figure 4.1.3) yields: R(t) (4.2) s(t)cos2(p + s(t-&)sin2p T(t) = [(s(t) s(t-&))/2]sin2(p For small öt compared to the period of s(t), the waveform of the radial component will be very similar to s(t), the radial waveform in absence of anisotropy, and the waveform of the transverse component will be approximately proportional to the time derivative of the radial component [equations (4.2) and figure 4.2.3. 1A]. For ( ØO and 900 the amplitude of the transverse component is zero (equation 4.2). Hence, for events with a back-azimuth identical to the fast or slow polarization direction, no transverse energy is observed. These events are referred to as non-splitting events. 4.2 Methods to investigate shear-wave splitting The analysis of shear-wave splitting can be performed with methods such as particle motion analysis, waveform cross-correlation (Bowman and Ando 1987; McNamara and Owens 1993) and tangential energy minimization (Silver and Chan 1988, 1991). These methods are similar in that they attempt to estimate the orientation and the degree of anisotropy by determining the fast polarization direction and the time delay between fast and slow direction of the S wave. In this section, these three methods are described. 4.2.1 Particle motion analysis The polarization anomalies caused by anisotropy are observed as departures from rectilinearity in particle motion diagrams. The particle motion produced by shear wave splitting is elliptical and initially polarized in the fast S-wave direction. Resultant particle motion anomalies are used in a number of techniques (Bowman and Ando, 1987; Silver and Chan, 1988, 1991; Savage et al., 1989, 1990a; Shih et al., 1989) as the main diagnostic property of shear-wave splitting. 17 In the particle motion analysis method, the initial particle motion orientation is measured directly from the horizontal particle motion plot of the radial and transverse components and taken as the angle () of the fast polarization direction clockwise with respect to north. The seismograms are then rotated onto a coordinate system defined by . In this coordinate system, the time delay (&) between fast and slow direction can be directly measured, e.g., using cross-correlation. The time window of the seismograms must be long enough to include both the beginning of the fast and the end of the slow wavelet. The time lag producing maximum cross-correlation is taken as a measure for &. To test the accuracy of the estimates of and öt, the seismograms are shifted by & with respect to each other and rotated back the radial/tangential coordinate system. If the estimates of and & are correct, the "corrected" seismograms' particle motion is significantly more linear than the original. 4.2.2 Waveform cross-correlation The waveform cross-correlation method takes advantage of the fact that the waveforms of the split S phase will be most similar in the fast/slow coordinate system [equations (4.1); Bowman and Ando, 1987]. To obtain are rotated in increments from 00 and &, horizontal seismograms to 180°. At each increment the two seismograms are shifted with respect to each other (by &) and cross-correlated. The rotation angle and time shift with the maximum absolute cross-correlation value are taken as estimates of 4 and 4.2.3 Tangential energy minimization The tangential energy minimization method was originally proposed by Silver and Chan (1988). Since every P to S converted phase is radially polarized, energy appearing on the S wavets tangential component is evidence for presence of a nonhomogeneous structure or anisotropy. Silver and Chan (1988) assume the transverse energy is primarily due to anisotropy. To obtain and öt, following steps are performed for all possible and & pairs: the original N and E component seismograms are rotated through a candidate and shifted by a candidate & with respect to each other, then rotated into the radialltangential coordinate system where the tangential energy is calculated; the candidate rotation angle and time delay that most effectively minimize the energy on the tangential seismogram are taken as estimate of 4.2.3.1. and &. The basic steps are displayed in figure 19 Figure 4.2.3.1 Energy minimization method using the Philippine Isi. event (ba.z 300.8°) recorded at station A24. (A) Radial and tangential component of the recorded SKS phase. The tangential component is very energetic and its waveform is proportional to the time derivative of the radial waveform. (B) Seismograms rotated to the fast and slow polarization directions. In this coordinate system & = 1.7 sec can be directly read of the diagram and waveforms are most similar. (C) Corrected seismograms in the radial/tangential coordinate system. The tangential component shows almost no energy. (D) and (F) show the particle motion plot of the uncorrected and corrected seismograms respectively. Note the elliptical particle motion in (D) versus the rectilinear, parallel to the radial component, particle motion in (F). (E) Contour plot of transverse component energy. The minimum energy for 4 = 69° and t = 1.7 sec is indicated by a star and the 95% confidence level is shown within the bold contour line. A C - Radial - Fast - Slow - Tangential 0 20 10 30 0 seconds 20 10 Radial Tangential 30 0 20 10 seconds 30 seconds D F C.) I 2.0 ci) E 1.5 I > 1.0 ci) 0 :: i' -90 -75 -60 -45 -30 -15 0 I I 15 30 45 60 75 90 Fast Angle (deg) Figure 42.3.1 C 21 5.0 ANALYTICAL PROCEDURES AND DATA Seismograms showing a clear SKS phase are most suitable for shear wave splitting analysis. The SKS can be observed for earthquakes which occur between 61° to 144° distances, but the most suitable distance range is from 85° to 1100. At distances closer than 85°, SKS is not well separated from the S phase, and at distances greater than 1100, the signal amplitude is generally too small. Furthermore, events deeper than 100 km are desirable in order to avoid complications in the SKS-waveforms caused by near-source surface reflections. A complete back-azimuth coverage is desirable in order to obtain a tight web of raypaths underneath a station. Common splitting measurements are probably caused by common raypaths, whereas variations are probably caused by mutually exclusive raypaths thus constraining depth and lateral variations of anisotropic regions more tightly (figure 5.1). The energy minimization method was chosen for the final analysis. For most events the splitting measurements were also performed using the cross-correlation method with very similar results. The energy minimization method was found to be more stable for events with back-azimuths near the fast or slow polarization directions and the computing time was shorter. As an example, the splitting results using both techniques for the Western Caroline Islands event are compared in table 6.1. All the seismograms used for the calculation of the splitting parameters were integrated, detrended, demeaned, and bandpass filtered from 1 to 30 sec. A time window of approximately 40 sec was cut around the SKS arrival and the ends were cosine tapered (10%). In performing the energy minimization, increments of 1° and 0.05 s for and & were chosen. In order to asses the uncertainty of each measurement, and to check for multiple minima, a contour plot of energy on the corrected tangential component was 22 plotted as a function of all tested 4-öt pairs (appendix A). The resulting 4-& estimate is shown as a star within a 95% confidence region. The array's configuration (i.e., consistency of results for neighboring stations) allows in some cases the incorporation of measurements with large 95% confidence regions, if the minimum -öt pair or the signature of the energy contour plot are similar to the estimates from the neighboring stations, which have well resolved results. From all the earthquakes recorded with the TORTISS array, only 11 showed SKS phases suitable for the evaluation of shear wave splitting (table 5.1, figures 5.3-5.13). Source regions for most of these events are the active subduction zones of the western Pacific rim. Figure 5.2 shows the back-azimuth coverage provided by these events. As in other similar studies, it is assumed that all transverse energy is caused by shear- wave anisotropy. A visual inspection of the seismograms of these earthquakes can by itself provide an indication on possible velocity anisotropy along the raypaths of the recorded waves. Transverse energy can be observed for eight events. Theory suggests that the closer the back-azimuth of an event is to the fast or slow polarization directions, the smaller should be the amplitude of the signal on the transverse component (equation 4.2). This can be observed best for the Fiji Isi. events (figure 5.6, figure 5.13) with backazimuths only 50 to 100 from the fast azimuth of polarization, which show very small amplitudes on the transverse compared to the radial component. Furthermore, for the San Juan event (figure 5.4), the Xinjiang event (figure 5.10) and the Vanuatu Isl. event (figure 5.12), which have a mutually perpendicular back-azimuths, no SKS arrivals are observed, indicating the back-azimuths of these events coincide with the fast and slow polarization directions. Comparing the radial and transverse waveforms for events with significant transverse component, one observes that the transverse component is indeed a time derivative of the radial component, as expected for an anisotropic medium (chapter 4.1). This can be observed very clearly in the seismograms from the Philippine Isi. (figure 5.3) and Taiwan (figure 5.5) earthquakes. 23 Tabele 5.1 List of events. M* refers to either surface wave, body wave or moment magnitude, whichever is bigger. Event ID Origin UT 1993.05.18 1993.06.08 1993.08.07 1993.08.07 1993.08.09 1993.09.06 1993.09.26 1993.10.02 1993. 10.13 1994.02.11 1994.03.31 10:19:35 23:17:41 00:00:38 17:53:27 12:42:50 03:55:58 03:31:19 08:42:35 02:06:00 21:17:36 22:40:53 Epicenter LAT, LON 19.814N 122.418E 31.595S069.222W 26.528N 125.612E 23.940S 179.793W 36.348N 070.840E 04.654S 153.253E Depth Mag. Dist. Baz km 186 113 158 560 233 073 10.359N 137.996E 033 38.165N 088.640E 033 05.909S 146.017E 024 18.864S 169.102E 250 21.980S 179.585W 591 M* 6.5 6.4 6.4 6.9 6.9 6.6 6.4 6.3 7.2 7.0 6.5 (0) 92.34 90.68 85.38 85.68 98.06 88.57 88.93 92.72 94.58 89.53 85.05 Comment (0) 300.79 136.28 303.00 230.25 348.54 262.11 283.40 335.32 266.27 242.83 232.72 Philippine Isl. Reg. San Juan Province, Arg. NE. of Taiwan S. of Fiji Isi. Hindu Kush Reg., Afgh. New Ireland Reg. Western Caroline Isi. Southern Xinjiang, China Eastern New Guinea Vanuatu IsI. Fiji Isi. Reg. 24 array IS S 1' "k I S - - - S - CORE - - Figure 5.1 The S-legs underneath the stations of the SKS raypaths form cones underneath the stations, if earthquakes from all back-azimuths are recorded. Assuming 100 incidence angle for SKS phases and a 4 km station spacing the common part of two nearby ray cones starts at about 22 km depth. Comparison of mutually exclusive areas sampled by two cones and the splitting results of two stations can constrain the depth of the anisotropic region. 25 Event Distribution with respect to Corvallis , Einth Kih T 4 "WtCaró1. Isl\ 5\V\\\ \\\\\ \ ,\ \c\ A\ I \4\\\\\\\\\\\\\\ \\\\\\\\\\\\ inea\\\\\\\"\ '¼'U\ \ \\\ - us \\\\\\\)?c,,X\\ \ \ \ \\\\\\\\\\\\X\K\\\\\ , and\\\\\\\\\\('c.'\\\\\\ \ \\\ \\'.\\\\\\'k\\\\>K\YS\\\\\\\\ \\\\\\\\\\\\\/(\\)S\\\\\\\\\\\. '\ e I I / \\ \\\'\\\\\\'\\\\X\\\\\\\\\\\\\\\\\ \\\'\\\\\\\V\\\)\\\\..'\"I'Z\\\ \\\ \\ \\\ \\\ \Ss .\\ \\\ / \\\ Vafl1àtVLs1\\\\\\\\\\\\\\\\\\\\\\ \'\\\\\\ \\\\ \\\ \\\\/ \\\\\\\Qc\\\\\\.\\\\\\\\\\\\\\\\'c\\\\ \\\\.\\\ \\\\ \ \\\ \\\\\\\\\\\\\\\\\\\\\\\\_>,,-'\\\\\\ \\ 'Fiji Ia1O.pd Reg.\\\\\\\.\-c-\\\\\\\\\\\ \,.\\\ \\\ 'S\ \\\\\ \\ all r Vlflce\\\\ \\\\\\\\\'4,\\\\\\\\\\\\\\\\\\\\\\\\\\\\S\\\\ \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\.\ \\ \\\\\\\\\\\ \\\\\\\\\\ \\\\\\\\\\\\\\s.4\\\\\\\\\\\\\\\\\\\'\\ \\\\\\\\\\ \\\\\\\\\\\\\\\\\\\\\\\\\\\\ \ \\\\\\\\\\\\ \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ \\\\\\\\\\\\\ \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ 5' \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \\\\\\\\\\\\\\ \\\\\\\\\\\\\\\\\\\\\\\\\\\\\ \ \ \.\ \ \ \\ \S \ \ \ \ 5'\\\\\\\\\\\\\\\\\\\\\\\\"S\\\\\\\\\\\\\S'\ \\\\\\\\\\\\ \\\',\,\\,'\\'\ Figure 5.2 Distribution of the events used in this study with respect to Corvallis. The map uses an azimuthal projection. The dashed circles indicate distances of 300, 60°, and 90° degrees. The dotted pattern indicates the northern and the lined pattern the southern group. Figures 5.3 5.13 The following figures show seismograms of all events used in this study. The first arrival in each seismogram is the SKS-wave. Figures with figure numbers ending in (a) show radial seismograms, and those ending in (b) show transverse seismograms. Seismograms in (a) have different scales from those in (b). Distance From The Coast (km) -NNNN)N)NN) N) N) )N)C) Cl Co 0 0) -N 0) 0 -4 CD 0 0) -N -N 01 0) CC) 0 Cl0ClClN)0ClClN)0ClC)N)0ClCl4.N)0ClClN)0ClClN)0ClCl4.N)0Cl0)3 -N -N 01 Cl N) N) 01 Cl - Cl CO -4 Cl Cl N) 0 Ho -N CD C', 11 1993/1 38 10:19:35.1 Radial SKSab-wave 1-30 s 19.831N 122.436E 187km M6.3 92.3/300.7 PHILIPPINE ISLANDS REGION. Figure 5.3a N) Distance From The Coast (km) C) 0 C) C) N) - 0) - N) 0 - C) 01 C) C) . .J C) C) N) 0 C) (C) - C) - C) 4 N) -' N) 0 N) C) - _ _ - - 0) 4- 01 C) C) C) 4 N) 0 C) C) J L C) - -L N) N.) N) CC) 0 0 N) 0 Cr) C) N) N) . N) 0) N) N) - © N) - C) N) 01 C) N) C) . N) -.J N) 0) N) C) N) N) 0) C) C) 4 0) CC) 0 N) 0 -Ho 3 CD C,) [!i 1993/138 10:19:35.1 Transverse SKSab-wave 1-30 s 19.831N 122.436E 187km M6.3 92.3/300.7 PHILIPPINE ISLANDS REGION. Figure 5.3b Distance From The Coast (km) - C C C C) N) 0) 4 . . N) C (0 01 0) 0) 4 N) C - - (0 C - N) (0 0) 4 N) 0 C CX) ND 01 -L 0) 0) - - - .- - - N) N) N) N) N) N) N) N) 01 0)0) -J (0 (C) C C - N) 0) 01 N) 0 C C) ND C (0 0) 4 N) C C 0) . . ND 0) 4 N) N) N) ND CX) CX) (C) C 'J N) C C 0) 0) . N) C -Ho B CD (I) 1993/159 23:17:41 .4 Radial SKSab-wave 1-30 s 31.595S 69.222W 113km M6.4 90.4/1 36.2 SAN JUAN PROVINCE, ARGENTINA Figure 5.4a N) Distance From The Coast (km) rrc 3 CD (I) S. 1993/159 23:17:41 .4 Transverse SKSab-wave 1-30 31.595S 69.222W 113km M6.4 90.4/136.2 SAN JUAN PROVINCE, ARGENTINA Figure 5.4b S U- Distance From The Coast (km) 01 0 01 C) N) 4 OD 4 N) 0 4 01 01 0) 01 C) - ND 0 01 01 - CC) C C) 4 ND ND 0 - _ ND 01 - - C) 4 01 0) 0) ND - 0) 0 CD J C) - - 01 CC) 0 0 0 0) C) ND - ND ND ND ND N) . ND 0) ND ND 4 N) 0 0) ND ND ND (Dl C) -4 0) 4 ND ND 01 ND CX) 0 0) ND OD 0) 4. CD 0 ND 0 Ho 2 CD Cl) 1993/219 00:00:37.5 Radial SKSab-wave 1-30 s 26.528N 125.612E 158km M6.4 85.3/303.0 NORTHEAST OF TAIWAN. Figure 5.5a Distance From The Coast (km) N) 0 -10 2 CD 0) 1993/219 00:00:37.5 Transverse SKSab-wave 1-30 s 26.528N 125.612E 158km M6.4 85.3/303.0 NORTHEAST OF TAIWAN. Figure 5.5b I') N) 4. Hc CD C') 1. -L - -L ..L L L ...L N) Distance From The Coast (km) N) N) N) N) 1993/219 17:53:27.3 Radial 23.904S 179.793W 560km M6.9 85.6/230.2 SOUTH OF FIJI ISLANDS. Figure 5.6a N) Distance From The Coast (km) - - -L N) . . N) N) N) N) N) N) N) N) N) N) N) N) N.) 0 -10 3 CD C') S. 1993/219 17:53:27.3 Transverse SKSab-wave 1-30 s 23.904S 179.793W 560km M6.9 85.6/230.2 SOUTH OF FIJI ISLANDS. Figure 5.6b Distance From The Coast (km) C) C) C) C) N) . 0) . -P N) 0 C) 01 C) C) - -4 N) C) C) CD 0 C) C) C) 4 N) N) C) N) C) 0) C) - 01 N) C) C) -J C) C) C) C) 4 CD N) C) C) N)N)ON) 0 - N) 0)N)N)4 N) 01 CD C) 3 N) C) C) C) N) N)N) 4 N) C) -J C) C) N) C) C) N) CD C) 0) C) 4 N) C) Ic 3 CD Cl) S I. 1993/221 12:42:50.0 Radial sSKS 1-30 s 36.348N 70.840E 233km M6.9 98.4/348.4 HINDU KUSH REGION, AFGHANISTAN. Figure 5.7a 304 296 288 280 272 264 256 248 240 232 224 216 208 192 - 200 - 184 176 152 160 o 168 I- 120 128 136 E 144 L (I) 112 20 40 Time(s) 60 (I) 0 Cl) C,) U) 0) (0 I- I- L1 z 0 C, w I C,) 2 = co co 0) (0 E C1) 0(l) Lw siZ 0) 0)(0 r 36 N) C HC 3 CD (I) . S. - - 01 0) N) C C C 4 N) C) C C C C - ND - N)N)N)N)N)N N) 0) 4 01 C C C (C) C 0 - N) 0) N) C C C - N) C C C 4 N) C C C C -- Distance From The Coast (km) N) -.J C C (0 C - N) C C C Radial S 1993/249 03:55:58.0 4.654S 153.253E 33km M6.6 88.51262.1 NEW IRELAND REGION, Figure 5.8a Distance From The Coast (km) 0 Ho (D Ci-) i1 1993/249 03:55:58.0 Transverse SKSab-wave 1-30 s 4.654S 153.253E 33km M6.6 88.5/262.1 NEW IRELAND REGION, Figure 5.8b Distance From The Coast (km) -. r c - - 01 C) - C) - CD 0 -L - - - - N) N) c . cu - - C) C) - -' - r r N) -4 0 (0 Q Q _L N) N) N) ND 0) - 'D - N) 01 N) C) N) -.4 N) C) N) C) N) UD (0 C) Hc B CD C') jI S. 1993/269 03:31 :18.8 Radial SKSab-wave 1-30 s l0.359N 137.996E 33km M6.4 88.9/283.4 WESTERN CAROLINE ISLANDS. Figure 5.9a Distance From The Coast (km) C) C) C) C) N) . C) 4 ND 0 . 01 C) C) C) - C) C) CD 0 N) N) 0) N) 0 01 C) N) 0 01 C) -.4 - 01 C) N) 0 C) 01 .4 C) C) . (0 0 0 - N) 0 01 C) N) 4 0) N) 0 - C) CJ1 C) C) - -4 C) N) 0 N)r)c C) 01 (C) C C) - N) 0 lo B CD Cl) 1993/269 03:31:18.8 Transverse SKSab-wave 1-30 s 10.359N 137.996E 33km M6.4 88.91283.4 WESTERN CAROLINE ISLANDS. Figure 5.9b Distance From The Coast (km) L .. - -L .L -L . - .L N) N) N) N) N) N) N) N) N) N) N) N) N) 0) N) 0 Ho 2 CD (I) S. 1993/275 08:42:35.4 Radial SKSab-wave 1-30 s 38.165N 88.640E 33km M6.O 92.7/335.3 SOUTHERN XINJIANG, CHINA. Figure 5.lOa Distance From The Coast (km) L C) N) 0) - - 01 C) -J C) C) CD 0 0) C) A N) 0 0) C) A N) 0 0) C) C) - - -N)N)N)N)N)N)N)N)N)N)N)N)C) CD 0 C) A 01 C) C) CD 0 N) 0 0) N) 0 0) C) A N) 0 0) A N) 0 0) N) N) 0) A 01 C) C) N) 0 0) C) - -4 C) C) - N) 0) - C) C) J C) A I HoA 2 CD C,) S. 1993/275 08:42:35.4 Transverse SKSab-wave 1-30 s 38.165N 88.640E 33km M6.0 92.7/335.3 SOUTHERN XINJIANG, CHINA. Figure 5.lOb A N) N) C) HC) CD C,) C) C) C) C) 4 N) 4 - N) 0 C) 0) 01 C) C) ND -1 0) C) C) C) . (0 0) N) N) 0 N) C) 0) C) 4 - 01 N) C) C) J C) C) C) - C) C) C) C) - N) 0 0) CD Distance From The Coast (km) C) - N) - 0) N) - C 1993/286 02:06:00.2 Radial S 5.909S 146.017E 24km M6.7 94.5/266.2 EASTERN NEW GUINEA REG Figure 5.11a Distance From The Coast (km) 0) 0 0) N) o 0) - N) oi . . C) C) C) C) - - -i N) 0 C) C) 4 N) r' N) 0 0) O C) a) C) N) 0 0) 0-i - - C) C) 4 -N) N)N) N)N) N)N) (0 0 0 -' N) 0) 4 N) 0 0) C) N) 0 C) - - N) N) N)N) N) N) 0) C) . N) 0 C) 4 01 0) -4 C) C) C) (0 0 N) 0 B CD C/) 1993/286 02:06:00.2 Transverse SKSab-wave 1-30 s 5.909S 146.017E 24km M6.7 94.5/266.2 EASTERN NEW GUINEA REG., P.N.G. Figure 5.11b N) 0 -10 2 CD (1) CD 0 CD 0) 4 N) C CD 0) 4 1994/042 21 :17:36.0 NJ 0 CD 0) 0) - - ND 0 CD - _ 0) 4- - 0 N) CD N) 0) ND - N) N) N) N) 0 Radial N) Distance From The Coast (km) N) 0 CD . - 4 18.864S 169.1 02E 250km M7.0 88.1/241.1 VANUATU ISLANDS. Figure 5.12a Distance From The Coast (km) C) 0 C) C) N) 0) A A 01 N) 0 C) C) - C) - -J C) C) N) 0 C) (D 0 C) N) N) 0) 01 C) C) C) (C C) 0 - N) 0) A C) -J C) C) CD 0 A N-) 0 C) C) A N) 0 C) C) A N) 0 C) C) A N) 0 C) 01 0) A N) 0 0) C) A - -.J N-) 0 HoA (0 C/) [S.- 1994/042 21:17:36.0 Transverse SKSab-wave 1-30s 18.864S 169.102E 250km M7.O 88.1/241.1 VANUATU ISLANDS. Figure 5.12b Distance From The Coast (km) C) 0 C) II II - N) C) C) 4. N) 0 II II II - - C) (31 C) C) - -i C) 0 - C) N) 0 C) C) 4 N) N) 0 N) C) C 4 C) . 0-i C) C) N) 0 C) -J C) - N) N)N)N)N)N)N)N)N)N)N)C) o 01 C) 4 C) C) CC 0 0 oC) -C) N) 0) N) 0 0 C) N) 0 0 C) - - N) 0 -Ho 3 CD C,) rii L1 II II 1994/090 22:40:53.3 II II II II II II II II II II Radial SKSab-wave 1-30 s 21.980S 179.585W 591 km M6.5 83.8/230.9 FIJI ISLANDS REGION. Figure 5.13a Distance From The Coast (km) - C) 0 0 C) N) 0 . - 4 01 N) 0 0 C) C) 4 C) C) 0 C) - N) 0 0 C) 4 N) N) 0 N) C) 01 - 01 C) C) -.4 C) C) - N) C) - C) C) c N) C) C) C) - C) C) N) - 01 N) - - 0-1 C) C) -.4 CO C) - N) C) C) C) CO CD C) C) 4 N) 0 Ho, 2 CD (I) 1994/090 22:40:53.3 Transverse SKSab-wave 1-30 s 21.980S 179.585W 591 km M6.5 83.8/230.9 FIJI ISLANDS REGION. Figure 5.13b Co 6.0 RESULTS The analyzed events are listed in table 5.1. The measurements of shear-wave splitting for every event are presented in appendix A. Table 6.2 shows the estimates of 4 and öt for each station obtained as error-weighted averages of all the measurements done. The averaged results for each station are shown graphically in figure 6.1. The (-6t estimates are fairly consistent across the array. The arithmetic average of the weighted averages of all the stations is [70°, 1.61 sec]. Compared to worldwide splitting measurements, for which 2-sec delay times are about the maximum observed, this result is well above average. For comparison, delay times obtained from other subduction zones are not quite as large: e.g., 1.55 sec College, Alaska; 1.00 sec Longmire, Washington (Silver and Chan, 1991); 1.23 sec Naña, Peru; and 1.48 sec Erimo, Japan (Helffrich et al., 1994). Simple averaging of the station measurements does not take advantage of the unique set up of the array. With 4-km station spacing lateral variations in anisotropy can be charted. Figure 6.1 shows that the array can be roughly divided in three different regions according to gross anisotropic characteristics. The first region extends over the Coast Range from station AOl to about A14 (figure 6.1) with an average of [72°, 1.34 sec]. The delay times found for this region are the lowest in the array, and, compared to the other two regions, the measurements are less stable (i.e., have larger uncertainties; table 6.2). The second region extends over the Willamette Valley and the Cascades Range from about station A 15 to about station A36 with an average value of [66°, 1.74 sec]. This is the region with the most stable results. The third region is located east of the Cascades from about station A37 to station AS 15. Reliable splitting measurements were only obtained for stations A37 to A43 which have an average of [79°, 1.77 sec]. The reason for including 50 Table 6.1 Splitting measurements for event: W. CAROLINE ISL. 93/09/26 03:31:19 Mw=6.4 dist=88 .93 baz=283 .4 (1 s-30s) STA refers to the station, parameter () is the fast polarization direction (clockwise with respect to north), and öt is the delay time in seconds. (a) Cross-correlation method: STA 4) & A24 A25 A26 A27 A28 A29 A30 77.40 72.40 74.40 77.40 76.40 81.40 83.40 73.40 83.40 79.40 65.40 1.30 1.25 1.30 1.80 1.90 1.55 1.65 1.50 1.70 1.80 1.65 A31 A32 A33 A34 (b) Energy minimization method: STA A24 A25 A26 A27 A28 A29 A30 A31 A32 A33 A34 & 4) 68.00 53.00 48.00 56.00 56.00 53.00 54.00 57.00 64.00 73.00 60.00 -i-I- 18.00 +1- 14.00 +1- 13.00 9.00 +18.00 +1+1- 14.00 +1- 15.00 +1- 20.00 +1- 24.00 +1- 26.00 8.00 +1- 1.15 1.25 1.50 1.85 1.90 1.55 1.60 1.50 1.45 1.65 1.60 Tht 0.40 0.35 0.60 0.40 -i-I- 0.35 +1- 0.40 +1- 0.45 +1- 0.65 +1- 0.65 +1- 0.90 -i-I- 0.30 +1+1+1+1- 51 Table 6.2 Weighted average for all events STA refers to the station, parameter is the fast polarization direction (clockwise with respect to north), & is the delay time in seconds, G(1) and a& are the la uncertainties, and #_ev refers to the number of the events used to calculate the weighted average. STA AOl A02 A03 A04 A05 A06 A07 A08 A09 AlO All Al2 A13 A14 A15 A16 A17 A18 A19 A20 A21 A22 A23 A24 A25 A26 A27 A28 A29 A30 Mi A32 A33 A34 A35 A36 A37 A38 A39 A40 A41 A42 A43 4, 69.53 67.01 72.72 75.00 72.00 67.49 61.29 72.27 79.69 76.73 73.96 71.76 74.81 77.88 69.75 66.44 69.09 68.11 67.17 62.74 68.00 64.89 64.93 65.16 65.72 64.85 64.29 64.10 54.80 61.49 69.49 72.46 68.99 65.58 68.59 67.00 73.44 74.80 83.00 75.20 76.02 84.07 86.85 +1+1+1+1+1+1+1+1+1+1+1- 1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1- i-I+1+1+1+1+1+1+1+1+1+1- 04) & 5.86 7.94 6.90 9.00 8.00 20.38 43.04 6.29 3.98 7.55 8.41 4.83 4.43 4.78 3.67 14.87 4.03 3.33 2.68 2.59 7.00 1.94 2.61 1.22 1.38 1.10 1.20 1.35 1.00 0.89 1.22 1.44 1.60 +1+1+1+1+1+1+1+1+1+1- 1.51 1- 1.60 1.64 1.64 1.67 1.50 1.64 1.59 1.45 1.78 1.35 1.63 1.63 1.73 1.70 1.73 1.84 1.85 1.57 1.64 1.87 1.96 2.10 1.67 2.21 2.09 1.99 1.96 1.65 1.79 1.73 1.68 +1+1+1+1+1+1+1+1- 1.73 1.69 3.01 2.03 3.65 8.59 11.64 5.82 1.89 1.79 3.72 3.59 7.95 1.73 4.73 25.00 3.97 5.05 6.00 5.60 1.61 (Y& -i-I- +1+1+1+1+1+1+1+1+1+1+1+1- i-I+1+1+1+1+1- 1+1+1+1+1- 0.17 0.34 0.15 0.25 0.25 0.42 0.98 0.17 0.15 0.23 0.30 0.36 0.22 0.28 0.21 0.55 0.18 0.20 0.22 0.20 0.25 0.26 0.20 0.14 0.14 0.20 0.13 0.18 0.26 0.39 0.29 0.27 0.18 0.21 0.23 0.43 0.20 0.22 0.90 0.17 0.21 0.27 0.44 #_ev 4 3 2 1 1 2 2 5 4 3 3 5 5 4 5 4 6 6 5 5 1 4 5 6 6 6 6 5 5 3 4 4 4 5 5 2 5 3 1 5 6 6 4 WEIGHTED AVERAGE OF ALL EVENTS 45 N 3150 3050 2950 2850 2750 2650 2550 2450 2350 2250 2150 2050 1950 1850 1750 1650 1550 1450 1350 1250 1150 1050 44N 950 850 750 650 550 450 350 250 150 50 25 0 124W 123W 122W Figure 6.1 White vectors show the average of spitting measurements at every station. The orientation of the vectors corresponds to and the length to &. 121W Ui 53 stations A435 to A515 into the third region will be discussed later in this chapter. Compared to the other two regions, where the results within a region are fairly constant, gradually increases from 73° at station A37 to 87° at station A43 is observed. The averages for these three regions are displayed in a rose diagram in figure 6.2a. Correlating splitting measurements with back-azimuth of the earthquakes, one can divide the events in two groups. The first group covers events with back-azimuths from 1300 to 280° and the second group from 300° to 350° (hereafter called southern and northern group, respectively; figure 5.2). Rose diagrams for these two groups showing the average splitting values for the three regions are displayed in figure 6.2b,c. On average, the northern events show 8° higher and 0.16 sec higher 6t than the southern events. The three regions across the array mentioned above are observed for both groups, however, are less marked for the southern events (figure 6.3). The reason that the three regions are less marked for the southern events might be the fact that the southern events are less suitable (i.e., event depth < 50 km, event distance 80° - 90°) for shear-wave splitting measurements than the northern events, but most likely the reason is the difference in deep seated anisotropy sampled by northern and southern events. A visual inspection of the seismograms of events from non-splitting back-azimuths provides additional constraint for separation of results into northern and southern backazimuths and the three regions along the array. The fact that the non-splitting back-azimuth of the Xinjiang earthquake, part of the northern events, is 4° larger than the non-splitting back-azimuth of the Vanuatu Isl. earthquake, part of the southern events, supports the conclusion that for northern events is somewhat larger than the one for the southern events. Gradually increasing amplitudes on the transverse component along the array for the non-splitting Xinjiang event (figure 5. lOb) eastward from station A37 (i.e., third region, stations A37-A43), are consistent with the measured gradual increase of the third region. within 54 00 sec 00 sec 00 C sec Figure 6.2 Vectors showing the average splitting for the Coast Range (first region, black), Willamette Valley and Cascades (second region, dark grey) and east of the Cascades (third region, light grey). The orientation of the vectors corresponds to 4 and the length to &. Figure (a) is the weighted average of all, figure (b) for the "northern", and figure (c) for the "southern" events. 56 The 1994 Fiji Is!. earthquake (figure 5.13) was recorded only at the eastern end of the array (A375-A5 15) with a back-azimuth close to the fast polarization direction of the southern events leading to inconclusive splitting measurements for stations A375 to A475. For the last three stations at the eastern end of the array (A485, A505, A515; no data were available for A495), however, stable results were obtained. This suggests that the tendency of c to rotate clockwise from A37 to A43 continues to station AS 15. Therefore the third region can be expanded up to station AS 15. The only event recorded at the eastemmost part of the array (A445-A515), suitable for splitting measurements, was the 1994 Fiji Is!. earthquake. Additional information for this part the array is supplied by the Vanuatu Isl. event (figure 5.12), which has a non-splitting back-azimuth (243°), suggesting a 4 of about 63°. Finally, an event from Argentina (figure 5.4), the only event from the southeast (back-azimuth 136°) recorded on stations AOl to A43, has been analyzed. Although this event is close to the slow polarization azimuth, and does not produce a well resolved result, it suggests of approximately 55°. This value is compatible with the average result for the southern events. In order to obtain constraints on the thickness and depth of the anisotropic layers, the splitting parameters were calculated for different frequency bands for the Philippine Island event, which is the event with the most stable measurements. Eight different period bands were chosen: ls-30s, 3s-30s, 5s-30s, is-lOs, 3s-8s, 3s-7s, 3s-6s and 4s-6s. The splitting measurements for all bands are shown in figure 6.4. The biggest change (with 4 rotating anticlockwise) in splitting measurements is obtained while going from 3s-30s band to 5s30s band. Results for is-lOs band show a clockwise rotation compared to the ls-30s band. The most significant changes occur in the first and the third region of the array. For 3s-8s, 3s-7s, 3s-6s and 4s-6s bands the splitting results are essentially the same, however, compared to the bands containing periods between lOs and 30s, in the Coast Range and 3600 45 N 3300 3000 2700 2400 2100 1 800 1500 44 N 1200 900 600 300 0 124W 123W 122W 121W Figure 6.4 Vectors showing splitting results for different frequency bands for the Philippine Isi. event. Upper array shows results for periods between 5s-30s, 3s-30s, is30s, and 1-lOsec (clockwise from north at every station). Lower array (vectors are displaced from station locations by 0.2°) shows results for periods between 3s-8s, 3s-7s, 3s-6s, and 4s-6s (clockwise from north at every station). east of the Cascades, is rotated clockwise. In summary, the largest frequency-related changes in splitting are seen at both ends of the array for waves with periods between 3 and 8 seconds. In general, cutting out long periods (lOs-30s) causes 4 to rotate clockwise and cutting out high frequencies causes to rotate anticlockwise. The significance of these observations will be discussed in the next chapter. 59 7.0 DISCUSSION 7.1 Localization of anisotropy 7.1.1 Thickness of the anisotropic layer Assuming the observed splitting is due to a single anisotropic layer, its thickness can be estimated by L= öt vs/k, where & is the measured delay time, v is the isotropic shear velocity of the layer and k is the coefficient of anisotropy (Silver and Chan, 1988). To estimate k, we assume that the anisotropy is caused by lattice preferred orientation of anisotropic minerals, mainly olivine, which is most abundant in the mantle, and that the long axis [100] of olivine (figure 1.1) is oriented parallel to the extension direction and the short axis [0101 is oriented parallel to the maximum shortening direction (Christensen, 1984). In the case of simple shear, which is an appropriate mode of deformation for the differential motion along plate boundaries, the [100] axis is contained within the flow plane, and is parallel to the flow direction (Nicolas and Poirier, 1976). Estimates for the coefficient of anisotropy in mantle rocks based on calculations from petrofabric strain measurements are around 4%, and are not expected to be higher than 8% (Mainprice and Silver, 1993). Assuming k=4% and v=4.5 km/sec, the thickness of the anisotropic layer across the TORTISS array varies from about 112 km. in the Coast Range, to as high as 248 km underneath station A35. These variations seem to be related to the geology. The Coast Range shows anisotropic layer thickness values of around 130 km, underneath the Willamette Valley the thickness increases to about 180 km, and reaches its high (248 km) underneath the Cascades, to decrease eastward to about 180 km at station A43. However, the observed delay times, could be caused as well by a half as thick anisotropic layer assuming k=8%. One should consider that changes in splitting along the array could be also caused by two anisotropic layers. According to Savage and Silver (1993), splitting caused by a two- layer anisotropy shows a it/2 periodicity with back-azimuth. Because the splitting observations by TORTISS do not display this pattern, a simple two-layer anisotropic model does not seem to apply here. However, the earthquakes studied here have a rather poor azimuthal coverage. 7.1.2 Depth of the anisotropic layer It is generally assumed that the primary source of the observed shear-wave anisotropy is located in the upper mantle. According to tests done with olivine and perovskite under high pressure to simulate lower-mantle conditions, the lower mantle (beginning at about 700 km depth) appears to be isotropic (Silver et al., 1993). The contribution of crustal anisotropy to splitting delay öt was shown to be about 0.3 sec in magnitude, based on measurements from P to S conversions at the Moho (Silver and Chan, 1991; McNamara et al., 1989). Barruol and Mainprice (1993) modelled the seismic properties of the crust using hypothetical polycrystals with typical crustal compositions and commonly measured petrofabrics and suggest delay times of about 0.1-0.2 s per 10 km of crustal rock. If the entire observed anisotropy were confined to the crust, assuming the crust is 45-km thick, the derived k of 18% would be unreasonably high. However, variations in splitting measurements along the TORTISS array could be explained in part by variations in crustal anisotropy. Evidence for shallow crustal anisotropy comes from the examination of the quasi Fresnell zones in different frequency bands for stations with different observed splitting measurements. Assuming the distance between two stations showing significantly different splitting is 16 km, in order to be able to resolve this difference, the distances from the stations to the anisotropic source regions should differ by at least ?14 (using the idea of Fresnell zones [Burnett et al., 1958], X is the wavelength). Accordingly, the depth of the anisotropic source region for ? =4, 8, 12, 16, and 20 km would not be deeper than 3 1.5, 15, 9, 6, and 4 km, respectively. The wavelengths were calculated for v = 4 km/s and periods of 1, 2, 3, 4, and 5 s, which are the contributing periods responsible for variations in shear wave splitting along the array found in the frequency tests discussed in the previous chapter. The above calculation gives only a very rough estimate because the anisotropic source region is not a point; a shear wave must propagate over sufficient distance to accumulate observable splitting. The estimate of the depth of the anisotropic source region could have an error of the order of factor 2. Nevertheless, the above estimates indicate significant contributions to the shear wave splitting from the crust. The largest short-wavelength changes in splitting (öt as high as 0.3 sec) occur in the third region. A change in & of 0.3 sec can be explained by variations of k between 0% and 4% in a 30 km thick anisotropic layer. Hence, in principle, it is possible to explain most of the observed variations of splitting along the array by variations in the anisotropy of the crust. In order to estimate the maximum depth of the anisotropic region which causes the splitting differences between the first and the second region, the Fresnell zone concept is applied. Using the average splitting measurement for each region and half the distance between the centers of the regions (56 1cm), the depth of the source region would be not 62 deeper than 390, 195, 130, 95, 75, 62, 53, 45 km. for ?. = 4, 8, 12, 16, 20, 24, 28, and 32, respectively. This estimate allows for the depth of the source region of the observed change anywhere within the crust and the upper mantle. 7.2 Relation of anisotropy to tectonic processes Previous studies provide evidence that mantle anisotropy is caused by strain induced lattice preferred orientation of upper mantle minerals (e.g., Nicolas and Christensen, 1987; Silver and Chan, 1991). From the three principal strain directions (shortening, intermediate, and extension) extension is predicted to be parallel to the [100] axis of olivine (Nicolas and Poirier, 1976; Christensen, 1984; Nicolas and Christensen, 1987) This prediction is valid for almost all kinds of finite strain: uniaxial extension and shortening (pure shear) and simple shear, which is a deformation mode found in the differential motion between a plate and the underlying mantle or a underlying plate. In the case of simple shear, the [100] axis is contained within the flow plain parallel to the flow direction (Nicolas and Poirier, 1976). The same behavior is predicted for generally parallel to the [100] axis. In order to relate between stress and strain is needed, because (1) since it is to stress, a constitutive relationship is primarily related to finite strain (i.e., parallel to the axis of the extension direction). The exact dependency is not known, but it may be reasonably guessed at. The fast polarization direction should be either parallel or perpendicular to the maximum horizontal stress (maximum compression) direction Y, depending on the way the stress is generated (Silver and Chan, 1991). If the strains are produced by a basal shear stress, then 4 should be parallel to the Yhmax direction. If the 63 strains result from stresses parallel to the plane of the plate, the fast polarization direction should be perpendicular to the Yj direction. As the origin for strain responsible for the orientation of the fast polarization direction, the following possibilities should be considered: In the mantle, the relative (RPM) and absolute plate motion (APM) of the JdF and the North American plates (APM JdF plate: N76.5°E; APM N. A. plate: N240°E; Gripp and Gordon, 1990) and backarc corner flow, and, in the crust, the maximum horizontal stress ((Yhmax). Basal shear strain caused by APM seems to be responsible for the lattice preferred orientation of mantle minerals. Silver and Chan (1991) analyzed the relation between and APM in detail, and found that the hypothesis holds only in tectonically stable areas of fast moving plates. Comparing the APM direction (67.5°) and the RPM direction (65°) of the JdF plate (with respect to North America), with the average 4 (68.5°) obtained for the first two regions (west of the Cascades) of the TORTISS array, suggests a possible relation between plate motion and . In this case the strain responsible for the orientation of the minerals would be related to the mantle flow induced by the motion of the subducting plate. Comparing the APM of the North American Plate (240°) with 4 (79°) obtained for the third region, east of the Cascades, a relation between the two parameters might be possible as well, although not as marked as west of the Cascades. Generally, for splitting measurements done using longer periods, 5 sec and above, tends to rotate anticlockwise, i.e., closer to the APM of the respective plates, implying that deeper seated anisotropy is more closely related to APM. For short periods (ls-5s) 4 tends to rotate clockwise, i.e., closer to the east-west direction, again with largest magnitudes at the ends of the array, where as stated in the previous chapter the changes in may have a shallow, crustal origin. Comparing Yhmax provided by Zoback et al. (1989), which is about N-S oriented in Oregon, with the fast polarization direction at the ends of the array (almost E-W), calculated for short periods as mentioned above, a relation between the two parameters for crustal anisotropy (i.e., perpendicular to Yhmax) is seems possible. East of the Cascades, the fast polarization axes might be aligned by the strain induced by a possible backarc corner flow. Assuming a backarc corner flow one would expect a downward flow somewhere in the vicinity of the Cascades. Since the [100] axis of olivine aligns with the flow and the angle of incidence of SKS is nearly vertical, splitting due to such flow should be barely detectable. The observation of 2 sec split time delays in the vicinity of the Cascades argues against the significance of the corner flow mechanism. 7.3 Slab effect In a subduction zone, the knowledge of how a change in the dip of the slab would affect the splitting might be useful for the interpretation of the results. The effect of the change of slab dip can be estimated theoretically by varying the dip angle () of a plane (figure 7.3.1). The angle between the direction of the [100] axis of a olivine crystal and the dip direction of a plane in which it is imbedded is p. The angle between the surface projections of these two directions is a. Increasing the dip angle (), while keeping p constant, causes an increase in a. For example, assuming p of 20°, an increase in dip of 40° would increase a by 15° (figure 7.3.1). The relation between a and fast polarization direction, is = 90° a. , the observed Furthermore, since the [100] axis of olivine lays in the subducting slab, the angle between the [1001 axis of olivine and the ray propagation decreases for steeper slab angles and therefore the delay times should decrease with increasing dip angle. 65 Figure 7.3.1 The horizontal plane projections (bold) of vectors in the dipping planes. The relationship between the angles is: tan ix tamp / cosE. Evidence for a bend in the slab is brought by Nábëlek et al. (1993) in a receiver function study using data from the TORTISS array, who suggest a change in slab dip possibly around station A14. Hence, the observed change in the orientation of between regions "AO1-A14" and "A15-A36", (chapter 6) could perhaps be explained with a change in dip of the anisotropic layer. However, the delay times for the second region are higher than the delay times for the first region, which is in contradiction with our prediction. Larger delays might be caused by higher straining (i.e., higher anisotropy coefficient) within the steeper dipping slab, or by overlying anisotropic material which may be compensating for the decrease in the delay time. 7.4 Interpretation summary This chapter summarizes the most likely causes for the observed anisotropy along the TORTISS array. The summary is based on the discussions in the previous chapters and is presented for each of the regions of the array. Each region will be discussed in terms of deep (i.e., main, upper mantle) contribution and shallow (i.e., small, crustal) contribution to anisotropy. Spitting measurements throughout the array display a back-azimuth dependency. Therefore the events have been separated in two groups, the northern and the southern, within which splitting results are roughly consistent. The difference in splitting results for the two event groups is most likely caused by differences in deep anisotropy, somewhere between the CMB and the crust underneath the array, because a shallow source would most likely be sampled by both event groups. 67 7.4.1 Coast Range The major contribution to the anisotropy in the first region (A01-A14), the Coast Range, is most likely in the upper mantle. This estimate is based on the thickness of the required anisotropic layer (130 km) and the "Fresnell depth" of the Coast Range, considering the Coast Range as a region with uniform anisotropy. The mineral orienting strain is most likely caused by the APM of the JdF plate (i.e., the motion of the subducting slab). Because the changes in the observed splitting are on the order of measurement uncertainties, it is not possible to say if the crust has a significant contribution to the observed anisotropy. Nevertheless, because the thickness of the crust in the Coast Range is about 40 km (Trehu et al., 1994), a crustal contribution to delay times of up to about 0.3 sec is possible. 7.4.2 Willamette Valley and Cascades The major contribution to the anisotropy in the second region (A15-A36), the Willamette Valley and the Cascades, is most likely in the upper mantle as well, as suggested by the "Fresnell depths". The assumed thickness of the anisotropic body is on average about 180 km. The fast polarization direction appears to be related to the motion of the subducting slab. The crustal contribution for this region, if any, is not easily distinguished. Differences in splitting results between stations are very small, and might be just measurement uncertainties. Because the results are practically indifferent to the applied frequency bands, "Fresnell depths" are not constrained. The change in the average 4) from the first region to the second region might be due to a change in the dip of the slab. However, the delay time is increasing instead of decreasing as expected from the model simulating the change in anisotropy caused by a dip change in the slab. A more likely possibility is that the change in the average 4) for the two regions is due to crustal anisotropy or a combination of both mechanisms. The boundary between region 1 and 2 does not appear to be clearly correlated with principal crustal structures of the Coast Range and the Willamette Valley (Trehu et al., 1994). However, receiver function data suggest that the boundary roughly corresponds to the junction between the JdF plate's Moho and the North American plate's Moho (Nábëlek et al., 1993). 7.4.3 East of the Cascades The main contribution to the observed anisotropy in the third region of the array (A37-A5 15) comes most likely from the mantle as well. The strain responsible for the orientation of the minerals might be caused by the APM of the North American plate (N240°E; Gripp and Gordon, 1990). The fast polarization direction calculated for long periods (5-30s) is within 100 to the direction of the plate motion. Short period results show a more E-W tendency, indicating a possible relation of 4) to maximum horizontal stress in the crust. The crustal contribution in this region seems to be significantly higher than in the other two regions. The biggest changes in splitting results from station to station are in this part of the array. They are most likely caused by short scale crustal variations. 70 8.0 CONCLUSIONS The results of this thesis show that anisotropy varies across a subduction complex. The average observed splitting parameters, the fast polarization direction (4)) 700 and the time delay (&) 1.61 sec, are well above average. Worldwide the largest observed splitting measurements are of the order of 2 sec time delay. According to gross anisotropic characteristics, the array can be divided in three regions. The first region extends over the Coast Range and has average splitting values of [72°, 1.34 sec], the second region extends over the Willamette Valley and the Cascades with an average splitting of [66°, 1.74 sec] and the third region is located east of the Cascades with an average splitting of [79°, 1.77 sec]. The splitting results within each of the first two regions show little systematic variations. In the third region, a systematic rotation of 4) from about 70° to 87° from west to east is observed. The observed shear-wave splitting is assumed to be caused by strain induced lattice preferred orientation of mantle minerals, mainly olivine. The general trend of the 4) and & estimates is most likely due to the APM of the JdF plate (i.e., mantle flow induced by the motion of the descending slab) and the North American plate at the eastern end of the array. Variations in the 4)-6t estimates across the TORTISS array can be explained by variations in crustal anisotropy; however, only east of the Cascades (third region) can the contribution of crustal anisotropy be resolved. A change in dip of the descending slab might contribute to these variations as well. Backazimuth dependent variations can be best explained by deep anisotropy along the raypaths of the northern and southern event groups, anywhere between the CMB and the crust underneath the array. Future work should concentrate on the determination of crustal anisotropy from the P to S phase conversion at the North American and Juan de Fuca Mohos. This would 71 provide better constraints on the nature of the observed splitting changes across the array. TORTISS array data can be used for this task. Furthermore, better data between stations A43 and AS! 5 would be desirable and the investigation of the anisotropy east of station AS 15 would be interesting. 72 i:iu :i p (iItit1 :a,i Atwater, T , Implications of plate tectonics for the Cenozoic tectonic evolution of western North America, Geol. Soc. Am. Bull., 81, 3513-3536, 1970. Babuska, V., and M. Cara, Seismic Anisotropy in the Earth, Kiuwer, Dordrecht, 1991. Baldwin, E. M., Geology of Oregon, Kendal and Hunt, 1976. Barruol, G., and D. Mainprice, A quantitative evaluation of the contribution of crustal rocks to the shear-wave splitting of teleseismic SKS waves, Phys. Earth. Plan. mt., 78, 281-300, 1993 Barruol, G., and A. Souriau, Anisotropy beneath the Pyrenees range from teleseismic shear wave splitting: results from a test experiment, Geophys. Res. Lett., 22, 493496, 1995. Birch, F., The velocity of copressional waves in rocks to 10 kilobars, J. Geophys. Res., 65, 1083-1102, 1960 Bostock, M. G., and J. F. Cassidy, Variatons in SKS splitting across western Canada, Geophys. Res. Lett., 22, 5-8, 1995. Bowman, J. R., and M. Ando, Shear-wave splitting in the upper-mantle wedge above the Tonga subduction zone, Royal Astronomical Society Geophys. .1., 88, 89-111, 1987. Burnett, C. R., J. G. Hirschberg, and J. E. Mack, Diffraction and interference; in Handbook of Physics, Part 6, chapter 5, pp. 6.81-4 (ed. F. U. Condon and H. Odishaw): New York, Mc Graw Hill, 1958 Christensen, N. I. , The magnitude, symmetry and origin of upper mantle anisotropic based on fabric analyses of ultramafic tectonics, Geophys. J. R. Astron. Soc., 76, 89-112, 1984. Crampin, S., Effective anisotropic elastic constants for wave propagation through cracked solids, Geophys. J. R. Astron. Soc., 76, 135-145, 1984. Crampin, S., E. M. Chesnokov, and R. G. Hipkin, Seismic anisotropy The state of the art; II, Geophys. J. R. Astron. Soc., 76, 1-16, 1984. Duncan, R. A., and L. D. Kuim, Plate tectonic evolution of the Cascades arc-subduction complex, in Winterer, E. L., D. M. Hussong, and R. W. Decker, eds., The Eastern Pacific Ocean and Hawaii, The Geology of North America, Geol. Soc. Am., vN, 413-438, 1989. Harris, R. A., H. M. Iyer, and P. B. Dawson, Imaging the Juan the Fuca Plate Beneath Southern Oregon Using Teleseismic P wave Residuals, J. Geophys. Res., 96, 19,879-19,889, 1991. Hellfrich, G., P. Silver, and H., Given, Shear-wave splitting variation over short spatial scales on continents, Geophys. J. mt., 119, 561-5 73, 1994. 73 Leven, J. N., L. Jackson, and A. E. Ringwood, Upper mantle seismic anisotropy and lithospheric decoupling, Nature, 289, 234-239, 1981. Mainprice, D., and P. G. Silver, Constraints on the interpretation of teleseismic SKS observations from kimberlite nodules from the subcontinental mantle, Phys. Earth Planet. Inter., 78, 257-280, 1993. Makeyewa, L. I., L. P. Vinnik, and S. W. Roecker, Shear-wave splitting and small-scale convection in the continental upper mantle, Nature, 358, 144-147, 1992. McNamara, D. E., T. J. Owens, Azimuthal Shear Wave Anisotropy in the Basin and Range Province Using Moho Ps Converted Phases, J. Geophys. Res., 98, 12,00312,017, 1993. McNamara, D. E., T. J. Owens, P. G. Silver, and F. T. Wu, Shear Wave Anisotropy Beneath the Tibetan Plateau, personal communication, 1994. Nabelek, J., X-Q. Li, S. Azevedo, J. Braunmiller, A Fabritius, B. Leitner, A. Trehu, and G. Zandt, A High-Resolution Image of the Cascadia Subduction Zone From Teleseismic Converted Phases Recorded by a Broadband Seicmic Array, AGU abstract, 1993. Nicolas, A., and N. I. Christensen, Fonnation of anisotropy in upper mantle peridotites - A review, in Composition, Structure and Dynamics of the Lithosphere- Asthenosphere System, Vol. 16 (eds. Fuchs, K. and Froidevaux, C.), 111-123, (Am. Geophys. Un., Washington D.C.), 1987. Nicolas, A., and J. P. Poirier, Crystaline Plasticity and Solid State Flow in Metamorphic Rocks, 444 pp., John Wiley, New York, 1976. Owens, T. J., R. S. Crosson, and M. A. Hendrickson, Constraints on the subduction geometry beneath western Washigton from broadband teleseismic waveform modeling, Bul. Seis. Soc. Am., 78, 1319-1344, 1988 Rasmusen, J., and E. Humphreys, Tomagraphic Image of the Juan the Fuca Plate Beneath Washington and the Western Oregon Using Teleseismic P-Wave Tavel Times, Geophys. Res. Lett., 15, 1,417-1420, 1988. Savage, M. K., X. R. Shih, R. P. Meyer and R.C. Aster, Shear-wave anisotropy of active tectonic regions via automated S-wave polarization analysis, Tectonophysics, 165, 2 79-292, 1989. Savage, M. K., and P. Silver, Mantle deformation and tectonics: constraints from seismic anisotropy in the western United States, Phys. Earth. Planetary Interiors, 78, 207227, 1993. Sherrod, D. R., and J. G. Smith, Preliminary map of upper Eocene to Holocene volcanic and related rocks of the Cascade Range, Oregon, U.S. Geol. Sun'. Open File Rep., 89-14, 1989. Shih, X. R., R. P. Meyer, and J. F. Schneider, Seismic anisotropy above a subducting plate, Geology, 19, 807-810, 1991. 74 Silver, P. G., and W. Chan, Implications for continental structure and evolution from seismic anisotropy, Nature, 335, 34-39, 1988. Silver, P. G., and W. Chan, Shear Wave Splitting and Subcontinental Mantle Deformation, J. Geophys. Res., 96, 16,429-16,454, 1991. Silver, P.G., and S. Kaneshima, and C. Meade, Why is the the lower mantle so isotropic?, EOS Trans. AGU, 74, 1993 Trehu, A. M., I. Asudeh, T. M. Brocher, J. H. Luetgert, W. D. Mooney, J. L. Nabelek, and Y. Nakemura, Crustal Architecture of the Cascadia Forearc, Sience, 265, 237243, 1994. Vinnik, L. P., V. Fara, and B. Romanowicz, Azimuthal anisotropy in the Earth from Observations of SKS at GEOSCOPE and NARS broadband stations, Bul. Seism. Soc. Amer., 79, 1,542-1,558, 1989. Vinnik, L. P., and R. Kind, Ellipticity of teleseimic S-particle motion, J. Geophys. mt., 113, 165-1 74, 1993. Vinnik, L. P., L. I. Makeyewa, A. Milev, and A. Y. Usenko, Global patterns of azimuthal anisotropy and deformations in the continental mantle, Geophys. J. mt., 111, 433447, 1992. VanDecar, J. C., Upper-mantle Structure of the Cascadia Subduction Zone from Nonlinear Teleseismic Travel-time Inversion, thesis, UW, 1991. Wilson, D. S., Confidence Intervals for the Motion and Deformation of the Juan de Fuce Plate, J. Geophys. Res., 98, 16,053-16,071, 1993. Zoback, M. L., M. D. Zoback, J. Adams, M. Assumpacao, S. Bel, et. all, Global patterns of tectonic stress, Nature, 341, 291-298, 1989. 75 APPENDICES 76 APPENDIX A. 1 SPLITTING RESULTS, ANISOTROPY VECTOR MAP AND ENERGY CONTOUR PLOTS FOR: PHILIPPINE ISL. REG. 93/05/18 10:19:39 Mw=6.5 dist=92.34° baz=300.79° (1-30s) Table A. 1.1 Splitting measurements for the Philippine Isl. event. STA refers to the station, parameter (1) is the fast polarization direction (clockwise with respect to north), öt is the delay time in seconds, and a4) and a are the la uncertainties. SFA AOl A02 A03 A04 A05 A08 A09 AlO All Al2 Al3 A15 A17 Al8 A20 A21 A23 A24 A25 A26 A27 A28 A31 A32 A33 MS A37 A38 MO A41 A42 a 4, 73.00 67.00 73.00 75.00 72.00 75.00 80.00 77.00 74.00 72.00 74.00 67.00 74.00 70.00 59.00 68.00 70.00 69.00 69.00 68.00 67.00 68.00 70.00 68.00 69.00 68.00 74.00 75.00 76.00 79.00 83.00 +1- 7.00 +1+1- 8.00 7.00 9.00 8.00 7.00 4.00 8.00 9.00 16.00 7.00 5.00 7.00 5.00 5.00 7.00 6.00 5.00 4.00 6.00 4.00 5.00 6.00 6.00 4.00 4.00 5.00 5.00 5.00 7.00 9.00 +141+1+1+1+1+1+1+1+1+1+1+1+1- +1+1+1+1+1- +1+1+1+1+1#1-i-I- +1+1- 1.30 1.40 1.10 1.20 1.35 1.30 1.45 1.65 1.65 2.10 1.65 1.65 1.65 1.70 1.90 1.35 1.60 1.70 1.65 1.70 1.80 1.85 1.90 1.90 2.00 2.30 1.95 +1- 0.20 0.35 0.15 0.25 .iI- 025 +1- 0.20 0.15 0.25 0.35 0.90 0.25 0.25 0.25 0.25 0.35 0.25 0.25 0.20 0.20 0.30 0.20 0.25 0.30 0.30 0.20 +/+1+1- +1+1+141- 1. +1+1+1- +1+1-i-I- +1+1+1+1+1- 41+1+1+141- 2.05 1.90 -4-1- 1.85 1.75 -+/- --1- +1- 025 0.25 0.25 0.20 0.30 0.35 77 Figure A. 1.1 Anisotropy vector map for the Philippine Isi. event. The vector orientation corresponds to in (°) clockwise from north, and the size corresponds to the delay time in seconds. Figure A. 1.2 Contour plot of energy on the corrected transverse component for all (, St) pairs for the Philippine Isl. event. The absolute minimum is indicated by a star and the 95% confidence level is shown within the bold contour line. PHILIPPINE ISL. REG. 93/05/18 10:19:35 Mw=6.5 dist=92.34° baz=300.79° (is -30s) 3600 45 N 3300 3000 2700 2100 1800 1500 44 N 1200 900 600 300 0 124W Figure A.1.1 123W 122W 121W i H - : !FLfJ22J:. !PcH. '&Pi t JL:2J7 1 -I I I - - -. --I --S H i_' I - - ___ --I --a H I I - - - - S --S H -. I I - E'- --I S --I .1 A37 A35 A33 930518 0 0 U) a) a) a) E E E I>' >. > a) 0 a) a) a) a) a) a) a) a) 0 0 0 0 Fast Angle ( deg) Fast Angle (deg) Fast Angle (deg) A40 A38 0 0 a) a) a) a) a) a) a) E E E >. >.. >' a) a) a) a) a) a) a) 0 oI Fast Angle (deg) A42 0 a) a) a) E Iii a) a) 0 Fast Angle (deg) Figure A.12 Continued Fast Angle (deg) I Fast Angle (deg) APPENDIX A.2 SPLITTING RESULTS, ANISOTROPY VECTOR MAP AND ENERGY CONTOUR PLOTS FOR: TAiWAN 93/08/07 00:00:37 Mw=6.4 dist=85° baz=303° (ls-30s) Table A.2. 1 Splitting measurements for the Taiwan event. STA refers to the station, parameter (1) is the fast polarization direction (clockwise with and G6t are the icY uncertainties. respect to north), & is the delay time in seconds, and STA AOl A02 A03 A05 A06 A08 A09 AlO All Al2 A13 A14 A15 A16 A17 A18 A19 A20 A22 A23 A24 A25 A26 A27 A28 A29 A30 Mi A32 A33 A34 MS A37 A38 A39 A40 A41 A42 A43 & 4) 67.00 62.00 63.00 76.00 72.00 63.00 68.00 76.00 76.00 73.00 73.00 79.00 67.00 72.00 70.00 76.00 65.00 61.00 65.00 69.00 67.00 66.00 67.00 66.00 69.00 72.00 73.00 77.00 68.00 74.00 72.00 74.00 78.00 75.00 83.00 81.00 82.00 87.00 81.00 +1- +/+1- +1- 4/4/+1+1- +/- 4/*1+1+1+1- 4/+/4/4/4/+/4/+1+1+1+1+1- 4/+1- 4/4/4/. 4/+1- +/+1+1+1+1+1- 20.00 90.00 41.00 90.00 23.00 18.00 90.00 27.00 26.00 23.00 23.00 6.00 16.00 36.00 10.00 42.00 11.00 10.00 10.00 33.00 10.00 12.00 28.00 7.00 12.00 30.00 19.00 90.00 24.00 90.00 13.00 33.00 24.00 20.00 25.00 13.00 24.00 14.00 18.00 1.10 0.95 1.00 0.65 4/- 1.05 1.15 4/- 0.95 1.20 1.20 +1- 1.35 1.40 1.60 1.70 1.70 1.80 1.35 1.40 2.00 1.45 1.40 1.60 1.65 1.70 1.90 1.90 1.75 1.80 1.60 1.90 1.75 1.70 1.85 1.70 1.50 1.65 1.70 1.90 1.85 1.60 -i-I- 4/- 141 +1- 4/4/4/4/+1+1- +1+1- 4/iI+1- 4/+1- 4/4/4/+1- 4/+1- 4/- 0.50 2.05 1.00 2.35 0.45 0.55 2.05 0.65 0.60 0.55 0.60 0.30 0.65 1.30 0.40 1.50 0.45 0.75 0.40 1.10 0.45 0.60 120 0.35 0.65 1.20 0.85 4/- 1.55 1.05 1.70 *1- 0.45 +1- 4/. 1.15 4/- 0.90 0.55 0.90 +1- 4/4/+1- 4/+1- 030 0.95 0.60 0.60 Figure A.2. 1 Anisotropy vector map for the Taiwan event. The vector orientation corresponds to in (°) clockwise from north, and the size corresponds to the delay time in seconds. Figure A.2.2 Contour plot of energy on the corrected transverse component for all (tv, öt) pairs for the Taiwan event. The absolute minimum is indicated by a star and the 95% confidence level is shown within the bold contour line. NORTHEAST OF TAIWAN 93/08/07 00:00:37 Mw=6.4 dist=85.38° baz=303° (30s highpass) 3600 45 N 3300 3000 2700 2400 2100 1803 1500 44N 1200 900 603 303 0 124W Figure A.2.1 123W 122W 12VW I 0, 16 0, E A03 A02 3O8O7 "0 a, I>' a, U, a) a) E E !k1 ! I- > >S (U (0 (0 a) a) 02 I:L .90 .75 -60 -45 - .15 0 15 30 45 60 75 00 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A07 A06 A05 3.' 2.1 6.) a) a) U) (6 a) a) a) E u E E (U (U a) a) 20 I: >' a) LV > 11 O .0 9 .90.7590.45.30.150 153049607590 .95.75.00.45.30.150153045607595 .45.30.150 153045007500 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) AlO A09 I 0) a) 2.0 A U, (I) 07 a) a) E E is E I- >' (U (U a) a, 0.5 IL 0.0 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A13 Al2 All C) a, (6 a) a) E t a) a) >' >' 20 U, E E ) >' ;1 (0 IC a) 0) Fast Angle (deg Figure A.2.2 0) Fast Angle (deg Fast Angle (deg A15 A14 930807 03 0 a, a, a, C-, a 0 a, a a E a, E F E F >. I>. a CD a, a, a, 0 0 Fast Angle (deg) Fast Angle (deg) A19 S 5.1 25 0 a, 23 a a, 13 E * F a> ID a, 0 051 0_a -90 -75 .40 .45 .30 .15 . A20 0 15 30 45 00 75 01 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A22 23 1.2 a, E F C-, C.) a 0, U) a, 2.0 a, a, 13 E E > I>S F >' a a 1.0 a, 0 0, o 0 0.5 D.0 .95 .75 .40 -45 -30 -15 0 15 30 45 00 75 50 .90 -75 -40 -45 .30 -15 Fast Angle (deg) 0 15 30 45 90 75 90 Fast Angle (deg) Fast Angle (deg) 3.0 . A26 A25 A24 3.0 I 231 23 0 a, 20 2.0 a a, EisU F . E F Oil a 0 0 -90.75-00-45-30-150 153045007590 Fast Angle (deg) Figure A.2.2 Continued 75 Fast Angle (deg) - as 30 Fast Angle (deg) A29 A28 A27 930807 C) C) 2$I I 24 a I a In EuI Ei.s >'lI E > a a .I.0L I V I - Fast Angle (deg) -75 - A30 .45 - -is 0 Is 30 45 40 75 ID Fast Angle (dog) Fast Angle (deg A32 A31 2$ Li 2O a 1.0 '.0 CI) * a a Els IL\J E L5 a LO '- >. >-, a 0$ a a '5 SC -90 -75 -40 -45 -30 -15 0 15 30 45 40 75 a E 3 45 .5 '5 IL 90 .75 .40 -45 -30 -15 Fast Angle (deg) AM A33 0 15 30 45 ao 75 90 Fast Angle (deg) Fast Angle (deg) A35 - I a CO a E I.- >' CO 0) O v,. -90 .75 .40 .45 .30 .15 0 15 30 45 00 75 10 -75 -00 .45 -30 -IS 91 Fast Angle (deg) 0 15 30 45 40 75 90 Fast Angle (deg) A37 A39 a CO ID E >. (0 a p Fast Angle (deg) Figure A.2.2 Continued Fast Angle (deg) FiiiiiI1 Fast Angle (deg [SDJ - - - - - Fast Angle (deg a a E P a a Fast Angle (deg Figure A.2.2 Continued - - - -' ---- - - - Fast Angle (deg - - - -' _I-_ Fast Angle (deg - - APPENDIX A.3 SPLITTING RESULTS, ANISOTROPY VECTOR MAP AND ENERGY CONTOUR PLOTS FOR: SOUTH OF FIJI ISLANDS 93/08/07 17:53:27 560 6.9 DIST: 85.68° BAZ: 230.25° Table A.3. 1 Splitting measurements for the 93 Fiji Is!. event. STA refers to the station, parameter is the fast polarization direction (clockwise with respect to north), & is the delay time in seconds, and a4) and a& are the la uncertainties. (1) STA AOl A02 A05 A06 A08 A09 AlO All Al2 A13 A15 A16 A17 A18 A19 A20 A22 A23 A24 A25 A29 A31 A32 A33 A34 A35 A37 A38 A40 A41 A42 A43 64.00 57.00 60.00 59.00 64.00 60.00 64.00 61.00 60.00 62.00 57.00 57.00 75.00 59.00 53.00 43.00 54.00 51.00 48.00 40.00 50.00 56.00 56.00 55.00 59.00 72.00 56.00 62.00 64.00 68.00 59.00 64.00 +1+1+1+1+1+1+1+1+1+1+1+1+1-i-I-i-I-i-I- +1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1- a4 & 9.00 21.00 4.00 5.00 5.00 52.00 12.00 9.00 14.00 5.00 5.00 6.00 45.00 5.00 4.00 90.00 90.00 90.00 90.00 90.00 89.00 90.00 3.00 11.00 69.00 14.00 4.00 85.00 73.00 19.00 10.00 10.00 2.40 3.00 2.85 2.60 2.30 2.45 1.95 2.05 1.70 1.95 3.00 3.00 0.75 2.45 3.00 3.00 3.00 3.00 3.00 0.40 3.00 2.70 3.00 2.30 0.90 1.35 3.00 1.15 0.65 1.05 3.00 2.10 +1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1+1-i-I-i-I- +1+1+1+1+1+1+1+1+1+1+1+1- 1.20 2.50 0.60 0.95 0.65 2.00 1.05 1.00 1.30 0.80 1.50 1.90 1.10 0.75 1.30 2.95 2.90 2.95 2.95 2.60 2.85 2.15 1.55 1.80 2.10 0.70 0.85 1.85 2.35 0.80 2.20 1.00 APPENDIX A.4 SPLITTING RESULTS, ANISOTROPY VECTOR MAP AND ENERGY CONTOUR PLOTS FOR: HINDU KUSH 93/08/09 12:42:50 mb=6.3 dist=98° baz=349° (ls-30s) Table A.4. 1 Splitting measurements for the Hindu Kush event. STA refers to the station, parameter is the fast polarization direction (clockwise with respect to north), & is the delay time in seconds, and a and a& are the la uncertainties. (I) & STA Al2 A13 A14 All A18 A19 A20 A22 A23 A24 A25 A26 A27 A28 A29 MO A3l A32 A33 A34 A35 A37 A41 A42 A43 72.00 73.00 74.00 70.00 70.00 68.00 70.00 65.00 64.00 65.00 66.00 67.00 66.00 63.00 63.00 61.00 72.00 73.00 69.00 67.00 72.00 73.00 84.00 87.00 88.00 +1-4]- +1-4]- W-41-4]- +1- +/-+14]- +1-4]- +1+1+1-i-I-i-I- +1i-I- 6.00 12.00 65.00 69.00 5.00 3.00 4.00 2.00 3.00 2.00 2.00 4.00 3.00 20.00 45.00 53.00 32.00 2.00 2.00 5.00 9.00 2.00 1.90 1.95 1.90 1.05 1.75 +/- 2.40 2.05 2.00 -i]4]- 1.10 1.40 1.70 1.20 +1- 0.80 0.60 0.65 0.45 0.45 0.30 0.35 0.45 0.50 0.80 +1- 1.05 +1- 1.00 -4]- -1-1- 2.05 2.15 -4]- 1.95 1.80 1.25 1.05 +1- 0.75 1.90 3.00 2.70 2.10 -4-1- -4]- +1+1+1+1- 130 0.95 0.45 0.65 1.85 -i-I- 1.15 -i-I- 0.90 .4]- -4]- +1- 90.00 11.00 -i-I- 6.00 3.00 i]- -4]- 1.85 2.80 3.00 3.00 -if- +1- +1- 1.90 1.90 1.25 91 Figure A.4. 1 Anisotropy vector map for the Hindu Kush event. The vector orientation corresponds to in seconds. in (°) clockwise from north, and the size corresponds to the delay time Figure A.4.2 Contour plot of energy on the corrected transverse component for all (4, t) pairs for the Hindu Kush event. The absolute minimum is indicated by a star and the 95% confidence level is shown within the bold contour line. HINDU KUSH 93/08/09 12:42:50 nth=6.3 dist=98° baz=349° (is - 30s) 3600 450 N 3300 3000 2700 2400 2100 1800 1500 44°N 1200 900 600 300 0 124°W Figure A.4. 1 123°W 122°W 121°W 93 A03 A02 AOl 930809 C) 0) In C) a) C) E E a) '3 I- > > 10 a) a) a a) a 1.) a) (I) a) E I- >. a) a) . Fast Angle (deg) A07 C) a) a) 0) E I a) a) O 5 - -75 -SI) -40 -30 IS 0 IS 30 45 60 Fast Angle (deg) AlO 2.5 C.) 2.0 a) E '.s I- a) 1,0 a 0.5 0.0 Fast Angle (deg) Figure A.4.2 Fast Angle (deg) AlS A14 0 a, U a) a, a, a, a, E E F I- >' a, a, a, 0 0 L1)] Fast Angle (deg) A18 0 C) a, a, a, a) a, E E a) I- >' a, a, a, a) 0 0 - ') i'j I .0 i I____________________ 0 a, a) E I- - a, a, 0 0 L, a) a) a, E a, a, a I 0a, a, a, E F 11 a, a, 0 -90 -75 -60 -45 -90 -15 0 15 90 45 60 Fast Angle (deg) Figure A4.2 Continued 75 90 A26 930809 U a) 0, a) E F- a) 0) Fast Angle (deg) A29 3.0 2.5 i U 2.0 a) E 1.5 I- - a) 10 a 05 0.0 Fast Angle (deg) - A32 (3 a) 0) a) E FC, a) a Fast Angle (deg) A35 C) a) Co a) E F- a) a) a ---: .75 -60 -45 -30 -15 0 15 30 45 60 Fast Angle (deg) Figure A.4.2 Continued 75 90 Fast Angle (deg) iI\. V W A40 439 930809 U C) a a a a C., a E E F>' >. 0 0 a a a a .90 -75 .00 -45 -30 -15 0 15 30 45 90 75 90 Fast Angle ( deg) Fast Angle (deg) A43 A42 C) C.) a a a 2 U) a E a E I F > > a a a. 0 0 Ui oo .75 -00 -45 .30 .15 Fast Angle (deg) Figure A.4.2 Continued 0 15 30 45 00 75 90 Fast Angle (deg) 97 APPENDIX A.5 SPLITTING RESULTS, ANISOTROPY VECTOR MAP AND ENERGY CONTOUR PLOTS FOR: NEW IRELAND REGION 93/09/06 03:55:58 Mw=6.6 dist=88.57° baz=262° (ls-30S) Table A.5. 1 Splitting measurements for the New Ireland event. STA refers to the station, parameter 4) is the fast polarization direction (clockwise with and & are the la uncertainties. respect to north), & is the delay time in seconds, and & STA AOl A02 A07 A08 A09 AlO All Al2 A13 A14 A15 A16 A17 A18 A19 A20 A22 A23 A24 A25 A26 A27 A28 A29 A30 A33 AM A35 A36 A38 A40 A41 A42 72.00 73.00 69.00 58.00 67.00 71.00 62.00 65.00 66.00 70.00 67.00 70.00 67.00 70.00 67.00 61.00 59.00 55.00 59.00 55.00 53.00 54.00 58.00 64.00 70.00 58.00 49.00 61.00 67.00 71.00 71.00 75.00 74.00 +1- 1+1-W- +1+1+1+1+1+1+1+1-i-I- +1+1- 11-1-1- +1+1+1+1+1-i-I- -i-I- +1+1+1+1+1-i-I- -i-I- +1- +1+1- 1.55 1.75 0.85 0.85 +1- 2.15 1.10 1.15 1.75 1.10 1.55 1.50 1.55 1.65 1.55 +1+1+1- 1.55 +1+1+1+1- 0.85 2.10 +1+1- 15.00 90.00 2.15 90.00 55.00 67.00 42.00 58.00 17.00 27.00 90.00 65.00 90.00 8.00 23.00 10.00 8.00 18.00 28.00 9.00 12.00 18.00 7.00 10.00 33.00 77.00 63.00 37.00 58.00 73.00 21.00 90.00 90.00 90.00 1.25 1.80 2.00 2.00 1.50 1.30 1.85 1.80 1.70 2.00 2.10 1.85 1.65 1.25 1.20 1.45 1.35 2.10 1.20 2.10 2.00 1- -4-!- +1+1+1+1+1+1+1+1- 1+1+1+1+1- 1+1+1+1+1+1- 130 1.35 1.05 1.50 1.40 1.50 0.90 1.25 1.00 0.70 0.65 0.80 0.50 0.55 0.65 0.35 0.70 1.15 1.55 1.75 0.95 1.45 1.65 1.40 1.80 2.05 1.95 Figure A.5.1 Anisotropy vector map for the New Ireland event. The vector orientation corresponds to 4 in (°) clockwise from north, and the size corresponds to the delay time in seconds. Figure A.5.2 Contour plot of energy on the corrected transverse component for all (, 6t) pairsfor the New Ireland event. The absolute minimum is indicated by a star and the 95% confidence level is shown within the bold contour line. NEW IRELAND REGION 93/09/06 03:55:58 Nw=6.6 dist=88.57° baz=262.1° (30s highpass) 3600 45 N 3300 3000 2700 2400 2100 1800 1500 44N 1200 900 600 300 0 124W Figure A.5.1 123W 122W 121W 100 AOl 930906 2$ 2$ __ 1k 0 E 1.01 I 1 o ft 00 2.0 III i A03 A02 3.0 3.0 Ii 45 ID 113 E o 0 WIl >.- 11(11 1J!III Ca A'fflflIl a) 0.0 0.0 .00.75.00-45.30.150153043007500 .00.75.00-40-30-100103045007500 Fast Angle (dog) Fast Angle (deg) Fast Angle (deg) A07 A06 A05 0 0.) a, 63 CO 00 (I) U) 0) 00 E E Cs 00 CO Ca 00 0 a) E I > 0 a) o ii Fast Angle (dog) Fast Angle (deg) Fast Angle (dog) AlO A09 A08 a.) 0) 5) CO C') 00 03 E E 0$ 0 a) U) a) E I- I>' >-. 00 CO CO 02 0 00 0 0 .90 -IS -Go .45 .30 -IS 0 15 30 45 00 75 a) 90 .90 .75 .00 .45 .30 .15 0 IS 30 45 00 75 90 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) All Al2 3.0 A13 3r 3.) 2$ 0 C-) 00 2J 002.) 2.0 Cl) U) 0) ID Eu Eu Ei.s U2$ > >S CO o 0.5 05 - 0) 0 0.) 02 ox -90 -15 .00 -45 .30 -15 0 IS 30 45 Fast Angle (deg) Figure A.5.2 60 75 90 Fast Angle (deg) Fast Angle (dog) 101 U6 &15 A14 930906 1 0 0) (0 0) E I (0 I!à 0) a Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A19 A18 A17 3.0 25 UI2.6 C-, C) 2.0 25 5, a, 0) >, (0 1.0 - E E E '.5 15 a 0) a 05 0.5 0.0, 0.0 .00 75 -60 .45 .30 -15 0 15 30 45 60 75 90 Fast Angle (deg) A20 3.0 2.5 C) C, 0) 2.0 (0 II) 0) E E 1.6 I- F- i1IIlk4:' I. i tfJ1J I > (0 1.0 0) a 0.5 0.0 Fast Angle (deg) A24 3.0 l j 2.5 2.0 E 1.6 1.0 - a 0) 0.5 0.0 Jr .90 -75 .60 -45 -30 -(5 0 IS 30 45 60 75 90 Fast Angle (deg) Figure A.5.2 Continued 102 A29 27 930906 3-c U C) C) a) 63 U) 6) 63 63 23 03 E E F: >, F- (5 (5 a) a) a E >. >-. 1-c 0-C Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A32 A31 A30 3.0 2.5 2.0 C) U (0 (0 a) a) a) E E El.5 F: >' >, 0.5 00 Jib -60 .75 -60 .45 -30 -15 0 IS 30 45 60 75 CO (0 a a(0 90 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A35 A34 A33 3.0 2.5 C-) 2.0 0) E 1.5 I- 03 -90.75.00.45.30-IS 0 (53045607590 III'Ii .9045-60.45-30.150 (53045607590 .90.754045.30.150 153045607590 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A39 A38 A36 3.0 3.0 2.5 2,5 2.0 2.0 Ei.5 E'.s 3.0 2.5 Il 2.0 / I It Ei.s I I F: ,_____// 1.0 - / a i \__/ 0.5 00 53 a) a 0.5 0.5 0.0 -90.75 .60.45 -30 .55 0 IS 30 45 60 75 90 Fast Angle (deg) Figure A.52 Continued 1.0 . a .90.75-60-45 -30-IS 0 IS 30 45 60 75 Fast Angle (deg) 60 00 .90-7540.40 .30.15 0 15 30 45 66 Fast Angle (deg) 75 90 103 A40 930906 A41 A42 104 APPENDIX A.6 SPLITTING RESULTS, ANISOTROPY VECTOR MAP AND ENERGY CONTOUR PLOTS FOR: W. CAROLINE ISL. 93/09/2603:31:19 Mw=6.4 dist=88.93° baz=283.4° (ls-30s) Table A.6. 1 Splitting measurements for the W. Caroline Isi. event. STA refers to the station, parameter is the fast polarization direction (clockwise with respect to north), & is the delay time in seconds, and Y() and a& are the icT uncertainties. (1) & STA AOl A08 A09 Al2 A15 A16 A17 A18 A19 A20 A22 A23 A24 A25 A26 A27 A28 A29 A30 A31 A32 A33 A34 A35 A36 A37 A38 A39 A40 A41 A42 A43 A44 78.00 80.00 84.00 75.00 59.00 57.00 59.00 45.00 69.00 42.00 43.00 69.00 68.00 53.00 48.00 56.00 56.00 53.00 54.00 57.00 64.00 73.00 60.00 67.00 67.00 81.00 71.00 41.00 58.00 73.00 77.00 81.00 69.00 +1+1+1- 4/-i/- 4/-+1- +1+1+1- +1+1+1+1+1- 4/4/+1.- +1+1+1- +1- 4/4/+1+1+1+1- #/4/+1- 4/+1- 54.00 36.00 26.00 61.00 21.00 23.00 11.00 13.00 18.00 15.00 40.00 42.00 18.00 14.00 13.00 9.00 8.00 14.00 15.00 20.00 24.00 26.00 8.00 45.00 8.00 6.00 34.00 35.00 14.00 8.00 26.00 41.00 31.00 1.55 *1- 0.95 +1+1- 1.35 0.60 1.55 1.45 1.45 1.45 1.05 1.65 1.00 0.85 1.15 1.25 1.50 1.85 1.90 1.55 1.60 1.50 1.45 1.65 1.60 1.05 2.15 2.35 0.90 1.45 1.45 1.75 1.30 1.55 1.30 i/ 4/- +1- 4/+1- +1- +/+1- 4/+/+1+1- 4/+1- 4/+1- 4/+1+14]- +1- 4/4]- +1+1-4/- +1- 4/+1+1- 0.95 1.50 1.10 2.40 0.70 0.80 0.40 0.60 0.45 1.05 1.00 0.75 0.40 0.35 0.60 0.40 0.35 0.40 0.45 0.65 0.65 0.90 0.30 1.80 0.45 0.65 0.85 1.55 0.40 0.40 0.90 1.30 0.90 105 Figure A.6. 1 Anisotropy vector map for the W. Caroline Isi. event. The vector orientation corresponds to 4 in (°) clockwise from north, and the size corresponds to the delay time in seconds. Figure A.6.2 Contour plot of energy on the corrected transverse component for all (, öt) pairs for the W. Caroline Is!. event. The absolute minimum is indicated by a star and the 95% confidence level is shown within the bold contour line. W. CAROLINE ISL. 93/09/26 03:31:19 Mw=6.4 dist=88.9° baz=283.4° (is -30s) 3600 45 N 3300 3000 2700 2400 2100 1800 1500 44 N 1200 900 600 300 0 124W Figure A.6.1 123W 122W 121W 107 A03 A02 t101 930926 C-) a) Co a) E P >, Co a) 0 1 :r -00 .75 .00 .45 .30 .15 0 15 30 45 80 15 50 - 0 .79 .80 .45 .30 .15 A04 0 15 30 45 60 75 9 Fast Angle ( deg) Fast Angle (deg) Fast Angle (deg) A06 05 3.0 IT 25t 0 0 a) 2C CO a) a) P I- Eu E CO a) 15 a) 0 0 01 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A09 A08 A07 3.0 3.0 3.0 2$ 2$ 2$ 2.0 2.0 E1.5 E'.s E1.5 P P P > > 0 0 o 00 00 .95 .75 .90 .45 .30 .15 0 15 30 45 60 75 .90 -75 .00 .45 .30 -55 90 0 15 30 45 66 75 0 _ E 1.51 I 0 Ill 2llL o.41J 0011 .90 I 1_sill i.oI-) 001 , .75 -60 -45 .30 .15 0 55 30 45 Fast Angle (deg) Figure A.6.2 5)5 75 50 IS 30 45 60 75 90 Al2 01 IlI III 0 3.0 3.0_ 2$ > -75 .60 .45 .30 .15 Fast Angle (deg) All AlO 3.0_ . 0.3 -90 90 Fast Angle (deg) Fast Angle (deg) E l.5F U,.,. 0.5 0.5 0.5 P 1.0 . 1.0 . 5.0 .95 .75 .50 .45 .30 .55 0 15 30 45 60 75 Fast Angle (deg) 95) 2$ 0 E is P > - 1.0 0 os 00 .90 .75 .5)0 .45 .30 .15 0 IS 30 45 Fast Angle (deg) 60 75 90 A15 A14 C) 20 II) E is I > - -90 -75 .60 .45 -30 -15 2 0 15 1.0 30 45 60 75 9 1I.III.II.i Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) U6 A18 A17 C) 0) U) 03 E > (9 03 -30.75.4(45.30-150 153040007530 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) U9 A22 A20 3.0 25 C) 0) (15 0) C) ri 215 02 E is E I- I- >' >.. (0 0) c:1 '-I 0.5 0.0 A25 A24 A23 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) 3.0 IT1fl 2.5 C-) 215 05 E is 10 Os 015 Fast Angle ( deg) Figure A.6.2 Continued Fast Angle ( deg) Fast Angle (deg) i09 p.D 32 2 (3 0) 0)20 00 03 03 E 50 05 i1'J .60 o j530 .90 e pge (de9 p29 31 p.3O ._-rgT1 3.0 (2 0322 Si) is5 5$ 0) eg 59 0 0$ .50 deg 1 faSt (P 03 I- 0 L_--5.00 p34 33 3.0 39 3.0 25 2$ 2.0 'p 0) (deg') fast deQ') 1 V1 ('29 Si) 29 0$ is5 02 0 0 0.5 0$ 2Va 0.0 .15 00 gte e eg) 36 3S 3.0 3.0 2$ 22 0) 03 r '5 '5 is 0 0 i) 0 0$ deg p.6.2 Cot 110 A40 A39 A38 930926 33 '.5 25 25 (3 a, 65 23 05 20 SC a, E E ii 05 E .5 (a 1-3 a, .5 F F F ISO '3 a, CD c '.5 '3 03 03 P.O .90 -70 60 -45 -30 -15 0 15 30 45 90 75 90 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A43 A42 A41 33.-.- 2311 C., V a, In a, F F 115 E F - F F >. Ca o4 > 'U a) 0j9L -00 .75 .60 .45 -30 .15 0 15 30 45 60 75 90 Fast Angle (deg) A44 as Cl, a) E I>. CO a) Fast Angle (deg Figure A.6.2 Continued -90 .75 .00 -45 -30 -15 0 IC - CD 15 30 45 Fast Angle (deg 60 70 91 !..II.Ii ast Angle (deg) 111 APPENDIX A.7 SPLITTING RESULTS AND ENERGY CONTOUR PLOTS FOR: XINJIANG, CHINA 93/10/02 08:42:35 M=6.3° dist=92.72° baz=335.32 Table A.7. 1 Splitting measurements for the Xinjiang event. STA refers to the station, parameter ( is the fast polarization direction (clockwise with and a& are the 1c uncertainties. respect to north), öt is the delay time in seconds, and The baz of this event coincides with the direction of slow polarization leading to a null measurement. STA AOl A02 A03 A05 A06 A07 A08 A09 AlO All Al2 A13 A15 A16 A17 A18 A19 A20 A22 A23 A24 A25 A26 A27 A28 A29 A30 A32 A33 AM A35 A36 A37 A38 A39 A40 A41 A42 A43 A44 & 4) -34.00 -19.00 -19.00 -9.00 56.00 68.00 -33.00 -41.00 -39.00 81.00 78.00 81.00 37.00 -18.00 -19.00 83.00 57.00 -16.00 66.00 -34.00 -29.00 75.00 73.00 -36.00 87.00 89.00 -28.00 73.00 82.00 55.00 -85.00 -34.00 -14.00 -73.00 80.00 82.00 83.00 -78.00 80.00 -85.00 +1- +141#1- 4141+1- +1+1+141+141+1-i-I- +1- 1+1- 41+1- +1+1+1+1+1+1- -il- -/+141- +1+1- 1+1+141+1+1+1+1- -ii- 2.50 1.30 +1- 1.95 +1- 2.20 2.20 2.95 90.00 4.00 9.00 90.00 90.00 90.00 90.00 61.00 75.00 59.00 72.00 90.00 90.00 90.00 90.00 90.00 90.00 73.00 90.00 77.00 0.50 3.00 2.75 0.80 0.80 3.00 +1- 2.00 90.00 66.00 11.00 66.00 40.00 90.00 80.00 39.00 89.00 21.00 90.00 90.00 33.00 22.00 7.00 66.00 11.00 20.00 18.00 41*1- 1.05 +1- 0.85 +1- 1.05 1.15 1.05 +1- +1- 1.95 1.30 1.95 1.55 1.75 0.80 3.00 2.45 +1- 2.20 +1+1- 2.40 +1- 1.40 +1- 0.60 2.40 +1- 1.30 +1- 3.00 +1- 1.10 +1- 3.00 0.50 1.50 +1- 1.25 0.60 0.85 3.00 1.65 1.20 2.15 0.90 0.95 2.05 0.90 3.00 2.35 1.05 1.40 1.70 1.40 -i-I- +1-i-I- +1+1+1- i/+1- +1+1+1- 41- 1.60 1.60 2.40 1.35 1.70 2.95 1.90 0.95 2.50 1.50 0.80 1.85 0.90 2.95 1.40 1.15 0.90 0.45 2.05 +1- 1.05 +1- 0.85 2.05 0.65 41+1+1+1+1+1- 1.95 0.40 1.30 0.80 112 Figure A.7. 1 Contour plot of energy on the corrected transverse component for all (, öt) pairs for the Xinjiang event. The absolute minimum is indicated by a star and the 95% confidence level is shown within the bold contour line.The baz of this event coincides with the direction of slow polarization leading to a null measurement. 113 rn ::i A03 A02 AOl 931002 11.1 0) 0 (3 5) 20 6) 03 90 0) 5) 5) E E E 15 I- I > >, I >, CO C0 0) 0 1.0 5) 0 0 0 :: .90 .75 .60 .45 .30 -15 0 15 30 45 60 75 -90.75-60-45-30-150153045907590 90 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A06 A05 A04 3_a I 2.5 (3 C) 0) C0 C) 0) 03 E E is 2.0 2.0 0) E 1.5 I I CO 0) 0 Os 0.5 0.0 .90 .75 .00 .45 -30 -15 0 15 30 45 60 75 90 -90 -75 -60 .45 -30 .15 15 30 45 A09 A08 A07 0 Fast Angle (deg) Fast Angle ( deg) Fast Angle (deg) ::riru 0 C) (3 0 0 CO 2.0 C') a) 0) 'C) E E E i-1 I I I- >' >' (0 CO 03 a, 1.7 0) 0 0 0 ill I Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) Al2 All AlO 3.0 2.5 2.0 C) U 113 0) U) 61 0 03 E E > I>, I (5 CO 0) 0 0 0 0 05 0.3 90 75 90 .45 .30 .15 0 15 30 45 Fast Angle (deg) Figure A.7.1 60 75 90 Fast Angle (deg) Fast Angle (deg) 60 75 90 114 A14 A13 931002 a.) a.) a.) a) a) U) U) U) a, U) 09 F F E >' > U) U) U) a, a, .90-75.60.45.30.150159045907590 .90-75.60-45-30-150159045907590 Fast Angle (deg) Fast Angle (deg) U Fast Angle (deg) A18 A17 A16 3.0 3.0 2.5 2.0 2.0 0.5 0.0 90 75 .60 .45 - U .15 0 IS 36 45 60 75 90 U) a, Eis E F: 1.0 ii 6.0 00 .90 .75 .40.45 .30 .15 6 15 90 45 90 75 00 :r'. .90 .75 .60 .45 .30 .15 0 15 90 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A20 A19 a., a) U) a, E F: > U) .90-75.4045.90-IS 0 Fast Angle (deg) .... A23 A22 ! 15 30454075 95 ii\ .90 .75 -40 -45 .30 .15 0 15 30 45 40 75 90 Figure A.7.1 Continued -90 .35 .40 .45 90 15 3045 6075 90 Fast Angle (deg) A24 1 'V .V \ m!k Fast Angle (deg) .90-75 .60 .45 .30.15 0 Fast Angle (deg) .15 0 15 30 45 Fast Angle (deg) 50 75 90 Fast Angle (deg) 115 3.0 2.5 V : - b (5 (3 (3 0) as U) U) ((1 0) as 0) E E 0) >' (a (U as as 0 as as >' A36 A35 A34 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) 2.5 (3 2.0 0) E is I>' 1.0 0) Os I! .90 .75 .90 .45 .30 .)5 Fast Angle (deg) Figure A.7. 1 Continued Fast Angle (deg) 0 15 30 4 Fast Angle (deg) 116 A37 931002 A39 38 23 .5 (3 03 U) 23 63 03 E E 13 a "p (0 F F F .0 a E >.. 66 13 0) 63 iii. Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A42 A41 A40 33 3.0 2D 6) 0, 4 a E Ets E1.5IQ F F F >. a 25U. 03 .1O 6) I 0 0.0 .00 .70 .00 -40 -30 -10 0 15 30 A44 A43 3.0 3X 23 2 0 2X 20 a a, Eu E 13 F I- >. (0 0.5 03 Fast Angle (deg) Figure Al. 1 Continued 45 60 75 90 Fast Angle (deg) Fast Angle (deg) 0 . .9. ast Angle (deg 0 03 ___ OJ -90 .15 .60 .45 .30 -10 0 15 30 45 60 70 Fast Angle (deg) 90 117 APPENDIX A.8 SPLITTING RESULTS, ANISOTROPY VECTOR MAP AND ENERGY CONTOUR PLOTS FOR: EASTERN NEW GUINEA REGION 93/10/13 02:06 mb=6.5 dist=95° baz=266° (is30s) Table A.8. 1 Splitting measurements for the Eastern New Guinea event. STA refers to the station, parameter 4 is the fast polarization direction (clockwise with and & are the la uncertainties. respect to north), öt is the delay time in seconds, and & STA AOl A06 A07 A08 A09 Al2 A13 A14 A15 A16 A17 A18 A19 A20 A21 A22 A23 A24 A25 A26 A27 A28 A29 A32 A34 A35 A37 A38 A40 A41 A42 A43 79.00 51.00 59.00 59.00 70.00 74.00 77.00 76.00 75.00 74.00 70.00 55.00 56.00 50.00 88.00 59.00 60.00 62.00 59.00 56.00 59.00 57.00 51.00 62.00 66.00 63.00 65.00 78.00 77.00 55.00 54.00 62.00 3.00 0.70 +/- 49.00 0.90 +1- 1.75 1.15 1.10 31.00 53.00 13.00 7.00 8.00 6.00 24.00 12.00 24.00 12.00 7.00 90.00 23.00 13.00 7.00 7.00 10.00 0.85 +1- 0.60 1.05 1.65 +1+1- 1.20 1.00 2.10 2.05 2.05 1.40 1.60 +1- 0.90 0.95 0.85 1.15 1.30 1.55 +1+1- +1- 6.00 +1+1- 44.00 +1+1+1+1+1+1+1-W- +1+1+1+1- *1- */+1- +1W+1+1-W- +1+1- */+1- JW- *1*141- 8.00 14.00 13.00 52.00 10.00 26.00 8.00 11.00 9.00 29.00 46.00 49.00 +1- 1111- 3.00 +1+1- 1.35 1.70 +1+1- 2.10 1- 1.95 1.75 1.65 +1+1+1+1+1+1+1+1+1- 3.00 1.60 1.10 1.70 1.70 1.75 2.10 1.80 0.95 0.70 0.85 +/+1- 4/+1- 1.20 0.80 0.50 0.40 0.30 0.60 0.80 0.65 0.55 0.45 0.50 0.45 0.60 0.40 1.35 0.60 0.90 0.55 1.20 0.95 0.60 0.95 0.95 118 Figure A.8. I Anisotropy vector map for the Eastern New Guinea event. The vector orientation corresponds to 4 in (°) clockwise from north, and the size corresponds to the delay time in seconds. Figure A.8.2 Contour plot of energy on the corrected transverse component for all (4, öt) pairs for the Eastern New Guinea event. The absolute minimum is indicated by a star and the 95% confidence level is shown within the bold contour line. EASTERN NEW GUINEA REGION 93/10/13 02:06:00 mb=6.5 dist=95° baz=266° (is - 30s) 3600 45 N 3300 3000 2700 2400 2100 1800 1500 44N 1200 900 600 300 0 124W Figure A.8.1 123W 122W 12VW 120 F03 A02 AOl 931013 3.0 :r 2.5 (.3 (-3 C.) C) CO 2.0 2.0 C) a) 9) E E E 1.5 I- I- I- > >' (C 10 - 1.0 a a) a a 0.5 I 0.0 .15 .00 .45 .lfl .15 0 IS 30 45 60 75 9 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A06 A05 A04 C.) C.) C.) 9) a) C) 'C) Cl) (I) 9) C) C) E E E I F I >. >' (Ii 0) CO 69 0) a C) a a .30.75,60.45.30-ISO 153045607590 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A09 A08 A07 I 0 (-3 (3 C) 50 9) (13 II) E 2.) C) C) E E I>' I- >. 69 (0 C) a 1,) a) 5) a a 01 0,) Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) Al2 All 10 3.0 2.5 0 C-) C) (3 a) 2.0 Cl) (13 E C) a) C) l\' 'W''i E E 1.5 I I- > CC (C C) C) a a a I.30 '75 '60 -45 -30 .15 Fast Angle (deg) Figure A.8.2 0 15 30 45 Fast Angle ( deg) 60 75 90 -90 .75 -60 -4) -30 -15 0 15 30 45 Fast Angle (deg) 60 75 90 121 A15 A14 30 2.5 0 0 a) (0 2.0 03 0) E E is F- >' 0) Ce 10 a 0.5 O 0.5 0.0 Fast Angle (deg) Fast Angle (deg) A18 A17 3.0 2.5 0 2.0 0 0 0) in 05 03 0) a) 0) E E E is I>' I- 1.0 I- 5, 'a Ca 03 a 0) .75 .60 .45 .30 .15 IS 30 45 60 75 90 45 60 75 90 60 75 90 A21 A20 iii ! 0 Fast Angle (deg Fast Angle (deg) 3.0 2.5 0 a) 2.0 In a) E to E I- > 5, a 1.0 a 0) a 05 00 -90 .75 .60 .45 .30 -15 0 15 30 Fast Angle ( deg) Fast Angle (deg) A24 A23 3.0 2.5 0 6) 0 2.0 U) 0) 0) E is E I- I5, > (a 1.0 0) a) a a 0.5 _1.r 4._ 00 .90 .75 .60 .45 .30 .15 Fast Angle (deg) Figure A.8.2 Continued 0 15 30 45 Fast Angle (deg) 122 A27 A26 A25 931013 3.) 2.0 0 0 0.) 5) U) 0) 2.0 0) C) 03 E E E 1.1 I- > - > >' a (0 I-c C) C) C) CD CD CD 0.2 0.0 .95 .75 .60 .45 .35 .15 60 75 90 30 45 60 75 90 30 45 33 30 23 23 0 0 0.) a) U, 2.0 2.0 a) 0) 0) E 1.5 F- E E 1.5 I- I>' >.. >-' - 15 A30 A29 A28 0 Fast Angle (deg) Fast Angie (deg) Fast Angie (deg) a 1.0 1.0 C) C) C) CD CD CD 03 0.5 0.0 00 .90 .75 .60 .45 .30 .15 7 15 A34 A32 3.0 3.0 2.5 2.5 2.0 2.0 2.0 E is E is E is 11 0 Fast Angle (deg) Fast Angie (deg) 0 C) I>' 1.0 0) C) CD CD CD 0.5 0.5 0.5 0,0 0.0 0,0 5045 -60 .45 . 5 .15 70 45 -90-75-60.45.30-15 0 15)0456075 90 60 75 U 7 Fast Angle (deg) Fast Angle (deg Fast Angle (deg) A37 A36 A35 3.0 2.5 0 0 0.) a) 0) 0) 6) 2.0 0) 0) C) E 1.5 I>, E E F: >' I3-. a a 1.0 CD 02 C) CD CD 0.5 0.0 -90 .75 -60 -45 -90 .15 Fast Angle (deg) Figure A.8.2 Continued 0 15 30 45 Fast Angle (deg) 60 75 90 Fast Angie (deg) 123 A40 C) a) 9) 3) E I- > (3 0) a I:L1 0.75-00.4540-150 153045607590 Fast Angle (deg) Fast Angle (deg) A43 442 0.) (0 a) E I- >' (a a II) Fast Angle ( deg r U 2.5 0) E 1.5 F- > 'C 3) a .90 .75 -60 -45 -30 -(5 0 IS 30 45 60 Fast Angle (deg) Figure A.8.2 Continued 75 90 Fast Angle (deg) 124 APPENDIX A.9 SPLITTING RESULTS, ANISOTROPY VECTOR MAP AND ENERGY CONTOUR PLOTS FOR: FIJI ISLANDS REG. 94/03/01 22:40:53 591 6.5 baz=230.9° dist=83.8° Table A.9. 1 Splitting measurements for the 94 Fiji event. STA refers to the station, parameter is the fast polarization direction (clockwise with are the 1c uncertainties. respect to north), öt is the delay time in seconds, and cvq and Results suggest that baz of this event is very close to the direction of fast polarization at A375, but is gradually changing towards the end of the array. STA A375 A385 A395 A405 A425 A435 A465 A475 A485 A505 A515 63.00 56.00 63.00 51.00 68.00 62.00 77.00 74.00 67.00 72.00 71.00 18.00 90.00 90.00 90.00 8.00 1- 90.00 +1- 26.00 +1- 12.00 i-I- 18.00 +1- 10.00 +15.00 +1+1+1+1+1- 3.00 3.00 1.45 3.00 2.15 2.25 1.05 2.30 1.70 1.55 1.65 +1+1+1+1+1+1+1- 1+1+1- 1- 2.30 2.95 1.55 2.95 0.85 1.45 0.65 0.95 1.30 0.55 0.35 125 Figure A.9. 1 Contour plot of energy on the corrected transverse component for all (4, öt) pairs for the 94 Fiji event. The absolute minimum is indicated by a star and the 95% confidence level is shown within the bold contour line.Results suggest that baz of this event is very close to the direction of fast polarization at A375 (elongated energy contours parallel to the time axis), but is gradually changing towards the end of the array. 126 A395 A385 A375 940331 30 3.0 3.0 2.5 2.5 2.5 2.0 2.0 2.0 11.5 E13 E 1.0 ______ >. to . .0 1.0 0) 0.5 0.5 :: 0.0 0.0 -90-75-60-45-30-150153045607500 -50-75-6045-30.150153045607590 90.7560-45-30-150 153045607530 Fast Angle (deg) Fast Angle ( deg) Fast Angle (deg) A425 A415 A405 3.0 2) 0 0 6) 0 2) a, 6) Cl) 0) 03 03 E E ' E > >, I- >' (0 (0 IC 07 03 0) 0.) 0.0 Fast Angle (deg) Fast Angle (deg) Fast Angle (deg) A475 A465 A435 3.0 3.0 2.5 2.5 2.5 U (3 C-) 2.0 2.0 2.0 0) 03 0) E is E is E '.5 I- I- I- > 1.0 - 6) 1.0 0.5 0.5 0.5 L1I 0.0 VETh.'- ,- 75, .1I. I. i---- I TU - % I Et I/I :l. i1iR I - ; ____ -:.